Home

ATHENA

1. nternational Figure 2 16 Structure Created by Etch Statement Deckbuild ATHENA Etch Etch Method Etching Machine Geometrical type All Left Right Above Below or thickness Any shape Material a Polysilicon Rip ShCgRee Gers User defined material Eh fecatien urd 3 Yhichness pmi i Arbitrary points x 0 20 y 1 x 0 4 y 1 Insert 7 x 2 y 1 To X location 0 20 000 d 10 00 Y location 1 000 10 00 Comment Poly definition WRITE Figure 2 17 Arbitrary Etch Silvaco 2 21 ATHENA User s Manual If this input file fragment is runned instead of the previous one using the INIT statement from the History capability the structure after this etch sequence will appear as displayed in the right hand plot in Figure 2 16 ATHENA etches all polysilicon material within the specified polygon The polygon etch can consist of any number of points If you use the Insert button an additional point will appear after the currently selected point An additional option for geometrical etching is a dry etch with a specified thickness This can be used for spacer for
2. nternational 1994 f Figure 2 23 MOSFET Structure with Electrodes You can now use the ATHENA Electrode menu see Figure 2 24 by selecting Commands gt Structure gt Electrode To set an electrode at a specified position select the Specified Position button type in the X Position e g 0 9 and Name for example source and press the Write button The following statement will appear in the input file ECTRODE NAME SOURCE X 0 9 Similarly specify the drain electrode ECTRODE NAME DRAIN X 0 9 The polysilicon gate electrode specification has the same format For this structure it can be done the same way as for source or drain ELECTRODE NAME GATE X 0 0 Silvaco 2 27 ATHENA User s Manual If the polysilicon layer is not the topmost layer at x 0 the Y Position can be specified In this case check the Y Position checkbox and type in a y coordinate within the polygate layer e g 0 2 If Y is not specified and the electrode is not on top ATHENA will look for the electrode in the underlying layers If it fails an error will be reported To specify a backside electrode select Backside from the Electrode Type field and type in a name see also Figure 2 24 iy Deckbuild ATHENA Electrode Elect
3. MATERIAL specifies MATERIAL1 for which parameters to be set see Section 6 2 9 Standard and User Defined Materials for the list of materials MATERIAL specifies MATERIAL2 for which parameters to be set see Section 6 2 9 Standard and User Defined Materials for the list of materials DIFF 0 DIFF E SEG 0 SEG E TRN O and TRN E specifies the diffusion coefficients of oxidant in MATERIALI and the boundary coefficients transport and segregation from MATERIAL1 to MATERIAL2 as defined above DIFF 0 is the diffusivity pre exponential factor in cm sec DIFF E is the energy in eV The transport coefficient represents the gas phase mass transfer coefficient in terms of concentrations in the solid at the oxide gas interface the chemical surface reaction rate constant at the oxide silicon surface and a regular diffusive transport coefficient at other interfaces The segregation coefficient is 1 at the oxide gas interface it is infinity at the oxide silicon interface and is a regular segregation coefficient at other interfaces Note Oxidant in materials other than oxide is allowed to diffuse and segregate but its concentration is then ignored for instance no oxynitridation Silvaco 6 75 OXIDE ATHENA User s Manual HENRY COEFF Henry s coefficient is the solubility of oxidant in MATERIAL1 measured in cubic centimeter
4. Description This statement is used to simulate deposition of specified material on the exposed surface of the current structure MATERIAL specifies the material to be deposited see Section 6 2 9 Standard and User Defined Materials for the list of materials NAME RESIST specifies the type of photoresist to be deposited THICKNESS specifies the deposited layer thickness in microns SI_TO_POLY specifies that crystalline silicon will be deposited only over crystalline silicon while polysilicon will be deposited elsewhere TEMPERATURE specifies deposition temperature used by STRESS HIST model The temperature is also used for surface diffusion simulation during ELITE deposition Grid Control Parameters DIVISIONS specifies the number of vertical grid spacings in the layer In some cases it is important to control the number of grid points in a conformally deposited layer since this also controls the accuracy of subsequent processes SPACES is an alias for DIVISIONS The default is 1 Note The default for DIVISIONS is 1 This typically needs to be increased for all deposition steps If DIVISIONS is set too low to maintain grid integrity in a non planar deposition ATHENA will attempt to recover by increasing DIVISIONS automatically ATHENA will echo the number of DIVISIONS finally used to the run time output DY specifies the nominal spacing in the layer Units are microns YDY sp
5. Table 6 1 Types of Parameters Type Description Bea ae Example Character An alphabetic alphanumeric or Yes OUTFILE MOS STR numeric string Integer Any whole number Yes DIVISIONS 10 Boolean A true or false condition No OXIDE or OXIDE f Real Any real number Yes C BORON 1 5e14 Any parameter that does not have a logical value must be specified in the form PARAM VAL where PARAM is the name of the parameter and VAL is the value of the parameter Boolean parameters must be separated from other parameters or commands with a space For example in the statement line DEPOSIT NITRIDE THICK 0 35 the NITRIDE parameter has a Boolean value true and the THICK parameter has a value of 0 35 real Many parameters are provided default values If a parameter is not specified its default value will be used Table 6 1 explains the different types of parameters which may be used when preparing an ATHENA input deck The command language of ATHENA is not case sensitive and can be entered using either upper case or lower case letters 6 1 1 Abbreviations It is not always necessary to input the entire statement or parameter name ATHENA only requires that you input enough letters to distinguish that command or parameter from other commands or parameters For example DEPO can be used to indicate the DEPOSIT command Silvaco 6 1 Overview ATHENA User s Manual 6 1 2 Continuati
6. Silicon oxide Interstitial Vacancy Silicon nitride KSURF 0 LORIOT 1 0x1070 Silicon oxynitride KSURF 0 1 0x1070 1 0x107 3 Silicon gas KSURF 0 1 0x10704 7 0x10708 KSURF E eV 0 0 4 08 KPOW 0 1 0 1 0 All parameters for other combinations are 0 0 B 8 Defect Growth Injection Interface Parameters Table B 18 Defect Growth Injection Interface Parameters Silicon oxide Interstitial Vacancy THETA 0 3 67x10708 0 0 THETA E eV 0 902 0 0 GPOW 0 0 0 1 0 GPOW E eV 0 0 0 0 VMOLE 5 0x102 5 0x1024 All parameters for other combinations are 0 0 B 9 Material Parameters Table B 19 Material Parameters Parameter Silicon Poly Oxide Oxynitride Nitride Photo Alumin NI O 127 3 9x1016 3 9x1016 1 0 1 0 1 0 1 0 1 0 NI POW 127 IES s 0 0 0 0 0 0 0 0 0 0 I E 127 0 605 0 605 0 0 0 0 0 0 0 0 0 0 EPS 11 9 11 9 3 9 fies s 1 0 1 0 Silvaco ATHENA User s Manual This page is intentionally left blank B 10 Silvaco Appendix C Hints and Tips This appendix is a collection of answers to commonly asked questions about the operation of ATHENA Some of these questions and answers have been previously published in articles in The Simulation Standard Silvaco s trade publication The original articles can be viewed at Silvaco s home page at
7. PACE MULT lt n gt INTERVAL R lt n gt LINE DATA SCALE lt n gt FLIP Y EPTH STR lt n gt WIDTH STR lt n gt Description This command sets up the mesh from either a rectangular specification or from a previous structure file The statement also initializes the background doping concentration in all regions Material Related Parameters MATERIAL specifies the material to be initialized see Section 6 2 9 Standard and User Defined Materials for the list of materials ORIENTATION specifies the substrate orientation Only 100 110 and 111 are recognized The default is 100 ROT SUB specifies the major flat of the silicon substrate It is measured in degrees from the external x axis of the crystallographic coordinate system By default ROT SUB 45 i e it represents the 101 plane This parameter is used only in BCA implantation module C FRACTION specifies the fractional component of the first element of a ternary compound substrate i e Al is the first component for AlGaAs The fractional component of the second component i e Ga is the second component for AlGaAs is 1 C FRACTION This parameter is valid for standard ternary materials AlGaAs and InGaAs or user defined ternary materials with the following standard names AllnAs InGaP GaSbP GaSbAs InAlAs InAsP GaAsP HgCdTe InGaN and AlGaN Dopant Related Parameters C IMPURITIES specify the uniform impu
8. Description This command specifies parameters of the base mesh used for initial grid generation SURELY specifies the location of surface The default is y 0 0 um SURE DY specifies the local grid spacing in y direction at SURF LY ACTIVE LY EPI LY SUB LY specify another three base line location at some critical region of the device structure to be fabricated ACTIVE DY EPI DY SUB DY specify the local grid spacing at the ACTIVE LY EPI LY SUB LY The units are all in microns BACK LY and BACK DY are the location of bottom in the structure to be fabricated and the BACK DY is the local grid spacing in the BACK LY location Examples The following example assigns the initial base line for the substrate materials It places the base line at y 1 0 y 2 0 y 10 0 with a local grid spacing of 0 01 um 0 5 um 1 0 um and 10 um separately The device dimension in the y direction is specified as y main 0 y max 100 This depth however is only for reference The real depth and width of the device structure will be assigned in the INITIALIZE command BASE MESH SURF LY 0 0 SURF DY 0 01 ACTIVE LY 1 0 ACTIVE DY 0 5 EPI LY 2 0 EPI DY 1 0 SUB LY 10 0 SUB DY 10 0 BACK LY 500 BACK DY 100 For more examples see BASE MESH and INITIALIZE Silvaco 6 15 BAS E PAR ATHENA User s Manual 6 8 BASE PAR Sy Description n
9. Silvaco 6 57 LAYOUT ATHENA User s Manual Note You can only use the RINGW GAP parameter in the IMAGE statement PHASE specifies the phase shift produced by the feature 180 lt PHAS IDTH and MULTIRING parameters for proximity printing model specified by the E lt 180 The default value is 0 TRANSMIT specifies the intensity transmittance of the feature 0S lt TRANSMIT lt 1 The default value is unity INFILE specifies the name of the MaskViEws layout file This file contains the mask information The MASKVIEWS mask information is a set of polygons for each mask layer with attributes transmittance and phase shift for each polygon MASK specifies the name of mask to be used for image calculation SHIFT MASK specifies the name of the additional mask layer usually phase shifting layer Examples The following statement describes a mask feature that is 2 microns in the x dimension and 0 4 microns in the z direction and rotated by 45 with respect to the x axis LAYOUT X LO 1 X HI 1 Z LO 0 2 Z HI 0 2 ROT ANGLE 45 TRANSMIT 1 For more examples IMAGE ILLUMINATION PROJECTION ILLUM FILTER PUPIL FILTER ABERRATION and the VWF INTERACTIVE TOOLS USER S MANUALS 6 58 Silvaco LINE 6 33 LINE LINE specifies a mesh line during grid definition Syntax LINE X Y LOCATION lt n gt
10. Monte Carlo Implant Example This example specifies a 300keV boron implant at zero degrees tilt and rotation Accurate modeling of such implants is only possible in the BCA model Since ion channeling is highly dependent on the tilt angle IMPLANT BORON DOSE 1E13 ENERGY 300 BCA TILT 0 ROTATION 0 Implant Damage Example This example implants phosphorus and invokes the unit damage model The UNIT DAMAGE model creates an interstitial profile scaled to the implant doping profile DAM FACTOR is used here to specify that the interstitial concentration will be ten times less than the doping throughout the depth of the implant profile IMPLANT PHOSPHORUS DOSE 1E14 ENERGY 50 UNIT DAMAGE DAM FACTOR 0 1 6 48 Silvaco IMPURITY 6 29 IMPURITY IMPURITY specifies impurity parameters Note This statement supersedes the older syntax using separate statements for each impurity type The ARSENIC ANTIMONY BORON INDIUM and PHOSPHORUS statements should no longer be used Syntax IMPURITY I IMPURITY DONOR ACCEPTOR NEUTRAL MATERIAL AT NUMBER lt n gt AT MASS lt n gt DIX 0 lt n gt DIX E lt n gt DIP 0 lt n gt DIP E lt n gt DIPP 0 lt n gt DIPP E lt n gt DIM 0 lt n gt DIM E lt n gt DIMM 0 lt n gt DIMM E lt n gt DVX 0 lt n gt DVX E lt n gt DVM 0 lt n gt DVM E lt n gt DVMM 0 lt n gt DVMM E lt n gt DVP 0 lt n
11. The following changes are made in the METHOD statement The LOWTHER parameter is removed It had been set to TRUE as default all along Obsolete parameters SU MOD GRIFFIN MOD and V LOOP SINK are removed Obsolete diffusion model POWER is removed It was effectively equivalent of FERMI model All parameter names related to vacancies and interstitials are standardized to VACANCY and INTERST i e C VACANCY F INTERST and so on Removed obsolete PAUSE command DECKBUILD has built in Pause capability Fixed long time broken command CPULOG Removed obsolete command ECHO PRINTF has the same capabilities Removed obsolete DEFINE and UNDEF commands DECKBUILD has extensive SET capability Removed obsolete command ECHO PRINTF has the same capabilities Renamed SET command to SETMODE to distinguish with DECKBUILD s SET Similarly UNSET is renamed to UNSETMODE Old names are also available as synonyms The composition fraction the C FRACTION parameter could be now specified INITIALIZE and DEPOSIT statements not just for standard ATHENA ternary material AlGaAs and InGaAs but also for the user defined materials corresponding to the following standard SILVACO ternary materials AlInAs InGaP GaSbP GaSbAs GaSbAs InAlAs InAsP GaAsP HgCdTe InGaN and AlGaN D 6 Silvaco ATHENA Version History D 6 ATHENA Version 5 6 0 R Release Notes D
12. ANGLES specifies the angle parameter used by the model PLANETAR C AXIS specifies the central axis length used by the models CONICAL and PLANETAR P AXIS specifies the planetary axis length used by the models PLANETAR and CONICAL DIST PL specifies the distance from wafer to planetary axis used by the model PLANETAR SIGMA DEP specifies the surface diffusion parameter used by the models UNIDIRECT DUALDIRECT HEMISPHERIC PLANETARY CONICAL MONTE1 and MONTE2 6 86 Silvaco RATE DEPO SIGMA 0 and SIGMA E specify pre exponential coefficient and activation energy of temperature dependent surface diffusion The temperature is specified on the DEPOSIT command SMOOTH WIN and SMOOTH STEP specifies a window size in microns and a number of smoothing passes for the simple geometric deposit smoothing algorithm MCSEED specifies a seed to be used for random number generation in the Monte Carlo deposit models MONTE1 and MONTE2 STICK COEF specifies the sticking coefficient for the MONTE1 model Unitless which must be between 0 0 and 1 0 Examples The following statement defines a machine named TEST that deposits silicon nitride with a rate of 1500 A minute using the CVD model with step coverage of 75 RATE DEPO MACHINE TEST NITRIDE DEP RATE 1500 A M CVD STEP COV 75 For more examples see DEPOSIT Silvaco 6
13. A bug in ELITE deposition of oxide when voids are formed has been fixed A bug in boundary conditions during impurity diffusion from ambient has been fixed Specified ambient concentration is now guaranteed at the surface points A bug in image simulation for the case of contact printing has been fixed D 7 ATHENA Version 5 4 0 R Release Notes D 7 1 SSUPREM4 Diffusion Simulation Features 1 Transient impurity activation model is implemented Parameters of the models TRACT 0 TRACT E and TRACT MIN are specified in the IMPURITY statement See the Section Transient Activation Model in Chapter 3 SSUPREM4 Models The Interface Trap Model is implemented The model simulates effect of the dose loss at the silicon oxide interface See the Interface Trap Model Dose Loss Model Section in Chapter 3 SSUPREM4 Models for more information A special capability to simulate Boron diffsuion in SiGe is implemented It includes a feature to deposit the SiGe layer with graded Ge content and two empirical models which modify Boron difusivity as a function of Ge content Implant Simulation Features 1 Old Monte Carlo simulation capabilities both for amorphous and crystalline materials are phased out 2 Binary Collision Approximation Monte Carlo module capabilities now supercede those of old MONTE models and are to be used for all non analytical implant simulations 3 All parameters related to old M
14. Table B 2 Parabolic and Linear Rate Constants for Wet Ambient Parameter Value IN L O pm min 3 45x104 IN L E eV 1 6 LIN H O ym min 2 95x10 JIN H E eV 2205 i BREAK C 900 B 1 3 Orientation Factors For Linear Coefficients both Ambients Table B 3 Linear Coefficient Orientation Factors Orientation Value For lt 100 gt orientation unitless ORI FAC 0 595 For lt 110 gt orientation unitless ORI FAC 0 833 For lt 111 gt orientation unitless ORI FAC 1 0 B 1 4 Pressure Dependence For Dry Oxidation DRY L PDEP 1 0andP PDEP 1 0 0 75 and P PDEP 1 0 For Wet Oxidation WET L PDEP Table B 4 Thin oxide coefficients only for dry ambient Orientation THINOX 0 THINOX E eV THINOX L p THINOX P 14 u2 min lt 111 gt 5 87 x 106 2732 0 0078 1 0 lt 110 gt 5 37 x 104 1 80 0 0060 1 0 lt 100 gt 6 57 x 106 2 37 0 0069 120 B 2 Default Coefficients B 1 5 Chlorine Dependence Table B 5 Chlorine dependence of dry coefficients for three temperatures 900 1000 1100 C HCL PC HCL LIN HCL PAR 900 1000 1100 900 1000 1100 0 0 1 0 L O 1 20 1 0 1 0 1 0 1 0 2 75 25 1 621 1 083 1 658 1 355 3 0 PJS 486 2 207 125 1 840 1 490 5 0 hers 486 2 207 1 444 2 075 1 641 7 0 1 75 486 2 207 1 639 2 332 1 816 10 0 ters 486 220
15. where Ccp is the total particle chemical concentration J is the flux of mobile particles OV is the gradient operator and S accounts for all source and sink terms The difference between the total chemical concentration and the actual mobile concentration is described in a later section entitled Section 3 1 6 Electrical Deactivation and Clustering Models In semiconductor diffusion problems there are generally two contributors to the particle flux The first contributor is an Entropy Driven Term which is proportional to the concentration gradient of mobile particles The coefficient of proportionality D4 is called the diffusivity The second contributor is a Drift Term which is proportional to the local electric field Notice that if there are several types of electrically charged species present this term establishes a coupling between them since all charged particles both contribute to and are influenced by the local electric field The Flux Term J4 can be written as J4 D C VC CyoE 3 3 where C4 designates the mobile impurity concentration o is the mobility and E is the electric field It should also be observed that Equation 3 3 is non linear since both the diffusivity D4 and the electric field E in general depend on the concentration of all present species In thermodynamical equilibrium the Einstein relation relates mobility and diffusivity through the expression D tTa Silvaco 3 3 ATHE
16. 6 2 7 Execution Control Statements These statements control some execution capabilities Some of them are useful only in a batch mode when ATHENA is run outside DECKBUILD COMMENT is used to document the input file CPULOG instructs ATHENA to output CPU statistics FOREACHIEND specifies the command looping facility HELP prints summary of statement names and parameters OPTION specifies the level of run time output PRINTF parses a string or expression and places result into standard output QUIT terminates execution of ATHENA SETMODE sets execution mode parameters SOURCE causes ATHENA to read statements from the specified file UNSETMODE sets execution mode parameters to false 6 2 8 Obsolete Statements The following statements existed in earlier versions of ATHENA Their capabilities were substituted either by superior capabiliites of VWF INTERACTIVE TOOLs or included in other more advanced or generic statements ANTIMONY is substituted by the I ANTIMONY parameter in the IMPURITY statement ARSENIC is substituted by the I ARSENIC parameter in the IMPURITY statement BORON is substituted by the I BORON parameter in the IMPURITY statement COLOR all plotting capabilities are now provided by ToNYPLOT CONTOUR all plotting capabilities are now provided by TonyPLoT ECHO is a synonym of the PRINTF statement DEFINE is substituted by the SET capability in DECKBUILD
17. The machine is modeled with a hemispherical deposition model The deposition rate is 1 micron minute The logical parameter U M specifies what units are used in this case microns per minute Finally the angles of incidence of the hemispherical deposition with respect to the surface normal are specified with the ANG El and ANGLE2 parameters You can modify these characteristics of the machine PE4450 by copying the specification to the input file and using an ASCII editor For example RATE DEPO MACHIN E PE4450 ALUM INU U M SIGMA DEP 35 HEMISPH ANGLE1 72 ANGLE2 70 E D EP RAT redefines machine PE4450 to have a deposition rate of 0 5 micron minute 2 60 Silvaco Tutorial Defining ELITE Deposition Machines You can define your own deposition machine using the ATHENA Rate Deposit menu Figure 2 44 To open this menu select Process Deposit Rate Deposit in the Commands menu Machine definition requires the specification of five general parameters and one or several model specific parameters The general parameters that must be specified are the following e Machine name e g TESTO1 This parameter uniquely identifies the machine e Material name e g aluminum A user defined material e Machine model type e g unidirectional You can select one of six models by pressing the appropriate button e Deposition rate units specif
18. 4 8 Silvaco ELITE Models Central Axis C AXIS Source Figure 4 6 Geometric of Source to Substrate in a Conical Evaporator 4 3 7 Conical Deposition To use this model specify the CONICAL parameter in the RATE DEPO statement The Conical model is a simplified version of the Planetary model with and r 0 the substrate always sees a symmetrical cone source In this type of configuration the integral of the above two equations can be evaluated analytically and expressed in the following simple closed form _ RR IW L L i R x y l tan n ax z tan nin aW RNY z i 4 10 If expression under either square root in this formula is less than 0 it is set to exact 0 L R LW L i 2 R x y ST asin E tan A asin tan nin 4 11 NR W R L R Silvaco 4 9 ATHENA User s Manual In the Conical model the parameter ANGLE1 y and other parameters are C AXIS and P AXIS as shown in Figure 4 6 4 3 8 Monte Carlo Deposition There are two models that are invoked by specifying the MONTE1 or MONTE2 parameters in the RATE DEPO statement The parameters SIGMA DEP DEP RATE and ANGLE1 MONTE1 invokes the Monte Carlo based deposition model which you can use to model low pressure chemical vapor deposition LPCVD 89 90 Since the radicals are incident on the substrate with non zero thermal velocities they may be re emitted from the surface before
19. Example setting numerical techniques The following statement specifies that minimum fill reordering should be done and that the entire system should be solved using a conjugate residual technique with three back vectors The initial time step should be 0 1 seconds and time should be integrated using the TRBDF parameter The FERMI model should be used for diffusion and the COMPRESS model for the oxide growth METHOD MIN FILL CG BACK 3 INIT TI 0 1 TRBDF FERMI COMPRESS Example setting diffusion model for power devices The following step specifies that a simple diffusion model should be used appropriate for power electronic devices METHOD POWER DIFFUSION TEMP 1000 TIME 300 NITROGEN Example setting diffusion models for RTA The following statement invokes all 311 cluster models for RTA simulation It must be set before the IMPLANT statement that generates the cluster damage METHOD NEWTON FULL CPL CLUSTER DAM I LOOP SINK HIGH CONC BACK 6 IMPLANT DIFFUSE Silvaco 6 69 MOMENTS ATHENA User s Manual 6 37 MOMENTS MOMENTS specifies tables and spacial moments used in analytical implant models Syntax OMENTS SVDP_TABLES STD_TABLES USER_STDT USER_TABLE lt c gt ATERIAL I IMPUTITY DOSE lt n gt ENERGY lt n gt RANGE lt n gt STD DEV lt n gt GAMMA lt n gt KURTOSIS lt n gt STD DEV LGAMMA LK
20. Note Since ATHENA parser doesn t recognize parameter names that begin with numerals non standard names are used for Silicon Carbides STC_6H STC_4H and SIC_3C Standard names 6H SiC 4h SiC and 3C SiC are used for these materials outside ATHENA e g TonyPlot and ATLAS Insulators OXIDE OXYNITRIDE NITRIDE PHOTORESIST Metals ALUMINUM TUNGSTEN TITANIUM PLATINUM COBALT Silicides WSIX Tungsten Silicide TISIX Titanium Silicide PTSIX Platinum Silicide COSTX Cobalt Silicide Special Materials GAS is used only in the IMPURITY INTERSTITIAL and VACANCY statements to specify some parameters i e segregation at exposed boundaries BARRIER is a fictitious material It can be specified only in DEPOSIT and ETCH statements and serve as a masking material Silvaco 6 7 ATHENA Statements List ATHENA User s Manual User defined Materials User defined materials can be specified by MATERIAL lt c gt where lt c gt could be a single word MATERIAL OXIDE1 or any string in double quotes as MATERIAL MY INSULATOR The user defined material with the names exactly corresponding to SILVACO standard material names are saved in SILVACO Structure Files as those standard materials and will be recognized as such by other tools e g DEVEDIT and ATLAS The fo
21. Silvaco E 8 Bibliography 10 11 12 13 14 15 16 17 18 19 20 J D Plummer M D Deal and P B Griffin Silicon VLSI Technology Fundamentals Practice and Modeling Prentice Hall 2000 G F Carey W B Richardson C S Reed and B J Mulvaney Circuit Device and Process Simulation Mathematical and Numerical Aspects John Wiley and Sons 1996 C C Lin and M E Law 2 D Mesh Adaption and Flux Discretizations for Dopant Diffusion Modeling IEEE Trans CAD v 15 p 194 1995 C C Lin and M E Law Mesh Adaptation and Flux Discretizations for Dopant Diffusion Modeling NUPAD V p 151 1994 D Mathiot and J C Pfister Dopant Diffusion in Silicon A Consistent View Involving Nonequilibrium Defects J of Appl Phys v 55 p 3518 1984 P M Fahey Point Defects and Dopant Diffusion in Silicon Ph D Thesis Integrated Circuits Laboratory Department of Electrical Engineering Stanford University June 1985 M E Law Two Dimensional Numerical Simulation of Dopant Diffusion in Silicon Ph D Thesis Department of Electrical Engineering Stanford University 1988 J A Van Vechten and C D Thurmond Entropy of Ionization and Temperature Variation of Ionization Levels of Defects in Semiconductors Phys Rev B v 14 p 3539 1976 R B Fair ed F F Y Wang Concentration Profiles of Diffused Dopants in Silicon Impurity Doping Process in Silic
22. e Angadd the sum of Cergy lt n gt and Cergyn lt n gt for each i k pair It corresponds to the energy angle distribution of the particles e Angtot the sum of Angadd for each k it corresponds to the angle distribution of particles e Erel the result of i Nrow 1 2 Nrow corresponding to the medium energy in eV of the 50 normalized intervals 0 0 0 02 0 02 0 04 0 98 1 0 OUTF ANGLE specifies the name of an output file in which energy angular ion flux distribution is saved The distribution can be plotted using ToNYPLOT ER LINEAR ER INHIB ER COVERAGE and ER THERMAL specify surface kinetics model to be used Simple linear Adsorbed inhibiting layer threshold coverage and thermal spike models correspondingly Default is ER LINEAR K I specifies the plasma etch rate linear coefficient related to the ion flux K F specifies the plasma etch rate linear coefficient related to the chemical flux 6 92 Silvaco RATE ETCH K D specifies the plasma etch rate linear coefficient related to the deposition flux SPARAM specifies S parameter of threshold coverage and thermal spike models THETA specifies theta parameter of threshold coverage and thermal spike models IONFLUX THR specifies the flux threshold value below which the flux is not considered for etching Default is 0 0 MAX IONFLUX specifies a multiplier for ion flux generated by the plasma etching machine Default is 1 0 MAX CHEMEFL spe
23. 20 SILVACO International 2004 Figure 3 15 VV Ratio versus Doping Concentration TonyPlot V2 8 18 A A Files View Plot Tools Print Properties Help X Antimony 2 Arsenic G Boron Phosphorus Oxidation Conditions 950 C 20 min DryO2 amp 3 g E a 2 lelh oping Concelftthtion rem 3 20 Click to place P changes alignment or drag to get leader SILVACO International 2004 Figure 3 16 Simulated Silicon Dioxide Thickness vs Doping Concentration for Common Silicon Dopants 3 56 Silvaco SSUPREM4 Models 3 3 5 Parabolic Rate Constant For long oxidation times and high temperatures the oxide growth is parabolically related to the oxidation time The diffusion of oxidant in the oxide is the determining factor in describing the growth kinetics For these times and temperatures the oxide thickness can be approximated as xo Bt 3 165 where B is called the parabolic rate constant and is given by Equation 3 135 When using this equation the oxidant diffusivity Dg is determined from specified values for Cc N yz and experimentally determined values of B The parabolic rate constant has been determined to have dependencies on the ambient pressure and the chlorine content during oxidation and is given by B B Bp Bucy 3 166 where PARL exp PARLE T lt P BREAK b B 3 167 PAR H Oep PARLE T gt P BREAK b B is determin
24. ATHENA User s Manual 112 D J Kim et al Development of Positive Photoresist IEEE Trans Electron Devices v ED 31 p 1730 1984 113 C A Mack PROLITH A Comprehensive Optical Lithography Model SPIE Proc v 538 p 207 1985 114 P Trefonas III et al New Principle For Image Enhancement In Single Layer Positive Photoresist SPIE Proc v 771 p 194 1987 115 Y Hirai et al Process Modeling For Photoresist Development And Design Of Drl sd double Resist Layer by a Single Development Process IEEE Trans on CAD v CAD 6 p 403 1987 116 J Van Roey J vander Donk P E Lagasse Beam propagation method analysis and assessment J Opt Soc Am Vol 71 No 7 p 803 July 1981 117 J Z Y Guo F Cerrina Modeling X ray proximity lithography IBM J Res Develop Vol 37 No 3 p 331 May 1993 118 J W Goodman Introduction to Fourier Optics McGraw Hill New York 1969 119 B Kernighan and D Ritchie The C Programming Language Prentice Hall 1988 120 D B Kao J P McVittie W D Nix and K C Saraswat Two Dimensional Thermal Oxidation of Silicon I Experiments IEEE Trans Electron Dev ED 34 p 1008 1987 121 R B Fair Impurity Doping Process in Silicon Ed F F Y Wang North Holland Amsterdam 1981 122 C L Chu Characterization of Lateral Diffusion in Silicides IEDM Tech Digest p 245 1990 123 P H Langer and J I Goldstein Boron Autodoping D
25. Control Print Figure C 8 Junction depth of an arsenic implant after a fixed diffusion as a function of DAM FACT and KSURF 0 Silvaco C 11 ATHENA User s Manual Question Which are the key parameters for tuning RTA simulations when using the new Stanford diffusion models in ATHENA version 4 0 Answer For RTA applications it is recommended to use the new set of models from Stanford University included in ATHENA version 4 0 These models include effects of 311 defect clusters dislocation loops and high concentration effects To enable all these models the syntax used is METHOD FULL CPL CLUSTER DAM I LOOP SINK HIGH CONC The syntax METHOD NEWTON is also recommended to improve the speed of simulations Since these models are an extension of the existing FULL CPL models many of the same tuning parameters apply Previous simulations 311 have shown how the surface recombination rate of interstitials KSURF 0 is a key tuning parameter for reverse short channel effect where damage enhanced diffusion is significant This is also true in the 311 cluster models In RTA simulations with the FULL CPL model all point defects are created by the implantation They are at a maximum at t 0 of the RTA and their concentration decays rapidly with time due to diffusion and recombination A very important effect of the 311 cluster model is that the free point defect concentration is not created at the time of the impl
26. Figure 4 5 illustrates the planetary evaporation system By inspecting this system you can be convinced that the rotation of the planet along the system central axis has no effect on the deposition rate For simplicity s sake you can calculate the growth rate by holding the planet stationary and by rotating only the source along the axis of the planet see Figure 4 5 The growth rate is derived according to the following equations r rLtanA LW Lsec A Ltan Asin p Lcos p tan Ry xX 2 d 4 8 R 7 L 2rLtan A N R WIR r Ltan Ay Lee R 7 rLtanA LW Lsec Ay L tan Asin B Lcos p 5 ja R r ei Deter Al E EW R Or Lan A where e 6is the incident angle of the vapor stream e fis the tilt angle of the planet plane e A 56 8 e ris the distance between the position of the wafer and the planet axis e R L and W are the parameters dependent on the system dimensions Silvaco 4 7 ATHENA User s Manual Central Axis Substrate N Figure 4 5 Illustration of Planetary Evaporator Using the planetary model of ELITE you can observe asymmetries both in edge coverage and the depth of cracks produced by the particular location and orientation of a specimen in a planetary system Figure 4 6 shows the following planetary model parameters Yy ANGLE2 B ANGLE3 Y DIST PL P AXIS C AXIS and ANGLE1 ANGLE1 is used to calculate DIST PL P AXIS tan ANGLE1
27. K M Beardmore and N Gronbech Jensen Efficient Molecular Dynamics Scheme for the Calculation of Dopant Profiles Due to Ion Implantation Phys Rev E 57 1998 p 7278 J M Hern ndez Mangas J Arias M Jaraiz L Bail n and J Barbolla Algorithm for Statistical Noise Reduction on Three Dimensional Ion Implant Simulations Nucl Instr Meth in Physics Research B 174 2001 p 433 438 W Bohmayr A Burenkov J Lorenz H Ryssel and S Selberherr Trajectory Split Method for Monte Carlo Simulation of Ion Implantation IEEE Transactions on Semiconductor Manufacturing v 8 p 402 1995 P Glasserman P Heidelberger P Shahabuddin and T Zajic A look at Multilevel Splitting Technical report RC 20692 IBM Research Division T J Watson Research Center Yorktown Heights New York 1997 S E Hansen and M Deal SUPREM IV GS Two Dimensional Process Simulation for Silicon and Gallium Arsenide Integrated Circuits Laboratory Stanford University 1993 O Madelung Ed Semiconductors Basic Data Springer Verlag 1996 R Anholt et al Ion Implantation Into Gallium Arsenide J Appl Phys v 64 p 3429 1988 K Rajendran Simulation of Boron Diffusion in Strained SiGe Epitaxial Layers Proc SYSPAD p 206 2000 R F Lever Boron Diffusion across Silicon SiGe Boundaries J of Appl Phys v 83 p 1988 1998 J L Ngau P B Griffin and J D Plummer Modelling the Suppression of
28. Note The EXTRACT statement is supported under DECKBUILD and is fully documented in the VWF INTERACTIVE TOOLS MANUAL USER S MANUAL VOL I Silvaco FOREACH ATHENA User s Manual 6 22 FOREACH FOREACH specifies the command looping facility Syntax FOREACH NAME COMMANDS END LIST Description This command is used to specify input loops FOR is equivalent to FOREACH When the loop executes NAME will consecutively take on each value in LIST and exit the loop after assuming the last value COMMANDS will be executed once for each value in LIST NAME is set to a value in LIST using the shell define function LIST is a set of strings separated by commas or spaces The values in LIST can be delimited by either commas or spaces LIST can also take the following numerical operator form START TO END STEP VAL where START is a numerical start value END is the last value and VAL is the amount to increment at each iteration Examples The following statement will increment val from 1 0 to 10 0 in steps of 0 5 This loop will be executed 19 times FOREACH VAL 1 0 TO 10 0 STEP 0 5 ECHO VAL END Note Command line continuation using the backslash character P indicator is not supported in the FOREACH statement Note This statement is not supported within the VWF Automation Tools The Automation Tools contain a separate and more powerful capabilitie
29. RATE POLISH 6 52 RATE POLISH RATE POLISH specifies the polishing parameters for a chemical mechanical polishing CMP module Syntax RATE POLISH MACHINE lt c gt MATERIAL NAME RESIST lt n gt A H A M A S U S U M U HIN M SOFT RATE HEIGHT FAC lt n gt LENGTH FAC lt n gt KINETIC FAC lt n gt MAX HARD lt n gt MIN HARD lt n gt ISOTROPIC lt n gt a Description This command sets the parameters for the CMP machine used in the POLISH statement The parameters must be set for each material to be polished There are two polish models hard and soft that can be used together or separately Define these models by specifying their parameters MACHINE specifies the machine name MATERIAL specifies material for which parameters of the CMP machine to be applied see Section 6 2 9 Standard and User Defined Materials for the list of materials NAME RESIST is the user defined photoresist to be polished A H A M A S U H U M U S and N M specifies that the rates are in Angstroms per hour Angstroms per minute Angstroms per second microns per hour microns per minute microns per second and nanometers per minute respectively SOFT RATE is the rate for the soft polish model HEIGHT FAC is the vertical deformation scale for the soft polish model Units are microns LENGTH FAC is the horizontal deformation scale for the soft polish model
30. The boundary interface conditions for vacancies are set similarly to those for interstitials except that the VACANCY statements should be used instead of all INTERSTITIAL statements Dislocation Loop Based Enhanced Bulk Recombination A topic of some debate in recent literature has been the creation of Dislocation Loops Currently the exact physical nature of these defects is still under investigation It is believed that they arise from amorphizing implants and only exist at the edge of the amorphous layer It has been suggested that these loops grow through the absorption of interstitials during oxidation and perhaps shrink by the emission of interstitials when annealed in nitrogen Due to the lack of proper scientific description only a simple recombination model has been implemented into ATHENA This model introduces an additional sink of interstitials that is described by the following expression x Rj damalpha C C 3 52 where damalpha is a parameter you can set on the INTERSTITIAL statement This additional recombination is only applied to a region of the amorphizing implant controlled by the parameters MIN LOOP and MAX LOOP in the DISLOC LOOP statement As an example the following statements will produce a region of dislocation loops where the as implanted phosphorus concentration is between lel6 cm and 1le18 cm The damalpha parameter is then set to 1e8 in this region For example DISLOC LOOP MIN LOOP 1lel6 MAX
31. The following is a short description of the cosine law deposition This is a simple model that accounts for metallization due to evaporation The cosine law deposition model is based on the following assumptions e The mean free path of atoms or particles is much larger than the distance between the source and the substrate e The source to substrate distance is large compared to the surface topography e The film grows in the direction toward the vapor flux e Shadowing effects must be included 4 4 Silvaco ELITE Models The magnitude of the film growth rate follows the cosine distribution law which says that deposited film thickness grows at a rate proportional to cos w where o is the angle between the vapor steam and the normal surface The sticking coefficient can be used as a tuning parameter It is assumed to be 1 0 for deposition on the cold substrates at 300 K Z D A Bs 2 Angle conventions Figure 4 2 Step Profile with a Unidirectional Source 4 3 4 Dual Directional Deposition This model is invoked by specifying the DUALDIRECT parameter in the RATE DEPO statement In this type of source each point in the unshadowed region views the vapor streams arriving from two different directions and assumes the diffusion length of deposited material large compared to the features see Figure 4 3 Growth rate is given as R x y 0 if point x y is
32. 0 0 cee e eee eee eee 3 89 3 6 Deposition Modelska aiia arse dia ahaha achat ake 0 alas alas ioa aSa beara Bok nadia ape cee ahaa ed Bc 3 91 3 6 1 Deposition of Doped Layers cas 4 wan eae cian ienh a Rls veh ern pie Vader Parma date oi aches 3 91 3 6 2 Grid Control During Deposit nonnes enaena ak etree den 3 91 3 6 3 Epitaxy Simulations m ia eta hiss Shr te A E tal eeu BUN soe Ri A A 3 91 3 7 ELCHING Models sit cs cc oti be toi si inatt n i ne e AEP A E a ia 3 92 3 8 Compound Semiconductor Simulation 0000s cece eee eee eee eens 3 93 3 81 Diftusion MOdelSs 2 ccy bey is pele ee oe eerie eae tein phot pele e eee eects 3 93 3 8 2 Implantation Mod ls cient ee tai ae etcetera tact EE ENEE ite hea ate tae ava ge 3 94 3 9 SiGe SiGeC Simulation cio icccvecv sake cows sscwavenweeeaed E EA EA DENENA E EEA 3 95 3 9 1 Deposition of SiGe SiGeC Epitaxial Layer n 0 c ccc e een eens 3 95 3 9 2 Boron Diffusion in SIGE SIGE Gs sic caret oh Sete nich Seu See tues eh ket oe ecb eh es 3 95 3 9 3 Boron Transient Diffusion Suppression by Carbon Incorporation Models 0 00sec eee eee 3 96 STOP OUESS MOGI ie ieoi rend isd a de etal open deta tape nee beret een ese axe 3 97 Chapter 4 ELITE MOQEIS i nindian Satine a a een iea nwa aca thes teat aes aR 4 1 AV Overview 25 03 natra takai Aa a ace seein de es ee eae ee eed ote tees 4 1 4 2 String Algorithm oore E an den Siete Re ekw lk be rene Core wt oote tL eeews OYE SE e
33. 05 Microns 0 6 o7 0 8 0 9 1 SILVACO International 1996 al 1 Figure 2 71 Too Dense Mesh Causes Too Much CPU Time during Subsequent Simulation 2 92 Silvaco Tutorial Reasonable Adjacent Triangle Ratios 3 i Large Adjacent Triangle Ratios L a B A GRAD SPACE D C D RATIO BOX Figure 2 73 Base Mesh Formation Silvaco 2 93 ATHENA User s Manual TonyPlot 2 5 1 File z view v Plot Tools 7 Print Properties T Help v ATHENA Structure with Automatic Base Mesh Generation ee Boron cm NNWAKAN ARASAN N 4 K SNE SY fe ae NE SRLN a 08 1 12 14 Printing complete SILVACO International 1995 Figure 2 74 Automatic Base Mesh Generation TonyPlot 2 5 1 ATHENA The adaption of lon Implantation on New Base Mesh INN RENAA INRA RANNA i N aTa 1717 Waa a Ni ZN ZR ZS SZ Dx VAVAVAVAVAVAVAVAVAVA SDR DADA DADS Wak KI E R22 Vad SSSScs Printing complete SILVACO International 1995 HH Figure 2 75 lon Implantation Adaption on New Base Mesh 2 94 Silvaco Chapter 3 SSUPREM4 Models 3 1 Diffusion Models The diffusion models in ATHENA describe how implanted profiles of dopants defects see the Note below redistribute themselves during thermal treatment due to concentration gradients and internal electric f
34. 102 H H Hopkins On the Diffraction Theory of Optical Images Proc Roy Soc v A 217 p 408 1953 103 H H Hopkins Applications of Coherence Theory In Microscopy And Interferometry J Opt Soc Am v 47 p 508 1957 104 J Tsujiuchi Image Forming Performance of Projection Systems Japanese J of Appl Phys v 4 Suppl I Proc Conf On Photographic And Spectroscopic Optics p 251 1965 105 W H A Fincham M H Freeman Optics Butterworths London Ch 15 1980 106 J Van Roey J van der Donk P E Lagasse Beam Propagation Method Analysis and Assessment J Opt Soc Am v 71 p 803 1981 107 J Z Y Guo F Cerrina Modeling X ray Proximity Lithography IBM J Res Develop v 37 p 331 1993 108 A Erdmann C L Henderson C G Wilson W Henke Influence of Optical Nonlinearities of the Photoresist on the Photolithographic Process Basics SPIE Proc v 3051 p 529 1997 109 A Erdmann C L Henderson C G Wilson R R Dammel Some aspects of thick film resist performance and modeling SPIE Proc v 3333 p 1201 1998 110 D A Bernard Simulation Of Post Exposure Bake Effects On Photolithographic Performance of a Resist Film Phillips Journal of Research v 42 p 566 1987 111 F H Dill W P Hornberger P S Hauge J M Shaw Characterization of Positive Photoresist IEEE Trans Electron Devices v ED 22 p 445 1975 Silvaco BIB 5
35. ERF Q lt n gt ERF DELTA lt n gt ERF LBB lt n gt ERF H lt n gt Description All parameters relating to oxidation are specified in this statement Oxidation models are specified in the METHOD statement All oxidation models are described in Chapter 3 SSUPREM4 Models Section 3 3 Oxidation Models To properly set values for most coefficients you need to know whether to use wet or dry oxidation and to know the substrate orientation Note If a required parameter is omitted e g orientation when a linear rate coefficient is being specified then the statement is ignored without warning Oxide Growth Rate Parameters DRYO2 WETO2 specifies the type of oxidation to which specified coefficients apply Required for everything except for one dimensional coefficients and the volume ratio ORIENT is the substrate orientation the coefficients specified apply to the required for orientation factor see ORI FAC and thin oxide coefficients Only 100 110 and 111 are recognized The default is 100 LIN L O LIN L E LIN H 0 LIN H E L BREAK and L PDEP specifies the linear rate coefficients B A A doubly activated Arrhenius model is assumed L BREAK is the temperature breakpoint between the lower and higher ranges in degrees Celsius LIN L 0 is the pre exponential factor in microns min LIN L E is the activation energy in eV for the low temperature range LIN H 0 and LIN H E are the corresp
36. LBB EOX for instance would give a lateral spread equal to the field thickness similar to the Hee Gook Lee model with a spread of one ERF H is the ratio of the nitride lifting to the field oxide thickness It corresponds to the Guillemot H parameter except that it is normalized to the field oxide thickness It is specified as an expression of Eox eox Tox en NIT THICK specifies the nitride thickness to substitute for the parameter EN 6 76 Silvaco OXIDE Note The nitride thickness is user specified in the OXID ERFG model uses both oxide and nitride thickness These values are not inferred from the structure Instead the E statement and the oxide thickness is computed by adding the total oxide grown and the initial user specified oxide thickness If the structure has more than 20 angstroms the default of native oxide on it when diffusion begins then thickness must be specified Beware of this when continuing a diffusion by any means e g after reading in a previous structure Examples The following modifies the parabolic oxidation rates for 100 silicon in a dry oxygen ambient OXID E DRY ORI 100 PAR L 0 283 333 PAR L E 1 17 Note If a required parameter is omitted e g orientation when a linear rate coefficient is being specified then the statement is ignored without warning The following set the native oxide thickness at 10 Angstroms OXI
37. Properties Help m ATHENA SIMULATION FROM FIGURE 1 WITH LOWER TAU 311 0 20 lt _ As Implanted RTA15 sec RTA1 min amp 4 RTAS min G 4 _ furnace anneal 60 min Phos Conc cm 3 SILVACO International Figure C 10 The effect of lower TAU 311 0 is to speed up the diffusion over the initial time period Silvaco C 13 ATHENA User s Manual Question How can I determine implant range for non standard materials such as silicides or photoresist Answer The analytical implant tables in ATHENA SSUPREM4 cover implantation of the common silicon dopants B P As Sb In into the commonly used set of materials in semiconductor processing Silicon SiO2 Si3N4 polysilicon aluminum For other materials or implant species the lack of complete data means full analytic tables are not available The only alternative approach was to use Monte Carlo MC Implant simulation Implantation using MC with the crystalline model is usually required for silicon implantation For realistic 2D cases these implants may take up to 30 minutes to run In order to overcome this problem an alternative approach is available in ATHENA This approach uses MC implant in 1D mode to run implantation simulations into the material of interest Then the analytical implant moments are extracted from the implanted doping profile These analytical moments can be used in a MOMENTS statement to
38. RATE DEVELOP specifies development rates and other photoresist parameters RATE ETCH specifies the etch rate for machine etches RATE POLISH specifies polishing parameters for definition of a polishing machine SILICIDE sets the coefficients for silicidation reactions TRAP sets the coefficients of trap kinetics VACANCY sets the coefficients of vacancy kinetics Silvaco 6 5 ATHENA Statements List ATHENA User s Manual 6 2 5 Special DECKBUILD Statements These statements invoke special operations when running under DECKBUILD For more information on these statements see the VWF INTERACTIVE TOOLS manual AUTOELECTRODE defines layout based electrodes EXTRACT extracts parameters GO indicates interfacing between simulators MASK performs photoresist deposition and etching through the MASKVIEws interface SET sets the value of a user defined variable SYSTEM allows execution of any UNIX C shell command within an input file TONYPLOT creates a plot using TONYPLOT 6 2 6 Post processing Statements Starting from version 4 0 all internal plotting capabilities of former SUPREM IV have been depreciated Enhanced superior capabilities are available through TonyPLoT and other VWF INTERACTIVE TOOLS Only the following two post processing statements remain PRINT 1D is used to print the values data points and profile information SELECT allows a variable to be chosen as the z coordinate for the PRINT ID command to follow
39. SILVACO ATHENA User s Manual SILVACO 4701 Patrick Henry Drive Bldg 1 Santa Clara CA 95054 Telephone 408 567 1000 Internet http Awww silvaco com April 14 2008 ATHENA User s Manual Copyright 2008 Silvaco 4701 Patrick Henry Drive Building 6 Santa Clara CA 95054 Phone 408 567 1000 Web www silvaco com Notice The information contained in this document is subject to change without notice Silvaco Data Systems Inc MAKES NO WARRANTY OF ANY KIND WITH REGARD TO THIS MATERIAL INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTY OF FITNESS FOR A PARTICULAR PURPOSE Silvaco Data Systems Inc shall not be held liable for errors contained herein or for incidental or consequential damages in connection with the furnishing performance or use of this material This document contains proprietary information which is protected by copyright laws of the United States All rights are reserved No part of this document may be photocopied reproduced or translated into another language without the prior written consent of Silvaco Data Systems Inc VIRTUAL WAFER FAB VWF MANUFACTURING TOOLS VWF AUTOMATION TOOLS VWF INTERACTIVE TOOLS DECKBUILD TONYPLOT DEVEDIT TONYPLOT3D DEVEDIT3D MASKVIEWS ATHENA SSUPREM4 MC IMPLANT OPTOLITH ELITE MC DEPO ETCH SSUPREM3 SPDB ATLAS S PISCEs BLAZE BLAZE38D GIGA2D 3D MIXEDMODE2D 3D SIC FERRO QUANTUM2D 3D LUMINOUS2D 3D LED VCSELS LASER TFT2D 3D
40. The following example sets the moments for boron implantation into the user defined material SAPPHIRE MOMENTS MATERIAL SAPPHIR RANGE 0 098 STD For more examples see 1 MP LANT E I BORON DOS DEV 0 045 GAM E 1 6e12 A 0 04 EN ERGY 25 KU RTOSIS 3 5 Silvaco OPTICAL ATHENA User s Manual 6 38 OPTICAL OPTICAL sets the optical parameters of materials for OPTOLITH Syntax OPTICAL MATERIAL NAME RESIST lt c gt LAMBDA lt n gt I LINE G LINE H LINE DUV LINE REFRAC REAL lt n gt REFRAC IMAG lt n gt DELTA REAL lt n gt DELTA IMAG lt n gt aE Description This command sets the optical parameters reflective index and extinction coefficient or REFRAC REAL and REFRAC IMAG for each material at a particular wavelength If photoresist is used NAME RESIST must also be specified MATERIAL specifies the material for which the optical parameters to be set see Section 6 2 9 Standard and User Defined Materials for the list of materials NAME RESISTspecifies the name of the photoresist I LINE G LINE H LINE DUV LINE and LAMBDA specifies the line or the wavelength Units for LAMBDA are microns WAVELENGTH is an alias for LAMBDA REFRAC REAL specifies the real component of the refractive index REFRAC IMAG specifies the imagina
41. range between 700 and 1200 C Outside of this range the diffusion coefficients may be inaccurate and numerical difficulties may occur during simulation For ramped thermal step a synonym T START can be used T FINAL specifies the final temperature for ramped thermal steps Synonym is T STOP T RATE specifies the ramp rate in C minute for ramped thermal steps THICKNESS specifies thickness of epitaxially grown layer Units are microns GROWTH RATE specifies epitaxial growth rate It is applicable only when THICKNESS is not specified Units are um minute Doping Related Parameters C IMPURITIES specify the concentration of the impurity in the epitaxially grown layer in cm You can specify more than one of these parameters to define materials doped with multiple impurities F IMPURITIES can be specified only together with the corresponding C IMPURITY e g F BORON and C BORON This parameter generates the linearly graded concentration of the specified impurity in the epitaxially grown layer where C IMPURITY specifies concentration at the bottom of the layer and F IMPURITY specifies concentration at the top of the layer C INTERST specifies the concentration of interstitials in the epitaxially grown layer in cm F INTERST can be specified only together with C INTERST This parameter generates the linearly graded interstitial concentration in the epitaxially grown layer where C INTERST
42. 1 orp 22 ty 5 51 The solution for given initial conditions i e amplitude A immediately after thin mask should be equal to the mask transparency T x y can now be expressed as a convolution of A with T 2 x X 0 Y A x y Z T X Y exp gad tO ta dX dY 5 52 This expression exactly corresponds to the Fresnel diffraction 5 16 Silvaco OPTOLITH Models 5 7 3 Simulation Method The Beam Propagation Method BPM is very efficient for tasks where a one way propagation of electromagnetic field is considered Common description of BPM can be found in 116 117 A simplified approach applicable to proximity printing conditions i e a gap space filled with homogeneous matter so refractive index does not depend on coordinates in the gap is described here The initial distribution of amplitude A in x and y in the plane perpendicular to the propagation direction z can be obtained easily as a complex mask transparency function Ay A x y z 0 5 53 Then by applying the 2D Fourier transform over x and y you can obtain the angular distribution Alko kp k 0 AG y z O expli k x k y dxdy 5 54 In this angular spectrum domain the relation between A and its z derivative from Equation 5 48 is given by the following expression 118 OA a ee az Oke k z ijk k kA ko k z 5 55 It means that for each particular direction a specific phase shift has to be applied to get the
43. 6 seconds N2 str 1 min N2 str 10 mins N2 str n S 2 5 5 T ta ke S T Drd pi G pad TREET EART SS e pee el Aa te Sell iene Fi al 1 2 3 4 5 6 7 8 9 10 Depth Into Surface um Click to place P changes alignment or drag to get leader SILVACO International 1996 Figure 2 26 Interstitials can move far into the substrate even after a short 10min anneal 2 32 Silvaco Tutorial Figure 2 26 shows typical diffusion profiles of interstitials after a 1e15 cm3 20keV Boron implant at various anneal times After only a 10 minute anneal the interstitials have diffused 8um into the substrate Interstitials like dopant require a concentration gradient in order for overall diffusion to take place For example if the concentration gradient of interstitials is removed by having too shallow a substrate depth the concentration of interstitials will start to pile up because they are no longer being removed through diffusion into the bulk of the substrate If the level of modeled interstitials becomes too high the diffusion of dopant even near the surface of the substrate will also be too high and the simulation will be inaccurate TonyPlot V2 6 6 File 7 View 7 Plot va Tools Fa Print Fa Properties 7 Help T EFFECT OF SUBSTRATE DEPTH 20keV Boron 1e15 cm2 gt X Boron cm3 1e19 z 1e18 E 1e17 E 1e16 E 1e15 E 1
44. BPM ceeeeee 5 9 5 17 D 9 Adaptive Meshing Bimolecular Recombination eeeeeeeeeeeeeeeeneeeesneeeteneeeees 3 30 Adaptive Meshing Control ccccccsessseseeseeseeseseeseeeees 2 89 Binary Collision Approximation BCA ssessssssserssersersees 6 45 Base Mesh Formation 0 c ccecceessesessesessesesseseseeseeeeees 2 89 Bipolar Process FIOW cseseseeseeeeseeseteeeeteeeeeeeenees 2 48 53 Heat Cycle eeinetan 2 87 88 CONCUSSIONS eroien a aeaa d e a anadai 2 53 Interface Mesh Control ccccececeseseseeseceresesteeereees 2 89 94 The Base Current Profile Low Injection cse 2 52 53 lon Implantation sesiis 2 86 The Base Current Profile Medium Injection 2 51 52 Advanced Diffusion Model Examples Tuning Base and Collector Currents All Regions 2 49 Cowern s Experiment ccccccsscscsesesssecscseseseececseseseeees 3 39 Tuning the Base Current All Regions sssssseseeeeeees 2 49 Implantation Diffusion Experiment cccceeseneeees 3 41 43 Tuning the Collector Current All Regions ssssssseeees 2 50 Pelaz Experiment ccccsccsscssscsscsseseccseeseceseeseeeseeeees 3 40 Buzz Saw Model e r e r ee E EE 4 21 Pr deposition s i enea e 3 36 37 See also Hard Polish Model Advanced Diffusion Models eessceeeeseeeeeseeeeeeneeeeeaes 3 23 43 Classical Model of Dopant Diffusion CNET 2 3 24
45. C 5 3 4 Self interstitial generated ed C E core jf 4 by the silicon implant paces 700 C celeste 200 C 600 C 20h J SA Simulation with full PLS model z Full PLS model 18 j I aml J a z S s E 5 1 a 2 g amp vw i 13 1 i 8 v f al n 5 7 i Od i 16 0 0 2 0 4 0 8 1 1 2 14 16 18 2 10 10 10 Depth um time s Click to place P changes allanment or drag to gat leader SILVACO International 2003 Click te place P changes alignment or drag to ost leader SILVACO International 2003 Figure 3 5 Simulation of the Cowern experiment and extraction of the evolution of the supersaturation during the annealing Experimental data are from 20 Figure 3 5 shows the free interstitial supersaturation behavior with time This quantity is simply calculated using the ratio between local effective interstitial concentration and its equilibrium value From the experimental point of view this quantity can be related to the boron diffusion enhancement with respect to its thermal equilibrium diffusion The supersaturation evolution curve exhibits three parts e The first step characterized by a high supersaturation value corresponds to a large acceleration of the dopant diffusion and can be explained by the presence of small clusters e The second step exhibits a plateau slightly decreasing with time It is explained by the competitive growth between lt 311 gt defects known as the Ostw
46. II Reflecting a Structure in the Y Plane using the Mirror Parameter This tutorial process simulation has been building one half of a MOSFET like structure At some point in the simulation you will need to obtain the full structure This must be done before exporting the structure to a device simulator or setting electrode names In general structure reflection should be performed when the structure ceases to be symmetrical e g a tilted implant an asymmetrical etching or a deposition takes place or when a reflecting boundary condition no longer applies to the side which is going to be the center of the structure This example will explain how to mirror the structure at its left boundary To mirror the structure select Structure Mirror in the Commands menu Figure 2 21 iy ATHENA Mirror Mirror Figure 2 21 ATHENA Mirror Menu Then press the Write button to write the following statement to the input file STRUCT MIRROR LEFT The resulting structure is shown in Figure 2 22 Silvaco 2 25 ATHENA User s Manual TonyPlot 2 2 1 File ji View 7 Plot 7 Tools 7 Print 7 Properties 7 Help 7 ATHENA Full structure after mirror operation is Hh Hn KRE atgrie naw rt E i ct ETETE betet rte crt Cet rth te eine pit Hilt Eki aE chet etre tate ti Crs cut cs cite Hiti ERR y i tet fee tat Haiti Ail ih ite tet cit i cs i H arip
47. LINE X LOC 0 SPA 1 TAG LEFT LINE X LOC 1 SPA 0 1 LINE X LOC 2 SPA 1 TAG RIGHT LINE Y LOC 0 SPA 0 02 TAG SURF LINE Y LOC 3 SPA 0 5 TAG BACK Silvaco 6 59 LINE ATHENA User s Manual Note It is difficult to predict how many lines are going to be generated in each interval The initial mesh specification is quite important to the success of the simulation Use the geometric mode by specifying the NO IMP parameter on the INITIALIZE statement to perform a fast simulation without impurities to determine if the grid spacings are appropriate For more examples see I TALI E RE ION BASE ESH and BAS E PAR 6 60 Silvaco MASK 6 34 MASK MASK deposits and patterns photoresist or artificial masking material barrier via the MASKVIEWS interface Syntax MASK NAME lt c gt REVERSE DELTA lt n gt Description MASK is used in DECKBUILD to provide interface to Silvaco s general purpose layout editor MASKVIEWS When you specify a mask statement ATHENA will deposit photoresist and pattern it by etching The etched pattern is determined by selected cut line in MASKVIEWS See the VWF INTERACTIVE TOOLS USER S MANUAL VOL I for a complete description of this feature NAME specifies the name of the layer that defines the photoresist patterning Mask names must appear inside of double quotes This name must corres
48. OTFT OLED NOISE DEVICE3D THERMAL38D ATLAS INTERPRETER MERCURY FASTBLAZE FASTNOISE FASTGIGA FAST ATLAS C INTERPRETER MOCASIM VICTORY HARM ZENITH VISION MIXSIM TCAD DRIVEN CAD SIMULATION STANDARD CONNECTING TCAD TO TAPEOUT AND TCAD OMNI are trademarks of Silvaco Data Systems Inc All other trademarks mentioned in this manual are the property of their respective owners 2008 by Silvaco Data Systems Inc Silvaco jii How to Read this Manual Style Conventions Font Style Convention Description Example This represents a list of items or e Bullet A terms e Bullet B e Bullet C 1 This represents a set of To open a door 2 directions to perform an action 1 Unlock the door by inserting 3 the key into keyhole 2 Turn key counter clockwise 3 Pull out the key from the keyhole 4 Grab the doorknob and turn clockwise and pull gt This represents a sequence of File gt Open menu options and GUI buttons to perform an action Courier This represents the commands HAPPY BIRTHDAY parameters and variables syntax New Century Schoolbook This represents the menu File Bold options and buttons in the GUI New Century Schoolbook This represents the equations abc xyz Italics This represents the additional 077 7 Note important information Note Make sure you save often when working on a manual NEW CENTURY SCHOOLBOOK IN SMALL CAPS This represents the n
49. Parameters related to Grid Control during ETCH ETCH EPS sets a tolerance on the grid movement during ETCH statements This parameter is defined in relative units The default is 10 that corresponds to about 10 Angstroms Reducing this number will allow sub 10A etches to be exact But the possibility of small triangles being created during etches is high if the parameter is set too low This parameter should not be set to zero Parameters used in the Adaptive Meshing Module ADAPT specifies that the adaptive meshing should be performed on the IMPLANT DIFFUSE or EPITAXY statements the default is false DEPO SMOOTH specifies that the mesh smoothing should be performed after each DEPOSIT statement 6 68 Silvaco METHOD ETCH SMOOTH specifies that the mesh smoothing should be performed after each ETCH statement DIFF SMOOTH specifies that the mesh smoothing should be performed after each DIFFUSE statement STEP SMOOTH specifies that the mesh smoothing should be performed after each time step on each DIFFUSE statement Miscellaneous Parameters STRESS HIST specifies that stresses to be calculated during etching deposition diffusion and epitaxy process steps Example setting tolerances The following statement specifies that the arsenic equation should be solved with a relative error of 1 and concentrations below 1 x 10 can be ignored METHOD ARSEN REL ERR 0 01 ABS ERR 1 0E9
50. Silvaco D 3 ATHENA User s Manual D 4 ATHENA Version 5 10 0 R Release Notes D 4 1 General Features 1 2 Added multiple features to provide compatibility with TSUPREM4 and TSUPREMB see Appendix E TSUPREM4 and TSUPREM3 Compatibility Features LEX_LM and OMNI licensing D 4 2 SSUPREM4 1 10 11 12 13 BCA module has now two different engines for ion trajectory calculations One of these engines is 2 to 4 times faster than another In some cases the faster engine might be slightly less accurate The new parameter FAST for IMPLANT BCA statement is introduced This parameter allows you to specify which engine to be used during current Monte Carlo simulation If FAST t rue default the fast engine is used If FAST false the slower potentially more accu rate engine is used The default version of parameter DIVERGENCE the alias is BEAMWIDTH in the IMPLANT statement has changed from 0 to 1 0 ion beam divergence is very difficult to achieve A typical ion beam divergence of industrial implanters is 1 to 1 5 New parameter IV SCALE is introduced in the IMPLANT statement to control estimation of after implant interstitial and vacancy distributions from BCA damage calculations using parameter DAMAGE DAM FACTOR 0 0 can now be specified in the IMPLANT statement This is used with Advanced Diffusion Module DifSim Wafer miscut feature is implemented for BCA implant in crys
51. TAU 311 E lt n gt Description This command specifies the scaling of 311 clusters during a subsequent IMPLANT step and the time constant for the dissolution of clusters into free interstitials Note This command will only work if you switch on the 311 cluster model with the METHOD CLUSTER DAM command I IMPURITY specifies an impurities to be used for the 311 cluster scaling see Section 6 2 10 Standard Impurities for the list of impurity names that can be used e g I BORON MATERIAL specifies a material in which the scaling takes place see Section 6 2 9 Standard and User Defined Materials for the list of materials Default is SILICON MIN CLUST and MAX CLUST define two values of implanted dopant concentration Clusters will be placed between these two dopant concentration levels only These parameters are used to control the scaled position of clusters during ion implantation Typically MIN CLUST is the background doping level MAX CLUST is the dopant concentration required to amorphize the substrate CLUST FACT specifies the ratio between the concentration of clustered interstitials and the implanted dopant concentration TAU 311 0 lt n gt and TAU 311 E lt n gt specify the time constant in seconds for the dissolution of clusters into free interstitials TAU 311 0 is the pre exponential linear coefficient and TAU 311 E is the exponential coefficient used to control
52. This is because stopping powers and range parameters are different in different materials This section will describe the implant scaling methods available in ATHENA DOSE MATCH The Dose Matching Method was historically the first and is the most widely used 50 method The Dose Matching Method is selected by the DOSE MATCH parameter default in the IMPLANT statement With this method the segment of the profile within it layer is calculated by C x pfx txofp 3 194 where f x is the distribution function specified for this implant Gaussian Pearson or Dual Pearson with moments corresponding to the it layer x is the distance from the surface to the top of the jth layer k lt i 1 x gt a th 3 195 k 1 x is the thickness of the k layer Xr is the effective thickness evaluated from eff k lt i 1 cod So 3 196 0 ei where b is the portion of the total implant dose which is consumed in the k layer Obviously for the first layer xef 0 and x 0 3 70 Silvaco SSUPREM4 Models RP SCALE and MAX SCALE The other two methods for analytical calculation of implantation profiles in the layered structures are projected range depth scaling set by RP EFF or RP SCAL in the IMPLANT statement and maximal depth scaling set by the MAX SCALE parameter These two methods differ from the dose matching method in the way the effective depth x is calculated and in the normalization of the partial profile
53. This line will be the new surface ETCH DRY THICK 0 1 Physical Etch Example The following sequence defines an etch machine named PLASMA that performs reactive ion etching of silicon The machine is applied to etch the current structure for 10 minutes RAT ETCH E ETCH MACHINE PLASMA1 SILICON AACHINE PLASMA1 TIME 10 MINUT U M RII E ISOTROPIC 0 1 DIR ES ECT 0 9 Note The program can be sensitive to grid placement It often helps to prepare the initial grid by having a vertical grid line exactly at the etch coordinate for geometric etches For example see RATE ETCH Silvaco EXPOSE 6 20 EXPOSE EXPOSE runs the exposure module of OPTOLITH EXPOSE INFILE lt c gt PERPENDICUL PARALLEL X CROSS 2Z CROSS CROSS VALUE lt n gt DOSE lt n gt X ORIGIN lt n gt F F JAATNESS lt n gt NUM REFL lt n gt RONT REFL lt n gt BACK REFL lt n gt ALL MATS lt n gt MULT EXPOSE POWER MIN lt n gt Description This command defines the parameters associated with and performs two dimensional exposure INFILE is the name of an input file that contains a user aerial image cross section data file This file has the form lt wavelength in microns gt lt number of data pairs gt lt x location in structure gt lt relative image intensity gt PE
54. Units are microns KINETIC FAC is the Kinetic factor soft polish model The vertical polish rate increases as the surface becomes more vertical MAX HARD is the maximum rate for the hard polish model Corresponds to a pattern factor of zero MIN HARD is the minimum rate for the hard polish model Corresponds to a pattern factor of one ISOTROPIC specifies the isotropic etch rate used by the POLISH model Examples The following statements describe a polishing machine named CMP for nitride and oxide RATE POLISH MACHINE cmp NITRIDE SOFT 4 N M HEIGHT FAC 0 02 ENGTH FAC 80 KINETIC FAC 10 RATE POLISH MACHINE cmp OXIDE SOFT 25 HEIGHT FAC 0 02 ENGTH FAC 30 KINETIC FAC 10 For more examples see POLISH and RATE ETCH Silvaco 6 95 REGION ATHENA User s Manual 6 53 REGION REGION specifies a material to be assigned to a defined mesh region Note Typically the REGION statement is not required since initial substrate material is specified on the INIT statement Syntax REGION MATERIAL XLO lt c gt YLO lt c gt KHI lt c gt YHI lt c gt Description This command specifies the material in a rectangular mesh REGION statements should follow LINI statements Material must be specified for every triangle in a mesh Therefore for each rectangula mesh there must be at least one REGION statement specifying which mat
55. ambient is present method two dim Use before implant doses less than 1e13 cm and for oxidations method full cpl Use before implant doses greater than 1e13 cm cluster dam high conc 2 4 4 Changing the Method Statement During the Process Flow It has previously been stated that the disadvantage of using the most advanced and complex models is the time involved during diffusion cycle simulation Accordingly there is an incentive during complex process simulations to switch back to a simpler model during a diffusion cycle when the majority of the damage created by a previous implant has been annealed We will show you when to switch to a simpler model If the process being modeled has involved implantation or oxidation at any stage we advise not to use the fermi model An exception to this would be in some power devices with very long diffusion times where the exact nature of surface damage would have little impact on the final distribution of the dopant and simulation time is at a premium In reality for most small geometry processes the question of switching models becomes one of when to add a new method statement that changes from METHOD FULL CPL CLUSTER DAM HIGH CONC to METHOD TWO DIM after a high dose implant Switching guidelines A simple guideline to follow when to switch method statements during a process flow is by switching back to the TWO DIM model if the anneal temperature is gr
56. atoms of the material The number of AT NUM parameters specified must correspond to COMPONENTS AT MASS 1 AT MASS 2 AT MASS 3 and AT MASS 4 specify the atomic masses of the constituent atoms of the material in atomic mass units The number of AT MASS parameters specified must correspond to COMPONENTS ABUND 1 ABUND 2 ABUND 3 and ABUND 4 specify the relative fraction of the constituent atoms of the material The number of ABUND parameters specified must correspond to COMPONENTS Note At least one parameter from each of the four lines above are required to define materials for Monte Carlo implants Parameters Related to REFLOW Calculations REFLOW specifies that the material will flow when a DIFFUSE step including REFLOW is defined GAMMA REFLO specifies the surface tension parameter used in the reflow calculation Units are dyne cm The material viscosity VISC parameters will also affect the reflow rate Parameters Related to the Grid Control NO FLIP specifies that triangle flipping procedure should not be applied to the specified material Parameters Related to the Boron Diffusion Model in SiGe SiGeC NIFACT SIGE specifies the linear coefficient for Ge dependency formula of intrinsic carrier concentration for Boron diffusion model in SiGe SiGeC EAFACT SIGE specifies the linear coefficient for Ge dependency formula of intrinsic carrier concentration for Boron diffusion model in SiGe SiGeC NIFACT SIC specif
57. calibration so this can allow the interstitials to diffuse 10 um along the surface from both the source and drain ends without effecting diffusion near the center of the device In summary tuning THETA 0 involves the comparison of modeled and measured threshold voltage data for a long gate length device THETA 0 can be rapidly tuned by taking a one dimensional 1D vertical cutline through the center of the gate and doing a 1D process simulation You can either tune THI ETA 0 manually or by using the Optimize function in DECKBUILD Theta 0 is tuned until the measured and simulated data of the long channel threshold voltage correspond The fine tuning of THI simulation ETA 0O is performed by using a full 2D Figure 2 34 shows a typical dependence of extracted threshold voltage on the Theta 0 tuning parameter Realistic values of THI curve is due to rounding errors in the 1 to the automatic and independent mesh generated in the ETA 0O correspond to the rising part of the curve The glitch in the EXTRACT statement used to calculate the threshold voltage due EXTRACT statement The mesh can be changed from its default value shown here to eliminate this effect But close examination reveals that the error is only a few millivolts off which is accurate enough for most process parameter extractions TonyPlot V2 6 6 File View Plots Tools Print Properties Help 7 gt g
58. e C and Co are the particle concentrations in the immediate vicinity of the interface in the regions 1 and 2 e hj is the interface transport velocity e M39 is the segregation coefficient The transport velocity and segregation coefficients are temperature dependent parameters defined through the following Arrhenius expressions TRN E hy TRN 0 exp ZRN E 3 13 12 exp kT M SEG 0 exp EE 3 14 You can specify the parameters TRN 0 TRN E SEG 0 and SEG E in the IMPURITY statement All parameters are specified for only one direction which is from region 1 to region 2 The following is an example of the syntax used to change the segregation coefficients between oxide and silicon Two material names separated by a to indicate the combination and the ordering of materials for which these parameters are specified IMPURITY I PHOSPHORUS SILICON OXIDE SEG 0 30 TRN 0 1 66E 7 3 6 Silvaco SSUPREM4 Models Interface Trap Model Dose Loss Model You can simulate the effect of dose loss at silicon oxide interface by specifying the DOSE LOSS parameter in the METHOD statement This model is based on the theory that the dopant diffusing through silicon oxide interface can be trapped into the trap sides located at the interface 10 11 A modified equation for impurity flux is used in the Dose Loss Model Fi g S C_ 1 B Cr Crmax MnCr 3 15 where m 1 2 correspond to Si and
59. http www silvaco com Question Simulating the whole process in ATHENA may take a long time How can the process flow be checked or tuned quickly Answer Several methods are available in ATHENA that enable you to do quick look and see simulations of a complex process flow Deciding which method to use in a given situation depends on the particular items of interest Three modes that can be useful are outlined below 1 1D Mode This is used to perform 1D analysis at any x location in the 2D structure This mode can be invoked from the ATHENA Mesh Initialize menu Figure 2 10 by selecting the 1D box under Dimensionality The X Position item of the menu will become active so you should choose the x location at which the 1D analysis will be performed These changes in the menu will add two parameters ONE D and X LOCAT lt real gt to the INITIALIZE statement ATHENA automatically takes into account all masking and etching steps at the specified location This mode is particularly useful for optimization and process tuning For example it can be used to rapidly check MOS source drain junction depth or the intrinsic base profile of a BJT 2 Geometrical Mode In this mode all impurities are turned off by checking the No Impurities box in the ATHENA Mesh Initialize menu This will add the NO IMPURITY parameter to the INITIALIZE statement disabling all implantation and dopant diffusion steps Impurity diffusion which usually limits
60. lt n gt Description This statement specifies the aberration coefficients in the power series expansion of the wave aberration function Each coefficient is entered in fractions of a wavelength in the range 0 lt C lt 0 5 X FIELD and Z FIELD define or change the position in the image field for which the irradiance distribution is to be computed Note that the position is expressed in fractional field coordinates so that the values for the x and z directions vary between 1 0 and 1 0 SPHERICAL specifies 0 40 the amount of third order spherical aberration present in the power series expansion of the wave aberration function of the optical projector COMA specifies 1C31 which is the amount of third order coma present in the power series expansion of the optical projector ASTIGMATISM specifies 2C22 which is the amount of third order astigmatism present in the power series expansion of the optical projector CURVATURE specifies 2C20 which is the amount of third order field curvature present in the power series expansion of the optical projector DISTORTION specifies 3C11 which is the amount of third order distortion present in the power series expansion of the optical projector FIFTH SEVENTH and NINTH specify the aberration order Coefficients for only one aberration order can be specified on a single statement C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 C15 C16 C17 C18 C19
61. yA ye ki vA a 3 99 Frenckel Pair Recombination During annealing many of the interstitials and vacancies recombine either at the surface or in the bulk The driving force for this reaction is to change both interstitial concentration and vacancy concentration V toward their equilibrium concentrations i s and Ven Moreover it is clearly shown that defect recombination strongly depends on the impurity concentration Silvaco 3 29 ATHENA User s Manual The following reaction is also considered kr V v 0 3 100 kvr The recombination rate can be written as follows cs sea ihe ler aa 3 101 Here nS keu dz a i a 3 102 S and k 47a D Dy 3 103 where ag 2 35 A is the distance between two separated silicon atoms in crystal Bimolecular Recombination As an alternative to the direct recombination of point defects in reaction of Equation 3 100 it is also possible for J and V to recombine through reactions such as Kv Nae aV SA SA kav kar v 3 104 In these cases the annihilation of the Frenckel pairs implies a dissociation of a dopant defect pairs Recombination at the Surface Understanding of the mechanisms that determine the interaction of interstitials and vacancies with the interfaces is getting more important because the implantation energies of dopants and the temperatures for the thermal treatments become lower and the devices are fabricated closer to the sur
62. 0 ccc eee eee eee 6 4 6 2 2 Structure and Mesh Manipulation Statements 0 cece eee eee eens 6 4 6 2 3 Simulation Statements lt 3 ise bec beret bewie beat eeld bedbeeet hoist iii e di ewks edt 6 5 6 24 Model StAleMONIS 6 ata we ek terre the aed baarre ses es bebe edel ert al eee r G 6 5 6 2 5 Special DECKBUILD Statements n on tack tamed cps aia seis eso e ake erase eas 6 6 6 2 6 Post processing StAteMeNIS x27 dance ud oa Pa ieee Gare a wer seeeati doe nh Dea achexed yaks de 6 6 6 2 7 Execution Control Statements 229 22 s is aetaveed dee ehentowdald Dies seat ass ge Aeneid 6 6 6 2 8 Obsolete Slatements o 23 5005 eoriet a e n 26 ices Sale E a Gee tae een tas 6 6 6 2 9 Standard and User Defined Materials 0 0 ccc cece e eens 6 7 6 2 10 Standard Impres ins Sanco Cee dae Mens aks Cart et Ee et Ra ad ae eae a oes 6 8 6 3 ABERRATION 22 25 i3v se teneugaieSe sa dee yad soe tvenew aes Geaee domes ENE Cares ene wee 6 9 64 ADAPT MESH oc occ ctuue ont or EnA EEEE A EREET seated EERE DESAN voceweumebe ca oy 6 11 6 5 ADAPT PAR Ari cise ope ieee eh pac ew bY Ree oe ao Sees Sees 6 12 6 67 BAKE ie ida ee ented oh Ua we cee eee Vice nee aw eared EE Umea ee eee teh Gore Lee 6 14 6 7 BASE MESH ii0852e Geue Ml cet ive ae tee cedar we dwiaged See ade suet seaeddeseuseetewies 6 15 6 8 BASE PAR cc ironies nianna orian E aa DADE E eGo ewereaeetee eds EE Se OE DADEA ss 6 16 GOP BOUNDARY i raaa A D E EE AEA ne aia I AEA DA E
63. 04 3E15 0 06 1 5E15 Or 1 7E15 0 2 9E15 0 4 2 6E15 In the following example the PROFILE statement will read in a 1D Silvaco s standard format SSF file All doping and layer information will be preserved This allows you to start a simulation in for example SSUPREM3 and finish it in ATHENA The ATHENA grid must be set up in the conventional manner first The PROFILE statement will then include any overlying layers that may have been deposited or grown in creating the SSUPREMS3 structure The value LAYER lt n gt DIV controls the number of grid points in the overlying layers The default grid spacing generated for overlying layers is 0 05 um PROFILE MASTER INF SSUPREM3 STR LAYER1 DIV 3 LAYER2 DIV 6 Silvaco 6 81 PROFILE ATHENA User s Manual The first layer above the substrate will have 3 vertical grid spacings and the second layer above the substrate will have 6 vertical grid spacings The file SSUPREM3 STR must be a SSF file The following is a list of special cases and their solutions e If a SSUPREMS structure is deeper than the ATHENA structure the PROFILE statement will extend the value of the bottom grid point e Ifa SSUPREMS structure is shallower than the ATHENA structure the PROFILE statement will clip the ATHENA profile e Loading a SSF file works only with a bare silicon wafer as a starting point If you try to use some other material for a substrate the results are u
64. 1 SPAC 0 1 LINE Y LOC 0 00 SPAC 0 03 LINE Y LOC 0 2 SPAC 0 02 LINE Y LOC 1 SPAC 0 1 The first line GO ATHENA is called an autointerface statement and tells DECKBUILD that the following file should be run by ATHENA Defining the Initial Substrate The LINE statements specified by the Mesh Define menu set only the rectangular base for the ATHENA simulation structure The next step is the initialization of the substrate region with its points nodes triangles background doping substrate orientation and some additional parameters To initialize the simulation structure select ATHENA Command Menu Mesh Initialize and the Mesh Initialize Menu will appear see Figure 2 10 Deckbuild ATHENA Mesh Initialize Material 7 Silicon Orientation Impurity Antimeny Arsenic Phosphorus silicon zinc Selenium Beryllium Magnesium Aluminum Gallium Carbon Concentration By Concentration By Resistivity 30 1 0 9 9 Exp E 14 atom cm3 Dimensionality Aute 10 Cylindrical Pasifibii Grid scaling factor 1 0 1 0 r Campasifion frackign GEC hit No impurities L Comment Initial silicon structure WRITE Figure 2 10 Mesh Initialize Menu Background doping can be set by clicking on the desired impurity box e g Boron The background impurity concentration specification will then become active If the None box is checked the concentration information will become inactive
65. 3 58 Assuming that Equation 3 58 is always in equilibrium the following equation describes the relationship between the chemical and the active arsenic concentration Ci ee tO Se Nee 2 a 3 59 chem act clust act TN n Since there is a cyclic dependency between the active arsenic concentration C and the carrier concentration n an initial guess for the value of n has to be made at the start of the simulation In addition C Cc is by definition set to unity when running the Fermi Diffusion Model see Section 3 1 2 The Fermi Model The clustering coefficient Cry is temperature dependent according to the following equation CTN E i Cry CTN 0exp NE n 3 60 Here the CTN 0 and CTN E parameters can be defined in the IMPURITY statement You can specify CLUSTER ACT CTN 0 and CTN E parameters for other acceptors in the IMPURITY statement But be aware the model isn t elaborated for other impurities and these parameters are unknown The model can also be empirically used for acceptors e g indium The following equation is based on acceptor interstitial clusters with empirical parameters CTP 0 and CTP E 2c 2 Baye S Chem Cail i CoP GF a The Semi empirical Activation Model based on Table B 14 in Appendix B Default Coefficients which is used for all other dopants except arsenic uses a two step scheme to calculate the active dopant concentration 1 The program i
66. 30 times the concentration of phosphorous in oxide at equilibrium IMPURITY I PHOSPHORUS SILICON OXIDE SEG 0 30 0 SEG E 0 0 The following syntax sets the temperature dependent impurity activation of Indium in Silicon IMPURITY I INDIUM SILICON SS TEMP 800 SS CONC lt VAL1 gt SS CLEAR IMPURITY I INDIUM SILICON SS TEMP 900 SS CONC lt VAL2 gt IMPURITY I INDIUM SILICON SS TEMP 950 SS CONC lt VAL3 gt Note The transport and segregation coefficients TRN 0 TRN E SEG 0 and SEG E are known to be inaccurate for some values of concentration material combinations and temperature ranges If the simulation is inaccurate consider these coefficients for calibration For more examples see DIFFUSE METHOD INTERSITITIAL and VACANCY Silvaco 6 51 INITIALIZE ATHENA User s Manual 6 30 INITIALIZE INITIALIZE specifies the initial starting material and background doping levels LOADFILE synonym for this statement 4 a isa Syntax INITIALIZE MATERIAL ORIENTATION lt n gt ROT SUB lt n gt C FRACTION lt n gt C IMPURITIES lt n gt RESISTIVITY lt n gt C INTERST lt n gt C VACANCY lt n gt BORON PHOSPHORUS ARSENIC ANTIMONY NO IMPURITY ONE D TWO D AUTO X LOCAT lt n gt CYLINDRICAL INFILE lt c gt STRUCTURE INTENSITY S D
67. 6 1 SSUPREM4 Diffusion Simulation Features 1 Time for diffusion and epitaxy can be specified in seconds minutes or hours New parameters SECONDS MINUTES and HOURS are added to the DIFFUSE and EPITAXY statements MINUTES is the default Diffuse time output is now presented in a new standard Total time is in hours minutes seconds hh mm ss t and time increment is in seconds The POLY DIFF diffusion model can be now applied to Poly regions formed during Si deposition over non silicon surfaces using the SI_TO_POLY parameter see below Implant Simulation Features 1 Now the damage in BCA model is calculated strictly using the modified Kinchin Pease damage model The Damage Amorphization model Implant Damage is based on the concept of critical energy density model while the damage generation rate vacancies and interstitials is based on the modified Kinchin Pease model The energy dependence of lattice disorder is analyzed with respect to spatial density of deposited energy substrate temperature and ionization events The statistical sampling method is introduced for BCA ion implantation simulation This method increases statistics for low probable events which results in better quality of ion implant profile tails Using the sampling method allows you to reduce calculation time between 5 and 100 times without reducing statistical accuracy of resulted profiles The method is switched on by using the SAMPL
68. 81 6 45 PROJECTION 2 0 arini cd ee pecan D E DAD ETA E AAEE EEE AA LA DAt 6 83 6 46 PUPIL FILTER ii h ha e a A aa a ea eee a Aa 6 84 647 QUT oee a E e a a vent E EAA E a E E 6 85 6 48 RATE DEPO gereint e n aa ewe Ue a A a E N 6 86 6 49 RATE DEVELOP seni rebar EEE ved DENE EPERE EE RAEE th REDEN DE LER DENER EA E PARA 6 88 6 50 RATE DOPE e aa aa estat ie sie a E as Siew crea aot E a aA ga A aa ee 6 90 6 51 RATE ETCH ccs cnas i amn naa aai a E AA A A A AAE 6 91 6 52 RATE POLISH niia A a a a a E O aa a A ee Mk aad ak 6 95 6 53 REGION ciei aana t a a a ENEA AO EE AEA AAAA a eee OAE AAAA RAAS 6 96 6 54 RELAX e ia iA unaa a Meant oh ae AT AEE a A Wadia aaa A e ah 6 97 0 55 SELECT ara ates seth A A AAEE EE AA E EAA A ee NA 6 98 000 SET enir EEEE EEA O EEA E E EAE E E OEE EE O ATEA 6 100 6 57 SETMODE 2 6 once lauans vel deeiexenbawetiies wide ena verem ten ese A 6 101 6 58 SILICIDE 2 cccctereeareuterscabeeeer ame aed Dae ieee eek EAA EA AE ERA 6 102 6 59 SOURCE vata cleat ra a a ge a a a are eee awed eae ae reas 6 103 6 60 STRESS ic cvedarse caveat yneniie ee veneeediencretyoessitaedeseyakiese eee E aA 6 104 6 61 STRETCH arer iana tausews a eives GovedadaysS Hon tepee eaters a a ened 6 105 0 62 STRIP chee ives bs AANE eran TANA EA aan na autbea iota CE adh ewr EAE wan 6 106 6 63 STRUCTURE a a a he a a E ses Ge sea E eG pean eaten a aes 6 107 6 64 SYSTEM araia wend ih sens ce eat pratt cen th see aoe ts bho eee aeeme eats 6 109
69. 87 RATE DEVELOP ATHENA User s Manual 6 49 RATE DEVELOP RATE DEVELOP sets development rate and exposure bleaching parameters for each type of photoresist in OPTOLITH Syntax RATE DEVELOP NAME RESIST lt c gt G LINE H LINE I LINE DUV LINE LAMBDA lt n gt A DILL lt n gt B DILL lt n gt C DILL lt n gt E1 DILL lt n gt E2 DILL lt n gt E3 DILL lt n gt RMAX MACK lt n gt RMIN MACK lt n gt MTH MACK lt n gt N MACK lt n gt RO TREFONAS lt n gt Q TREFONAS lt n gt RO HIRAI lt n gt RC HIRAI lt n gt ALPHA HIRAI lt n gt R1 KIM lt n gt R2 KIM lt n gt R3 KIM lt n gt R4 KIM lt n gt R5 KIM lt n gt R7 KIM lt n gt R8 KIM lt n gt R9 KIM lt n gt R10 KIM lt n gt C1l EIB lt n gt C2 EIB lt n gt C3 EIB lt n gt DIX E lt n gt R6 KIM lt n gt CO EIB lt n gt DIX 0 lt n gt Description This command sets the development rate parameters and exposure parameters for each type of photoresist These statements can be entered into the athenamod file so that the parameters are loaded each time ATHENA is started NAME RESIST is the photoresist name for this set of parameters G LINE H LINE LLINE DUV LINE and LAMDBA specify the wavelength for each set of photoresist parameters A DILL B DILL and C DILL are the A B and C constants
70. A T 0 for A lt A sD where A r is the concentration of the precipitate and is the effective length of capture The solid A ee 3 108 E DAA Ag A T solubility A T is defined using Arrhenius expressions in the corresponding dopant mod file 3 2 3 Interstitials Clusters Model IC Point defects in crystalline materials inherently have high free energy Free interstitials in silicon are thermodynamically unstable because of their unpaired electron orbitals and induced lattice strain At high concentrations the interstitials clusters are formed to reduce free energy Many of interstitial cluster species have been observed for many years e g 311 defects and dislocation loops The interstitial cluster configurations are believed to occur mainly in ion implanted silicon The formation and dissolution of interstitial clusters are simulated to correctly predict TED To activate the interstitial clusters models use the following statement METHOD PLS IC During a typical rapid thermal annealing various type of clusters small clusters 311 defects perfect and faulted loops evolve according to a competitive growth mechanism named Ostwald ripening The driving force for this evolution is the reduction of the formation energy per interstitial of these clusters as they grow in size and change their crystallographic structures In IC model a cluster containing n interstitials evolve to a cluster of size n 1 by int
71. AE 6 17 6 TOS CLUSTER oie e a Meets eae wand aed E E ee a ie A is ee me we NA 6 18 6 11 COMMENT isis act cd e A arte ah ee eater iat dk Dag eS Need tet ath a att irs 6 19 6 12 GPULOG iirinn aaa e GA ctor at xeatel eluate de a Like E E EA 6 20 6 13 DEPOSIT oa cet hare cat ite cacao a ape ame atta aa E A EA ue Sah nee ath a sien tia aces Seale he 6 21 6 14 DEVELOP is 2 hesitant wii Eae e A Mean i ania ctamene oe aude ean ke AAEE ANELA 6 24 6 15 DIFFUSE cee irae ote an oie ty cea beer A a a E Geter E reeset ews 6 25 6 16 DISLOG LOOP rretan exe wise se weed seepenas eedawe tote te bisa ere ne ca a aN 6 28 6 17 ELECTRODE lt ociccccsiiieactert reset evireecgeceasteewis eae ed EE AEE AEREA RT 6 29 6 18 EPITAXY aira cid betel caee ed eis a liek pias eae tas oe eee eS 6 30 6 19 ETOH oriri coven en Dare Sete ve Reed Deck ee oF sien Ure eee Ve Ree Seve eee 6 32 6 20 EXPOSE eos enaid a a deetedt el se ade gase ea a seul eee ties a e 6 35 6 21 EXTRACT casi ives g Leth atti e ee Mihaly EE et tale edad irate weal uals AAGA 6 37 6 22 FOREACH a cae dd et edad made ade sien DA Ea UTA meus cts EA AASE Ea EA aAA 6 38 6 23 GO icc elas weer aise ewes Da ie eae widen a ea ks te es i neers Mae iS wwe 6 39 624 IE Pe east Meany Dates eave a ea a ees tne trate a tate aa ata arate A een nate the late A eins een uated a Gate bere 6 40 6 25 ILEUM FILTER eee sec tod ne ad te path toby Nears taeda ed a E e oe Ha a E EA 6 41 6 26 ILLUMINATION a a a Gk eal A aia Say Case
72. ALBl lt n gt MC ALB2 lt n gt MC PLM ALB lt n gt MC NORM T1 lt n gt MC NORM T2 lt n gt MC LAT T1 lt n gt MC LAT T2 lt n gt MC ION CU1 lt n gt MC ION CU2 lt n gt MC PARTS1 lt n gt MC PARTS1 lt n gt MC ANGLE1 lt n gt MC ANGLE2 lt n gt Description This statement is used to define parameters and the machine name for one of four etch models available in ELITE MACHINE specifies the machine name for the RATE ETCH statement MATERIAL specifies material for which parameters of the etch machine to be applied see Section 6 2 9 Standard and User Defined Materials for the list of materials NAME RESIST specifies the name of photoresist to be etched WET ETCH RIE PLASMA and MC PLASMA specify a particular model for the machine definition Parameters used for RIE and WET ETCH models A H A M A S U H U M U S and N M specifies that the etch rates are in Angstroms per hour Angstroms per minute Angstroms per second microns per hour microns per minute microns per second and nanometers per minute respectively DIRECTIONAL specifies the directional component of the etching rate used by the RIE model The ionic etch rate is the contribution of the ions to the chemically oriented etching mechanisms The ions are assumed to have an anisotropic angular distribution specified by divergence parameter I
73. ATHENA predicts the physical structures that result from processing These physical structures are used as input by ATLAS which then predicts the electrical characteristics associated with specified bias conditions Using ATHENA and ATLAS makes it easy to determine the impact of process parameters on device characteristics Silvaco 1 3 ATHENA User s Manual ATHENA can also be used as one of the core simulators within VIRTUAL WAFER FAB VWF VWF makes it convenient to perform highly automated simulation based experimentation VWF is used in a way that closely resembles experimental research and development procedures Therefore it links simulation closely to technology development resulting in greatly increased benefits from simulation use For more information about VWF see the VWF AUTOMATION CALIBRATION AND PRODUCTION TOOLS USER S MANUAL 1 2 2 The Value Of Physically Based Simulation Physically based process simulators predict the structures that result from specified process sequences This is done by solving systems of equations that describe the physics and chemistry of semiconductor processes Detail analysis of various aspects of process simulation can be found in 1 and 2 Physically based simulation provides three major advantages it is predictive it provides insight and it captures theoretical knowledge in a way that makes this knowledge available to non experts Physically based simulation is different fro
74. Adaptive meshing Process Structure File 1 0 Notes Models Templates Extract Farse Deck paste init pause clear J restart kill Stop None ee ee eee eee eee ee eee Thu May 11 17 05 32 1995 Executing an host scorpio B ATHENA gt a ATHENA started ATHENA Figure 2 3 Commands Menu Now you can specify the initial rectangular grid The correct specification of a grid is critical in process simulation The number of nodes in the grid N has a direct influence on simulation accuracy and time A finer grid should exist in those areas of the simulation structure where ion implantation will occur where p n junction will be formed or where optical illumination will change photoactive component concentration The number of arithmetic operations necessary to achieve a solution for processes simulated using the finite element analysis method could be estimated as N where a is of order 1 5 2 0 Therefore to maintain the simulation time within reasonable bounds the fine grid should not be allowed to spill over into unnecessary regions The maximum number of grid nodes is 20 000 for ATHENA simulations but most practical simulations use far fewer nodes than this limit To create a simple uniform grid in a rectangular 1 um by 1 um simulation area click on the Location field and enter a value of 0 0 Then click on the Spacing field and enter a value of 0 10 Then click on the
75. Boron Transient Enhanced Diffusion in Silicon by Substitutional Carbon Incorporation Japanese of Appl Phys v 90 p 1768 2001 R Jewett A String Model Etching Algorithm M S Thesis University of California Berkeley 1979 W G Oldham et al A General Simulator For VLSI Lithography And Etching Processes Part II Application To Deposition And Etching IEEE Trans on Electron Devices v ED 27 p 1455 1980 A R Neureuther C H Ting and C Y Lin Application of Line Edge Profile Simulation to Thin film Deposition Process IEEE Trans on Electron Devices v ED 27 p 1449 1980 A R Neureuther Basic Models And Algorithms For Wafer Topography Simulation Problems and New Solutions for Device and Process Modeling Ed J J H Miller Boole Press Dublin p 99 1985 A R Neureuther Algorithms For Wafer Topography Simulation NASECODE IV Dublin Ireland p 58 1985 BIB 4 Silvaco Bibliography 88 SAMPLE User Guide Department of Electrical Engineering and Computer Sciences UC Berkeley 1991 89 M Sikkens et al Opt Eng v 25 p 142 1986 90 R N Tait T Smy and M J Brett A Ballistic Deposition Model for Films Evaporated Over Topography Thin Solid Films v 187 p 375 1990 91 R N Tait S K Dew T Smy and M J Brett Ballistic Simulation of Optical Coatings Deposited Over Topography SPIE Proc v 1324 p 112 1990 92 R N Tait T Smy and M J Brett Simu
76. C s Experimental data are from 29 The result obtain with the full PLS model is in good agreement with experimental data As expected most of the arsenic at concentration above the solid solubility limit precipitate quickly and consequently immobilize the dopant Silvaco 3 43 ATHENA User s Manual 3 3 Oxidation Models The fabrication of integrated circuit microelectronic structures and devices vitally depends on the thermal oxidation process for the formation of gate dielectrics device isolation regions spacer regions and ion implantation mask regions Particularly the precise control of silicon dioxide thickness as device geometries continue to scale to sub micron dimensions In SSUPREM4 silicon thermal oxidation is modeled when a DIFFUSION statement contains a DRYO2 WETO2 F 0O2 or F H20 parameter Oxidation takes place when there is an interface between silicon or polysilicon and silicon dioxide or a silicon polysilicon surface is exposed to an oxidizing ambient SSUPREM4 simulates polysilicon oxidation in a very similar manner as silicon almost all oxidation parameters for polysilicon are the same as for silicon SSUPREM4 also allows oxidation completely through a silicon polysilicon layer This is very important in processes e g poly buffered LOCOS in which polysilicon regions are completely consumed during oxidation Because exposed silicon surfaces usually have a thin native oxide layer SSUPREM4 automati
77. CLI KAR FILT ER parameter is used to remove all pre existing filters and sources 2 78 Silvaco Tutorial N Sigma 0 6 Figure 2 63 Annular Source 2 9 4 The Projection System The Projection System is defined using two statements PROJECTION and PUPIL FILTER The PROJECTION command is used to define the numerical aperture and flare of the projection system The PUPIL FILTER command describes the shape of the projection system and the possible filters of the projection system The shape of the projector pupil can be square or circular The circular pupil has the option of having a Gaussian or anti Gaussian transmittance profile Filtering of the Fourier spectrum can be performed by using annular filters The filters have a multiplicative effect on the transmittance and phase in the projector pupil The following example creates an opaque square at the origin PUPIL FILTER SQUAR PUPIL FILTER SQUAR m E m E INNER RAD 0 0 OUTER RAD 0 1 TRANS 0 0 This creates the following projection pupil Figure 2 64 4 Figure 2 64 Projection Pupil Silvaco 2 79 ATHENA User s Manual The maximum extent of the projector pupil plane is 1 or 1 in both dimensions A filter exceeding these dimensions will be ignored and a warning will be issued 2 9 5 Imaging Control The image calculation is done by the IMAGE command and its associated param
78. DDC and SS specify advanced diffusion models see Chapter 3 3 2 Advanced Diffusion Models If only the PLS parameter is specified the classical dopant diffusion model will be used The parameters IC and VC will invoke additional interstitial and vacancy clustering models The DDC parameter switches on the dopant defect clustering model The SS parameter will include the solid solubility model Note These advanced diffusion models can be used only for boron phosphorus and arsenic in silicon technologies Also these models cannot be used when oxidation or silicidation or both occur during the simulated diffusion step There are also some limitation on complexity of the 2D structures which can be handled by the solver In most cases when the solver cannot handle the structure materials impurities or other conditions it returns control to standard diffusion models CLUSTER DAM specifies that the Stanford 311 cluster model is enabled allowing a scaled profile of 311 clusters during a subsequent implant Only use this model when FULL CPL is also specified It further causes a transient dissolution of the 311 clusters leading to bulk interstitial injection The CLUSTER statement is used to set parameters for this model Note For correct operation set METHOD CLUSTER DAM FULL CPL before the IMPLANT statement that generates the 311 clusters HIGH CONC specifies that extra dopant concentration dependent point defect recombi
79. DU Mtx y Rmin 5 41 a 1 M x y n 1 n a 1 My 5 42 where the parameter n is a selectivity parameter describing the sensitivity of the developer to the exposed photoresist The M parameter is the threshold PAC concentration The R parameter is the development rate of a completely exposed resist The parameter Rmin is the development rate of totally unexposed resist 5 6 4 Trefonas Development Model The Trefonas development rate model 114 requires only two parameters R x y Ry 1 M x y 5 43 where R is the development rate for unexposed photoresist and q is sensitivity 5 6 5 Hirai s Development Model The development rate model by Hirai 115 is very similar to the one by Trefonas The rate function of the Hirai model is given by R x y R 1 M x y Re 5 44 where R is the development rate for fully exposed photoresist Rc is the rate for unexposed resist material and a is a reaction constant 5 14 Silvaco OPTOLITH Models 5 7 Proximity Printing Alongside with the standard projection imaging simulation OPTOLITH includes an additional module for simulation of proximity printing The proximity 1x1 printing i e the imaging without any reduction lens is used to print relatively big features of a micron scale This method is still practical and in some cases could be cost effective Internally projection and proximity printing modules use different simulation techni
80. E GPOW 0 GPOW E and VMOLE parameters can be specified in the INTERSTITIAL and VACANCY statements As a general rule the ratio 0 Kpyr Ksgyppr should be maintained reasonably constant during calibration The entities v and U _mq have the same meaning as described for surface recombination The maximum interface velocity Vi max cannot be manipulated directly and will change only when oxidation characteristics change The TIME INJ parameter in the VACANCY or INTERSTITIAL statement activates the time dependent injection model It is defined as T Giy Alt t 3 43 where is the total diffusion time in seconds and A t and Tp ow are free parameters used for calibration purposes and are calculated according to the following equations A A A dexp 44 3 44 exp kT T0 ty T0 0exp T2 3 45 0 exp kT Pog TPOW dexp POE 3 46 Table 3 5 shows all user specifiable model parameters for point defect boundary and injection conditions Silvaco 3 13 ATHENA User s Manual Table 3 5 Parameters for specifying point defect boundary and injection conditions Entity Pre exponential factor Activation Energy Keourr KSURF 0 KSUREF E KRAT KRAT O0 KRAT E Kpow KPOW 0 KPOW E A A O A E to t0 0 tO E Tpow TPOW 0 POW E HETA O0 HETA VMOLE VMOLE Gpow GPOW 0 GPOW E KSURF KSURF 0 KSURF E Vacancies The dif
81. END statements for looping The example below shows the input language used to perform the loop PRINTF ATHENA gt SWING PRINTF 16 2 2 gt SWING PRINTF THICKNESS gt SWING PRINTF CDS gt SWING FOREACH J 0 1 TO 0 5 STEP 0 25 INITIALIZE INFILE ANOPEX15 STR DEPOSIT NITRIDE THICK J DIV 1 MIN SPACE 0 01 DEPOSIT PHOTORESIST NAME RESIST 2Z2ZZ THICK 1 DIV 30 MIN SPACE 0 01 EXPOSE DOSE 150 NUM REFL 3 NA 0 FRONT REFL 1 BAKE DIFF LENGTH 0 05 STRUCTURE OUTFILE ANOPEX15 J STR2 DEVELOP MACK TIME 45 STEPS 9 SUBSTEPS 10 STRUCTURE OUTFILE ANOPEX 15 J STR3 PRINTF J Z222Z GAS 1 4 J GAS 222 1 4 J gt SWING D This creates an output file called SWING The first command writes the name of the framework The second command writes the number of rows number of columns and number of titles see the TONYPLOT chapter in the VWF INTERACTIVE TOOLS USER S MANUAL VOL I The FOREACH statement signals the beginning of the loop The END statement terminates the loop J is the parameter to be varied in the loop In this case it is the thickness of the nitride layer The final PRINTF statement prints the data to the file First the thickness J and then the CD at y 1 4 J In the DECKBUILD input file enter the command tonyplot da SWING and a plot of the swing curve will appear This command can also be written in the input file aft
82. For more examples see INTERSTITIAL and VACANCY Silvaco 6 111 UNSETMODE ATHENA User s Manual 6 67 UNSETMODE UNSETMOD unsets execution mode parameters defined in the SETMODE statement Note When SET variable value is used in DECKBUILD it is impossible to UNSETMODE the variable Syntax UNSETMODE NOEXECUTE ECHO Description This command turns off the following execution mode parameters The SETMODE statement allows you to turn on the same parameters NOEXECUTE puts all entered statements into a check only mode If this flag is on ATHENA will only check the syntax of the input commands and not actually run them ECHO instructs ATHENA to echo all input lines Examples The following turns off statement echoing UNSETMODE ECHO Note UNSET is a synonym for this command Note The parser does not recognize abbreviated forms of these commands It requires that you enter NOEXECUTE and ECHO verbatim For more examples see SETMODE 6 112 Silvaco Appendix A C Interpreter A 1 C Interpreter Overview ATHENA has a C language interpreter C Interpreter that allows you to modify the models contained in ATHENA In order to use this capability write a C language analytical function describing the model If yow re not familiar with the C language then we suggest that you r
83. INDIUM is substituted by the I INDIUM parameter in the IMPURITY statement 6 6 Silvaco ATHENA Statements List e LABEL all plotting capabilities are now provided by TonyPLOT e PAUSE is substituted by the Pause button in DECKBUILD e PHOSPHORUS is substituted by I PHOSPHOR parameter in the IMPURITY statement e PLOT 1D all plotting capabilities are now provided by TONYPLOT e PLOT 2D all plotting capabilities are now provided by ToNyPLOT e PLOT 3D all plotting capabilities are now provided by ToNYPLOT e UNDEFINE is substituted by the SET capability in DECKBUILD e VIEWPORT all plotting capabilities are now provided by ToNyPLoT 6 2 9 Standard and User Defined Materials Different materials can be specified as parameters in various statements Both standard and user defined materials are available in ATHENA The generic name MATERIAL appeared in a statement syntax description signifies that you can only specify one of standard material names from the list below or user defined material The generic name MATERIALS appeared in a statement syntax description signifies that you can specify one or several standard and user defined materials The following shows the standard material names currently available in ATHENA Semiconductors SILICON POLYSILICON GAAS ALGAAS INGAAS SIGE INP GERMANIUM SIC_6H SIC_4H STC 3G
84. Insert button and the line parameters will appear in the scrolling list Note ATHENA coordinate system has positive x axis pointed to the right along the structure surface and positive y axis pointed down to the depth of the structure Silvaco 2 9 ATHENA User s Manual In the same way set the location of a second X line to 1 0 with a spacing of 0 1 You can either set the values by dragging a slider or by entering a number directly Now select the Y direction and set the lines with the same values as the X direction You can now add the comments at the Comment line The ATHENA Mesh Define menu should appear as shown in Figure 2 4 Deckbuild ATHENA Mesh Define Direction Location Co y loc 0 00 spac 0 1 Insert y loc 1 spac 0 TED F Delete Ww Location 1 0 00 H _ 113 00 Spacing 0 1 5 00 Ta 1 00 Comment Uniform grid Figure 2 4 ATHENA Mesh Define Menu You can now write the menu prepared mesh information into the input file But first preview the rectangular grid by selecting the View button and the View Grid window Figure 2 5 will appear Notice that vertical and horizontal grid lines are distributed uniformly and the 121 points and the 200 triangles will be generated 2 10 Silvaco Tutorial 121 points 200 triangles Figure 2 5 View Grid Window A uniform grid such as the one shown in Figure 2 5 is inefficient for performing complex simulations Therefore th
85. LOOP le1l8 PHOSPHORUS INTERSTITIAL SILICON DAMALPHA 1le8 IMPLANT PHOSPHORUS DOSE 1e15 ENERGY 120 This model should only be chosen when either the Two Dimensional or Fully Coupled Model is also used The two things to apply to this model are as follows e an implant has to create an amorphous layer e immediately after the implant there is an anneal in a wet ambient Even when these two criteria are met it is suggested to only apply this model when it is needed to match experimental results The Steady State Diffusion Model The Steady State Diffusion Model is a variant of the two dimensional diffusion model which assumes the point defect profiles are in a steady state It is turned on with the METHOD STEADY command Silvaco 3 15 ATHENA User s Manual Important Notes about Defect Diffusion Point defects have larger diffusivities than dopants and may therefore diffuse down to the bottom of the structure during a simulation If the simulation structure is too shallow you may get an unrealistic high defect concentration in the regions where dopant profiles are present and consequently too much dopant diffusion Therefore you may need to extend the depth of the simulation space to provide an adequate sink for the point defects To determine the depth of the structure you can estimate the characteristic defect diffusion lengths using I DyAt 3 53 where Dy is the defect diffusivity and At is the
86. METHOD PLS statement Boron In the case of boron these clusters are named boron interstitial clusters BIC The BIC species B Im consists of n atoms of boron and m atoms of silicon self interstitials In absence of any direct experimental data concerning the exact composition of these clusters BIC structure and charge states are chosen according to recent ab initio theoretical calculation 18 Various possible path are considered for these clusters a given cluster can grow or dissolve by the addition or release of a silicon self interstitial or a boron interstitial pair Figure 3 2 a Figure 3 2 BIC reaction paths 5 z 2 2 0 By default the DDC model is based on the formation and dissociation of four BIC s species BIg Bol BI and B43 Thus the following reaction are added 3 34 Silvaco SSUPREM4 Models q Ket koa BI PSB B BI PB 3 120 ket kor Bale ket 0 Bol B B L y B l BI gt Bil 3 121 kB l KB gls By default the model assumes that dopants just after implantation are inactive As this model does not assume any local equilibrium between each species the activation of the dopant will gradually evolve with time The kinetic constants k and k for each reactions are defined as D l BI A AR Dep ksi 67 n xp a T aia 3 122 Rog and Ep B Im are respectively the effective capture radius and the bindin
87. MTTYPE in the SILICIDE statement The modeling and understanding of silicide growth is nowhere near as developed as for oxidation But you can consider simulation of silicidation process similar to that of oxidation It starts with insertion of a thin 0 002 microns initial layer of silicide on the boundary between silicon or polysilicon and corresponding metal During each time step growth velocities are calculated for each point at both metal silicide and silicon or polysilicon silicide interfaces The growth velocity at the i interface point is calculated as follows dx n aoe ORAA 3 172 dt ENS where k is the interface reaction rate coefficient N 7 is the number of silicon or metal molecules per unit silicide material and C is the silicon or metal concentration n is the interface normal vector which points towards the silicon poly or metal side Similarly to oxidation this equation can be solved by applying an initial boundary condition x x at t 0 The solution is X X ttr 3 173 B B A where parameters B 2D C Nj and B A KC N are equivalent to Deal Grove coefficients of classical oxidation model The silicide growth data indicates that for most silicides the rate limiting step is diffusion of silicon This simplifies the Equation 3 174 to x Bt 3 174 The silicide growth rates parameters are extracted from experimental data for TiSi 41 42 and CoSi2 43 For two ot
88. Model The Dill model 111 uses the parameters E E9 and E3 Surface induction effects are not considered The bulk development is given by R x y exp E E M x y E M x y 5 34 and for M x y lt 0 4 5 6 2 Kim s Development Model The Kim model 112 describes the development rate through the function d R induction l 1 R5 RK5 Rs M x Y z exp 5 36 4 1 R y9 5 37 BD T Ey ME R R Silvaco 5 13 ATHENA User s Manual M x y M x y exp R3U M x y 5 38 R x y Rinduetion Reulk oa R x y Ri juction Reulk ot where Rp is the bulk development rate and Ry duction 18 the surface induction factor The limiting development rate values are R and Ro respectively for completely exposed and unexposed resist The function R nduction y is an empirical relationship describing the reduced dissolution rate at the surface of a resist layer and is a function of the normal distance from the original surface of the resist d y and the amount of remaining PAC M x y The parameter R4 is the characteristic length along this path for the induction effect The parameters Rs and R are respectively the ratio of the surface rate to the bulk rate for a completely exposed resist and the ratio of surface rate to bulk rate for an unexposed resist 5 6 3 Mack s Development Model The Mack model 113 describes the development rate through the function n R x y ee at
89. Neutrals seceeeseeeeeeneeeseeeeeeneeeees 4 15 lon and Neutral Fluxes 00 0 eeeeeeesseeeeeeeeeseeeseneeeees 4 15 17 Polymer FIUX S lt cccsecscescsecietoranacenerasoessescneayeseerenaaieeess 4 17 Rates nivenr aerate ath alee ees 4 18 Monte Carlo Implant Models Amorphous Material esccesssesesseeessneeeesneeereees 3 84 85 C Interpretet E A TETA 3 89 Cluster Model imen tena r e 3 88 Crystalline Material ceeceescesseeeeeeeeseeteaeeteeeeeaeeeaees 3 85 Damage Accumulation Model 22 cceeeseeseeeeeees 3 81 82 Dislocation Loops Model ceeseeeeesreeeseeeeeeneeeees 3 88 89 Electronic Stopping siiv scsieslecs nian nate ec eas 3 80 81 Implantation Geometry eeeeeceeeeeeeeeeeeeeeeneeteeeeeaeens 3 82 84 Interatomic Potential ceccecceseseeeesceeseneeeeseeeeeneenees 3 79 lon Implantation Damage eeeceeesseeeeeneeeesneeeeeneeeee 3 87 Nucl ar Stopping ersi atid A eatnes 3 77 79 Physical Problems escceessceeeeeeeeeneeeeeneeeesneeeeeneeeees 3 77 PluS 1 Model sececcsstcccscte tusteaShsoseeesticasiesdepiacageseteress 3 87 88 Statistical Sampling c ceeceeeceeeeeeeeeeeteeteeeeeeeesees 3 85 87 OTO POEA ETA TT epee eae ele 3 77 Monte Carlo Implant Module ccceeeeeesteeeeeeeeeeeteeteeeeeee 6 45 MOSFET Process Flow Gonesse Mek A ea ee 2 47 INputest eet its ee ee ee 2 42 43 PMOS TUNING seerne Aes eects valde 2 46 Predictive Powers of Tuned Proce
90. Oxide Implantation Through Thermally Grown Oxides and Dopant Loss During Subsequent Annealing Frequently dopants are implanted through thermally grown oxide layers It is important to have a proper grid spacing in the oxide through which the dopant is implanted for two reasons First this will aid in determining the proper dopant profile in the oxide layer and the underlying silicon Secondly proper gridding is required to resolved the dopant diffusion in the oxide during subsequent processing steps During annealing the dopant will diffuse in SiO and silicon and eventually evaporate into the ambient at the gas SiO interface If proper gridding is not supplied in the growing oxide layer the amount of dopant evaporating can be underestimated yielding a larger dose retained in the silicon substrate The mechanism is similar to what was described in the earlier sections There may not be any grid points in the interior of the growing SiOz layer The problem is again remedied by specifying more grid layers to be added as the SiO layer grows 3 60 Silvaco SSUPREM4 Models Figure 3 18 shows a comparison of the resulting arsenic profiles in silicon using the default grid spacing and a corrected grid spacing in the growing SiOz layer For this experiment a silicon dioxide layer was thermally grown Arsenic was ion implanted through the SiO Silicon structure A subsequent annealing step followed which results in the profiles sh
91. PLS in the METHOD statement Parameters Related to File Output DUMP and DUMP PREFIX specify that a structure file be output at every DUMPth time step The files are readable with the STRUCTURE statement or can be displayed using ToNYPLOT The names will be of the form DUMP PREFIX lt time gt str where lt time gt is the current total time of the simulation in minutes TSAVE and TSAVE MULT specify that intermediate structure files be output when the advanced PLS diffusion model is used The structure files named DUMP PREFIX lt time gt str will be output at time TSAVE TSAVE MULT k 0 1 2 where time is in seconds The default value for the parameter DUMP PREFIX is at The parameter TSAVE MULT should be greater than 1 0 Parameters Related to the Model Files for Advanced Diffusion Models B MOD PMOD AS MOD IC MOD and VI MOD specify direct paths to boron mod phosphorus mod arsenic mod i mod and defect mod files correspondingly By default these files are in SILVACO lib athena lt version_number gt common pls directory You can modify your own mod files inside directories specified by these parameters Miscellaneous Parameters NO DIFF specifies that impurity diffusion be neglected during the calculation This can be used to observe oxidation or silicidation geometry without unnecessary timesteps related to impurity diffusion REFLOW specifies that a surface tension based refl
92. Point Defects pinipan aaeeea ui oen Ea 3 26 27 ATHENA Features and Capabilities ceeeeeeteseeeeeneeeees 1 2 3 CNET Generation Recombination Terms ATHENA Input Output c cccccscccsccsecssesseceecsecssessesseessnenee 2 7 Formation of Pairs 2 cecs sstscssegecugbeatanteisstiseuesaiess 3 29 pit hs etna neh S 2 7 Frenckel Pair Recombination sesssseeessseeenseee 3 29 30 ON EEEIEE ste E AE AAT 2 8 Compound Semiconductor Simulation Standard Structure File Format SSF escesseeeseeeeeeeeee 2 8 Diffusion Models e ceesseeeeeneeeeeseeteseeesseeeeenetereaes 3 93 ATHENA OPTOLITH lon Implantation Models c eee eee eee teeeeeeeeeeees 3 94 CD Extraction Smile Plots And Looping Procedures 2 84 85 Concentration Jump Condition 0 2 eeeeeeeeeeeeeseeeeeeeeeeeeeeeeneees 3 4 Illumination SYStOM ssssssssceeeesessesereseeseeteanesn 2 77 79 Continuity Equation ase cetecile elaa ian Mele AG Ta ed lea rate 3 2 Imaging GCOntOl Sch ee icestert a aa keeled 2 80 82 See also Diffusion Equation MASK s 211 NEET A ARE 2 74 77 Correct Substrate Depth Modelling Material Properties s sssssssssseressererrerirrertnrerrnrerrnrerenrerenne 2 82 Diffusioma e ea tO a a a A ea betes 2 37 38 Projection SyStem sssssssssiseseseeeessssssannteeeeten 2 79 80 fomnlmplantalions 20vac5it Ai aA he NS ix 2 34 36 Structure Exposure cesceesceeseeeeseeeseeteteesseeeeaeetaees 2 82 84 ANONaMIMp
93. See this section for advice on selecting the appropriate pull down menu from DECKBUILD The default method for oxidation is Compress In SSUPREM4 examples there are a number of examples which illustrate the use of different models for different processes and structures In our previous example described in the Simulating Diffusion Section on page 2 37 if the next temperature step is going to be at a constant temperature of 1000 C in dry O2 with 3 of HCL in the ambient select the Dry O2 box and set HCL equal to 3 in the Ambient section of the Diffuse menu The following input file fragment will appear GATE OXIDE DIFFUSE TIME 60 TEMP 1000 DRYO2 PRESS 1 00 HCL PC 3 If the ambient is a mixture consisting of more than one oxidant the total oxidation rate will depend on the combined effect of all species in the ambient To specify the contents of the ambient mixture select the Gas Flow button in the Ambient section and an additional ATHENA Gas Flow Properties Menu Figure 2 31 will appear Gy Deckbuild ATHENA Gas Flow Properties H2 Flow lfm 0 0 o0 i 200 0 H20 Flow l m 5 WAOE _ 200 0 HEI Flow sccm 0 0 w _ _ 1000 0 N2 Flow lfm f CH 2700 0 02 Flow lfm 0 a M 2700 0 Figure 2 31 ATHENA Gas Flow Properties menu If the Gas Flow components are selected as shown in Figure 2 31 the following statement will be generated GATE OXIDE DIFFUSE TIME 60 TEMP
94. SiO2 Mm Cmss CTmax M12 M1 Mo and S 10S9 The Cr parameter is the real density of occupied trap sites at the interface and is found by solving the following equation i es Fe op 3 16 or where Cymax 6 8 x 10 4cm for phosphorus and 2 x10 4cm for other dopants The dose loss transport coefficient S4 is calculated through the following Arrhenius expression S TRNDL O exp Ene 1 kT where the TRNDL O and TRNDL E parameters can be specified in the IMPURITY statement 3 1 4 The Two Dimensional Model In this model the point defect populations are directly represented and evolved in time If there is a super supra saturation of point defects it will affect the dopant diffusivity through a simple scale factor which goes to unity as the actual defect concentration approaches the equilibrium defect concentration Therefore with equilibrium defect profiles the Two Dimensional Model merely reduces to the Fermi Model although in a more computational inefficient manner since solving for point defects is not required The pair coupling between defects and dopants in this model is assumed to be one way The diffusion of dopants is highly influenced by the diffusion of point defects while the diffusion of the point defects however is regarded as totally independent of dopant diffusion Stated in physical terms this corresponds to a pairing between defects and dopants with zero binding energy To turn on the Two Dimensional M
95. a solution created by ATHENA e DEVEDIT to generate an updated mesh and export the mesh and doping back to ATHENA or any other simulator For more information on structure files see Saving a Structure File for Plotting or Initializing an ATHENA Input file for Further Processing Section on page 2 28 2 3 3 Creating An Initial Structure This section will describe how to use DECKBUILD s Commands menu to create a typical ATHENA input file The goal of this section is not to design a real process sequence but to demonstrate the use of specific ATHENA statements and parameters as well as some DECKBUILD features to create a realistic input file You can find many realistic process input files among the examples and use them as a starting point in your process simulation Defining Initial Rectangular Grid Once DECKBUILD is running and the current simulator is set to ATHENA see the VWF AUTOMATION CALIBRATION AND PRODUCTION TOOLS USER S MANUAL for more information open and pin the Commands menu as shown in Figure 2 3 Then select Mesh Define and the ATHENA Mesh Define Menu will appear We recommend that you pin this popup because it will be used often in designing an initial mesh 2 8 Silvaco Tutorial Deckbuild 3 5 3 Beta NONE dir tmp_mnt main lucky stacy i File 7 View v CEdit 7 Find 7 Main Control 7 Commands 7 i Tools 7 CA ne ATHENA Mesh Define Mesh Initialize
96. and grid information generated by MASKVIEWS Press the Preview button and the Display Masks Window will appear as shown in Figure 2 52 AS Maskviews ATHENA cutline 1 Y1 K2 Y2 2 30 12 10 4 90 12 19 Depth 1 00 microns Figure 2 51 ATHENA Cutline Popup 2D masks from 2 3 12 1 to 4 9 12 1 L fo ots tho 125 150 1 45 zbo Figure 2 52 Display Masks Window 2 68 Silvaco Tutorial The additional information on the number of lines points and triangles is also displayed in this window If the grid does not appear as shown in Figure 2 52 select the Options Grid box and the Display Masks box in the Properties menu Figure 2 53 ns Maskviews Properties Category defaults Angle constraint None 45 amp 90 90 Options Wi Grid Wi Hints Pointer Center origin L Display labels fy Display masks W Simulator 7 ATHENA Cutline edit Wi Write Y grid Wi Request filename Group buffer all layers edit layer only File loader list files only add directories Cutline preview show line only Figure 2 53 Properties Menu To select another cutline location press the Done button in the Athena Cutline Popup and repeat the cutline selection process for the desired cutlines one at a time If you re not satisfied with the grid you can modify the X or Y or both settings You can then preview the modified grid without selecting another cutline For example if the Spacing at edge in
97. and will appear grayed out from the rest of the menu Select the desired concentration using the slider e g 3 0 and select an exponent from the Exp menu e g 14 This will give a background concentration of 3 0e14 atom cm3 You can set background concentration using the By Resistivity specification in Ohmecm For this tutorial check the 2D box in the Dimensionality field This will run the simulation in a two dimensional calculation The other items in this menu will be discussed in Section 2 8 Using Advanced Features of ATHENA 2 14 Silvaco Tutorial Note Two dimensional mode is used in this tutorial to demonstrate 2D grid generation and manipulation In most cases however it is unnecessary to change the Auto default in the Dimensionality item of the Mesh Initialize menu ATHENA will begin in 1D and will automatically switch to 2D mode at the first statement which disrupts the lateral uniformity of the device structure This generally results in massive savings of computation time You can now write the mesh initialization information into the file by pressing the Write button The following two lines will appear in the Deckbuild Text Subwindow INITIAL SILICON STRUCTURE INIT SILICON C BORON 3 0E14 ORIENTATION 100 TWO D Now run ATHENA to obtain the initial structure Press the Run button on the DECKBUILD control The following output will appear in the simulator subwindow ATHENA gt NON UNIFORM GRID ATHE
98. antimony and boron For the segregation calculation the file name for model substitution is set on the DIFFUSE statement with the string parameter PSEG CALC lt filename gt This syntax is valid for all of the above with the string parameters being PSEG CALC AS SEG CALC SB SEG CALC and B SEG CALC The activation calculation can also be accessed by the C Interpreter for phosphorus arsenic antimony and boron For the activation calculation the file name for model substitution is set on the DIFFUSE statement with the string parameter PACT CALC lt filename gt This syntax is valid for all of the above with the string parameters being PACT CALC AS ACT CALC SB ACT CALC and B ACT CALC All of these parameters can be used at the same time or separately as desired Templates for all these functions are located in a file called athena lib located in the directory SILVACO lib athena common A sample function is given for each of the diffusion coefficient calculations segregation calculations and activation calculations All these functions should have different names The template file is copied to the current directory by typing athena T lt filename gt in a C shell D 13 2 ELITE Capabilities CHEMICAL and DIVERGENCE parameters have been added to the RIE model on the RATE ETCH statement These account for ions that hit the structure at other than normal incidence A Gaussian distribution of ions as a function of the angle is assumed DIVE
99. area of the effective source for which y x9 zg has non zero values Silvaco 5 5 ATHENA User s Manual For this purpose Equation 5 19 is put into the form rusvi rx 29 Pepu v degi 5 20 2 where 1 1 H nd I 2 Dxgyousv 5 fa xoy yo Ky explilu x vy dxdy 5 21 xp Zo u v is proportional to the intensity at the point u v due to a wave of unit irradiance passing through x9 zo of the effective source In the case of an annular shaped source xg Za has the form ee ee 0 for xo Zo lt amp 2 2 XXo Zo 4 1 for xo Zo lt 1 5 22 x 2 0 for x9 Z gt 1 where s is the fractional radius of the centered circular obstruction in the exit pupil of the condenser lens For a circular exit pupil s becomes zero Equation 5 20 is the principle relation of a generalized Abbe theory where the image formation under partially coherent illumination of the object is accounted for by a combination of coherent imaging processes for perpendicular and obliquely incident illuminating plane waves on the object Since only the image irradiance is of interest it can be determined without using of coherence theory 103 For the computation the whole source is divided into a number of luminous point sources considering the imaging due to each source as an independent coherent image formation process The contributions from each point source do not interfere so the net image irradiance is the sum of the irradiance
100. area produces additional diffusion in the center that is not seen for longer channel devices This phenomenon explains some of the reverse short channel effects seen in certain processes Silvaco C 5 ATHENA User s Manual Reverse Short Channel Effect n MOS Channel profile versus Gate Length Active Boron em3 ft jij 1 L 1 0um x L O 8um L O0 7um x L 0 6 um per ey 0 3 0 4 Depth Figure C 2 Enhanced diffusion of MOS channel profile Question I use SSUPREM4 for process simulation but I need more realistic models for deposition and etch How can I use the ELITE module of ATHENA to do this How does the interface from ELITE to SSUPREM4 work Answer ATHENA is a general purpose two dimensional process simulator that includes modules for implant diffusion and oxidation for silicon and compound semiconductors SSUPREM4 topography ELITE and lithography OPTOLITH This means that it is simple to include physical etch or deposition steps using ELITE models in an existing SSUPREM4 input file As device dimensions shrink the need for more physical simulation of the deposition and etch steps in a process increases ELITE provides these physical deposition and etch models SSUPREM4 users can only use conformal deposition and geometrical etch features built into ATHENA These simple models may not be sufficient to describe certain steps in the process satisfactorily For example in a typica
101. based on models that are as reliable as possible It has become clear that the abnormal behaviors of dopant diffusion in silicon are caused by non equilibrium point defects These are induced by the diffusion process itself emitter push effect caused by high concentration of phosphorus diffusion or injected into the substrate by external treatments such as oxidation or silicidation Otherwise they result from the ion implantation used to introduce the dopants into the silicon substrate With the necessary decrease of the thermal budget due to the shrinkage of the device dimensions these transient phenomena become key issues for accurate dopant diffusion simulation Classical Interstitial Dopant defect model Clusters modal Clusters model Full and physical model Figure 3 1 The model consists of three parts the classical dopant diffusion model the interstitials clusters model and the model of mixed dopant clustering The new model of dopant diffusion implemented in ATHENA is called PLS and was developed in collaboration with CNRS Phase Strasbourg France CEA LETI Grenoble France and SILVACO This model is up to date with actual physical models and contains only physical parameters 18 It consists of three parts the classical dopant diffusion CDD model the interstitials clusters IC model and the model of mixed dopant defect clustering DDC This section describes the three parts of the model and how the new model differs fro
102. can also use to integrate along a specified line The value printed is the value that has been selected see Section 6 55 SELECT X VALUE specifies the x coordinate of a vertical cross section along which the selected values are to be printed Units are microns Y VALUE specifies the y coordinate of a vertical cross section along which the selected values are to be printed Units are microns MATERIAL specifies the selected values in the named material at the interface with another material named by MATERIAL are to be printed see Section 6 2 9 Standard and User Defined Materials for the list of materials ARCLENGTH is only relevant when printing along an interface If ARCLENGTH is chosen the printed ordinate is the arclength measured in microns along the boundary from the left most point of the curve If ARCLENGTH is not chosen the x value of the interface location is printed The coordinate of the left most point is equal to its x coordinate in the mesh layers LAYERS instructs the selected print variable to integrate in each material it crosses The integrated value and material width is reported Zero crossings of the variable are treated the same as material interfaces X MIN and X MAX specify the minimum and maximum positions along the cross section to be printed FORMAT changes the print format for the variable using standard format expressions of the C language Default is 16e Examples T
103. ccc en een n ene nes B 4 B 1 9 Stress dependent Growth Model Coefficients 0 c cece eee eens B 4 B 1 10 Mechanical Parameters For Stress Calculations 00 c cece cee eee eeeee B 4 B 1 11 Linear Coefficients Of Thermal Expansion 0c cece eee e eee eens B 5 B 1 12 Volume Expansion Ratio vess03sbevew eels teed ieee even bepiad coda den enenbewlaedveien B 5 B 2 Impurity Diffusion Coefficients 0 cece eee eee e enn eeee B 5 B 3 Impurity Segregation Coefficients cece cece teen eee eee ene eeeeee B 6 B 4 Interface Transport Coefficients 00 c cece ete eee e eee eee eee eens neee nee eeaeee B 7 B 52 Solid solubility In SICON ci beech iene rea hares wierd ade E aed eae B 7 B 6 Point Defect Parameters 00 cece eee eee B 8 B 7 Defect Interface Recombination Parameters 0ccceeee cece eee eee e een eee teens B 8 B 8 Defect Growth Injection Interface Parameters 00ce cece eee eee eee eee een neneeeee B 9 B 9 Material Parameters ss lt 0 dra ld acotd a aced eae a dla We A doa Vier aa eee ae ea B 9 Appendix C Hintsand MWS itccce eee see teotane peered eee NN ete tee ate OOE ye C 1 Appendix D ATHENA Version HIStOry oia sec nnn tecutr eins cee tae ten ten enn D 1 D 1 ATHENA Version 5 16 0 R Release Notes cc cece eect e eee eee ee eee n eee teeeenee D 1 Ditch SSUPREM4 Features neei aei pa eames eid Sard 6 an wh ae bad wate oat a D 1 D 4 2 Optolith F at
104. corresponds to redeposition ER ER polymer PF lt 0 4 24 The corresponding ejection rate EJR is equal to the etch rate of polymer EJR ER polymer 4 25 When calculated ER polymer is larger than polymer flux the actual etch rate is positive EP m i ER ER m PF 5 2 4 26 ve 2 BP ooma i n The corresponding ejection rate is calculated as follows EJR PF ER 4 27 4 18 Silvaco ELITE Models C Interpreter You can use the C Interpreter to introduce different etch and ejection models The following parameters are passed to the C Interpreter file and can be used for implementing the models number of ion types the four characteristics of ion fluxes for each ion type Equations 4 19 4 22 PF and surface material m Returned parameters are ER and EJR For example you can simulate the wet etching by setting the etch rate to a constant positive value depending only on the surface material In this case the trajectory tracing part of the model is not needed The number of trajectories can be set to one Uniform deposition can be simulated by the setting of a negative constant etch rate and by specifying the redeposited material other than polymer in the ETCH statement If the fluxes are not used as in the wet etching simulation the void formed will eventually be filled with the deposited material because inside the C Interpreter there is no way to determine if the current surface segment belongs to the vo
105. could be found from tan a tan cos 3 204 When the FULLROTAT parameter is specified in the IMPLANT statement ATHENA calculates superposition of 24 implants with rotation angles equal to 15n and doses equal to 24 The implantation front perpendicular to the direction is divided into a number of slices N usually gt 100 of width a The implant concentration in each grid point i with coordinates x y is calculated by the summation of contributions from each slice k Cx y gt C x 3 205 I lt k lt N The contribution from each slice C is calculated by integration of the point source 2D frequency function F p x y with the starting point at the intersection of the normal n to the central of the slice with the structure surface over slice width a 2 City frp d t di 3 206 a 2 3 72 Silvaco SSUPREM4 Models where d is the depth along implant direction i e distance between the starting point and the projection of the point i on the vector n and t is the transversal distance i e distance between the point i and the vector n See Figure 3 21 implant Figure 3 21 Integration Geometry for the Convolution Method Depth Independent Lateral Distribution The simplest type of the 2D frequency function is a product of longitudinal function fix which can be a Gaussian Equation 3 178 Pearson Equation 3 191 Dual Pearson Equation 3 193 and depth independent transve
106. defined materials can be also specified The following chemical elements are typical dopants in these compound semiconductors Si C Se Be Mg Ge and Zn 3 8 1 Diffusion Models The default diffusion model in compound semiconductors is the same as the Fermi Model with electric field effect used for silicon in SSUPREM4 see Section 3 1 2 The Fermi Model All diffusivity parameters from Table 3 2 can be specified for each dopant in all compound materials But only reasonably calibrated set of diffusion parameters exist for GaAs 77 You should perform calibration or all other materials Note More advanced diffusion models TWO DIM and FULL CPL can be potentially specified for compound semiconductors The point defects kinetics however is largely unknown which means that extensive research and calibration is needed It was determined in 77 that n type dopants Si Se and Ge in GaAs diffuse through the Ga Vacancy Mechanism while p type dopants Be Mg Zn and C diffuse through the Ga interstitial mechanism This means that for donors in GaAs the diffusivity is calculated as follows n 2 7 _ D lonor 7 Day Day Dyy 2 3 239 i i To look up these diffusivity terms see Table 3 2 The intrinsic carrier concentration n is calculated by Equation 3 9 with the parameters NI 0 NI E and NI POW taken from 78 The experimental data cited in 77 show that diffusivity of Si and Se in GaAs can be considered as conc
107. dopant atom cannot diffuse on its own it needs the assistance of a point defect a silicon self interstitial or a lattice vacancy in the near vicinity as a diffusion vehicle If there is a non vanishing binding energy between the two they can move as one entity a pair through a number of jumps and inversion cycles before eventually breaking up When speaking of dopant diffusivity within the scope of these models one actually means the diffusivity of the pair as a whole A point defect however can either diffuse freely or as a participant in a dopant defect pair The diffusivity of a free point defect can actually be different from the diffusivity of a point defect pair All diffusion models in ATHENA also use the concept of Chemical and Active Concentration Values The chemical concentration is the actual implanted value of the dopant but when dopants are present at high concentrations clustering or electrical deactivation can occur so that the electrically active concentration may be less than the corresponding chemical concentration This is described Section 3 1 6 Electrical Deactivation and Clustering Models ATHENA creates structures that can have multiple materials and interfaces such as the polysilicon oxide silicon interface in MOSFETs Each interface within ATHENA has boundary or interface conditions that model impurity segregation The model details are described later in this chapter You should however be aware that the g
108. during the analytical ion implantation process simulation step by scaling their distribution densities to the implanted profile Plus 1 Model The first damage model is related to free point defects Here interstitials are scaled to the as implanted dopant profile with the scaling parameter DAM FACTOR lt n gt This model is invoked with the UNIT DAM flag on the IMPLANT line For example IMPLANT PHOS DOSE 5E14 ENERGY 45 UNIT DAM DAM FACTOR 0 001 Silvaco 3 87 ATHENA User s Manual This model is known as the Plus 1 model In the case of low implantation doses the value for DAM FACTOR has been suggested to be equal to unity Although perhaps valid at low doses the related and subsequent diffusion mode METHOD FULL CPL is not required in most cases Therefore this combination is an impractical approach Recent research on RTA diffusion models e g Stanford s 311 Cluster model has introduced other forms of damage Thus lowering the dependency of free point defects being initially set at a Plus 1 scaled profile The DAM FACTOR parameter when used with the 311 Cluster model should have a far lower value in the order of 0 001 Note that this is an extremely sensitive parameter when studying shallow junction formation so use it carefully 311 Cluster Model The 311 Cluster model introduces a bulk injection source of interstitials in addition to any other free point defects sources Clusters are int
109. eae reteten ite AHH Ct feck ESTETI reise tite Le i it ct cht Cat etier it Hi sii yt CHHT fF fi celal atte it Microns LYACO International 13994 4 Figure 2 22 ATHENA Reflect Capability The left half of the structure is a complete mirror copy of the right part including node coordinates doping values and so on Beware of rounding errors when mirroring If the boundary of reflection is not smooth to within 0 1 angstroms some points will be duplicated Specification of Electrodes in ATHENA The ultimate goal of an ATHENA simulation is usually to create a device structure material layers plus doping which then can be used by a device simulator usually ATLAS for electrical characterization Although ATLAS is able to specify the locations of electrodes in many cases specifying electrodes must be done in ATHENA For example it is impossible to specify an electrode location in ATLAS when the electrode does not consist of straight segment
110. etching the liquid attacks the surface and removes soluble products This reaction produces volatile by products which are removed by a vacuum pump Physical reactions do not take place A barrel plasma reactor achieves such conditions usually at low powers and moderate pressures Due to the chemical reaction isotropic profiles develop with mask undercutting and circular cross sections fiso bisects the angle 7 between line segments _ Initial line of action Isotropic advance r fiso Figure 4 8 Segment Point in Case of Isotropic 4 4 2 RIE Model In the Reactive Ion Etching RIE model the etching process is divided into the two adjustable components isotropic etching and anisotropic etching Each of these components is characterized by empirical etch rates riso and rq The ratio ry Dee ee 4 13 Friso r dir defines the measure of anisotropy The isotropic component r models chemically reactive etching which results in profiles with undercut and circular cross sections For A 0 the process is completely isotropic Under isotropic conditions the string points are advanced at the constant rate r in the direction of the perpendicular bisector of the adjacent segments see Figure 4 8 4 12 Silvaco ELITE Models The anisotropic etch rate component r is proportional to the cosine of the angle between the flux direction and the surface normal the perpendicula
111. for the Dill exposure model E1 DILL E2 DILL and E3 DILL defines the E1 E2 or E3 parameter for Dill s development rate function These parameters are dimensionless RMAX MACK RMIN MACK MTH MACK and N MACK are the constants for the Mack development model RMAX MACK specifies the development rate of the fully exposed resist RMAX MACK must be specified in microns sec RMIN MACK specifies the development rate of the unexposed resist RMIN MACK must be specified in microns sec MTH MACK is the threshold normalized PAC concentration MTH MACK is dimensionless N MACK specifies the developer sensitivity N MACK is dimensionless RO TREFONAS and Q TREFONAS are constants for the Trefonas development model RO TREFONAS specifies a development rate constant RO TREFONAS must be specified in microns sec Q TREFONAS specifies a development rate constant RO HIRAI RC HIRAI and ALPHA HIRAT are constants for the Hirai development model RO HIRAI specifies the development rate of the fully exposed resist material RO HIRAI must be specified in microns sec RC HIRA specifies a development rate for unexposed resist RC HIRAI must be specified in microns sec ALPHA HIRAI specifies a dimensionless reaction constant R1 KIM R2 KIM R3 KIM R4 KIM R5 KIM R6 KIM R7 KIM R8 KIM R9 KIM and R10 KIM are constants for the Kim development model R1 KIM corresponds to the dissolution rate of the resist material if it has been fully exposed that is i
112. fragment appears in the input deck DECKBUILD will then generate the next sequence of the etc etc etc etc etc etc etc etc etc etc etc etc N N N DEPO PHOTO THICK 1 EXPOSE MASK POLY DEVELOP photo start x 0 100000 y 1000 cont x 0 100000 y 1000 cont x 0 800000 done x 0 800000 photo start x 1 cont x 1 200000 cont x 1 800000 done x 1 800000 photo start x 2 cont x 2 200000 cont x 3 100000 done x 3 100000 y 1000 y 1000 200000 y 1000 y 1000 y 1000 y 1000 200000 y 1000 y 1000 y 1000 y 1000 ETCH statements Silvaco E 5 ATHENA User s Manual If you use negative photoresist the photoresist will be removed underneath all opaque regions EXPOSE MASK POLY DEVELOP DEPO PHOTO NEGATIVE THICK 1 In this case DECKBUILD will generate an alternative sequence of the ETCH statements etch photo start x 0 800 etch cont x 0 800000 y 1 etch cont x 1 200000 y 1 etch done x 1 200000 y etch photo start x 1 800 etch cont x 1 800000 y 1 etch cont x 2 200000 y 1 etch done x 2 200000 y 000 y 1000 000 000 1000 000 y 1000 000 000 1000 Note sec files generated by SILVACO s MaskViews tool provide superior capabilities in the simulation grid generation and mask processing control E 6 Aliases and substitutions for some statements AMBIENT is alias for O
113. from each source point The normalization used throughout this investigation is that the mask is illuminated with unit irradiance so that the ideal image has unit irradiance where unit magnification is assumed Therefore the brightness of the source decreases as its size increases Equation 5 20 is the principle equation of the algorithm which is used for studying the influence of annular apertures The object spectrum see Equation 5 11 is calculated analytically and the coherent image see Equation 5 18 is calculated using a Fourier Series approach The shape of a single mask feature must be rectangular This is because the Fourier transform for a rectangular feature is calculated based on an analytical formula Since the Fourier transform is linear you can compose arbitrary shaped mask features from the rectangular components The object spectra of the single mask features components are simply added up The treatment can then be considered as appropriate and no numerical discretization errors in the size and placement of the mask features can occur Note You may use MASKVIEws to create or import masks of any arbitrary shape The mask layout will be sliced divided on rectangular elements when it is imported into OPTOLITH OPTOLITH can import masks containing any number of mask elements 5 6 Silvaco OPTOLITH Models 5 3 Optical System Figure 5 2 shows the optical system used by OPTOLITH The meshes in the Fourier
114. growth dependent generation model VMOLE is the lattice density of the consumed material the units are cm THETA 0 specifies the pre exponential constant for the fraction of consumed atoms injected as interstitial or vacancy THETA E specifies the activation energy for the fraction the units are eV GPOW 0 and GPOW E specify pre exponential constant and activation energy of the power parameter of the growth injection formula WETO2 DRYO2 specify whether the parameters THETA 0 and THETA E are for wet oxidation or dry oxidation The default is DRYO2 REC STR and INJ STR allow you to specify experimental models for interstitial or vacancy recombination or injection at interfaces Three macros are defined for use T is the time in seconds and X and Y is the coordinates If these are specified they are used in place of any default model A 0 A E T0 0 T0 E TPOW 0 and TPOW E specify parameters for time dependent injection model TIME INJ A O is the pre exponential constant for the injection rate and A E is the corresponding activation energy T 0 and T E are the pre exponential constant and activation energy for the time constant in the time dependent injection formula TPOW 0 and TPOW E are the pre exponential constant and activation energy for the power constant in the time dependent injection formula Silvaco 6 55 INTERSTITIAL and VACANCY ATHENA User s Manual
115. growth of boron doped silicon on top of silicon at a rate of 0 5 um per minute The deposit thickness is time x rate 5 um EPITAXY TIME 10 TEMP 1150 C BORON 5E14 GROWTH RATE 0 5 Time and Temperature Example The following statement will deposit 64m of epitaxial silicon on top of silicon over 10 minutes Phosphorus is out diffused during the processing The number of vertical grid points in the completed epitaxial layer is set with the DIVISIONS parameter The syntax is similar to the DEPOSIT statement EPITAXY THICK 6 TIME 10 TEMP 1180 C PHOS 1 5E14 DIVISIONS 20 Non uniform Grid Control Example The following statement performs epitaxial process with a non uniform vertical grid spacing The vertical grid spacing will be 0 5 um at a distance of 5 um below the final surface The epitaxial layer will be subdivided into 40 sublayers EPITAXY THICK 10 TIME 30 TEMP 1100 DY 5 YDY 5 0 DIVISIONS 40 For more examples see DEPOSIT and DIFFUSE Silvaco 6 31 ETCH ATHENA User s Manual 6 19 ETCH ETCH simulates an etch process Syntax ETCH MATERIAL NAME RESIST ALL DRY THICKNESS lt n gt ANGLE lt n gt UNDERCUT lt n gt LEFT RIGHT ABOVE BELOW P1 X lt n gt P1 Y lt n gt P2 X lt n gt P2 Y lt n gt START CONTINUE DONE X lt n gt Y lt n gt INFILE lt c gt TOP LAYER NOEXPOSE MACHINE lt c gt TIME lt n gt HOURS MINUTE
116. in ek ceive ere 3 1 Parabolic Rate Constant Chlorine Dependence es eeceseeeseeeseeeeeeeeeneees 3 57 Pressure Dependent iisisti isisisi inona 3 57 Pearson Differential Equation ceceeesseessseeeeeneeeeeeneereenes 3 67 Photoresist Bake imins hi einnsean os acter ie 5 12 Physically Based Simulation ccescseeeeseeeeeeeeneeeeeeeeeeeee 1 4 Plasma Etching Model Dopant Enhanced Etching ceesseeesseeeesneeeeeneeeeeeees 4 14 PLS Diffusion MOdels c ceeecceeeteeeeeseeeesseeeeneeeneeeeas 3 23 43 PLS Diffusion Models Examples ccesceseeeeseeeeeeees 3 37 43 See also Advanced Diffusion Models Examples point defect anana en e a 3 1 Proximity Printing Lithiography eiai Hee ieee a 5 15 Simulation Method eeeceeeseeseeeeeeeeneeeeeeeeeeesseeeeneeeaees 5 17 Theory Of is leet iain aA 5 15 5 16 Index 3 Silvaco ATHENA User s Manual R Rapid Thermal Anneals RTA Epitaxy E ETT 2 40 41 Noies a a MM rake eet Ea 2 38 Oxidation ieie octets ea teks oh our cols bates art an 2 39 Recombination at the Surface GNEV E EEE Boh As es oN 3 30 31 Retlow Model 3 2 ci6c eset heheh r a aaa 4 20 reverse short channel effect PROCES sie ies P led Oh ON Ce heat Moh lence C 10 RTA Diffusion Modeling eeseeeeseeeseneeeseeeeeeneeeseeeteaes 3 18 S Second Order Fick s Equation cescessceeeseeeeeeeeeeeseeeeeeeeeees 3 3 SiGe Process Simulation DIFFUSE iii a
117. incorrect use of the METHOD statement will invalidate the rest of the following section Calibrating an ATHENA input file for a typical MOS process flow involves using the device simulator ATLAS since electrical measurements from the MOSFETs in question often represents the majority of the physical data available for calibration This can be thought as a paradox since ATLAS would also have to be correctly calibrated The reason that this doesn t present a problem is discussed below An important point to remember when using Technology Computer Aided Design TCAD is that the most critical task is to accurately model the process flow Note For accurate MOSFET simulation you should invest 90 of the time in achieving an accurate process simulation while only investing 10 of the time in fine tuning the device simulation The reason for this especially for silicon technologies is that the device physics in general is understood For silicon not only is the physics well understood it is also well characterized so most of the default values in ATLAS will be correct Therefore the calibration of an ATHENA process file does not involve the calibration of well known quantities such as diffusion coefficients Instead the calibration involves variables that are process and production line dependent For example the damage caused by an implant cannot be determined exactly since it is dose rate dependent and can be influenced by beam heating
118. linear solution for flow velocities and stresses and then uses the stresses obtained to calculate the reduction factors for oxidant diffusivity Dg oxide viscosity u and the interface reaction rate constant k as follows x P V o o Dp Dp exp j 3 147 ee ee Pare mew alae 3 148 WV sinh 2kT i _ D Ne k k exp 3 149 where i is the iteration Vg V V and V are the activation volumes in specified in the OXIDE statement tis the total shear stress r 1 ho ae eae 3 150 2 XX YY xy o is the normal component of the total stress o o ie o oe 20 nn 3 151 r xx xX yy y xy X y o is the tangential component of the total stress O 6 ie Pa 20 nn 3 152 t XX y yy x XY X y where n and n are the x and y components of the unit vector normal respectively The reduced parameters feed back to the next iteration This process continues until the accuracy criterion is met Fast convergence of this process is not guaranteed Oxidation calculations by the stress dependent model usually take much more CPU time than the Compress Model Figure 3 11 shows the resulting structure from a LOCOS oxidation step using the stress dependent Viscous Model Silvaco 3 49 ATHENA User s Manual ATHENA Stress Dependent Viscous Model Materials SiO2 Silicon Si3N4 1 0 8 0 6 04 02 Midons 02 04 0 6 Figure 3 11 Resulting Structure from a LOCOS Oxidation step using the Stress D
119. new parameter specifies maximum ration between adjacent mesh lines in x direction It is equivalent to the same parameter in the MESH statement DY RATIO new parameter specifies maximum ration between adjacent mesh lines in y direction It is equivalent to the same parameter in the MESH statement IN FILE is alias for INFILE RATIO is an alias for INTERVAL R There are several additional ways to specify initial substrate doping You can specify the impurity name by using IMPURITY lt impname gt parameter where lt impname gt could be boron phosphor arsenic and antimony You can specify the corresponding doping either by I CONC lt conc gt or I RESIST lt resistivity gt Alternatively you specify the concentration of an individual impu rity by using BORON lt conc gt PHOSPHOR lt conc gt and so on Boolean parameters RESISTIVITY and CONCENTRATION specify which method of initial doping specification to be used A one dimensional grid structure can now be specified without using LINE statements or their equivalents MESH MASK or loading of a sec file from MASKVIEWS The syntax of the INITIALIZE statement of SSUPREMSB and its derivatives can be used e DX specifies the nominal grid spacing in the initial grid e MIN DX specifies the minimal grid spacing e SPACES specifies the number of the grid spaces in the initial structure e THICKNESS specifies total thickness of the initial structure e TI
120. of a microlithographic exposure system are approximated by A RES k WA 5 1 and DOF kz A 5 5 2 NA where is the wavelength of the exposing radiation NA is the Numerical Aperture of the imaging system and k and ko are process dependent constants Typical values for k are 0 5 for a research environment and 0 8 for a production process the value usually assigned to ko is 0 5 We shall discuss the basic assumptions upon which the model rests Next we shall derive the principal equations used for calculation of the image irradiance distribution for objects illuminated by partially coherent light The treatment presented here assumes the radiation incident on the object to be quasi monochromatic which means that the spectral bandwidth is sufficiently narrow so that wavelength dependent effects in the optics or in diffraction angles are negligible The source is of a finite spatial extent so that the advantages of spatial incoherence are realized in imaging The mask is completely general in that phase and transmission are variable but it must be composed of rectangular features The calculation of the diffraction phenomena is based upon the scalar Kirchhoff diffraction theory Since the dimensions of the mask are almost the same as the illumination wavelength we can ignore any polarization taking place as the radiation propagates through the mask We assume scalar diffraction which means neglecting the vector nature o
121. of process simulation software This section highlights some of the most common barriers encountered using process simulation to model oxidation steps and describes how to overcome those barriers with the proper methods for simulating these oxidation steps One of the most common errors made in simulating oxidation steps is improperly gridding of the oxide structure Improper gridding can result in jagged oxide shapes and errors in resolving impurity distributions As the oxide layer is growing grid points are added at predefined spacings As silicon is being consumed dopants are transported across the Si SiOs interface It is important to obtain a well gridded oxide to properly account for dopant redistribution during the oxidation step Growing Thin Oxides A typical application where thin oxide growth is important is during a gate oxidation step of a MOSFET which has a highly doped polysilicon gate By default SSUPREM4 uses a grid spacing of 0 1 microns in the growing oxide layer Thus one grid layer will be added in the growing oxide every 0 1 microns or 1000 angstroms This grid spacing is appropriate for field oxidations and hence the reason it is the default grid spacing in the growing oxide layer Using the default grid spacing in the oxide for typical gate oxidations in today s MOS technology results in no grid being added in the interior of the SiO layer With no grid present in the oxide to resolve the dopant diffusion in the oxid
122. of the time required to oxidize the thickness of one grid layer which should elapse before resolving the flow field Usually REDO OXIDE is much less than OXIDE GDT which is an upper bound on how long the solution should wait It is mainly intended to exclude solving oxidation at each and every one of the first few millisecond time steps when defects are being tracked TRBDF and FORMULA specify the time integration method to be used The TRBDF parameter indicates that a combination trapezoidal rule backward difference should be used The error is estimated using Milne s device The FORMULA method allows you to specify the time step directly as a function of time t previous time step dt and grid time gdt This option is primarily for testing The TRBDF method is the default Parameters related to OXIDATION models ERFC ERFG ERF1 ERF2 COMPRESS and VISCOUS are oxidation models see Chapter 3 SSUPREM4 Models Section 3 3 Oxidation Models The ERFC parameter indicates that a simple error function approximation to a bird s beak shape should be used The ERF1 and ERF2 models are analytic approximations to the bird s beak from the literature The ERFG model chooses whichever of ERF1 or ERF2 is most appropriate All erf models are applicable only to the simplest case of oxidation to the right of the mask edge All relevant parameters in the OXIDE statement must be explicitly specified when using any of the ERF models The COMP
123. optional optional smooth step no optional optional optional optional optional optional optional optional optional To use the planetary model either the ANGLE1 or the DIST PL parameter must be specified These parameters are mutually exclusive 2 62 Silvaco Tutorial Defining ELITE Etch Machines An ATHENA ELITE etch machine can be defined using the ATHENA Rate Etch Menu Figure 2 45 To open this menu select Process gt Etch gt Rate Etch in the Commands menu The machine definition requires the specification of four general parameters and one or several model specific parameters The general parameters that must be specified are as follows e Machine name e g TEST02 This parameter uniquely identifies the machine e Material name e g silicon A user defined material can also be specified e Machine type e g Wet Etch You can select one of three models by pressing the appropriate button e Etch rate units specifier e g A min You can select one of seven unit specifiers from the menu One or several model specific parameters are attributed to each model For example only the ISOTROPIC rate parameter is required for the Wet Etch model Table 2 5 indicates which parameters are required for each of the three models The Parameters for Specific Machine Type section of the Rate Etch menu includes only those parameters which are relevant to the selected model If the ATHENA Rate Etch Menu is set as
124. orientation of the substrate There three major planes regarding ion implantation in crystalline materials mainly e the implantation plane a e the surface plane e and the simulation plane B The implantation plane is where the initial beam of incoming ions lays in It equivocally defines the direction of the incoming beam tilt and rotation If the orientation of the surface plane is 100 which is the only substrate orientation available currently in the Binary Collision approximation implantation module BCA or CRYSTAL parameters the offset of the rotation angle is the direction lt 101 gt on this plane This means that the tilt angle 9 specified by the TILT parameter in the IMPLANT statement will be the polar angle in laying this plane while the rotation angle specified by the rotation parameter will be the difference of azimuths of the line where the implantation plane a crosses the surface plane and the direction lt 101 gt See Figure 3 23 Note Presently the surface orientation the ORIENT parameter in the INITIALIZE statement does not have any affect in the crystal Monte Carlo module and the surface orientation is always 100 3 82 Silvaco SSUPREM4 Models implantation plane surface plane 100 101 major flat I simulation plane or projection i e TONYPLOT s plane Figure 3 23 Implantation geometry The simulation projection plane is where all data regarding the simu
125. poly thick 0 50 c phosphor 5 0e19 divisions 10 dy 0 02 ydy 0 2 4 min spaced 001 struct outfile manual_ 4 str tonyplot manual_ 3 str manual_ 4 str set manual_ 2 set etch poly right p1 x 0 3 co paste init pause clear J restart kill Stop No Thu May 11 17 02 41 1995 Executing an host scorpio ne a ka ATHENA gt sa ATHENA started ATHENA l Figure 2 1 Main Deckbuild Window 2 1 2 Loading And Running ATHENA Standard Examples DECKBUILD makes it possible to load and run a number of example simulation input files To access the ATHENA examples select Main Control gt Examples and the Deckbuild Examples Window will appear See Figure 2 2 Groups of DECKBUILD examples are listed in the Section menu and are grouped according to the simulator or simulation topic the example demonstrates The Sub section menu lists individual example input files To run examples select one of the sections e g ATHENA_IMPLANT in the Section menu This will open a list of input file names Short descriptions of the examples will appear in the Examples Window Select one of the input files using the Sub section menu or by double clicking on the input file name and a description of the selected input file will appear Press the Load Example button to load the selected input file into the Deckbuild Text Subwindow bottom panel of the window The input file along with other files associat
126. processes accurately e Diffusion calculation has been modified to allow the previous discretization or Rex Lowther s dis cretization method The improved Lowther discretization can be accessed using the parameter LOWTHER on the METHOD statement e The oxidation gridding algorithm has been modified to allow a thin grid at the initial oxidation and a coarser grid throughout subsequent oxidation This technique is designed to create a fine grid dur ing gate oxide and similar growth steps but coarse grid for thicker oxidations The parameter GRI DINIT OX on the METHOD statement sets the value of the initial grid thickness A similar capability for silicidation is available via the GRIDINIT SIL on the METHOD statement e Substrate orientation can now be specified on the INIT statement to set the orientation of trench sidewalls This effects oxidation and Monte Carlo implantation e The SSUPREM4 MaskViews interface has been replaced by the MaskViews cutfile capability All references to the SSUPREM4 MaskViews interface in the INITIAL statement will be ignored dur ing calculation and will produce a warning message e The regrid capability has been replaced by the functionality of DEVEDIT REGRID statements in SSUPREM4 input will be ignored during calculation and will produce a warning message e Regional attribute information can now be set in SSUPREM4 Currently the attributes that are set by SSUPREM4 are only electrode names e Poly Oxidat
127. proposed by Brandt and Kitagawa 64 Their stopping power S ml of the medium for an ion is in the first approximation proportional to dx a mean square effective ion charge They derive the effective stopping power charge of a projectile Z from a given ionization state q If a fractional effective charge of an ion with the given ionization state q is defined as Z S 1 2 Eh AE le 3 229 where Sg a7 the stopping power for bare nucleus Brandt and Kitagawa theories produces the following simple expression for the fractional effective charge of an ion bagr eu n FONAN 3 230 where e q Z N Z isthe fractional ionization 3 80 Silvaco SSUPREM4 Models e Nis the number of electrons still bond to the projectile nucleus ag and rg are Bohr s radius and velocity k F and v p are Fermi wave vector and velocity For the screening radius A Brandt and Kitagawa assume exponential electron distribution which becomes 2 3 hea 3 231 ZU N 7N The only undefined quantity C is of about 0 5 and somewhat depends on the target The degree of ionization g can be expressed as r 0 92v oes 3 232 vz q 1 ef where v p 1 r is the relative velocity between the projectile and the target electrons which are calculated as follows 3vp 2v 1 1 v 1 1 for V lt Vp 3 233 j 4 3v 15 v v E v i 25 for V vp 3 234 1 Damage Accumulation Model The present mod
128. reactions Equations 3 120 and 3 125 3 2 6 Typical Examples The following will show each part of the PLS model validated using specific experimental results The CDD model is tested using a simple predeposition step The interstitial cluster part of the model is validated using the Cowern s experiment 20 and the mixed cluster part of the model is then analyzed using the Pelaz experiment 24 To illustrate the improvements given by the PLS model we perform simulations within a very broad range of experimental conditions from a standard implantation and diffusion step to a state of the art RTA Predeposition Boron As a first indication that the PLS model is able to handle the complex couplings between boron and the free point defects we show in Figure 3 3 the result of the conventional predeposition steps simulation diffusion with a constant surface concentration 3 36 Silvaco SSUPREM4 Models TonyPlot V2 8 18 A File 7 View Plot Tools Print Properties gt Help 7 file Miew v Plot Tools v Print v Properties Help PLS Model Boron Predeposition Csurf 6 0e19 cm 3 T 850C t 4hours z f 5 z a a Click to place P changes alignment or drag to get leader SILVACO International 2004 Figure 3 3 Simulation of Boron predeposition using the CDD model at various temperature time annealing and surface concentration Crosses pluses and squares are expe
129. res i nidi earnen area oad e cece whee tree baa sa here wee ine AREIA D 2 D 2 ATHENA Version 5 14 0 R Release Notes 0 cece cece eee e eee eee eee eee nnn D 2 D2 1s SoUPREMS Features piaren sipna wa todas basaeh alae aha ke tae RR E E h D 2 D 2 2 ELITE Features i erii i E ea ais eee is Pela ee ee D 2 Di2 3 OP TORI Features tars nina ciate cea a Reed eR ene eee E ead eae ae D 2 D 3 ATHENA Version 5 10 7 R Release Notes 0 cece cece tee e eee eee ene eens D 3 D 4 ATHENA Version 5 10 0 R Release Notes cc cece cece e eee e eee ee eee n eee neeenee D 4 D 4 1 General Featur S so cnte den AA aa A AE AA AE sup Asha alp Apc anne aE ee ala D 4 D42 SOUP REA Eei a A o e a aa e ae eaa ed es Reh o ea D 4 DA Ee PRT m A N E E E E AEE E A E A N E te od D 5 D 5 ATHENA Version 5 8 0 R Release Notes uuusununnunnnnnnnrnnnnrnnnnnnrrnrnenernenennan D 5 D5 TSSUPREM aa a Soak ah cae Ate ard barca Ana ien Se aed aes and ea A tate SNe se wc ad oh tat D 5 D 5 2 ELITE Capabilities ss ir Sei pete ec tle ede tees pee few deni ener ees D 6 D 5 3 OPTOLITH Capacitles crates seer atp anon eae E beled ane y nae sa EEEN eR eRe ce waaay D 6 D 5 4 Miscellaneous Features and Bug FixeS 2 000 e cece eee eee eee eee eee ee eeeee D 6 D 6 ATHENA Version 5 6 0 R Release Notes ccc eee e eee ee eee eee ee ene n eee eeeeenee D 7 D 6 1 SSUPREM4 cues a ais hel ad eee eee pote le eer peated EE Ena D 7 D 6 2 ELITE Capabilities fai inulin
130. s law as F D VC 3 129 eff where Dy is the effective oxidant H20 or Og diffusivity in the growing SiO layer C is the oxidant concentration in the oxide 3 44 Silvaco SSUPREM4 Models The reaction at the Silicon or Polysilicon SiO interface between silicon and the oxidant is expressed as F kCn 3 130 where e kis the apparent surface reaction rate constant e C is the oxidant concentration at the Silicon or Polysilicon SiOg interface e n is a unit vector normal to the Si SiO pointing toward the silicon layer Under steady state conditions the three fluxes are equal F F F F 3 131 By dividing the flux by N4 the number of oxidant molecules incorporated in a unit volume of SiO and considering one dimensional growth the growth rate of the oxide layer is given by dxo _ F 3 132 dt N where XQ is the oxide thickness From Equation 3 127 and Equations 3 129 3 131 Equation 3 132 can be expressed as a PE E 3 133 dt A 2x9 where A 2D A 3 134 effk h C B 2D p gt 3 135 TN Equation 3 133 is modified for thin oxides less than 500 A as follows dxo__ B R 3 136 dt A 2x9 where R is calculated according to 32 R THINOX 0exp OX op 2 pers 3 137 kT THINOX where P is partial pressure THINOX 0 THINOX E THINOX L and THINOX P parameters are specified in the OXIDE statement Silvaco 3 45 ATHENA User s Manual 3 3 1 Numerical Oxidati
131. selects that all process steps will be done in a full two dimensional calculation If the parameters are unspecified or AUTO is used ATHENA will then perform 1D calculation until a two dimensional calculation is required This is typically at the first ETCH statement which doesn t remove material across the whole width of the structure X LOCAT specifies the position within the defined 2D mesh for performing 1D simulation CYLINDRICAL specifies the boundary conditions for cylindrically symmetrical structure In this case the axis of rotation is X 0 0 and no negative x coordinates are allowed Parameters Related to Initialization from a File INFILE specifies a file name for reading This file must contain a previously saved structure or intensity distribution see Section 6 63 STRUCTURE IN FILE is a synonym for this parameter STRUCTURE and INTENSITY specify which type of file is to be initialized STRUCTURE is the default Grid and Structure Related Parameters SPACE MULT specifies a global spacing multiplier to be applied to the spacings defined on the previously specified LINE statements INTERVAL R is the maximum ratio between the distances of adjoining mesh lines The default is 1 5 RATIO is a synonym for this parameter LINE DATA specifies that locations of mesh lines be printed during execution SCALE allows an incoming mesh to be scaled The default is 1 0 FLIP Y is a Boolean parameter th
132. shown in Figure 2 45 the following RATE ETCH statement will be inserted into the input file TESTO2 ETCHING MACHINE RATE ETCH MACHINE TESTO2 SILICON U M WET ETCH ISOTROPIC 0 03 Deckbuild ATHENA Rate Etch GENERAL PARAMETERS Machine name TESTO Material Silicon Machine type wet Etch Etch rates are in Mimin PARAMETERS FOR WET ETCH MACHINE TYPE Isotropic rate 0 03 ooo O Comment TESTO etching machine tw RITE Figure 2 45 ATHENA Rate Etch menu If several materials are present in the structure to be etched etch rates for each material type should be specified in separate RATE ETCH statements Silvaco 2 63 ATHENA User s Manual Table 2 5 Allowable Etch Model Parameters Parameters wet etch rie isotropic yes yes directional no yes divergence no yes chemical no yes Using A Specified Etch Machine When etch rates for a specific machine are specified using RATE ETCH statements you can simulate the effects of the operation of this machine To simulate the etch process using a specified etch machine open the ATHENA Etch Menu and check the Etching Machine box The Parameters to Run the Defined Machine section will appear in the menu See Figure 2 46 Deckbuild ATHENA Etch Etch Method Etching Machine PARAMETERS TO RUN THE DEFINED MACHINE Machine name TESTO2 Ti
133. solid solubility concentration point to those already stored for the impurity Units for SS TEMP are C Units for SS CONC are cm CTN O and CTN E specify the vacancy clustering coefficients for the impurity CTN 0 is the pre exponential coefficient and CTN E is the activation energy By default these parameters are only used for Arsenic CTP 0 and CTP E specify the vacancy clustering coefficients ACT FACTOR specifies parameter for concentration dependent solid solubility activation model The value of ACT FACTOR must be between 0 8 and 1 0 TRACT 0 TRACT E and TRACT MIN specify parameters of the transient activation model Units for TRACT 0 and TRAT MIN are seconds Units for TRACT E are eV Interface Transport Parameters MATERIAL specify MATERIAL2 for the segregation and transport parameters on the boundary between two materials see Section 6 2 9 Standard and User Defined Materials for the list of materials 6 50 Silvaco IMPURITY SEG 0 and SEG E allow the computation of the equilibrium segregation concentrations SEG 0 is the unitless pre exponential constant SEG E is the activation energy in eV TRN O and TRN E allow the specification of the transport velocity across the interface given TRN 0 is the pre exponential constant units are cm sec TRN E is the activation energy units are eV TRNDL O and TRNDL E specify parameters of the Interface Trap Mode
134. specified in the INITIALIZE statement The default is 1 5 DY SURF specifies the grid spacing in the surface region i e between y 0 and y LY SURF FAST is equivalent to the AUTO default parameter in the INITIALIZE statement Note The MESH statement with Ly SURF and other related parameters cannot be used together with LINE Y statements GRID FAC specifies a global spacing multiplier which will be applied to all spacing parameters when you generate a grid with the INITIALIZE statement This parameter is equivalent to the SPACE MULT parameter of the INITIALIZE statement LY SUREF specifies the depth of of the surface region in the default vertical grid y direction LY ACTIV specifies the bottom of the active region in the default vertical grid y direction LY BOT specifies the depth of the bottom of the structure in the default vertical grid Silvaco E 3 ATHENA User s Manual E 4 Using MASK statement with the parameter IN FILE and XLINES for Automatic grid generation in the horizontal direction This capability can be used only if ATHENA runs within DECKBUILD If DECKBUILD encounters the MASK statement with the parameters IN FILE lt maskfile t11 gt and XLINES it recognizes that the file should be in Taurus Layout Mask Data format The following information from the maskfile t11 are used to build the grid e The scale factor units per micron e The minimum and maximum coordina
135. statement contains DIR Y direction This gives 638 grid points and a different pattern of elimination see the plot in the lower right corner of Figure 2 20 DIR E before This will give the following Relax statement 20 Grids after Various Relax Operations STRUCTUR 1 00 Y MIN direction in the area below y 0 3 select the X d the Area and Location Selections as THE 20 and the elimination in X and Y Figure 2 TDIR YSF 0 00 X MAX ER HALF OF 1 00 DIR X ELAX X MIN Y MAX F B RELAX LOWER HALF ONLY IN X RELAX LOW R In this case the number of grid points is 567 The grid above y If you don t want the grid to be relaxed above y Relax box This will give the following R To increase spacing only in the X Note The only difference is that instead of DIR Y lower left corner of Figure 2 2 24 Tutorial You can also apply several consequent RELAX statements to achieve grid elimination in different areas of the structure An important thing to remember about the RELAX capability is that it allows you to avoid creating obtuse triangles and avoid relaxing directly on the material boundaries This sometimes results in no relaxation or grid relaxation in a subset of the desired area The most desirable method for complete control over gridding is by using DEVEDIT as described briefly in this chapter and in the VWF INTERACTIVE TOOLS User s Manual Vol
136. states depends on the Fermi level location 8 39 and is given by ie oer Bear Ve Ve a aie where n is the electron concentration and n is the intrinsic carrier concentration and g JBAF EBK exp BARPE 2i 2 3 161 kT git BAP EBKexp PAR EPE 3 162 kT g JBAF EBK exp BAB NE 3 163 kT g BAF EBKexp BARNNE 3 164 aT where t o 6 and are fractions of the vacancy concentration which are positively double positively negatively and double negatively charged respectively Figure 3 15 shows a plot of V VA at 950 C for common silicon dopants Notice that for n type dopants V V increases as the doping concentration increases but V V remains essentially constant for the p type dopant The increase in V VA for n type dopants increases the linear rate constant This ultimately leads to thicker oxides when oxidizing highly doped n type substrates due to a higher availability of unoccupied silicon lattice sites vacancies for oxidant molecules to be incorporated The oxide thickness trend is shown in Figure 3 16 where the SiO thickness is plotted versus doping concentration for common silicon dopants Silvaco 3 55 ATHENA User s Manual TonyPlot V2 8 18 A els File v View 7 Plot Tools Print Properties 7 Help 7 CI 163 yoo T T gt x Antimony qJ 4 4 Arsenic G E Boron q 4 Phosphorus 1e1 oping Concelftthtion rem 3
137. that only the top layer of the etched material should be etched NOEXPOSE specifies that the new surface is not exposed for subsequent oxidation or deposition after geometrical etch Use this parameter to remove a part of the structure from the bottom or side of simulation Parameters used only with physical etching in the ELITE module MACHINE specifies the name of the etch machine that is to be run TIME specifies the time the etch machine is to be run HOURS MINUTES and SECONDS specify the units of the TIME parameter Parameters used only with RIE WET ETCH and PLASMA models DT FACT is used with ELITE type etch calculations By default the movement of a string node is limited to less than or equal to one quarter of the median segment length This is a good compromise between simulation speed and the danger of loop formation The optimization factor DT FACT must not exceed 0 5 You can however decrease it if necessary for more accuracy DT MAX is used to limit timesteps size By default the upper limit for the maximum timestep is one tenth of the total etch time specified This is a good compromise between calculation accuracy and calculation time But sometimes it is useful to adapt this value to the specific simulation problem Allowing the time steps to become greater gives a higher simulation speed but the accuracy may suffer For smaller time steps the simulation speed will decrease but the accuracy may be greater DX M
138. the IMAGE statement If specified the layout loaded with sec file generated in MASKVIEwS will be shifted so its center is in the point 0 0 the origin of coordinates for computational window This parameter should be specified when sec file is generated from the GDS2 file where absolute coordinates of mask features could be arbitrary Removed obsolete parameter NA in the EXPOSE statement Nonvertical light propagation was not implemented for non planar structures Added new standard impurity HYDROGEN The only practical application available in the moment is Monte Carlo ion implantation of hydrogen If you specify METHOD PLS before the Monte Carlo IMPLANT statement the initial distribution of impurity defect pairs will now calculate the same way as for analytical implants Fixed a bug for the ETCH START CONTINUE DONE sequence when etched window width is zero while using MASKVIEWS Fixed the bug for the case of deposion of ternary materials with variable composition fraction For example deposit material InGaAs thick 0 50 div 20 c fract 0 1 f fract 0 5 Fixed a bug in analytical ETCH with the ANGLE parameter In some cases when the THICKNESS parameter exceeded the total thickness of the structure a part of the etched layer was not removed Fixed a bug for the case when using EPITAXY C INTERST lt n gt after ATHENA starts and before any DIFFUSE statement
139. the calculation of the field inside the complex structure can be divided into several subsequent calculations of the plane wave propagation through the structure The final field distribution is the sum of distributions obtained for all separate plane waves The simulation algorithm is outlined as below 1 Simulation of the incident field propagation through the target resist substrate 2 For each segment of each interface e compute the direction of the reflection e align the coordinate system and simulation domain with the direction of the reflection e obtain the field distribution over the segment e recursion of the propagation simulation procedure for the field generated over the surface segment 3 Summation of the distributions obtained with all recursive steps The reflection from each segment of interface is computed using the preliminary obtained field on the segment as the initial incident field The recursion depth can be specified as a simulation parameter The Beam Propagation Method BPM 106 is used to simulate the field propagation The BPM can be used for different types of radiation e g UV EUV X ray as well as for multi exposure processes and multilayer and non linear resists There are three reason why we choose this method The first reason is because the diffraction of the field along the propagation is automatically taken into account The second reason is because it includes a capability to simulate no
140. the distance in microns that ETCH extends under a mask when dry etch is performed the default is 0 LEFT RIGHT ABOVE and BELOW provide a quick means of etching with a trapezoidal cross section The etch region will be to the specified side left right above below of the line specified by the coordinates given in P1 X P1 Y and P2 X P2 Y P1 X P1 Y P2 X and P2 Y allow you to specify a line for left right above below etching The P1 parameters are always required if left right above below are used The P2 parameters are required when the etch angle is non vertical Units are microns START CONTINUE and DONE specify an arbitrarily complex region to be etched You can combine several lines to specify the several points that make up the region See the examples X and Y specify a point in the start continue done mode of etch region specification Units are microns 6 32 Silvaco ETCH INFILE specifies that the etch profile will be taken from the filename specified by the INFILE parameter The specified file must have the following format x Yz Xp Yo X3 Y3 Xn Yn This will etch the region enclosed by the boundary coordinates within the file You can define any number of coordinates within the file This command is often useful for inputting data from digitized experimental profiles or external programs The closing line is automatically drawn from the final coordinate point to the initial point TOP LAYER specifies
141. the exposure calculation for repetitive line width calculations 2 9 6 Defining Material Properties There are two statements in ATHENA OPTOLITH that relate to material properties OPTICAL and RATE DEVELOP The OPTICAL command sets the complex index of refraction for a single material at a given wavelength The RATE DEVELOP command sets development rate parameters for each resist defined in the resist library Default values for these material parameters are located in the athenamod file which can be viewed in DECKBUILD by selecting the Models item from the Command menu You can change any of these parameters by entering the command with the new values For example to change the index of refraction of silicon at the wavelength of 0 365 um enter OPTICAL SILICON LAMBDA 0 365 REFRAC REAL 6 522 REFRAC IMAG 2 705 To enter resist parameters at wavelength of 0 407um use the RATE DEVELOP command as follows RATE DEVELOP NAME RESIST AZ1350J LAMBDA 0 407 A DILL 0 88 B DILL 0 077 C DILL 0 018 E1 DILL 5 63 E2 DILL 7 43 E3 DILL 12 6 Photoresist parameters for development or diffusivity DIX 0 DIX E can be entered separately from exposure parameters without specifying the wavelength The photoresist name must always be specified When specifying Dill exposure parameters A B and C specify the wavelength as these parameters vary with w
142. the grain growth parameters PD GROWTH 0 and PD GROWTH E are specified in the IMPURITY statement and t is the elapsed diffusion time 3 71 The segregation boundary condition at the polysilicon silicon boundary is also modified when POLY DIFF model is used The default segregation coefficient M79 in Equation 3 12 is increased by a factor which depends on the impurity concentration in the grain boundaries M M 1 PD SEG GBSI 3 72 i b ce 8 where M 9 is modified segregation coefficient and the PD SEG GBSI parameter is specified in the IMPURITY statement 3 22 Silvaco SSUPREM4 Models 3 2 Advanced Diffusion Models The ultimate goal of TCAD simulation is to compute the electrical characteristics of a given device by using only process related data as input parameters Since the electrical characteristics of the device are heavily dependent on the distribution of the electrically active impurity dopants resulting from the entire thermal processing sequence it is important that the diffusion models used in the process simulation are as accurate as possible This is particularly important for deep sub micron processes Therefore for these emerging technologies 2D or even 3D phenomena are expected to be of growing importance whereas there is presently no accurate technique to measure multi dimensional dopant profiles Consequently the active dopant 2D distributions can only be obtained by simulation
143. the highest temperature anneals a significant percentage of damage removal occurs in a fraction of a second Almost zero damage enhanced diffusion or total diffusion occurs in this instance and the anneal time to remove the damage is very short Extrapolating between these extremes provides a qualitative explanation of what occurs for intermediate temperature anneals Two important points have now been established 1 For sound device physics reasons most RTA processes consist of high temperature short duration anneals 2 Damage enhanced diffusion will only occur for a few seconds at typical RTA temperatures For accurate simulation of RTA the second point is most important and often wrongly neglected Suppose an RTA consists of a 10 second ramp up to 1000 C followed by a 20 second anneal and a 10 second cool down From the second point it is apparent that most of the Total Dopant Diffusion would have taken place during the Ramp Up Phase of the RTA Therefore always model the temperature ramp up accurately when simulating an RTA process In most cases the ramp down can be neglected since all the diffusion has already taken place at the beginning when the silicon was still damaged 2 38 Silvaco Tutorial 2 4 7 Simulating Oxidation It has already been stated that the pull down menu for simulating oxidations is the same as that for simulating inert diffusions described in the Simulating Diffusion Section on page 2 37
144. the implanted ion concentration multiplied by DAM FACTOR PRINT MOM prints out moments for all ion material combinations used in the analytical model In the case of Monte Carlo simulation it prints out moments calculated from the coordinates of ion in the standard structure file and can extracte them by the EXTRACT function DAM MOD specifies the name of the C Interpreter file which can be used to modify defect concentration models Parameters Applicable Only for Analytical Implant Models X DISCR specifies the width of slices along the direction of the ion beam used to calculate the implanted profile The value used is scaled relative to the lateral straggling of the current implant By default a slice width of between 0 1 and 0 2 of the average lateral straggle will be used This parameter allows you to override the internal selection of discretization along the implant front If the value of X DISCR decreases simulation accuracy and simulation time will increase LAT RATIO1 specifies a factor by which all lateral standard deviations for the first Pearson distribution would be multiplied Default is 1 0 LAT RATIO2 specifies a factor by which all lateral standard deviations for the second Pearson distribution would be multiplied Default is 0 2 Note The LAT RATIO parameters provide simple scaling of the default lateral standard deviation Use the MOMENTS statement for more complete lateral standard deviation modification
145. tion through a native oxide layer 0 5 1 5nm while the parameters at 10keV correspond to implantation into a bare silicon surface i e silicon wafer subjected to an HF etch less than two hours before implantation e For implant energies below 5keV the models for boron BFy and arsenic have not been verified experimentally The simulations in this range of implant energy are performed using an interpo lation between experimentally verified Dual Pearson parameters at 5keV and parameters based on UT MARLOWE estimates at 0 5keV e The SIMS measurements upon which these profiles are based have a concentration sensitivity limit in the order of 5x 10 to 2x1 ae increasing with dose from the implant The pro files have been extended below these limits following the trends that occur within the sensitivity limits of the SIMS e The screen oxide thickness range has been verified from 1 5 to 40nm only for boron and 15 80keV energy range But the oxide range has been extended to 50nm Silvaco ATHENA User s Manual Screen Oxide Thickness Parameter S OXIDE To specify screen oxide use the S OXIDE um parameterin the IMPLANT statement This thickness is specified independently of any actual surface oxide in the structure It is however possible to automate the extraction of the surface oxide thickness for use with the IMPLANT statement An example is supplied demonstrating this S OXIDE is another parameter for d
146. to the standard structure file in Version 3 of ATHENA the structure files created by Version 3 are not compatible with previous versions of ATHENA Structure files cre ated by old versions of ATHENA can be read by Version 3 of ATHENA Adaptive Meshing Capabilities A 2 D mesh adapting module has been incorporated into ATHENA The module is invoked by specifying boolean flag ADAPT on the METHOD statement preceding IMPLANT DIFFUSE or EPITAXY statements or by specifying boolean flag ADAPT on the ADAPT MESH statement to do stand alone mesh refinements A mesh smoothing algorithm has also been integrated into the module to improve the mesh quality after mesh adapting or after normal deposit etch oxidation silicidation diffusion process steps A set of parameters can be specified on ADAPT PAR statement to adjust the mesh adapting process The parameters available on the METHOD statement are as the following e Boolean ADAPT specify that the adaptive meshing should be performed on the following IMPLANT DIFFUSE or EPITAXY statements default false e Boolean DEPO SMOOTH specify that the mesh smoothing should be performed after each DEPOSIT statement e Boolean ETCH SMOOTH specify that the mesh smoothing should be performed after each ETCH statement Boolean DIFF SMOOTH specify that the mesh smoothing should be performed after each DIFFUSE statement e Boolean STEP SMOOTH specify that the mesh smoothing should be performed after each time s
147. total diffusion time Simulations show that a depth of 20 to 50 microns is required in most cases This restriction on the minimum structure depth poses a threat to computational efficiency whenever diffusion models that include point defects are employed But since the fine structure of the defect profiles near the bottom of the structure is not a feature of particular interest for processing purposes you can reduce the computational cost by making the grid very coarse in this region Time Step Control When using diffusion models that include the explicit representation and evolution of point defects be aware of time stepping issues Although step size control between iterations is fully automated you can still specify the size of the initial time step Different initial time step sizes can be specified for dopants and point defects respectively by using the parameters INIT TIME and PDINIT TIME For example the command METHOD INIT TIME 0 001 PDINIT TIME 0 001 would set the initial time step to 1 millisecond for both dopants and point defects Default values are INIT TIME 0 1 seconds and PDINIT TIME 1 0E 5 seconds Note There is no guarantee that the program will actually use these values For this initial time step the only purpose of these parameters is to make it feasible for you to give the program a hint about an appropriate initial time step size 3 1 5 The Fully Coupled Model The Fully Coupled Di
148. undoped silicon equal to 5 0 1022 cm For layers with graded germanium content use an additional parameter F GERMANIUM in the DEPOSIT statement The following statement DEPOSIT SILICON THICK 0 1 DIN 10 C GERMANIUM 1e20 F GERMANIUM 5e21 C CARBON 1e19 will create the 0 1pm layer of Si _ Ge C with constant carbon concentration of 1 0 1019 cm and with germanium concentration varying from 1 0 102 cm at the bottom of the layer to 5 0 102 cm at the top of the layer This corresponds to Ge content x varied from 0 2 to 10 3 9 2 Boron Diffusion in SiGe SiGeC The special model takes empirically into account experimental facts that boron diffusivity apparently decreases with germanium and carbon content The total boron diffusivity see Equation 3 10 decreases exponentially with the Ge content x and carbon content y 7 BP EAFACT SiGe y EAFACT SIGE DplSiz_y_yGe C DplSilexp 2P EAFACT SiGe ty BARACT SIGE 3 241 Ng kT Another effect taken into account in this model is variation of intrinsic carrier concentration n with x and y It is presumed that n increased with x and decreased with y n Si Ge C n Si qX NIFACT SIGE 7 X NIFACT SIC 3 242 1 x y x y 1 Noj Noi The user defined calibration parameters EAFACT and NIFACT for the above equations are specified in the MATERIAL statement for silicon As an alternative to Equations 3 241 and 3 242 you can use different dependencies for
149. value of A at a new z position A k ky z Az A k ky z exp ik Az 5 56 In this equation the z component of the wave vector is derived by k Sek 5 57 Now the inverse Fourier transform will produce the distribution of the field amplitude over x y plane at z Az i e at the gap distance from the mask Finally the required intensity distribution is the square of the module of the field amplitude A Silvaco 5 17 ATHENA User s Manual This page is intentionally left blank 5 18 Silvaco Chapter 6 Statements 6 1 Overview ATHENA executes a file that describes the process meshing and models to be used in a simulation The contents of the file are statements each of which prompts an action or sets a characteristic of the simulation This chapter is a reference to the command language that can be used to control ATHENA Throughout this manual we will refer to commands statements and parameters A line in an input file is referred to as a statement or statement line An ATHENA statement is specified in the general format lt COMMAND gt lt PARAMETERS gt lt VALUE gt where lt COMMAND gt is the command name lt PARAMETER gt is the parameter name and lt VALUE gt is the parameter value Four types of parameters are used in ATHENA Real Integer Logical and Character The space character is used to separate parameters from a command or from other parameters
150. vol 63 no 1 p 116 1988 M Yoshida E Arai H Nakamura and Y Terunuma Excess vacancy generation mechanism at phosphorus diusion into silicon J Appl Phys vol 45 no 4 p 1498 1974 S Solmi F Barualdi and R Canteri Diffusion of boron in silicon during postimplantation annealing J Appl Phys vol 69 no 4 p 2135 1991 B Colombeau Interactions entre les d fauts tendus et anomalies de diusion des dopants dans le silicium mod le physique et simulation pr dictives Th se de doctorat Universite Paul Sabatier Toulouse Septembre 2001 P Fastenko Modeling and simulation of arsenic activation and diusion in silicon Ph d thesis University of Washington 2002 B E Deal and A S Grove General Relationship for the Thermal Oxidation of Silicon J Appl Phys v 36 p 3770 1965 D Chin Two Dimensional Oxidation Modeling and Applications Ph D Thesis Department of Electrical Engineering Stanford University 1983 H Z Massoud Thermal Oxidation of Silicon in Dry Oxygen Growth Kinetics and Charge Characterization in the Thin Regime Technical Report Stanford Electronic Laboratories Stanford University 1983 H Eyring Viscosity Plasticity and Diffusion as Examples of Absolute Reaction Rate J Chem Phys v 4 p 283 1936 B E Deal Thermal Oxidation Kinetics of Silicon in Pyrogenic H20 and 5 HCI H20 Mixtures J Electrochem Soc v 125 p 576 1978 D W Hes
151. want to save the structure information generated after key process steps e g final structure To save or load a structure use the ATHENA File I O Menu See Figure 2 25 by selecting Commands gt File I O Specify a file name the file extension str is recommended for all ATHENA structure files and press the Save button The following line will appear in the input file STRUCT OUTFILE TUTOR STR You can reload this file tutor str back into ATHENA at any time during the current DECKBUILD session or during any subsequent session To reload the structure file press the Load button on the ATHENA File I O menu The following INIT statement will appear INIT INFILE TUTOR STR 2 28 Silvaco Tutorial Deckbuild ATHENA File I O flip y Intensity Format File name tuter str Figure 2 25 ATHENA File I O Menu Note Only the structure will be reloaded if ATHENA is restarted before this INIT statement Any parameters or coefficients that were set during previous simulations must be reset if they are needed This structure file can also be used by any device simulator or DEVEDIT Silvaco 2 29 ATHENA User s Manual 2 4 Choosing Models In SSUPREM4 2 4 1 Implantation Oxidation RTA Diffusion and Epitaxy This section describes how to simulate process steps e g implantation diffusion oxidation epitaxy and silicidation specific to the SSUPREM4 module of ATHENA Also discussed are
152. where silicon oxide and nitride coincide together at a point The default for the split angle is 22 5 The SPLIT ANGLE parameter for triple point oxidation is material dependent Specify the oxidizing MATERIAL1 without a and MATERIAL2 with a using the following format OXIDE SPLIT ANGLE 35 SILICON NITRIDE There are only three possible combinations and they are SILICON NITRIDE SILICON POLYSILICO and POLYSILION NITRIDE You can use this to control lateral encroachment during oxidation Parameters of the Analytical Oxidation Models ERF SPREAD and MASK EDGE are used only in the error function approximation to a bird s beak shape SPREAD is the relative lateral to vertical extension which defaults to 1 The fitting parameter makes the erfc bird s beak look realistic MASK EDGE is the position of the mask edge in microns and defaults to negative infinity Oxide grows to the right of the mask edge ERF Q and ERF DELTA are the DELTA and Q parameters for the erfg model Normally you don t need to change them but they are available if necessary ERF LBB is the length of the bird s beak and applies to the erfg model only It can be specified as an expression in Fox the field oxide thickness um eox the pad oxide thickness um Tox the oxidation temperature Kelvin and en the nitride thickness um The published expression can be found in the models file Specifying ERF
153. will stop units cm e Float EDGE MAX specifies the maximum edge length below which deleting points will stop units cm e Integer MAX POINT specifies the maximum number of points above which adapting will stop e Integer MAX LOOP specifies the maximum loop count above which adapting will stop effective D 16 Silvaco ATHENA Version History only with implant The parameters available on the ADAPT MESH statement are as follows e Boolean ADAPT specify that a stand alone adaptive meshing step should be performed to refine or relax the current mesh based on the material impurity specification given on ADAPT PAR state ment default false e Integer ADAPT COUNT specifies the number of adapting loops during the stand alone adaptive meshing operation default 1 e Boolean SMOOTH specifies to do stand alone annealing default false e Integer SMTH COUNT specifies the number of smooth loops during the smooth operation default 1 e Float ADD I LINE specifies that a mesh line is to be added at the interface between two materials as defined by the booleans MATERIAL1 and MATERIAL2 The line is added in MATERIAL 1 a dis tance ADD I LINE from MATERIAL2 Boolean SILICON OXIDE Specify materiall for ADD I LINE e Boolean SILICON OXIDE specify material2 for ADD I LINE SSUPREM4 Capabilities e Oxidation enabled for polysilicon diffusion model e Vacancy and interstitial diffusion in polysilicon have been deco
154. you can substitute this with conformal oxide deposition Select Oxide from the Material menu and set its thickness to 0 02 um It is always useful to set several grid layers in a deposited layer In this case at least two grid layers are needed to simulate impurity transport through the oxide layer In some other cases e g photoresist deposition over a non planar structure a sufficiently fine grid is needed to accurately simulate processes within the deposited layer There are also situations e g spacer formation when several grid layers in a deposited material region are needed to properly represent the geometrical shape of the region The grid in the deposited layer is controlled by Grid Specification parameters in the ATHENA Deposit Menu Set the Total number of grid layers to 2 add a Comment and click on the Write button The following lines will then appear in the Deckbuild Text Subwindow GATE OXIDE DEPOSITION DEPOSIT OXIDE THICK 0 02 DIVISIONS 2 The next step will be to deposit a phosphorus doped polysilicon layer of 0 5um thickness Select the material Polysilicon and set the thickness to 0 5 To add doping select the Impurities box The Impurity Concentration section will be immediately added to the ATHENA Deposit Menu See Figure 2 13 Silvaco 2 17 ATHENA User s Manual E Deckbuild ATHENA Deposit Material Palysilicon User defini Thickness um 0 50 0 00 mmm 100 Grid spec
155. 10 Resulting Structure from a LOCOS Oxidation step using the Compress Model 3 3 3 Viscous Model The Viscous Model solves the same flow equations as described in the previous section This model is activated by specifying the VISCOUS parameter in the METHOD statement prior to the DIFFUSE statement The Viscous Model calculates stresses in the growing oxide and creates almost the same shape for the silicon oxide interface as does the Compress model The stresses in the oxide are calculated as follows VISC 2 VISC 0 gt exp eE Ov Ov C56 eee 3 144 XX YY 1 2 POISS R x Ov a o o 2 VISCO ap ECE gt tme 3 145 xx yy kT amp ov oO E VISC x i o VISC 0 exp SCE 23 3 146 a 3 48 Silvaco SSUPREM4 Models where v and v are the x and y components of flow velocity v respectively VISC 0 and VISC E are the pre exponential and activation energy respectively for viscosity are specified on the MATERIAL statement The stress calculated by the Viscous oxidation model replace stress that may have been previously generated by the STRESS HIST parameter in the STRESS statement The stress dependent nonlinear model based on Eyring s work 33 allows a description of the real shape of LOCOS profiles with kinks on the interface The model is turned on by specifying the STRESS DEP parameter the OXIDE statement Using Equation 3 140 and Equations 3 143 3 146 the non linear solver first finds a
156. 1000 F H20 5 3 F HCL 0 06 F 02 8 0 PRESS 1 00 One or several impurities can be present in the ambient To set ambient in the Impurity Concentration section of the ATHENA Diffuse Menu See Figure 2 30 check the corresponding checkboxes and set the values using sliders and the Exp menus For example by selecting the appropriate boxes and values the following DIFFUSE statement will be inserted into the input file FIELD OXIDE DIFFUSE TIME 100 TEMP 850 T FINAL 1060 WETO2 PRESS 1 00 HCL PC 0 C ARSENIC 9 0E19 C PHOSPHOR 4 0E20 Several other parameters not included on the menu are available in the DIFFUSE statement Chapter 6 Statements Section 6 15 DIFFUSE The DUMP DUMP PREFIX and NO DIFF parameters can be useful DUMP and DUMP PREFIX can be used to make a movie using TONYPLOT The NO DIFF parameter specifies that impurity redistribution will be neglected This provides a good approximation for low temperature processes such as silicidation Silvaco 2 39 ATHENA User s Manual Several other model specification statements are important for diffusion processes These are as follows e IMPURITY INTERSTITIAL and other impurity and point defect statements which specify model parameters e g diffusivity or segregation of these species e The OXIDE statement which specifies parameters for different oxidation
157. 20 f 02 1 f N2 10 fHCL 5 Modify process description to produce desired characteristics at key locations asi a T pe K a oS Simulate the N Nt pt pt complete 2D P N process only once pt Nt P Substrate Figure 2 42 Use of One Dimensional Mode 2 8 2 Deposition and Wet Dry Etching using the Physical Models in ATHENA ELITE This section describes the deposition and etch capabilities of the ELITE module of ATHENA using ATHENA ELITE default machines To use ATHENA ELITE s physically based deposition and etch models use at least one of the following steps 1 Use one of the predefined machines described in Chapter 4 ELITE Models Section 4 4 Etch Models 2 Within your input file modify the specification of one of these predefined machines to fit your process 3 Define a custom machine 4 Invoke a prepared file that defines machines of interest For example one of the predefined machines is named PE4450 This machine deposits aluminum at a rate of 1 micron minute from a hemispheric source To simulate the effects of two minutes of operation of this machine open the ATHENA Deposit Menu and select the Machine checkbox The section PARAMETERS TO RUN THE DEFINED MACHINE will appear in the menu See Figure 2 43 In this section specify PE4450 as the Machine Name the Time of run 2 0 and the time units menu box beside the Time of run field as minutes We recommend that you spec
158. 201613x102 1150 204041x102 1175 2 6643x102 1200 2 9423x102 Silvaco B 7 ATHENA User s Manual Table B 14 Solid Solubility in Silicon 124 125 Temperature Boron Phosphorus Antimony C cm cm cm 1225 3 2387x10 1250 3 5536x102 6 6200x1019 1275 3 8876x107 B 6 Point Defect Parameters These parameters are for silicon and polysilicon only Table B 15 Point Defect Parameters 126 Bulk Parameters Interstitial Vacancy D O 600 0 0 1 D E eV 2 44 2 0 CSTAR 0 cm gt 5 0x1072 2 0x1023 CSTAR E eV 2 36 2 0 KR 0 3 16x10 3 16x10 KR E eV 2 44 2 44 Table B 16 Point Defect Parameters Charge State Information Interstitial Vacancy NEU 0 1 0 1 0 NEU eV 0 0 0 0 NEG 0O 5 68 5 68 NEG E eV 0 50 0 145 DNEG 0 0 0 32 47 DNEG E eV 0 0 0 62 POS 0 5 68 5 68 POS E eV 0 26 0 45 B 7 Defect Interface Recombination Parameters Table B 17 Defect Interface Recombination Parameters Silicon oxide Interstitial Vacancy KSURF 0 1 76x10 4 7 0x108 KSURF E eV 0 06 4 08 KRAT O0 1000 0 0 0 KRAT E eV 0 0 0 0 KPOW 0 0 5 1 0 KPOW E eV 0 0 0 0 B 8 Silvaco Default Coefficients Table B 17 Defect Interface Recombination Parameters
159. 25028 2 759 2102 B 1 6 Doping Dependence Of Oxidation Rate Table B 6 Doping Dependence of Oxidation Rate Parameter Value BAF EBK 241 6 BAF PE 0 46 BAF PPE 1 0 BAF NE 0 145 BAF NNE 0 62 BAF KO 2 6e3 BAF KE Tt See the bibliography reference 37 for more details B 1 7 Coefficients for the Analytical Guillemot Model l Table B 7 Coefficents for the Analytical Guillemot Model Parameter Value Spread 1 0 INITIAL 0 002 ASK EDGE 200 ERF Q 0 05 Silvaco B 3 ATHENA User s Manual Table B 7 Coefficents for the Analytical Guillemot Model Parameter Value BRE DELTA 0 04 ERF LBB 8 25e 3 1580 3 Tox Eox 0 67 eox 0 3 exp en 0 08 2 0 06 ERF H 402 0 445 1 75 en exp Tox 200 See the bibliography reference 40 for more details B 1 8 Numerical Oxidation Coefficients For Dry Oxidation HENRY COEFF 5E16 THETA 2 2E22 TRN 0 1E 3 For Wet Oxidation HENRY COEFF 3e19 THETA 2 2e22 TRN 0 let6 B 1 9 Stress dependent Growth Model Coefficients Ve 78 Vr 9 7 Vd 1 9 Vt 0 0 Dlim 1 0 See the bibliography reference 120 for more details B 1 10 Mechanical Parameters For Stress Calculations Table B 8 Parameters t
160. 3 1 In Table 3 1 the x designates the neutral charge state is a single negatively charged state is a double negatively charged state Table 3 1 Notational standards in diffusion literature Physical Entity Generic Symbol Replacement Values Dopant A B P As Sb Point Defect X IL V Charge State c X 3 2 Silvaco SSUPREM4 Models Many physical entities or parameters are temperature dependent In ATHENA this dependence upon temperature is modelled by the Arrhenius expression unless otherwise indicated OCT 0 0exp 25 3 1 where e Q 0 is the pre exponential factor e Q E is the activation energy e kis the Boltzmann constant e T is the absolute temperature Generic Diffusion Equation All diffusion models whether they are the Fermi the two dimensional or the fully coupled model follow the same generic mathematical form of a continuity equation A continuity equation merely expresses particle conservation that is the rate of change with time of the number of particles in a unit volume must equal the number of particles that leave that volume through diffusion plus the number of particles that are either created or annihilated in the volume due to various source and sink terms This basic continuity equation for the diffusion of some particle species C in a piece of semiconductor material is a simple Second Order Fick s Equation 9 oC seb VJ S 26
161. 3 86 Silvaco SSUPREM4 Models In ATHENA trajectory splitting is turned on with the SAMPLING command in the IMPLANT statement For more information this statement see Chapter 6 Statements Section 6 28 IMPLANT Theoretically sampling estimators are unbiased and consistent In practice however the estimate is obtained as the average of finite number of samples as opposed to the theoretical expectation which is obtained from the whole ensemble of sample paths including even the very unlikely ones Overbiasing can occur if the only goal in mind is to increase the probability of the event that needs to be analyzed further as is the case of trajectory splitting ion implantation simulations The result of overbiasing is usually the underestimation of the probability to be evaluated dopant concentration in case of ion implantation In fact it has been reported in 76 that when the splitting parameters are not consistent with the system s large deviations behavior the probability in question may be underestimated by several orders of magnitude This situation is almost present in ion implant simulators when treating multi layered targets and two dimensional layouts Therefore use splitting with caution In conclusion you should bare in mind that e increasing the probability of occurrence of the event to be analyzed is not always enough to guarantee variance reduction e trajectory splitting should be used carefully e co
162. 30min Simulation As implanted with SVDP Simulation with full PLS model Total Boron E Active Boron Simulation with classical IC model Simulation with classical model Concentration cm 0 0 1 03 04 02 Depth ym Click to placa P chargas alignment or drag to get bader SILVACO International 2003 Figure 3 7 Simulation of 35 minutes boron diffusion at 800 C after an implantation at 20 keV with a dose of 5 1014 cm Experimental data are extracted from 27 As shown in Figure 3 7 PLS model results in an excellent fit with the experimental annealed curves The comparison of chemical and electrically active boron concentration shows that even below the solid solubility most of the dopant stay inactive due to the formation of mixed dopant defect clusters By comparing results obtained with only CDD or IC model with the full PLS IC DDC it can be concluded that this type of clusters also consume free interstitials which leads to reduction of the boron diffusion This fact justifies one more time that all three parts of the model CDD interstitial clusters and mixed dopant defect clusters are needed in simulation of post implant diffusion Silvaco 3 41 ATHENA User s Manual Experiment with boron implanted at 2 keV A most aggressive technology simulation has been carried out with a 2 keV boron implant with a dose of 1014 cm followed by a RTA at 1000 C during 10 seconds This type o
163. 31 C Sacer Clustering DDO oa Changing the Method Statement During the Process Flow j E A Switching Guidelines cccsesesetseeeseteteteeseeeeeees 2 31 32 Solid Solubility saceczeess2ecvaseesd Seesezesstseiheeeh eases anak 3 31 Vacancy Cluster VC eesceesceseseseeeeeeeeeeneeeeeeteaeeeeeeeenees 3 33 Charge States See also PLS Diffusion Models CNET Rind Se aa Cet eae iian 3 24 25 Advanced Features Chemical and Active Concentration Values ecceeceeeeeeees 3 1 Deposition and Wet Dry Etching scsccscsseseeseseeseeee 2 59 64 Chemical Mechnical Polish CMP MESRV IB WS sineira etniaren teianei eaa tonade 2 65 73 Hard Polish Model sssssssssssssrrrrrreessssrrrrreeesssenrrreres 4 21 22 Structure Manipulation POO E NEEE TAAA D E 2 56 59 Soft Polish Model ni ern e a a a ia 4 23 24 Analytic Implant Models C Interpreter EEE OE A EE E E E E 3 89 4 1 9 A 1 2 Dual Pearson s siete wisi Me 3 68 CNET Charge States Gaussians sete erste het atid ated nce tie tesa 3 66 Dopant Defect pairs ccsecceceseeeeeeceneeeereeeners 3 25 26 P arSON en en ee Le 3 66 68 Point Defects idin 2a veil newindian 3 24 25 Screen Oxide Thickness Parameter S OXIDE 000 3 70 CNET Flux Equations SIMS Verified Dual Pearson SVDP Model seseee 3 69 Dopant Defects Pairs ssssssscsscsessessesssssssseeesees 3 27 28 Arrhenius XpreSsion ccesscceceeseceseeeeeseeeeeseneeseseeeeeseeers 3 3
164. 311 E must be specified as negative since the time constant decreases with rising temperature Silvaco 3 11 ATHENA User s Manual Interstitial Generation and Recombination at Interfaces Interfaces present a moving boundary problem during a thermal oxidation In this instance there will be a recombination rate at the interface which will vary as a function of the interface velocity Also as a consequence of the silicon being consumed there is a significant injection of interstitials into the substrate Within ATHENA this is modelled by an interstitial flux boundary condition as described by Hu 13 O Cr KIG GF G 3 35 where e Cris the projection of the interstitial flux vector on an inward pointing unit vector normal to the boundary e Kg is the effective surface recombination rate for interstitials e Gy is the generation rate at the interface of interstitials during annealing in an oxidizing ambient In other words Equation 3 35 shows that the number of interstitials generated on the surface minus the number of interstitials that recombine there must equal the number of interstitials that diffuse from the surface interface into the substrate The effective surface recombination rate Kg depends on the motion of the interface during oxidation according to vi Kpow Ks Kovar Kaar n i 3 36 Yi max where v is the velocity of the interface Vimax is the maximum interface velocity and the parameter
165. 4 14 4 57 Reflow Mod l rinira a a a E Wand Sn Wass E ale Wad a a are 4 20 4 6 Chemical Mechanical Polish CMP 0 ccceeee eee eee eee eee eee eee eens 4 21 4 6 1 Hard Polish Model 4 nye sire ute Fel BAS thos tet ae Sita eee eis Le Be he ah ease 4 21 4 0 2 SOM OSHIMOUEL 42g 28 ace ot OU ee Pee OG elites Mee ln Mee Laat OPS tet Lal ae So 4 23 Chapter 5 OPTOLITE MOdel v is cace as gendone cy eta cise oie ee ewe wee oa eek ve 5 1 BIPOVEIVIOWS natu aed choi ote Ae aide MU elt A tate eek Le cbt lane Make ie ciety 5 1 5 2 The Ima ing Module inuri rirerire etenin wer Uae wa tan od CRRA eae T A 5 2 5 3 Optical System ssis arena E a a Ea a EDA AEE ATE EA a AE aca E 5 7 5 3 T Dis retization EMG sri a aa a an a r A e EE Otic 5 7 We Me Sree reae cas tae See A ana a a Aa ek ot p O Re cone es cle EE AS E 5 7 5 33 Computation TMe aee 8 28 8 a Ve Race ace Boat cate ir e ek cae ee R 5 8 5 4 The EX DOS UG WOM Gece peste ari ao wh dchondae evn Shute a we Arteaga shag da era ane Gere age ad ate eae 5 9 5 5 Photoresist Bake MOGUIG icici s cccse snisce a Sasahe data Ge cea be date E a DER EA e Did 5 12 5 6 The Development Module aie wit waren Se ae ann a ae ee Sa ee Wwe 5 13 5 6 1 Dil s Development Model 020072 y etn ey eos ew daue deena eerie gis e eras yen yg bans 5 13 Silvaco vii ATHENA User s Manual 5 6 2 Kim Ss Development Model pict uk Sl ol Ue aot aN Oh eM Te TE iat Sa Ai 2 5 13 5 6 3 Mack s Development M
166. 45 Depth um Click te place P changes alignment or drag to get leader B SILVACO International 2003 Figure 3 6 Simulation of the Pelaz experiment using various parts of the PLS model Experimental Data are extracted from 24 The results obtained with the full PLS model correspond to the experimental data As expected the boron at the highest concentration stay immobile and are still inactive due to the formation of mixed BICs Moreover it is clear that only the full model can explain this behavior Therefore CDD and IC models cannot simulate this particular phenomenon without DDC model involved 3 40 Silvaco SSUPREM4 Models Implantation Diffusion Experiment The analysis above proves validity of each part and the entire PLS diffusion model In addition we present several simulations relevant to specific processes important in the modern VLSI technologies Experiment with boron implanted at 20 keV Boron implantation at 20 keV is widely used in silicon technologies Since the experiment setup involves low temperature such as 800 C the effects of boron immobilization and disactivation at high concentration take place Therefore you must use three parts of the PLS model K TonyPtot 28 14 lt 2 ray File view v Plot 7 _Teols 7 Print Properties Help 4 Boron Diffusion after implantation Medium energy implant furnace annealing 20 Ac implanted B 20keV SE14em Annealed at 800 C
167. 6 65 TONYPLOT a a Ss eRe hte te ie tok ee eR eet ee a EA 6 110 6 66 TRAP ascii eh chit a iets peated ik errs ay EE Maso veered goes 6 111 6 67 UNSEIMODE vus ineei a act e see ay eee cian Sate A die inte Mateo Desiree swe Sa aE 6 112 Appendix A G INIEI DICER seine rre was emanane cut ee eet heels Devore tess A 1 A 1 C Interpreter Overview 2 6 0 oe ec fon cee ben Vee cee ed cent ke dewe ed eine vs enwv nee ive der deus A 1 A2 EXAM Plen ronne ine e EE E A O OY A 1 Silvaco ix ATHENA User s Manual Appendix B Default Coefficients rinena tenset say san sie am oe tee OA aa eet B 1 B 1 Oxidation Rate Coetticient 235 20 sccsa es castes eee lee ees ide ete ese eee iad ee tae as B 1 B 1 1 Dry Ambient For lt 111 gt Orientation 5 ates oF ieee eth Raat nae Picard ot merece oes a l8 B 1 B 1 2 Wet Ambient for lt 111 gt Onientatonics 08 ccc divudcesesies tee balks iat Det dea se ieee des B 1 B 1 3 Orientation Factors For Linear Coefficients both Ambients 000 c cece eee eee eee B 2 B 1 4 Pressure Dependence naii io 8 5 2 Patt Gated Shi oe eee eee oe nae E EE es B 2 B 1 5 Chlorine Dependence ccccase ac kn cx neers ensiareet haha RAS San yard ea Aa ay A eevee eve Janke ts B 3 B 1 6 Doping Dependence Of Oxidation Rates fe cot dards tetaas arden pes eee B 3 B 1 7 Coefficients for the Analytical Guillemot Model 0 ccc e eee teeta nee B 3 B 1 8 Numerical Oxidation Coefficients 2 0 2 0 ccc
168. 66 DIP O em2 s 0 0 0 0 0 72 0 0 DIP E eV 0 0 0 0 3 46 0 0 DIM 0 cm s 15 0 12 8 0 0 4 44 DIM E eV 4 08 4 05 0 0 4 00 DIMM 0 cm2 s 9 0 0 0 0 0 44 2 DIMM E eV 0 0 0 0 0 0 4 37 CTN 0 cm s 5 19x10724 CIN E eV 0 60 FI unitless 0 05 0 20 0 94 Polysilicon DIX 0 em2 s 21 4 6 6 3 66 385 0 DIX E eV 3 65 3 44 3 46 Bi 6 6 HIP 0 cm s 0 0 0 0 72 0 0 0 Silvaco B 5 ATHENA User s Manual Table B 11 Impurity Diffusion Coefficients Parameter Antimony Arsenic Boron Phosphorus DIP E eV 0 0 0 0 3 46 0 0 DIM 0 cm s 1500 0 1200 0 0 0 443 9 DIM E eV 4 08 4 05 0 0 4 05 DIMM 0 cm2 s 0 0 0 0 0 0 4420 0 DIMM E eV 0 0 0 0 0 0 4 37 CTN 0 cm s 5 19 x 107 CTN 3 eV 0 60 Oxide DIX 0 cm s 1 31x101 1 75 3 16x10 4 7 6x1073 DIX E eV 875 4 89 35 53 3 2 Tungsten Silicide 122 DIX 0 cm s 2 6 2 6 OTO 4 2 DIX E eV 2 11 B14 1 17 2 14 Titanium Silicide 122 DIX 0 cm s 4 8 4 8 1 5x1077 392 0 DIX 0 eV 2 183 2163 20 2 64 Platinum Silicide 122 DIX 0 cm s 2 6 2 6 1 0x1073 4 2 DIX 0 eV 2 11 2 11 1 17 2 14 All other coefficients for refractory metals and their silicides are set to 0 0 B 3 Impurity Segregation Coefficients Table B 12 Impurity Segregation Coefficients Parameter Antimony Arsenic Boron P
169. ACANCY statement syntax is done regardless of physical meaning For example you can define vacancy injection during oxidation although default parameters are zero The models used here are involved in ongoing research Many of the parameters have unknown dependencies such as stress temperature starting silicon material and stacking fault density For more examples see DIFFUSE and TRAP 6 56 Silvaco LAYOUT 6 32 LAYOUT LAYOUT describes manual input of mask features for OPTOLITH Syntax LAY CLEAR lt n gt X LOW lt n gt Z LOW lt n gt X HIGH lt n gt Z HIGH lt n gt X TRI lt n gt Z TRI lt n gt HEIGHT lt n gt WIDTH lt n gt ROT ANGLE lt n gt X CIRCLE lt n gt Z CIRCLE lt n gt RADIUS lt n gt RINGWIDTH lt n gt MULTIRING PHASE lt n gt TRANSMIT lt n gt INF ILE lt c gt MASK lt c gt SHIFT MASK lt c gt Description This command is used to enter mask information for OPTOLITH Several LAYOUT statements can be used in sequence to define complete mask patterns All coordinate and size parameters are in microns LAY CLEAR specifies that the currently defined layout should be deleted prior to the execution of the new layout definition X LOW specifies the minimum x coordinate of the rectangular feature Z LOW specifies the minimum z coordinate of the rectangular feature X HIGH specifies the maximum x coordinate of the rectang
170. ACKSIDE XLO lt c gt YLO lt c gt XHI lt c gt YHI lt c gt Description EXPOSED surfaces correspond to the top of the wafer Only exposed surface have deposition or oxidation on top of them A surface created by etching will also be exposed unless the ETCH NO EXPOSE syntax is used REFLECTING surfaces correspond to the sides of the device and are also applicable to the backside as long as defects are not being simulated All surfaces default to REFLECTING BACKSIDE surfaces are physically identical to the reflecting surface with special meaning only when backside electrode is specified in the ELECTRODE statement XLO YLO XHI and YHI set the left right top and bottom bounds of the rectangle being specified The value string should be one of the tags specified in one of preceding line statements Examples The following lines define the top of the mesh to be an exposed surface and the bottom to be the backside BOUNDARY EXPOSED XLO LEFT XHI RIGHT YLO SURF YHI SURF BOUNDARY BACKSIDE XLO LEFT XHI RIGHT YLO BACK YHI BACK For more examples see REGION and INITIALIZE Silvaco 6 17 CLUSTER ATHENA User s Manual 6 10 CLUSTER CLUSTER specifies parameters of 311 cluster model Syntax CLUSTER I IMPURITY MATERIAL CLUST FACT lt n gt MIN CLUST lt n gt MAX CLUST lt n gt TAU 311 0 lt n gt
171. ATHENA the simulated Monte Carlo distributions are used to calculate an ion flux incident on the substrate surface This flux is then used to calculate an etch rate by integrating this flux over the window of visibility at each point on the surface The window of visibility is for point on a flat surface simply from 0 to 2 But for more complicated structures e g trenches points on the surface are shadowed and the window of visibility is reduced Currently only a simple linear surface kinetic model for etching is supported See Chapter 7 SSUPREM4 Models for a description of the RATE ETCH parameters required for plasma etch simulation 4 4 5 Monte Carlo Etching Model The shrinking critical dimensions of modern technology place a heavy requirement on optimizing the etching of narrow mask opening In addition the aspect ratio of etches has been increased requiring deeper etches along with the small ke s The simulation of these process requires more advanced techniques than the analytical rate based etching models described above A more complete treatment involving calculation of the plasma distribution and direct interaction of plasma particles with substrate materials is required 4 14 Silvaco ELITE Models The Monte Carlo etch module is implemented into ATHENA ELITE The main application of the module is simulation of plasma or ion assisted etching The module can take into account the redeposition of t
172. AYER1 DIV lt n gt LAYER2 DIV lt n gt LAYER20 DIV lt n gt Description This statement can be used to load a 1D stream of doping data into an ATHENA structure The data might come from a Secondary Ion Mass Spectroscopy SIMS profile or from a 1D simulation in SSUPREMSB Data is applied in 1D across the width of the mesh for subsequent 2D simulation INFILE specifies the name of the profile data file or Standard Structure File to be loaded MASTER or SSF indicates that the file to be loaded is an Silvaco Standard Format file Files generated by SSUPREM3 are in this format IMPURITY specifies the impurity type for profile data file Corresponding active impurity will be also added See Section 6 2 10 Standard Impurities for the list of impurities INTERST VACANCY CLUSTER DAM and DIS LOOP specify that profile data file includes a profile of interstitials vacancies 311 clusters or dislocation loops correspondingly LAYER1 DIV LAYER2 DIV LAYER20 DIV specifies the number of subdivisions for each layer when loading SSUPREM3 Structure files Examples An example of a PROFILE statement is given below PROFILE INF BORON SIMS BORON In this case the PROFILE statement specifies that only boron information will be added to the current working silicon structure The data file BORON SIMS should be in the following format THIS IS SIMS DATA 0 01 1815 0 02 1 1E15 0
173. As or user defined ternary materials with the following standard names AlInAs InGaP GaSbP GaSbAs InAlAs InAsP GaAsP HgCdTe InGaN and AlGaN GR SIZE specifies grain size in deposited polysilicon layer This parameter is recognized only when POLY DIFF model is specified in the METHOD statement Units are microns F GR SIZE can only be specified together with GR SIZE This parameter deposits polysilicon layer with grains linearly graded with their sizes where GR SIZE specifies grain size at the bottom of the layer and F GRAIN SIZE specifies grain size at the top of the layer Units are microns Parameters Specific to ELITE Depositions MACHINE specifies the name of the machine to be run for ELITE deposits The machine name must be specified in a previous RATE ETCH statement TIME sets the time in specified units the etch machine will be running HOURS MINUTES and SECONDS specifies the units of the TIME parameter Default is MINUTES N PARTICLE specifies the number of particle trajectories to calculate for the Monte Carlo deposit model OUTFILE specifies the name of the file to be written with Monte Carlo particle positions 6 22 Silvaco DEPOSIT SUBSTEPS specifies the number of timesteps made for each division of the deposit in the ELITE module VOID specifies that the voids formed during deposition are to remain unfilled with deposit material
174. Before discussing the simulation of physical processing using SSUPREM4 ELITE or OPTOLITH modules it s important to discuss structure manipulation statements that can precede or alternate with physical process steps Simple Film Depositions Conformal deposition can be used to generate multi layered structures Conformal deposition is the simplest deposit model and can be used in all cases when the exact shape of the deposited layer is not critical Conformal deposition can also be used in place of oxidation of planar or quasi planar semiconductor regions when doping redistribution during the oxidation process is negligible To set the conformal deposition step select the menu items Process Deposit Deposit from the Commands menu in DECKBUILD and the ATHENA Deposit Menu Figure 2 12 will appear 2 16 Silvaco Tutorial vy Deckbuild ATHENA Deposit Material Oxide say defined Thickness um 0 02 0 00 d H 1 00 Grid specification Wi Total number of grid layers 2 1 a 1 20 O Aipmnal grid sparing imi 0 16 O Grid spacing lication fumi O Minimum grid sparing tum O Stiniusei edge sporing ipm Composition fractions iniiai capevipa Pilea fractipi Pinal campas iipon fraction Comment Gate oxide deposition WRITE Figure 2 12 ATHENA Deposit Menu As shown Conformal Deposition is the default If it is known that the oxide layer thickness grown in a process is 200 Angstroms
175. Birigh shadow point izeft iright is a weighing factor based on the amount of shadowing at point i due to Silvaco 4 23 ATHENA User s Manual The kinetic factor is based on the following equation K 1 KINETIC FAC tana 4 39 This shows the effect of the parameter KINETIC FAC on the polishing rate The angle a is the local angle that is tangent to the polished surface The maximum allowable angle a is 89 9544 1 57 radians will avoid calculation errors Figure 4 14 demonstrates three regions where each of the components of the polishing rate would be large A_I Large K_I Large t _I Large Figure 4 14 Soft Polishing Model Areas where different components dominate 4 24 Silvaco Chapter 5 OPTOLITH Models 5 1 Overview The OPTOLITH module of ATHENA allows the use of sophisticated models for imaging photoresist exposure photoresist bake and photoresist development OPTOLITH includes a library of photoresists with default characterizations for development and optical properties These default characterizations can easily be tuned to adjust for variations that very typically occur from one facility to another This chapter describes the models and capabilities of OPTOLITH Silvaco 5 1 ATHENA User s Manual 5 2 The Imaging Module OPTOLITH includes an imaging module that utilizes the Fourier series approach The theoretical resolution RES and Depth Of Focus DOF
176. CALE gives the spread of the enhancement over solution values In other words how quickly does the enhancement or retardation factor reach its maximum S is the dopant value The positive value of ENH Max corresponds to enhancement while negative value corresponds to retardation For exponentially varying solutions e g oxidation stress and dopant concentrations both S and ENH MINC are taken to be log base 10 of their respective value Parameters of the model are specified in the RATE DOPE statement 4 4 4 Plasma Etch Model The plasma etch model in ATHENA is based on a Monte Carlo simulation of the ion transport from the neutral plasma or bulk denoted by its glow through the dark sheath surrounding the electrodes and walls and isolating the plasma Ions enter the sheath from the plasma and are then accelerated through the sheath due the electrical potential drop between the plasma and the electrodes The Monte Carlo simulation follows a large number ions in their transport through the sheath including collisions with other gaseous species present in the etch chamber The number of collisions encountered by a particular ion depends on both the ion mean free path a calculated quantity and the sheath thickness an user specified quantity To reduce the computation time ion trajectories are calculated independently and inter ion interactions are not considered in this version of the code In the current version of
177. CK DY Figure 2 69 Initial 1D Structure Relationship The BASE MESH parameters should be considered alongside the BASE PAR parameters When forming a base mesh there are three objectives to remember regarding the quality of mesh These objectives are as follows e 1D dopant information is neither lost in the 2D transition nor overly refined upon resulting in overly dense BASE M ESH See Figures 2 70 and 2 71 e Little or few flat triangles exist in regions and materials of importance See Figure 2 74 e The adjacent triangle ratio in both X and Y directions is not abrupt in spacial regions of importance to the device See Figure 2 75 Controlling the quality of the BASE MESH formed at the 1D 2D transition is achieved with the BASE PAR command parameter Specific materials can be assigned different parameters The GRAD SPACE parameter controls the Vertical Adjacent Triangle Ratio Quality while the RATIO BOXx parameter controls the lateral Adjacent Triangle Ratio These two statements can be thought of as operating upon the 1D and 2D simulation segments respectively during 1D simulation Only the adjacent spacing ratio can be controlled in the vertical profile with the GRAD SPACE parameter Silvaco ATHENA User s Manual Subsequently at the point of 2D transition quality for mesh density The INIT command includes the parameters WIDTH STR and D th
178. CTRONS electron concentration INTERSTITIAL interstitial concentration NI intrinsic electron concentration OXYGEN oxygen concentration PHOSPHORUS phosphorus concentration Sxx Sxy Syy components of stress in rectangular coordinates TRAP unfilled interstitial trap concentration VACANCY vacancy concentration x x coordinates y y coordinates X V x velocity Vivi y velocity 6 98 Silvaco SELECT Table 6 5 Select Functions Function Description abs absolute value active active portion of the specified dopant erf error function erfc complimentary error function exp exponential gradx numerically differentiates the argument with respect to x location grady numerically differentiates the argument with respect to y location log logarithm log10 logarithm base 10 lt mat1 gt lt mat2 gt returns the y value of the interface between lt mat1 gt and lt mat2 gt along a vertical slice at the given location scale scales the value given by the maximum value sqrt square root TEMPERATURE specifies the temperature at which expressions are evaluated It defaults to the last diffusion temperature This parameter has to be specified by default or explicitly when printing a net active concentration or preparing a ATLAS structure file Examples The following example will choose the base 10 logarithm of the arsenic concentration as the PRINT 1D variable SELECT Z
179. Capabilities e The default value for nitride viscosity has been changed from VISC 0 5e12 to VISC 0 1 8e15 This value is changed in the athenamod file using the following MATERIAL statement MATERIAL NITRIDE VISC 0 1 8E15 VISC E 0 VISC X 0 499 e The parameters WET and DRY were changed to WETO2 and DRYO2 on the INTERSTITIAL OXIDE and MATERIAL statements e The MOMENTS statement has been added to ATHENA to facilitate the entering of user defined moments for analytic implant The MOMENTS statement includes the following parameters mate rial SILICON impurity ARSENIC DOSE incident ion flux em ENERGY incident ion energy KeV RANGE projected range microns STD DEV standard Deviation microns GAMMA third moment KURTOSIS fourth moment SRANGE projected range for second Pearson microns SSTD DEV standard Deviation for second Pearson microns SGAMMA third moment for second Pearson SKURTOSIS fourth moment for second Pearson DRATIO dose ratio in the double Pearson formula D 14 Silvaco ATHENA Version History e The parameters WETO2 and DRYO2 were added to the INTERSTITIAL statement for THETA 0 and THETA E e A parameter FLIP FACTOR has been added to the METHOD statement to let the user change crite ria for controlling triangle flipping FLIP FACTOR is a measure of the obtuseness of the angles of the opposite nodes of a pair of triangles The default is 1e 6 It is unitless Four new materials h
180. Cc is the chemical concentration of the dopant and C4 is the equilibrium active concentration calculated either from solid solubility or clustering model for arsenic as defined in the previous section The parameter t is the time constant for activation which is a function of temperature and is calculated using the following Arrhenius expression t4 TRACT 0 exp FRACLE 3 64 kT The initial condition at time t 0 for Equation 3 63 is specified by C ets min C4 TRACT MIN n 3 65 t where n is the intrinsic carrier concentration Therefore implantation activation will occur immediately up to a level of TRACT MIN n after the active concentration is calculated according to Equations 3 59 and 3 62 To activate the transient activation model set the CLUST TRANS parameter in the METHOD statement and specify the TRACT 0 TRACT E and TRACT MIN parameters in the IMPURITY statement The defaults for B P As and Sb are TRACT 0 8e 16sec TRACT E 4 2 and TRACT MIN 1 0 TRACT MIN for Phosphorus is 2 0 3 20 Silvaco SSUPREM4 Models 3 1 7 Grain based Polysilicon Diffusion Model The mechanism for impurity diffusion in polysilicon is different than that of crystalline silicon Polysilicon has a micro structure of small compared to the interesting device regions crystalline regions called grains These grains are separated by grain boundaries which occupy a certain spatial volume and are connected to for
181. Changes in the ETCH statement cc cece eee e cece e eee e eee eee ene enna E 8 E 11 Changes in the STRUCUTURE SAVEFILE statement 0 ccc cece eee e eee eee eee eee E 8 Silvaco xi ATHENA User s Manual E 12 Changes in the IMPLANT statement 000 c cece eect eee eee eee ene neeee E 8 E 13 Changes in the ELECTRODE statement 000 ece eee e eee e eee een ee eeeee E 8 E 14 Changes in the METHOD statement 000 c cece eee ete eee eee eee eeee E 8 E 15 Changes in the MATERIAL statement 00 cece cece eee eee eee een een eneeeee E 8 xii Silvaco Chapter 1 Introduction 1 1 Athena Overview ATHENA is a simulator that provides general capabilities for numerical physically based two dimensional simulation of semiconductor processing ATHENA has a modular architecture that the following licensable tools and extensions e ATHENA This tool performs structure initialization and manipulation and provides basic depo sition and etch facilities e SSUPREM This tool is used in the design analysis and optimization of silicon semiconductor structures It simulates silicon processing steps such as ion implantation diffusion and oxidation e ELITE This tool is a general purpose 2D topography simulator that accurately describes a wide range of deposition etch and reflow processes used in modern IC technologies e OPTOLITH This tool performs general optical lithography sim
182. Conformal Deposition Example The following statement deposits a conformal layer of silicon dioxide 1000 Angstroms thick on the surface of the simulation structure It will contain 4 vertical grid points DEPOSIT OXIDE THICK 0 1 DIVISIONS 4 Example Depositing Doped User defined Material The following deposits a layer of a user defined material BPSG doped with boron and phosphorus DEPOSIT MATERIAL BPSG THICKNESS 0 1 DIV 6 C BORON 1e20 C PHOS 1e20 Grid Control Example The following statement deposits a conformal layer of silicon nitride with a thickness of 0 3um The grid spacing at the bottom of the layer is 0 01um and the layer will include 10 vertical sublayers DEPOSIT NITRIDE THICK 0 3 DY 0 1 YDY 0 3 DIVISIONS 10 ELITE Machine Deposition Example The following statements define a machine named MOCVD and use it to deposit tungsten with a thickness of 0 1m on planar areas and step coverage of 0 75 RATE DEPO MACHINE MOCVD DEP RATE 1 u m STEP COV 75 TUNGSTEN DEPOSIT MACHINE MOCVD TIME 1 MINUTE For more examples see RATE DEPO Silvaco 6 23 DEVELOP ATHENA User s Manual 6 14 DEVELOP DEVELOP runs the development module in OPTOLITH Syntax DEVELOP MACK DILL TREFONAS HIRAI KIM EIB TIME lt n gt STEPS lt n gt SUBSTEPS lt n gt DUMP lt n gt DUMP PREFIX lt c gt
183. Current layer Label Name Field Mis alignments Xx 0 0 y Delta CD 0 00 add 7 Delete Figure 2 60 Layers Popup If you select Dark the field background will be dark and the features will have the intensity transmittance T where T is user defined If you select Clear the intensity transmittance automatically becomes 1 T Only rectangular features are used in the imaging module MASKVIEWS automatically converts triangles or polygons to a set of parallel rectangles Finer resolution on these rectangles can be obtained by changing the resolution on the Screen popup under the Define menu 2 76 Silvaco Tutorial Mask Layout In the LAYOUT command each mask feature is defined with one command line For example LAYOUT X LO 0 5 Z LO 5 0 X HI 0 5 Z HI 5 0 TRANS 1 PHASE 0 defines a 1u wide line that is 10m long The mask has an intensity transmittance of one and a phase of 0 The LAYOUT command can be repeated as often as desired The number of mask features is limited only to the amount of memory available The LAYOUT LAY CLEAR command will remove all previous mask features from memory Overlapping mask features will cause an error The OPAQUE and CLEAR parameters can be specified in the IMAGE command This will not reverse polarity as it does in MASKVIEWS 2 9 3 Illumination System The Illumination System is defined using two statements ILLUMINATION and ILLUM FILTER ILLU
184. D E INITIAL 0 001 The following defines that stress dependent oxidation rates will be used with the viscous oxidation model METHOD VISCOUS OXIDE STR ESS D EP t For more examples see DIFFUS E and M ETHOD Silvaco POLISH ATHENA User s Manual 6 41 POLISH POLISH runs the chemical mechanical polishing CMP module Syntax POLISH MACHINE lt c gt TIME lt n gt HOURS MINUTES SECONDS DX MULT lt n gt DT FACT lt n gt DT MAX lt n gt Description This statement executes the chemical mechanical polishing module of ELITE The POLISH statement must be preceded by a RATE POLISH statement to define the polishing machine MACHINE specifies the name of the polish machine TIME specifies the time the machine is to be run HOURS MINUTES and SECONDS specifies the units of the TIME parameter DX MULT is the accuracy multiplier for ELITE polishes The discretization size used for the polish calculation will be multiplied by DX MULT For improved accuracy decrease the value of DX MULT For improved speed increase the value of DX MULT DT FACT controls the timestep size By default the movement of a string node is limited to less than or equal to one quarter of the median segment length This is a good compromise between simulation speed and the danger of loop formation The optimization factor DT FACT must not exceed 0 5 but ca
185. DAMAGE flag on the IMPLANT statement The model calculates the trajectory of secondary ions generated by the collision between the pri mary ion and crystal lattice atom REC FRAC controls the fraction of the secondary ions generated by primary ions to be simulated e Work in MC Implant has changed the results so that the peaks for crystalline and amorphous implants are now at the same position e Substrate rotation is now taken into account for Monte Carlo implants This parameter is set on the INITIALIZE statement and is called ROT SUB The default for ROT SUB is 45 degrees Silvaco D 17 ATHENA User s Manual e Access to implant parameters for electronic stopping have been added to the IMPLANT statement These parameters affect the electronic stopping model and the angle for the Monte Carlo implant First the BEAMWIDTH parameter has been added This parameter allows specification of the implant beamwidth in degrees When the BEAMWIDTH angle is specified the TILT angle is varied between TILT BEAMWIDTH 2 0 Each ion will have an angle somewhere in this range decided by a random number generator There are two electronic stopping models The first default model is a simple model that uses the atomic mass of the ion and the current ion energy after each colli sion to calculate the electronic stopping A parameter called PRE FACTOR has been added as a multiplier to the atomic mass factor The default value of PRE FACTOR 1 A paramete
186. Description This command runs the development module and enables the use of the option to select a development model MACK DILL TREFONAS HIRAI KIM and EIB specify the development model to be used TIME STEPS and SUBSTEPS are related parameters that control the string algorithm in development TIME is the total development time in seconds STEPS gives the number of times ETCH is to be performed SUBSTEPS controls string movement Each substep or string movement has a time duration of TIME STEP SUBSTEPS DUMP determines whether a structure is saved after each step of the development is completed DUMP PREFIX specifies the prefix name for the structure file to be saved The number of steps will be equal to the number of output files The files are readable with the STRUCTURE statement or can be displayed using TONYPLOT The names of the files will be of the form DUMP PREFIX sty where is the current development time Examples The following example dumps out five structure files to show the evolution of development using the KIM development model DEVELOP KIM DUMP 1 TIME 60 STEPS 5 For more examples see RATE DEVELOP 6 24 Silvaco DIFFUSE 6 15 DIFFUSE DIFFUSE runs a time temperature step on the wafer and calculates oxidation silicidation and diffusion of impurities DIFFUSION is a synonym for this statement Syntax DIFF
187. E T ae Rete e AA D 11 DATS BUTE merio r berth caste eh ion at cobain AERE OE Rh ne ahaa eed ee D 14 D2 OR TOUGH 2 cess ted a E stat tte tin aA anal AAA Sia aint aa ciat raae e ES D 14 D 13 ATHENA Version 3 0 1 R Release NoteS 0 cc cece cece eee e eee eee nent e eee ene D 14 Dig ASA PENA Capabilities tr isna tha Saree acer a rte N Sarat E tid Dace en turns Ak D 14 D 13 2 ELITE Capabilities 2g a eens aia nann a Bene age here EIEE eE EE E eae ap ens aie D 18 D 13 3 FLASH Capabilities 2 22 a eeadsy A addi EE Pedi the EET D 20 D3 4 OPTOLITH Capabilities os otieudscesdieute dete ae teases seen E E E D 20 D 13 5 Known Bugsa meyn eet lye weet elt they OENE EEA sleet eet EA EA D 20 D 14 ATHENA Version 2 0 savantxetavaeey sel a eaa be E NE EA eae D 20 D 14 1 ATHENA Capabilities n on D 20 D 14 2 SSUPREM4 Capabilities etc nnne D 21 D 14 3 ELITE Capabilities nunnan D 22 D 14 4 OPTOLITH Capabilities sy 9 oc tin i eS sa hh a Oecd eae Rais ka Races D 22 Data 5 REASH eO e ache Scie ok ok A E E cine pel E a he ot D 23 DAS ATHENA V rsion IO aea a teiaa a cnc dina a aaa Tas shan aia Sy atlas aunt a asa nada ean lew D 23 B16 SSUPREMA Version 0 0 iren a Pian a eee ee See be ye ea bbs Se D 24 D 17 SSUPREM4 Version 5 14 miesien piei a hice ee oo eve ad eee e fete eee een es D 25 D 18 SSUPREM4 Version SiT riisiin dads twa hedesiws ti isexs iad dubut veentaee es D 25 D 19 SSUPREM4 Version 5 0 cer 2c ocec casas ccedieeeee
188. E command or select TM by entering the PARALLEL parameter TE is the default The exposure dose is also defined in the EXPOSURE command in units of mJ cm using the DOSE parameter Exposures can be made with either coherent or incoherent sources Coherent sources are described by SIGMA 0 01 in the IMAGE command This defines a small enough source so that only one discretization point is included If a large SIGMA is defined and discretization of the source allows at least three source points in the x or z direction then three points from the source will be used in the bulk image calculation with equal weight given to each point The points chosen will be the central point and the outermost points or the dimension of the chosen cross section x or z If multiple sources are defined using the ILLUM FILTER command then the central point of each SOURCE defined is used for calculating the bulk image in the exposure The latter allows an arbitrary amount of source points to be simulated for the bulk image calculation This is done by specifying many small adjacent sources and one point will be taken from each source You can add bulk image exposures together by specifying MULT EXPOSE on repetitions of the EXPOSE command Any number of exposures can be added together The first EXPOSE statement should not contain the boolean parameter MULT EXPOSE because the preceding expo
189. EDGE MAX specifies the maximum edge length below which deleting points will stop Units are cm Default is 1 0 10 MIN ADD percent criteria to turn off implant adapt loop MIN ADD stops point addition in IMPLANT when the number of points added in the current loop is less than MIN ADD total number of points The default value for MIN ADD 0 05 MAX POINT specifies the maximum number of points above which adapting will stop Default is 20000 6 12 Silvaco ADAPT PAR MAX LOOP specifies the maximum loop count above which adapting will stop This is only effective with implant Default is 10 IMPL SMOOTH specifies which annealing algorithm to use after each adaption step Currently IMPL SMOOTH 0 corresponds to no annealing during IMPLANT IMPL SMOOTH 1 corresponds to Laplacian smoothing and dose conservation interpolation algorithm The default is IMPL SMOOTH 1 DIFF SMOOTH specifies which annealing algorithm to use after each adaption step Currently DIFF SMOOTH 0 corresponds to no annealing during DIFFUSE DIFF SMOOTH 1 corresponds to Laplacian smoothing and dose conservation interpolation algorithm The default is DIFF SMOOTH 0 IMPL SUB flag to do grid subtracting in implant adapt IMPL SUB is a boolean flag that stops point removal during IMPLANT adaptive meshing The default value for IMPL SUB false signifies that points are not being removed DOSE ERR specifies dose error for the refinement unrefinement DOSE M
190. ENSITY lt n gt Description ER DUV LIN E ARF LAS ER F2 LAS ER LAMBDA lt n gt I LINE G LINE H LINE KRF LASER alias DUV LINE ARF LASER and F2 LASER specify that the standard wavelengths of the illumination to be used The corresponding wavelengths are 0 365 0 436 0 407 0 268 0 193 and 0 157 microns LAMBDA defines or changes the source wavelength Only monochromatic sources are assumed for simulation that is only one wavelength can be specified The units are microns X TILT and Z TILT specify the tilt of the illumination system with respect to the optical axis of the projection system All values are to be entered in degrees INTENSITY defines or changes the absolute value usually set to one of the complex amplitude that is the intensity in the mask or reticle plane Examples The following statement defines i line illumination with X and Z tilt of 0 1 and an intensity of 1 ILLUMINATION I LIN For more examples see IMAGI E PROJI LAYOUT ECTI E X TILT 0 1 Z TIL1 ON ILLUM FILT 0 1 INTENSITY 1 ER PUPI L SEE LT ER Al ERRATION Silvaco IMAGE 6 27 IMAGE IMAGE calculates a one or two dimensional aerial image Syntax IMAGE INFILE lt c gt DEMAG lt n gt GAP lt n gt OPAQUE CLEAR DEFOCUS lt n gt CENTER WIN X LOW lt n gt WIN X HI
191. ERMI model When the interstitial concentration near the surface during a very long anneal has been reduced to only marginally above the background level at the anneal temperature concerned the method statement can be switched to METHOD FERMI to greatly reduce the simulation time The interstitial background level will be the level deep in the substrate where little damage has occurred 2 4 5 Modelling the Correct Substrate Depth An important and often overlooked aspect of the correct modeling of dopant diffusion is the choice of substrate depth It has been mentioned previously that the rate of dopant diffusion is highly dependent on the level of damage in the substrate Therefore the accurate modeling of dopant diffusion requires the accurate modeling of substrate damage particularly the movement of interstitials In general the interstitials created directly or indirectly by implantation and oxidation tend to diffuse much greater distances than the dopant The substrate depth chosen for modeling purposes must therefore be deep enough to allow the interstitial concentrations to return to background levels at the bottom of the simulated substrate even if no dopant diffusion occurs at this depth TonyPlot V2 6 6 File View Ploty Tools Print v Properties Help INTERSTITIAL DIFFUSION 20keV Boron 1e15 cm2 Xx Interstitial Clusters cm3 Interstitials cm3 as implant str
192. EXPITAXY statement IMPLANT MES specifies which adapting algorithm to use on IMPLANT statements Currently IMPLANT MES 0 corresponds to the University of Florida s algorithm This is the default Also currently this is the only recommended algorithm There are four other parameters on the METHOD statement that specify mesh smoothing They are as follows e ETCH SMOOTH specifies that mesh smooth operation will be performed after etch a e DEPO SMOOTH specifies that mesh smooth operation will be performed after deposit As e DIFF OOTH specifies that mesh smooth operation will be performed after diffusion e STEP SMOOTH specifies that mesh smooth operation will be performed after each diffusion time step These four parameters are currently set as default The ADAPT PAR statement is used to set parameters to adjust the mesh adaptation process The parameters available on the ADAPT PAR statement are the following Specify material regions to be adapted such as SILICON OXDIDE and POLYSILICON This may be one or several materials at a time The default impurities include such as I BORON or I ARSENIC Specify impurities to be adapted on This may be one or several impurities at a time The DISABLE parameter specifies materials impurities given disabled to be effective on mesh adapting or smoothing The MAX ERR parameter specifies the maximum errors allowed before adding points to the mesh unitles
193. F is ignored ATHENA uses SILVACO Structure File SSF format e XDX specifies the distance from the top of the initial structure at which nominal grid spacing is placed E 8 Changes in the DEPOSIT statement There are several additional ways to specify doping in the deposited layer You can specify the impurity name by the IMPURITY lt impname gt parameter where lt impname gt could be boron phos phor arsenic and antimony You can specify the corresponding doping either by I CONC lt conc gt or I RESIST lt resistivity gt Alternatively you specify the concentration of an individual impurity by using BORON lt conc gt PHOSPHOR lt conc gt and so on Boolean parameters RESISTIVITY and CONCENTRATION specify which method of the doping specification to be used ARC SPACKEH is an alias for MIN SPACE To provide compatibility with SSUPREM3 the following aliases have been introduced e DX for DY e XDX for YDY e MIN DX for MIN DY Silvaco E 7 ATHENA User s Manual E 9 Changes in the DIFFUSE statement There are several additional ways to specify impurity concentration in the ambient gas You can specify the impurity name using the IMPURITY lt impname gt parameter where lt impname gt could be boron phosphor arsenic and antimony You can specify the corresponding concentration by I CONC lt conc gt Alternatively you can specify concentration of an individual impurity by using BORON lt conc gt PHOS
194. GH lt n gt WIN 2Z LOW lt n gt WIN 2Z HIGH lt n gt DX lt n gt DZ lt n gt X POINTS lt n gt Z POINTS lt n gt N PUPIL lt n gt MULT IMAGE X CROSS Z CROSS ONE DIM Description This statement calculates a 2D aerial image and sets parameters that control the accuracy input and output of the imaging module The IMAGE statement accepts layout information created by MASKVIEWS INFILE is the name of the mask data file from MaskVieEws It contains coordinates of rectangular mask features as well as the transmittance and phase of each feature This file name usually ends with the extension sec Note For more information on the alternative method of loading MASKVIEWS layout information for image calculations see the LAYOUT statement DEMAG specifies demagnification factor If specified all elements of layout as well as all parameter of image window and grid will decrease GAP specifies the mask to wafer gap for the case of contact printing The units are microns OPAQUE and CLEAR specify the type of mask to be used The background will be opaque if you select OPAQUE while the mask features will be clear The background will be clear if you select CLEAR and the mask features will be opaque DEFOCUS is a user specified defocus parameter If lt 0 above the resist If gt 0 below the resist surface CENTER specfies that layout loaded using the INFILE para
195. H is created by pressing the Write File button in the MASKVIEWS window The Optolith Simulation Popup will appear Figure 2 59 Silvaco 2 75 ATHENA User s Manual ne OPTOLITH simulation control File name defaultseg Figure 2 59 OPTOLITH Simulation Control Popup Enter the desired file name which should end with a sec extension and proceed to the next step Note that at the bottom of the MAskVIEws window the message Select first corner of OPTOLITH simulation area will appear MASKVIEWS is now prepared for the selection of the image window The image window describes the area where intensity will be calculated Click on the desired area for intensity calculation to create the first corner of the OPTOLITH simulation area The message Select the other corner of OPTOLITH simulation area will then appear at the bottom of the MASKVIEWS window Click on the desired second corner Once this second point is selected the coordinates of the image window s lower left and upper right corners will be displayed in the OPTOLITH Simulation Control Popup Press the Write button to save the OPTOLITH mask file The input file created by MASKVIEWS is loaded into OPTOLITH by the IMAGE command which is described in Section 2 9 5 Imaging Control To modify the layers open the Define menu in the MaskViEws Window See Figure 2 48 and select Layers menu item The Layers Popup Figure 2 60 will then appear vy Maskviews Layers
196. IAL statements Table 3 3 shows the complete set of corresponding parameters For dopants the boundary and interface conditions are identical to the ones stated in the Fermi Model 3 8 Silvaco SSUPREM4 Models Table 3 3 Parameters for charge statistics and intrinsic point defect concentrations Entity ere Activation Energy neu NEU O NEU E neg NEG O NEG E dneg DNEG O DNEG E pos POS 0 POS E dpos DPOS 0 DPOS E Cy i CSTAR O CSTAR E Interstitials The interstitial profile evolves with the following continuity equation OC a VeJ Rg RR iin 3 24 where J is the flux of interstitials Rpg is the bulk recombination rate of interstitials Ry accounts for the capture or emission of interstitials by traps and R377 is the recombination rate of 311 clusters Each of these terms are described below The interstitial flux J is calculated according to 5 with C J D C Y 3 25 C which correctly accounts for the effect of an electric field on the charged portion of the interstitials C taking the gradient of the normalized interstitial concentration ri Dz is the diffusivity of free I interstitials Don t confuse it with the pair diffusivity D4 which was mentioned in Section 3 1 2 The Fermi Model Dy is calculated once again with the following Arrhenius expression D D dexp 2 3 26 where the pre exponential factor and
197. ID OXIDE to 0 01 This allows a better simulation of impurity segre gation and a more accurate prediction of the important surface doping concentration parameter under the gate These parameters should be chosen extremely carefully If you set a small value of GRID OXIDE for thick oxide it will result in a considerable slowing down because as this parameter is decreased time steps are shortened and more grid points are generated Question In some cases oxidation of a complex structure fails right in the very first time step How can this situa tion be fixed Answer ATHENA uses a special algorithm for depositing a native oxide layer on the oxidizing surface This algorithm sometimes fails when using highly nonplanar surfaces This can be fixed by the selection of a thinner native oxide using the INITIAL parameter in the OXIDE statement Default is 0 002 microns Decreasing this value down to 0 001 microns or even less may help overcome the problem Direct deposit of native oxide could also be used Question The relative oxidation rate of polysilicon compared to silicon varies depending on the properties of the polysilicon and the oxidizing ambient How is this modeled in ATHENA SSUPREM4 Answer The oxidation rate coefficients in ATHENA SSUPREM4 are specified separately for bulk silicon and polysilicon This allows you to tune the growth rates on the two materials independently For example to change the hi
198. IMPACT POINT lt n gt SMOOTH lt n gt SAMPLING DAMAGE TRAJ FILE lt n gt N TRAJ lt n gt Zl lt n gt M1 lt n gt wu MISCUT TH MISCUT PH Description This statement simulates ion implantation using different analytical and Monte Carlo models Model Selection Parameters GAUSS PEARSON FULL LAT MONTECARLO and BCA specify the implant model that is being used GAUSS selects a Gaussian distribution PEARSON selects the Pearson IV distribution or where available dual Pearson IV distributions FULL LAT is the same as PEARSON with lateral component of the 2D distribution calculated using all available moments instead of just a lateral standard deviation MONTECARLO synonym is BCA activates the Monte Carlo Implant Module which based on the Binary Collision Approximation CRYSTAL and AMORPHOUS specify whether or not the silicon lattice structure is to be taken into account during implant steps The statements are mutually exclusive and CRYSTAL is true by default For implants through thick screen materials you often need to specify AMORPHOUS to avoid incorrect channeling profiles e For analytical implant models these parameters select which set of tables are used for silicon implant ranges The CRYSTAL model uses the SVDP tables where available and is the default e For MONTECARLO or BCA mod
199. IN specifies minimum of dose level for grid refinement during adaptation DIFF LENGTH used to limit the activity of adaptation of grid during the simulation of dopant diffusion This parameter will allow the mesh to adapt only after a given diffusion length for a given dopant and will override any other adaptation triggers based upon gradient error estimates This is a useful control to limit the number of time steps Units are microns ANISOTROPIC is the flag used to maintain the mesh to be anisotropic The flag is material dependent Examples The following is an example of setting the adaptive meshing parameters during diffusion for Boron IMPLANT BORON DOSE 1E15 ENERGY 60 ADAPT PAR DIFF LEN 0 1 SILICON I BORON DIFFUSE TEMP 1000 TIME 100 NITROGEN For more examples see ADAPT MESH Silvaco 6 13 BAKE ATHENA User s Manual 6 6 BAKE BAKE performs post exposure or post development photoresist bake Syntax BAKE DIFF LENGTH lt n gt TEMERATURE lt n gt REFLOW TIME SECONDS MINUTES HOURS DUMP lt n gt DUMP PREFIX lt c gt Description This command runs a bake process using the diffusion length as the parameter that incorporates the bake temperature and bake time DIFF LENGTH specifies the diffusion length for the post exposure bake Default is 0 05 micrometers TEMPERATURE specifies the temperature of the bake pro
200. ING parameter in the IMPLANT statement Silicide Simulation Features 1 7 Silicide models have been revised Silicide growth rates are now based on experimental data for TiSix 41 42 New data for diffusivities and transport coefficients for B As Sb and P inTiSi2 129 130 New data for As 131 and improved implementation of segregation model at TiSi2 Si interface New ALPHA parameter has been added in the SILICIDE statement It is similar to the volume expansion parameter also called ALPHA in the OXIDATION statement It specifies the ratio between consumed silicon volume and volume of grown silicide The obsolete parameters DSV 0 DSV E NSILICON and NMETAL are removed More realistic silicide shapes near spacer corners are obtained This has been achieved by suppressing lateral silicide encroachment and empirical decrease of silicide growth rate near the spacer corners Silicidation process stops when the whole thickness of metal is consumed Etch and Deposition Features 1 The DRY geometrical etch is extended to include the ANGLE and UNDERCUT parameters These parameters allow you to obtain the etch regions with tilted sidewalls and undercuts under the material mask Selective deposition and epitaxy for crystalline silicon and polysilicon When silicon deposition and epitaxy is performed with the SI_TO_POLY parameter specified in the DEPOSIT or EPITAXY statement the crystalline silicon layer w
201. IO2 are used The only difference is that the default for LAT RATIO2 is 0 2 This is because the channelled portion of a 2D profile is obviously very narrow Parabolic Approximation of Depth Dependent Lateral Distribution It has been shown 51 52 53 and 54 that in general the transversal function f y is not independent of depth because there is considerable correlation between transversal and longitudinal motion of the implanted ions This correlation could be taken into account by using a transversal function with the depth dependent lateral standard deviation Gy x As it was shown in 52 and 54 if the spatial moments up to fourth order are used the best approximation for Oy x is the parabolic function A saa R RS 3 210 oy x Cy c p C4 x p In order to find the coefficients of the function two additional spatial moments should be used These are mixed skewness rey feo 0 8 dedy 3 211 and mixed kurtosis o0 22 Bey Ane y R y dxdy 3 212 00 The cg c1 and co parameters can be found by substituting Equation 3 207 into Equations 3 209 3 211 and 3 212 and taking into account in Equations 3 180 3 186 while integrating over x This results in the system of equations where you can find the following relations cy AY 1 B 3 213 AY Cy aR yB 3 214 P Cy Ary 3 215 AR 3 74 Silvaco SSUPREM4 Models where BH 1 V B 22 Ta 3 216 1 7 This pa
202. It defaults to true This is a highly recommended option since it can reduce the matrix sizes by a factor of two or more and operation speed is a function of the size of the matrix MIN FREQ is a parameter that controls the frequency of the minimum fill reorderings It is only partially implemented and has no effect on the calculation GAUSS and CG specify the type of iteration performed on the linear system as a whole CG specifies that a conjugate residual should be used BACK specifies the number of back vectors that can be used in the CG outer iteration The default is three and the maximum possible value is six Note A higher value of BACK will give faster convergence at the cost of more memory usage BLK ITLIM is the maximum number of block iterations that can be taken The block iteration will finish at this point independent of convergence TIME STE ERROR and NEWTON specify the frequency with which the matrix should be factored The default is TIME The TIME parameter specifies that the matrix should be factored twice per time step This option takes advantage of the similarity in the matrix across a time integration The ERROR parameter indicates that the matrix should be factored whenever the error in that block is decreasing The NEWTON parameter forces factorization at every NEWTON step DIAG KNOT and FULL FAC specifies the amount of fill to be included in the factorization of the matrix FULL FAC indicates that the entir
203. KBUILD without display and then logout from the system use the UNIX nohup command before the following DECKBUILD command line nohup deckbuild run ascii an lt input filename gt outfile lt output filename gt amp 2 2 4 Running ATHENA inside DeckBuild Each ATHENA run inside DECKBUILD should start with the following command line go athena A single input file can contain several ATHENA runs each separated with a go athena line Input files within DECKBUILD can also contain runs from other programs such as ATLAS or DEVEDIT along with the ATHENA runs Running a given version number of ATHENA You can modify the go statement to provide parameters for the ATHENA run To run version 5 8 0 R the syntax is go athena simflags V 5 8 0 R Starting Parallel ATHENA The P option is used to set the number of processors to use in a parallel ATHENA run only the MC Implant module is parallelized starting ATHENA release 5 16 0 R If the number of processors specified by P is greater than the number of processors available it is automatically reduced to this cap number If P parameter is not specified ATHENA will run on all available processors automatically To run ATHENA on 4 processors use the command go athena simflags V 5 16 0 R P 4 Running ATHENA with a user specified default parameter file ATHENA supports the use of multiple default parameter files These files have the default root filename athenamod To start ATHE
204. LINE X statements line x loc 0 000000 tag left ine x loc 0 800000 spac 0 010000 line x loc 0 950000 spac 0 080000 ine x loc 1 050000 spac 0 080000 line x loc 1 200000 spac 0 010000 ine x loc 1 350000 spac 0 080000 Silvaco E 4 TSUPREM4 and TSUPREM3 Compatibility Features Note MASK IN FILI ine x loc 1 650000 spac 0 080000 ine x loc 1 800000 spac 0 010000 line x loc 1 950000 spac 0 080000 ine x loc 2 050000 spac 0 080000 ine x loc 2 200000 spac 0 010000 ine x loc 3 000000 tag right E lt maskfile t11 gt cannot be used together with LINE X statements Note sec files generated by SILVACO s MaskViews tools provide superior capabilities in generating grid and mask processing E 5 Using mask information with the EXPOSE MASK lt maskname gt the statement sequence of specified mask The EXPOSE statement This capability can be used only if ATHENA runs within DECKBUILD If DECKBUILD encounters EXPOSE with the parameter MASK lt maskname gt it provides ATHENA with a ETCH statements which will remove photoresist below all transparent regions of the MASK lt maskname gt statement should be preceded by a D EPOSIT PHOTO statement and followed by a DEVELOP without parameters statement For example if you load the same Mask Data File as used above in the MASK statement the following input deck
205. LOG10 ARSEN The following chooses the difference between the phosphorus and an analytic profile as the PRINT 1D variable SELECT Z PHOS 1 0E18 EXP Y Y 1 0E 8 The following chooses the excess vacancy interstitial product as the PRINT 1D variable SELECT Z INTER VACAN CI STAR CV STAR ro Note When using log or log10 functions make sure the argument is positive and non zero For example always use log10 abs doping 1 For more examples see the PRINT 1D Silvaco SET ATHENA User s Manual 6 56 SET SET specifies strings or numbers for variable substitution Note This commands executed under DECKBUILD and is documented fully in the VWF INTERACTIVE TOOLS MANUAL VOLUME I Syntax SET variable lt value gt Numerical Variable Example The following statement defines a variable and performs an expression on it for use later within the ATHENA processing syntax SET MYDOSE 1e13 SET HALFMYDOSE MYDOSE 2 IMPLANT BORON DOSE HALFMYDOSE String Variable Example The following uses SET to define a string variable The saved file will be called mosfet_fred str SET MYNAME fred STRUCTURE OUTFILE mosfet_ myname str For more examples see EXTRACT 6 100 Silvaco SETMODE 6 57 SETMODE SETMODE specifies execution mode parameters Syntax SETMODE
206. MINATION defines the illuminating wavelength the possible x and z tilt of the optical system and the relative intensity which is usually set to 1 ILLUM FILTER defines the shape of the illumination system The general shapes available are CIRCLE SQUARE GAUSSIAN ANTIGUASS and SHRINC The extent of the source must be defined to be within a square centered at the origin as shown in Figure 2 61 The extent of the source is defined by the coherence parameter SIGMA SIGMA defines the radius for circular sources CIRCLE GAUSSIAN and ANTIGAUSS the x and y intercepts for square sources and the radius of each individual SHRINC source element as shown in Figure 2 62 In all cases anything outside of the square defined by SIGMA 1 will be ignored The SHRINC source position is defined by the RADIUS and ANGLE parameters as shown in Figure 2 62 The SHRINC source can be defined by the command ILLUM FILTER SHRINC RADIUS 0 25 ANGLE 45 SIGMA 0 1 Source Region Figure 2 61 Maximum Extent of the Source Region Silvaco 2 77 ATHENA User s Manual Arbitrary sources can be defined by using the ANGLI transmittance of each source element are controlled by the parameters PHASE Sigma 4 Sigma Sigma Fi Sigma N Z Sigma CIRCLE SQUARE Sigma Radius Angle SHRINC Figure 2 62 Three Different Source Types E and RADIUS parameters Phase and inten
207. MINUM A S SIGMA DEP 0 2 SMOOTH WIN 0 1 SMOOTH STEP 1 UNIDIREC DEP RATE 1000 ANGLE1 0 00 Silvaco 2 61 ATHENA User s Manual Deckbuild ATHENA Rate Deposit GENERAL PARAMETERS Machine name test Material 7 Aluminum User defined material Machine type Unidirectional Planetary Dualdirectional Conical CYD Hemispherical Simple MC Single Particle MC Custom Deposition rate 1 0 __ v u min Surface Diffusion 0 20 0 00 1 00 Smoothing window 01 5 00 Smoothing step 1 10 PARAMETERS FOR DUALDIRECTIONAL MACHINE TYPE Angle 1 deg 45 00 0 00 Angle 2 deg 45 00 90 00 Comment Deposit machine test WRITE Figure 2 44 ATHENA Rate Deposit Menu Table 2 4 Deposition Model Required Parameters Models Parameters CVD JUNI DUAL HEMI CONIC PLANET MONTE1 MONTE2 CUSTOM1 CUSTOM2 dep rate yes yes yes yes yes yes yes yes optional yes step cov yes no no no no no no no no no anglel no yes yes yes yes yes no yes yes no no angle2 no no yes yes no yes no no no no c axis no no no no yes yes no no no no p axis no no no no yes yes no no no no dist pl no no no no no yes no no no no no sigma dep no optional optional optional optional optional optional optional no yes smooth win no optional optional optional optional optional optional optional
208. N sets the minimum power accounted for in multiple reflections POWER MIN is used in a multiplicative format In other words if power attenuation due to 10 reflections is less than POWER MIN it will not be counted for calculation Silvaco 6 35 EXPOSE ATHENA User s Manual Examples The following statement loads a cross section of an aerial image that you can input It then runs the exposure module The number of reflections increases calculation time when it is set to a value greater than one EXPOSE INF IL E CROSS SI ECT NUM R EF L 3 The following command runs the exposure module for the z CROSS section of a two dimensional aerial image that has been previously generated The x value of the cross section is 0 1 EXPOSE Note The D For more examples see INIT EF OCUS parameter on the IMAGI command must be used in conjunction with the Z CROSS CROSS VAL 0 1 E statement must be used to do defocus exposure calculations The image EXPOSE command for a defocussed bulk image TALIZE and IMAGE Silvaco EXTRACT 6 21 EXTRACT The EXTRACT command is used to analyze the current structure or a previously saved file It can extract important parameters such as material thickness junction depth and peak doping levels It also includes electrical extractions such as sheet resistance threshold voltage and CV curves
209. NA User s Manual Substituting for o in Equation 3 3 is writing the particle charge as a signed integer Z4 times the elementary charge q giving us this Flux Expression I DKO WC Z C 4 3 4 In insulator and conductor materials the electric field is zero In semiconductor materials the electric field is given by E Vy I yn 3 5 qn where y is the electrostatic potential and n is the electron concentration If charge neutrality is assumed then the electron concentration may be rewritten as re poe MBNA Np x ge 3 6 2 2 i where Np and N4 are the electrically active donor and acceptor impurity concentrations and n is the intrinsic carrier concentration calculated as n NIO exp NLE aac 3 7 kT where you can specify the NI 0 NI E and NI POW parameters in the MATERIAL statement The electrically active and mobile impurity concentrations are equivalent Boundary conditions Boundary conditions within ATHENA are of mixed type and are expressed mathematically as a Cy B AC R 3 8 where a B are real numbers and C4 designates the flux of C4 across the boundary The right hand term R accounts for all source terms on the boundary Boundary conditions are applied at two main regions The first region is at the top of the simulation region the surface The second region is at the inter regional interfaces for which the species in question only has a meaningful existence in one of the region m
210. NA gt LINE X LOC 0 00 SPAC 0 10 ATHENA gt LINE X LOC 0 3 SPAC 0 02 ATHENA gt LINE X LOC 1 SPAC 0 1 ATHENA gt ATHENA gt LINE ATHENA gt LINE LOC 0 00 SPAC 0 03 LOC 0 2 SPAC 0 02 ATHENA gt LINE LOC 1 SPAC 0 1 ATHENA gt I TIAL SILICON STRUCTURE ATHENA gt INIT SILICON C BORON 3 0E14 ORIENTATION 100 TWO D ATHENA gt STRUCT OUTFILE history0l str KKK Z The line STRUCT OUTFILE history01 str is automatically produced by DECKBUILD through the history function This function provides an important service when debugging new files performing what if simulations and visualizing the structure at different steps of simulation This feature will be used throughout the tutorial Use any of the three methods to visualize the initial structure 1 Click on the Tools menu button DECKBUILD will then automatically save a temporary standard structure file and invoke ToNYPLOT with this file 2 Click on the Main Control button and the Deckbuild Main Control popup will appear Then click on the Plot Current Structure button DECKBUILD will then automatically save a temporary standard structure file and invoke TonyPLot with this file 3 Select highlight the name of a structure file history01 str in this case and click on the Tools or Plot Current Structure DECKBUILD will then start ToNYPLOT with the selected structure file After a short delay TONYPLOT will appear It will h
211. NA with athenamod 97 the syntax is go athena simflags modfile 97 Running ATHENA In Standalone Mode Without DeckBuild You can run ATHENA outside the DECKBUILD environment but we don t recommended it If you don t want the overhead of the Deckbuild Window use the No Windows Mode Many important features such as variable substitution automatic interfacing to device simulation and parameter extraction are unavailable outside the DECKBUILD environment To run ATHENA directly under UNIX use the following command line athena lt input filename gt To save the run time output to a file don t use the UNIX redirect command gt Instead specify the name of the output file athena lt input filename gt logfile lt output filename gt Note Some of the standard examples supplied with ATHENA will not run correctly outside of DECKBUILD 2 6 Silvaco Tutorial 2 3 Creating a Device Structure Using ATHENA 2 3 1 Procedure Overview ATHENA is designed as a process simulation framework The framework includes simulator independent operations and simulator specific functions that simulate different process steps e g implant RIE or photoresist exposure This section describes ATHENA input output and the following basic operations for creating an input file e Developing a good simulation grid e Performing conformal deposition e Performing geometric etches e Structure manipulation e Saving and loading st
212. NOEXECUTE ECHO Description This command turns on the following execution mode parameters The UNSET statement allows the same parameters to be turned off NOEXECUTE puts all entered statements into a check only mode If this flag is on ATHENA will only check the legality of the input syntax and not execute any statements ECHO instructs ATHENA to echo all input lines to the run time output Note that in DECKBUILD this is not required as all lines are echoed to the bottom run time window or run time output file by default Examples The following statement causes ATHENA to echo each command it receives SETMODE ECHO For more examples see UNSETMODE Note The parser does not recognize abbreviated forms of these commands It requires that you enter NOBXECUTE and ECHO verbatim Silvaco 6 101 SILICIDE ATHENA User s Manual 6 58 SILICIDE SILICIDE specifies the silicidation coefficients Syntax SILICIDE SILICON POLYSILICON TUNGSTEN TITANIUM PLATINUM COBALT WSIX TISIX PTSIX COSIX MATERIAL lt c gt SILICON POLYSILICO TUNGSTEN TITANIUM PLATINUM COBALT WSIX TISIX PTSIX COSIX MATERIAL lt c gt MTTYPE lt c gt MTTYPE lt c gt ALPHA lt n gt Description SILICON POLYSILICON TUNGSTEN GAAS TITANIUM PLATINUM COBALT WSIX TISIX PTSIX and MATERIAL specify the first materi
213. NT lt n gt ADAPT ADAPT COUNT lt n gt ADD I LINE lt n gt SENSITIVITY MATERIAL MATERIAL Description This statement runs the adaptive meshing algorithm or the smoothing algorithm in standalone mode SMOOTH flag to do mesh smoothing SMTH COUNT specifies the number of smooth loops during the smooth operation The default is 1 ADAPT flag to do stand alone mesh adapting Specifies that a stand alone adaptive meshing step should be performed to refine or relax the current mesh based on the material impurity specification given on ADAPT PAR command The default is False ADAPT COUNT specifies the number of adapting loops during the stand alone adaptive meshing operation The default is 1 ADD I LINE depth of the shadow interface mesh line in microns Add the mesh line at the interface between two materials as defined by the booleans MATERIAL and MATERIAL The line is added in MATERIAL at a distance ADD I LINE from MATERIAL SENSITIVITY specifies sensitivity of adaptation algorithm The lower value leads to grid with more triangles The default is 1 0 MATERIAL one of standard materials or user specified material see Section 6 2 9 Standard and User Defined Materials for the list of materials MATERIAL one of standard materials or user specified material see Section 6 2 9 Standard and User Defined Materials for the list of materials Examples The following statement will add a s
214. ON THICK 0 02 DIVISIONS 8 C BORON 5E18 C GERMANIUM 1E22 F GERMANIUM 1E21 DIFFUSE TIME 4 TEMP 650 NITRO PRESS 1 This section will give a brief explanation of the statements and possible variances A more detailed description of the individual parameters are given in the ATHENA notes files and in Chapter 6 Statements 2 7 1 METHOD Statement FULL CPL is the recommended diffusion model for boron to obtain the best accuracy It is however slower than other diffusion models due to the larger number and more complex inter relations taken into account MIN TEMP is required if the deposition temperature is below 700 C which is the minimum calibrated temperature for ATHENA in standard processing Select a MIN TEMP value below the deposition temperature otherwise no diffusion will be calculated The MODEL SIGE parameter invokes the silicon germanium models 2 7 2 MATERIAL Statement The following statement specifies reasonable user definable parameters for the SiGe models in the material silicon MATERIAL SILICON NIFACT SIGE 100 EAFACT SIGE 1 5 NO FLIP The NO FLIP parameter prevents automatic mesh optimization which preserves the user defined x grid spacing ATHENA often will try to remove what it believes are excessive grid points during DEPOSITION ETCH statements It is not advisable to remove mesh points in the base of an HBT which i
215. PH specify the wafer s miscut The explanation of these parameters is as follows Let s consider the internal coordinate system of the crystal structure xyz to be right hand oriented where y is the inward direction relatively to the surface You can then define the misorienting of the surface by tilting the wafer by MISCUT TH degrees in the xy plane and rotating it counter clockwise in the xz plane by MISCUT PH degrees if MISCUT PH is positive and clockwise if MISCUT PH is negative Remember that ROT SUB MISCUT TH and MISCUT PH are measured from the internal co ordinate system compared to the ROTATION parameter which is measured from the wafer s major flat defined by ROT SUB The simulation plane shown in Chapter 3 SSUPREM4 Models Figure 3 23 is defined by the ROT SUB parameter In the case of silicon carbide the simulation XY plane for 4H SiC is 1100 In other words if specified by ROT SUB 0 in the INITIALIZE statement then e a miscut of 8 towards the 1 20 direction i e in the 1 01 0 plane is specified by MISCUT TH 8 and MISCUT PH 60 e a miscut of 8 towards the 7010 direction ie in the 1 1 20 plane is specified by MISCUT TH 8 and MISCUT PH 90 TRAJ FILE specifies the name of the file in which ion trajectories calculated with the Monte Carlo BCA method are to be saved Note This parameter switches off statistical sampling if it s specifie
216. PHOR lt conc gt and so on The new parameter SS IMPURITY where the generic name IMPURITY could be substituted by any standard impurity name specifies that concentration of the named impurity in the ambient gas is set to its solid solubility in silicon at the current temperature E 10 Changes in the ETCH statement TRAPEZOI is an alias for the DRY If THICKNESS is not specified it assumed to be infinite E 11 Changes in the STRUCUTURE SAVEFILE statement OUT FILE is an alias for OUTFILE TIF DEVICE and MEDICI are ignored because ATHENA and other SILVACO TCAD tools use the universal SSF data format E 12 Changes in the IMPLANT statement D PLUS is an alias for PLUS ONE and UNIT DAMAGE D SCALE is an alias for DAM FACTOR IMPL TAB is ignored ATHENA usually uses the default set of implant tables You can define tables in the MOMENTS statement E 13 Changes in the ELECTRODE statement BOTTOM is an alias for SUBSTRATE E 14 Changes in the METHOD statement OX ADAPT IMP ADAPT and DIF ADAPT are aliases for ADAPT PD FERM1 is an alias for FERMI PD TRANS is an alias for TWO DIM PD FULL is an alias for FULL CPL E 15 Changes in the MATERIAL statement E FIELD specifies that the electric field terms are to be accounted for in the diffusion calculations This parameter is always set to TRUE in semiconductors POLYCRYS is an alias for the POLY DIFF parameter in the METHOD statement
217. Parameters for Carbon Effects in SiGeC KCARBON 0 and KCARBON E specify interstitial recombination rate in carbon sink KCARBON 0 is the pre exponential constant for the rate in sect and KCARBON E is the corresponding activation energy in eV DCARBON E specifies the coefficient of interstitial diffusion retardation in SiGe in presence of carbon impurity The units are eV Basic Example The following statement specifies the silicon diffusion and equilibrium values for interstitials INTERST SILICON DI 0 5 0E 7 D E 0 0 CSTAR 0 1 0E13 CSTAR E 0 0 Defect Injection during Oxidation Example The following statement specifies the oxide silicon interface injection for DRYO2 ambient is to be computed using the oxide growth velocity and with 1 of consumed silicon injected as interstitials INTERST SILICON OXIDE GROWTH VMOLE 5 0E22 THETA 0 0 01 THETA E 0 0 Surface Recombination Example The following statement specifies the surface recombination velocity at the nitride silicon interface is 3 5 x 10 cm sec INTERST SILICON NITRIDE KSURF 0 3 5E 3 KSURF E 0 0 KRAT 0 0 0 Experimental Injection Model Example The following statement describes an injection at the silicon oxide interface that exponentially decays in time INTERST SILICON OXIDE INJ STR 10 0E4 EXP T 10 0 General Comments The equivalence of INTERSTITIAL and V
218. Physical parameters that are specific for various materials such as viscosity and surface tension are also given in MATERIAL statement Reflow will proceed according to the time and temperature given in the DIFFUSE statement The finite element solver are invoked by specifying the VISCOUS parametrs and various numerical control parameters in the METHOD statement The viscous creep flow equations solved are as follows see 97 uv V VP 4 28 vey 1 2 p i u ene ee 4 30 2 1 v where V is the velocity P the pressure u the viscosity v the Poisson s ratio and E the Young s modulus The parameters v and E can be specified as POISS R and YOUNG Min the MATERIAL statement 4 20 Silvaco ELITE Models 4 6 Chemical Mechanical Polish CMP Chemical Mechanical Polish CMP is a module in ATHENA To run CMP you need to have the license to use ELITE CMP is used to model wafer planarization using polishing pad and chemical slurry characteristics CMP is also used to circumvent two major problems First the depth of focus of high numerical aperture lithography systems Second metal thinning that can occur over non planar topographies The CMP module that is incorporated into ATHENA has two distinct models The first is the Hard Polish or Buzz Saw Model 98 The second is the Soft Polish Model based on the work of J Warnock 99 To access these models use the ATHENA statements RATE POLISH and POLISH These st
219. Powers in Amorphous Materials and Range Validation ss cccssscesesereeeseeeeeseeeenesees 3 89 90 Two Dimensional Implant Profiles eeeeeeseeeeeeeees 3 72 75 lon Implantation Damage C lnterpret tes sisi tatititat dei gained ia Be ae 3 89 Cluster Model ists reier ss 3 88 Dislocation Loops Model eeseeceeesseeeesneeeesneeeeeneeeees 3 88 Plus Modelen kaiina a ae net 3 87 K Kinchin Pease model ccccccceeeeeeeeseessseeeeeeeeees D 7 D 8 Klaassen bandgap narrowing model ccceeeeeeeeeeeeeeeees 2 52 L Linear Rate Constant Chlorine Dependence ccecceescessteeeeeeeeeeeeneeseeeeeanensees 3 54 Doping Dependence ceeeeeeeeseeeeeeneeeesneeeteeeeees 3 55 56 Orientation Dependence eecececeeseeeeeteneeseeeetretees 3 52 Pressure Dependence eeecceeeeeeeeeseeeeeeneeeesneeeseneeeens 3 53 EPOVD EET E T 4 10 Lithography esiseina as eir ee 5 15 M MaskViews Generating Masks in ATHENA seceeeseeeeeeeeeeeeees 2 71 73 Initial Rectangular Grid ecceeeeeeeeeseeeeeneeeseeeeeeeeeeteaeee 2 65 Medium Injection Bandgap Narrowing Effects esscessseeeeseeesneeeeeneeeees 2 52 Poly emitter work function 00 eeeeeeeeteeeeeeneeeeeeeeteeeeees 2 51 Modelling the Correct Substrate Depth See Correct Substrate Depth Modelling Modified Gaussian Function MGF cesseesseeeeeeeeeeneees 3 75 Monte Carlo Etching Model Incoming lons and
220. RESS model regards the oxide as a compressible liquid The VISCOUS model treats the oxide as an incompressible viscous liquid Oxide is actually believed to be incompressible but the compressible model runs faster The default is the COMPRESS model Note For Hints on the use of the different oxidation models see Chapter 2 Tutorial Note Use of the VERTICAL model is not recommended in ATHENA LIFT POLY LIFT OXIDE and LIFT NITRID specifies that the polysilicon oxide and nitride materials can be lifted by oxidation or silicidation processes These are t rue by default but you set them to false to eliminate the lifting portion of the calculation for geometries where lifting is not expected to occur OX THRESH specifies that the oxidation threshold model is enabled This doesn t allow oxidation when the concentration of oxidant drops below a critical threshold value set by MIN OXIDANT on the OXIDE statement Silvaco 6 67 METHOD ATHENA User s Manual SKIP SIL is a Boolean parameter which controls the computation of stress in silicon SKIP SIL defaults to true stress can only be computed when the VISCOUS oxide model is used The silicon substrate is treated as an elastic solid subject to the tensions generated by the oxide flow Indiscriminate use is not recommended The silicon grid is usually much larger than the oxide grid and stres
221. RF lt n gt Description This statement allows you to increase grid spacing You can place the RELAX statement anywhere within the input file RELAX commands however are ignored if ATHENA is in 1D mode The RELAX statement also includes an algorithm for relaxing grid on the surface of the simulation structure MATERIAL specifies that RELAX will only apply to the regions of this MATERIAL see Section 6 2 9 Standard and User Defined Materials for the list of materials If MATERIAL is not specified RELAX will be applied to all materials in the box X MIN X MAX Y MIN and Y MAX specifies the corner coordinates of the RELAX box Units are microns Default is bounding box of the current simulation structure DIR X or DIR Y allow the direction of the grid relax to be controlled DIR X and DIR Y are true by default i e when the RELAX statement is encountered the grid is relaxed in both directions by default When DIR X or DIR Y is selected as false i e DIR X F or DIR Y F then the grid is only relaxed in the direction that is left as true SURFACE specifies to relax the surface grid DX SURF specifies a minimum size for surface segments Examples RELAX SILICON X MAX 1 Y MIN 0 This statement changes a grid over a rectangular area in silicon from the left side of a structure to 1 and from y 0 to the bottom of the silicon Note RELAX will not make any c
222. RGENCE is the standard deviation of this distribution in degrees CHEMICAL is the etch rate for this component of the RIE model D 18 Silvaco ATHENA Version History Chemical Mechanical Polish e Two models for chemical mechanical polishing have been added to ELITE They are the Burke model hard polish and the Warnock model soft polish The Burke model polishes the structure at a rate proportional to the pattern factor of the structure The Burke parameters MAX HARD and MIN HARD are the maximum and minimum polish rates and are entered via the RATE POLISH statement MAX HARD corresponds to a pattern factor of zero and MIN HARD corresponds to a pattern factor of one The actual polishing rate is calculated on the line between MAX HARD and MIN HARD depending on the pattern factor of the structure being polished The Warnock model has four parameters on the RATE POLISH statement SOFT sets the polish rate HEIGHT FAC is the vertical deformation scale in microns LENGTH FAC is the horizontal deformation scale in microns The polishing rates for tall features and holes are calculated using HEIGHT FAC and LENGTH FAC HEIGHT FAC measures how much the polishing pad will deform with respect to the height of the feature LENGTH FAC measures the distance the effect of a tall feature will be felt LENGTH FAC is a measure of the stiffness of the pad and the distance at which shadowing will be felt by a tall feature where HEIGHT FAC is a measure of the sp
223. RPENDICUL and PARALLEL specify TE mode or TM mode respectively PERPENDICUL is the default X CROSS and Z CROSS specify that the cross section is parallel to the x axis z constant and parallel to the z axis x constant respectively X CROSS is the default CROSS VALUE specifies the x or z coordinates of the cross section of the aerial image The default will be centered in the image window Units are microns DOSE specifies the exposure dose in mJ cm X ORIGIN locates the beam relative to the structure This allows the aerial image to be shifted if necessary Units are microns The default is 0 0 FLATNESS specifies the accuracy of the change in surface topography in degrees A value of zero specifies that all grid points will be calculated The default value is 0 25 In any case maintain the limits 0 lt FLATNESS lt 1 NUM REFL specifies the number of reflections to be considered FRONT REFL specifies that front surface reflection should be considered in the calculation The default is no front reflection BACK REFL specifies the back surface reflection The default is no back reflection ALL MATS specifies that intensity be displayed in all materials The default is photoresist only MULT EXPOSE is used to make multiple exposures MULT EXPOSE is specified on the second EXPOSE command for addition of exposures If MULT EXPOSE is not specified previous exposures will be erased POWER MI
224. S SECONDS DT FACT lt n gt DT MAX lt n gt DX MULT lt n gt MC REDEPO MC SMOOTH lt n gt MC DT FACT lt n gt MC MODFNAME lt c gt Description ATHENA provides two different etch simulation methods The first is geometrical etching available within any ATHENA module The second is physical etching available only in ELITE Parameters used for Geometrical Etching MATERIAL specify the material to be etched see Section 6 2 9 Standard and User Defined Materials for the list of materials If a material is specified only that material is etched even if other materials lie within the etch region If no material is specified all materials in the etch region are removed NAME RESIST specify the type of photoresist to be etched ALL specifies that all of the specified materials are removed DRY indicates that the resulting surface will replicate the exposed surface and will simply be lowered by a fixed depth of THICKNESS microns below the exposed surface If ANGLE or UNDERCUT or both is specified the shape of DRY etched region is modified accordingly TRAPEZOI is a synonym for this parameter THICKNESS specifies the thickness to be etched for the dry etch type Units are microns ANGLE specifies sidewall slopes in degrees 90 corresponding to vertical slope is the default UNDERCUT specifies
225. S N At each time step Equation 3 139 is solved The incremental oxide thickness grown is calculated by multiplying Equation 3 139 by the time step During the oxidation reaction silicon atoms bond with the oxidant to form the SiO compound Thus silicon material is removed during the oxidation process The ratio of the silicon thickness consumed to form a given thickness of SiOz is specified using the ALPHA parameter in the OXIDE statement Equation 3 138 is sufficient to describe the motion of the Si SiO interface if the oxide flow is in the same direction as the growth for planar oxidation structures In most structures of interest the oxide flow is two dimensional Therefore additional equations have to be solved Both the Compress and Viscous models calculate the two dimensional flow of oxide elements by solving a simplified hydrodynamic creeping flow equation 3 46 Silvaco SSUPREM4 Models 3 3 2 Compress Model In addition to solving Equations 3 138 and 3 139 a simplification of the hydrodynamic flow equation is solved to obtain the flow of oxide elements 34 The Compress Model is activated by specifying COMPRESS in the METHOD statement prior to a DIFFUSE statement The Compress Model is the default oxidation model in SSUPREM4 Neglecting the acceleration and gravitational terms in the hydrodynamic flow equation the creeping flow equation is given by uv V VP 3 140 where P is the hydrostatic pressure V is velocit
226. SNA AAAS PARES aN i NY Se Na RERET Bie aa Gaal LA a AAA Cee AA SAREE ARISE PS ARSARARARS ARI SSS SSNS RAR AREBRRARR ESS rs EEEE SARS AN BREN a ae a K FJ a w A F A i microns WACO International 1993 Figure 2 14 Grid Control for Deposition Silvaco 2 19 ATHENA User s Manual Simple Geometrical Etches The next step in this tutorial process simulation is to define the polysilicon gate definition Implant and thermal steps will be discussed in Section 2 4 Choosing Models In SSUPREM4 To set a geometrical etch step select Process gt Etch gt Etch from the Command menu of DECKBUILD The ATHENA Etch Menu Figure 2 Materia Thickness om Etch Method Etching Machine Geometrical type all Left Right Dry thickness Any shape Etch location um 0 3 0 00 d H 10 00 Athitrary points X fecation T YX fecation I Comment Poly definition 15 will appear I Polysilicon Figure 2 15 ATHENA Etch Menu The Geometrical etch is the default method Other methods will be discussed in Section 2 8 2 Deposition and Wet Dry Etching using the Physical Models in ATHENA ELITE Select Polysilicon from the Material menu This example will use a polysilicon gate edge at x 0 3 and set the
227. SOTROPIC specifies the isotropic etch rate used by the WET ETCH and RIE models The isotropic etch rate is the contribution of thermal atoms radicals and molecules coming out of the plasma These are assumed to have an isotropic angular distribution Therefore the isotropic etching may lead to an underetching of the mask CHEMICAL DIVERGENCE CHEMICAL is the etch rate in the RIE model normal to the ion beam when the DIVERGENCE is specified as non zero DIVERGENCE specifies the beam divergence used by the RIE model The angular distribution of the ions coming down to the wafer is Gaussian Silvaco 6 91 RATE ETCH ATHENA User s Manual Parameters used for Plasma Etch Model PRESSURE specifies the plasma etcher reactor pressure Units are mTorr Default 50 mTorr TGAS specifies the plasma etcher reactor gas temperatures Units are K Default is 300 K TION specifies the plasma etcher reactor ion temperatures Units are K Default is 300 K VPDC specifies the DC bias in the plasma sheath Units are V Default is 32 5 V VPAC specifies the AC voltage in the sheath bulk interface Units are V Default is 32 5 V FREQ specifies frequency of the AC current Units are Mhz Default is 13 6 Mhz LSHDC specifies the mean sheath thickness Units are mm Default is 0 005 mm LSHAC specifies the AC component of the sheath thickness Units are mm Default is 0 0 MGAS specifies the atomic mass of the gas at
228. SPACING lt n gt TAG lt c gt TRI LEFT TRI RIGHT Description This statement defines the position and spacing of mesh lines All LINE statements should come before the REGION and BOUNDARY statements which should then be followed by an INITIALIZE statement X and Y specify whether a mesh line is horizontal or vertical LOCATION specifies the location along the chosen axis in microns at which the line should be positioned The x coordinate increases from left to right the y coordinate increases progressing from top to bottom going into the substrate This is the opposite of normal Cartesian y axis progression which increases going upward SPACING specifies the local grid spacing in microns ATHENA adds mesh lines to the ones given according to the following recipe Each user line has a spacing whether user specified or inferred from the nearest neighbor These spacings are then smoothed out so no adjacent intervals have a ratio greater than the value given by INTERVAL R on the INITIALIZE statement default is 1 5 New grid lines are then introduced so that the line spacing varies geometrically from one end of the interval to the other Refer to the example below TAG labels lines for later reference by BOUNDARY and REGION statements The tag label can be any word TRI LEFT and TRI RIGHT can be specified in the LINE X statement to control triangle orientation in the initial grid Initial simulation area is divided in
229. See Figure 2 58 2 74 Silvaco Tutorial Figure 2 58 MASKVIEws Properties Popup showing the Simulation Menu Open the Simulator menu and select the ATHENA OPTOLITH menu item Customized controls for MASKVIEWS OPTOLITH will appear in the MAsKVIEWS window The colored buttons on the right side of the window are discrete controls for phase in degrees and intensity transmittance The buttons first appear as phase To change to transmittance open the Phases menu above the buttons and select the Transmittances menu item This will change the buttons from phase to transmittance controls Continuous controls for phase and transmittance are located directly below the colored buttons The mask can now be designed using the mouse driven line writer following the description outlined in the MASKVIEws chapter of the VWF INTERACTIVE TOOLS USER S MANUAL VOL I Once the mask is created it should be saved to a file with a name ending in a lay extension for future editing It is important to be aware that there are two types of files that can be saved from a MASKVIEws layout information The first type is the layout file This file includes the information about layers and mask features To store this information select the Files Save menu item in the MASKVIEWS screen The second type of file that can be saved from MASKVIEWS is a file that is similar to the layout file but is written to interface with ATHENA OPTOLITH The file to be used by OPTOLIT
230. Ss e gt z e pA Q pe T Q g Ss fai Ba uu SENSITIVITY of Vt to THETA O 20keV Boron 3 min Gate Ox 850C Threshold Voltage v 0 5e 08 1e 09 15e 09 2e 09 25e 09 3e 09 Theta 0 SILVACO International 1996 Figure 2 34 A Typical Dependence of Extracted Threshold Voltage Silvaco Tutorial 2 5 3 Tuning Implantation Parameters You can now tune two implantation parameters by using the threshold voltage versus gate length data The peak value of threshold voltage for a given process flow the reverse short channel effect will be a function of the initial implant damage caused by the LDD and source drain implants Since these implants have a high total dose and damage the tuning parameter here is the clustering factor In ATHENA this parameter is called CLUST FACT and is defined in the CLUSTER statement The higher the clustering factor the greater the damage and the greater the diffusion the greater the reverse short channel effect Figure 2 35 shows the effect on the threshold voltage of changing the CLUST FACT parameter for a typical process flow TonyPlot V2 6 6 Filey View Plot Tools Print Properties 7 Help 7 _ xl 4 EFFECT OF DAMAGE FACTOR ON TH RESOLD VOLTAGE ajeje 5min 850C nitrogen anneal V A 0 26 clust fact 0 0 clust fact 0 5 clust fact 1 0 clust fact 1 4 clust fact 2 0 S
231. T statement of DECKBUILD or both and the normal information messages generated by ATHENA The number of messages generated depends on the output mode chosen in the OPTION statement The QUIET mode is the default Minimum output is generated in this case all statements are echoed and the status of a time consuming simulation is reported The NORMAL option produces some additional output information including information about the current grid e g number of nodes or triangles VERBOSE and DEBUG modes are useful for debugging but these options produce too much output for any other purpose Standard Error Output consists of the warning and error messages describing syntax errors file operation errors system errors and internal inconsistencies Standard Structure File Format The main channel of information exchange between ATHENA and other simulators and tools is the Standard Structure File SSF format SSF is a universal file format used by a number of Silvaco simulation programs The STRUCTURE statement of ATHENA creates a Standard Structure File which contains mesh and solution information model information and other related parameters The saved Standard Structure File can be used by the following e ATHENA to re initialize the structure and continue process simulation e ATLAS or other device simulators to perform electrical analysis of the structure produced by ATHENA e TonyPLot to graphically display
232. TA Notes The usual reason for employing a Rapid Thermal Anneal RTA in a process flow is to anneal out damage in the substrate that has been caused by a previous process step usually an implant while at the same time minimizing dopant diffusion Dopant activation also occurs during this process These anneals are usually high in temperature and low in duration for sound device physics reasons Once again the key to accurate simulation of RTA lies in the accurate simulation of substrate damage behavior The role of interstitials in enhanced dopant diffusion has already been explained in Section 2 4 Choosing Models In SSUPREM4 to become familiar with the role of interstitials during process simulation The reason why an RTA usually employs high temperatures and short durations is because for a given high dose implant if an anneal duration is selected so that a fixed percentage of the damage is annealed the lower the anneal temperature the more dopant diffusion occurs The above statement requires an explanation since intuitively the opposite would seem more likely A descriptive explanation of what is happening can be informative if the two extremes of anneal temperature are considered For the lowest anneal temperatures the damage anneal rate is almost zero so dopant diffusion rates are enhanced by a factor of 1000 C or more for the long time periods required to remove the damage This results in high total dopant diffusion For
233. TOR parameter The effect of the damage on subsequent diffusions are modeled in ATHENA using the fully coupled diffusion model METHOD FULL CPL A previous Hints and Tips covered a description of this in the Simulation Standard February 1995 To model RSCE in ATHENA and ATLAS it is necessary to construct MOSFETs of different channel lengths This can be done either using the MASKVIEWS layout interface or using the STRETCH command in ATHENA or DEVEDIT The user should simulate the shortest channel length up until the polysilicon etch and stretch the device to the desired length The FULL CPL model is only required for diffusion after the source drain implants Figure C 6 shows the result of a threshold voltage simulation versus gate length for various values of implant damage vwF was used to automatically generate and run this experiment VWF handles the automatic interface to ATLAS and the extraction of the threshold voltages Looking horizontally along the y 0 line it is seen that with zero implant damage the threshold voltage decreases with decreasing length No RSCE is seen However as DAM FACT is increased the threshold voltage starts to rise before falling at very short lengths It is clear the size of the RSCE increases with implant damage factor It is also interesting to note that even the threshold voltage for the 20mm long device is affected slightly by the implant damage This is to be expected from Figure C 7 which shows p
234. This function is only available in the mesh based ATHENA framework and can not be implemented into a string based tool Void Formation Control Extra control has been added to allow the control of the formation of a void in the case of two encroaching CVD fronts D 12 OPTOLITH Image Routines Enhancement The algorithm in evaluating the aerial image of the mask has now been streamlined Approximately the speed improvement is equivalent to a change from n n to n log n For a complex mask the speed can be as high as 20X A minor bug in calculating the diffraction pattern has now been removed Exposure Routines Improvement Optolith Exposure now runs around 4 5 times faster than version 3 0 This has been achieved by restructuring the ray tracing algorithm used to expose a given non planar device structure In addition an error in setting up the boundary conditions for the electromagnetic wave has been corrected to yield the proper standing wave pattern The asymmetry in energy deposition for a symmetric structure has also been fixed New Material RSM Calibration System When used with the VWF system Optolith may be used to calibrate physical model parameters Example model parameters include A B C bleaching parameters and Development rate parameters for all Development rate models The system will fit simulation model parameters to a range of experimentally measured CD data D 13 ATHENA Version 3 0 1 R Release Notes D 13 1 ATHENA
235. TonyPlot Window select Plot Display and a popup will appear Then pull down the Group menu and select the Dose menu item This will group the set of plots for each exposure dose Silvaco 2 85 ATHENA User s Manual 2 10 Adaptive Meshing 2 10 1 Introduction to Mesh Adaption ATHENA has a built in mesh adaption module that automatically adapts the grid to dopant profiles Used together with implantation and diffusion the module can achieve more optimized accuracy of a given profile s representation for a given number of grid points This relieves you to some extent from the time consuming mesh generation task in the simulation structure preparation stage It will also improve the accuracy and speed of the subsequent diffusion oxidation epitaxy stages where impurity profiles change with time The algorithm used was suggested in 3 4 It uses an efficient local error estimator and a triangulation scheme suitable for complex two dimensional moving boundary problems Adaption During lon Implantation Ion implant is a common process step to introduce impurities into the substrate to form active device regions Prior to the implant step it is difficult to determine the required mesh density distribution because the exact dopant profile is unknown before processing Thus you can only estimate the profile and required mesh It s a time consuming process to specify mesh generation statements to create the mesh with a densi
236. ULT is the accuracy multiplier for ELITE etches The discretization size used for the etch calculation will be multiplied by DX MULT For improved accuracy at the cost of extra simulation time decrease the value of DX MULT Parameters used only with MC PLASMA model MC REDEPO specifies that redeposition of polymer should be simulated Default is true MC SMOOTH specifies level of smoothing of the surface MC DT FACT specifies time step control for Monte Carlo etching and redeposition MC MODFNAME specifies name of the C Interpreter file with user defined Monte Carlo etching and redeposition models Simple Geometrical Etch Example The following command etches all the nitride to the left of a vertical line located at x 0 5 ETCH NITRIDE LEFT P1 X 0 5 Silvaco 6 33 ETCH ATHENA User s Manual Arbitrary Geometrical Shape Etch Example The following set of commands etch the oxide in the square defined at 0 0 1 0 1 1 and 0 1 E dah E E CH CH CH CH OXIDE START X 0 0 Y 0 0 CONTINU E X 1 0 Y 0 0 CONTINU E X 1 0 Y 1 0 DONE X 0 0 Y 1 0 Be careful when using this style of syntax that the list of coordinates forms a regular polygon The closing line from the last coordinate pair to the initial point is automatically added Anisotropic Geometrical Etch Example The following command finds the exposed surface and lowers it straight down 0 1 microns
237. URTOSIS SKEWXY SRANGE lt n gt SSTD DEV lt n gt SGAMMA lt n gt SKURTOSIS lt n gt LSSTD DEV SSKEWXY SKURTXY SKURTT DRATIO lt n gt IGNORE_MOM Description Parameters Used to Select Moment Tables SVDP_TABLES specifies that the SIMS Verified Dual Pearson SVDP moments tables will be used with dual Pearson implant model Default is true See Chapter 3 SSUPREM4 Models Section 3 5 1 Analytic Implant Models STD_TABLES specifies that SVDP_TABLES are ignored and standard tables are used with the subsequent implant statements USER_STDT specifies the user defined moments file see the USER_TABLE parameter will be used with standard format You can find a template for the user defined moments file in lt install area gt lib athena lt version gt common userimp USER_TABLE lt c gt specifies the file that contains user defined look up implant parameter tables Implant Definition Parameters MATERIAL specifies the material for which the implant moments are set see Section 6 2 9 Standard and User Defined Materials for the list of materials I IMPURITY specifies the implanted impurity for which the moments are set see Section 6 2 10 Standard Impurities for the list of impurities I BF2 can be also specified DOSE is an incident ion dose em ENERGY sets the incident ion energy keV Parameters Used for Specification o
238. USE TIME lt n gt HOURS MINUTES SECONDS TEMPERATURE lt n gt T FINAL lt n gt T RATE lt n gt DRYO2 WETO2 NITROGEN INERT HCL PC lt n gt PRESSURE lt n gt F 02 lt n gt F H2 lt n gt F H20 lt n gt F N2 lt n gt F HCL lt n gt C IMPURITIES lt n gt DUMP DUMP PREFIX lt c gt TSAVE lt n gt TSAVE MULT lt n gt B MOD lt c gt p MOD lt c gt AS MOD lt c gt IC MOD lt c gt VI MOD lt c gt NO DIFF REFLOW Description This command specifies diffusion and or oxidation silicidation steps Any impurities present in the wafer are diffused if they have non zero diffusivities The oxidation and diffusion control parameters are contained in the associated METHOD OXIDE and SILICIDE statements Default coefficients are in the ATHENAMOD file available from the DeckBuild Commands menu under Models To change model coefficients refer to the appropriate IMPURITY statement for information Parameters to Define the Diffusion Step TIME specifies the amount of time for the diffusion step in specified units HOURS MINUTES and SECONDS specify the units of the TIME parameter Default is MINUTES TEMPERATURE specifies the ambient temperature in C This temperature should fall within the range between 700 and 1200 C Outside of this range the diffusion coefficients may be inaccurate and numeri
239. X 4 2 4 3 Deposition Model i lt c civiec casing deena Sear er ae ade meas ae ae ate alana 4 4 4 3 1 Conformal Deposition ssi oeubinseberi te eeiinnd wpe binds dewdimecbie edi ew ieetsweeies 4 4 4 3 2 CVD Deposition e derni enri rannesenseauidad edad ch e a pad ebecee ene 4 4 4 3 3 Unidirectional Deposition a5 255 5 nck ioe a een eae eek Seek onthe SENS oad seid eacct ek eM ace Ged 4 4 4 3 4 Dual Directional Deposition tae 8 octet kere Mosier darters aranaren 4 5 4 3 5 Hemispheric Deposition ss Bi tenet FU Me eat lB oh iadie te Bert ise h aa Bl 4 6 A 3 6 Planetary Deposition Sent otal oon cep shoes Lees eet ante hee Soe e ic ieee Sie oe aes 4 7 4 3 7 Conical Deposition nennen ane tna aunt Ae aa ok Sab Rey OS a atone an ee i 4 9 4 3 8 Monte Carlo Deposition uct te ri n Ck ie a Sete cae ee One eee al ead tie 4 10 4 3 9 Custom Deposition MOdelSyv ci 22h dice toasid ews ceden a teadereemeta bine eietia mee Lek 4 11 4 4 Eteh ModelS isinisi inisGuensaedrewkradedaswodateeneeg DWE EE AR e eee ead ee aa 4 12 4 4 1 Isotropic Etch Model esssciccsuensa tins cael ada ta eit ee edad Cate k esd PM RE Med adda Valet 4 12 AA OERIE MOG sctee te ci oie o ee ee elie ee a aa aa EEEE REE ent 4 12 4 4 3 Dopant Enhanced Etching 2 e0ac2esawer ee u att esate aay awe ce eed wate Rha wee ee eee 4 14 4 4 4 Plasma Etch Mod l sasira aisee mre acana peste AEE Aa wicca E E ceeds 4 14 4 4 5 Monte Carlo Etching Model ununun annaa
240. XIDE DIFFUSION is alias for DIFFUSE Note PRINT 1D issues a warning message if there is no SELECT statement prior to it ELIMINATE is not used in ATHENA and therefore is ignored A warning is then issued The RELAX statement should be used instead The RELAX capability is similar but more flexible since it can be used in anywhere in the input deck ELECTRICAL and MOBILITY are not used in ATHENA and therefore are ignored A warning is issued The EXTRACT capabilities of DECKBUILD should be used instead EXTRACT is not used within ATHENA and therefore is ignored DECKBUILD has superior extract capabilities The EQUATION REAC Therefore they are ignored LOADFILE is alias for INITIALIZE TION and INTERMEDIATE statements are not part of ATHENA PLOT in TSUPREMS decks is ignored and TONYPLOT should be used instead PLOT 1D PLOT 2D PLOT 3D CONTOUR LABEL and COLOR statements are depreciated in ATHENA Warnings are issued TONYPLOT should be used instead PRINT in TSUPREMS decks is alias for PRINT 1D SAVEFILE is alias for STRUCTURE Di Silvaco E 6 TSUPREM4 and TSUPREM3 Compatibility Features E 7 Changes in the INITIALIZE statement Boolean parameters lt 100 gt lt 110 gt and lt 111 gt that specify crystalline orientation of the silicon substrate are aliases for ORIENTATION 100 ORIENTATION 110 and ORIENTATION 111 DX RATIO
241. _val Vacancy concentration double CL_val 311 Cluster concentration double DL_val Dislocation loops concentration if mater only in Silicon The function modifies 1 interstitial generation model The interstitials are generated only in unamorphized layer where damage is less than 0 1 of atomic density of Si if implanted_dam lt 5e21 T val implanted_conc return Silvaco A 1 ATHENA User s Manual The function receives the following input values e material index e xandy coordinates of the point in the structure where damage is calculated e implant concentration at this point e implant damage but only when the Monte Carlo BCA model is used for the current implant calculations This allows you to return the values of interstitial and vacancy concentration and concentrations of 311 clusters and dislocation loops If one or few of return values are not modified in the function the corresponding concentrations will remain unchanged after the implant The function then needs to stored as a file i e damage 1ib The model stored in the function can then be activated by specifying the DAM MOD DAMAGE LIB parameter in the IMPLANT statement Note Prior to ATHENA version 5 4 0 R DAMAGEMOD FN DAM MOD was a parameter in the MOMENTS statement When you execute the IMPLANT statement using analytical or Monte Carlo models the specified C Interpreter function w
242. a delves 2 55 6 25 27 MATERIAL 2s scesessvseneeerteesapieithmexteectesseeteette 2 54 6 62 63 METHOD siei eea 2 54 6 64 69 SiGe SiGeC Simulation ccscceceeceeeeseeeeseeeeeeeeeeees 3 95 96 Silicidation Model cccccsscccccesssseeceecssseeeseeeessaeees 3 64 65 Solid Solubility essai ioneina aneneen B 7 8 See also Deactivation Threshold SSUPREM aisnean eeaeee aren inai 2 30 41 3 1 96 Calibrating ATHENA for a Typical Bipolar Process FIOW moscoc aeaaea ai aeaea 2 48 53 Changing the Method Statement During the Process FIOW ccccceeeseseseeeesssssesenaeaeeeeeeeeeeess 2 31 32 Choosing an Appropriate Model sesceeeeeeeeeenees 2 30 31 Modelling the Correct Substrate Depth eeeeeeeees 2 32 38 Multiple Models eeecceceseeeeeeseeeeseneeseeeeeeeeenenesenenesees 2 30 Process SLEDS set ieceaazepskeactcndeacevssanscsnadstgcssnssseaisatveresetse 2 30 Rapid Thermal Anneals RTA cseseeeeesseeeeneeeneetaes 2 38 SiGe Process Simulation cccccecsseeeeseeeeeeeteeeees 2 54 55 Standard Examples ccccccccsseecesseeeeeeeeeeeseeeessneeesseees 2 2 4 Statistical Sampling rare event trajectory splitting technique seeeeeeeeees 3 85 FANG OVOMS 2 222 228 502 255 beves cadet cc Sen caes eecetedecdes etesekicdetadechs 3 85 resta ss oso Soe xen tans celrtater geveec dh E EE cadet 3 85 trajectory splitting w2 023 sece eves eet ees 3 85 Std tables 4 2
243. a tee enw reed teres 2 43 2 5 3 Tuning Implantation Parameters 7s 02ecee yi ay Way eet ee laws eed aes eg bes 2 45 254 TUNING DIMUSION Parameters 10sec etna den crore ree aioe ted pens erehenss Hee Ren Pee Beds 2 46 2 5 5 Related Issues on using the Device Simulator ATLAS for MOS Process Tuning 05 2 46 2 6 Calibrating ATHENA for a Typical Bipolar Process FIOW ccseeee eee eee eee ee een eee 2 48 2 6 1 Tuning Base and Collector Currents All Regions 0c ana 2 49 2 6 2 Tuning the Base Current All Regions oceans cha tied Doon ie ed eae we ay 2 49 2 6 3 Tuning the Collector Current All RegionS ccc cece eee nents 2 50 2 6 4 The Base Current Profile Medium Injection 0 0 cece eect eens 2 51 2 6 5 The Base Current Profile Low Injection 0 0 c cece eee eee eee eee eens 2 52 2 7 Using ATHENA for Simulating SiGe Process 000s cece cece e eee eee eee eee eens 2 54 ZF ASI METHOD St tement sacs i r riria nae le le cede once nade E EEE RETE ete ath ee ey cack 2 54 2 7 2 MATERIAL Statement ienien ee eds bed be eee eked bee ee eh E eds vas ees 2 54 Silvaco v ATHENA User s Manual Deb DEPOSIT SIAlOMENE E EO E E ost oly ahaa he alates E ha AT otal a 2 54 2 7A DIFFUDE Statement 2 aoe alt Late eae ety eed tee aces A A Goee bee G twee ee 2 55 2 8 Using Advanced Features of ATHENA 00 cece e eee e eee e eee eee nn eens 2 56 2 8 1 Struc
244. abilities such as CD extraction for generating SMILE plots This section of the tutorial describes ATHENA OPTOLITH input output and the following basic operations for creating a typical input file for optical lithography e Creating an input mask using MASKVIEWS or the LAYOUT command e Designing custom or standard illumination systems e Projection Fourier plane filtering e Imaging controls e Properties of materials e Structure exposure post exposure bake and development e CD extraction SMILE plots and looping procedures This section of the tutorial assumes that you are familiar with the general operation of ATHENA This includes familiarity with the command language used to generate structures as well as a general knowledge of the use of the VWF INTERACTIVE TOOLS Specific features that refer particularly to OPTOLITH will be explained here 2 9 2 Creating A Mask A mask can be created using the MASKVIEWS tool supported by the VWF INTERACTIVE TOOLS or by using the LAYOUT command MAsKVIEWS facilitates the creation of complicated masks and can import different mask data formats such as the GDS2 stream format In the case of simple masks containing one or two features it may be simpler to use the LAYOUT command MaskViews Once you select MaskViews from the Tools MaskViews menu press the Start MaskViews button and the MASKVIEWS window will appear Then press the Properties button and the MAsKVIEWS Properties popup will appear
245. actions are added A 2 14 v O S yO V P ie P 7 3 117 k3 ky ky Therefore the VC model consists of Equations 3 113 and 3 118 OV m ot GRycim GRyem 1 3 118 According to the following reactions Equations 3 115 and 3 117 there are two kinds of summands in the generation recombination part GRycm K VV m 1 ANTA Ve GRycm n k y k y 3 119 m m m 1 m 1 The equations for vacanies and intersititials will also contain the additional terms from Equation 3 119 Silvaco 3 33 ATHENA User s Manual 3 2 5 Electrical Deactivation and Clustering Models DDC At high doses of dopant the electrically active concentration may be less than corresponding chemical concentration The impurity atom becomes activated inside semiconductor only if it is incorporated into a substitutional lattice site In this case the activated atom will contribute with a carrier to either the valence band an acceptor impurity or the conduction band a donor impurity It has been observed that even below solid solubility a significant dopant concentration can stay inactive This effect can be explained by the formation of immobile dopant defect clusters which is described by the DDC model This is the third part of the PLS model The model strongly depends on the nature of the dopant and therefore is presented separately for each type of dopant below To activate the DDC model add the DDC parameter to the
246. activation energy D 0 and D E can be set in the INTERSTITIAL statement The bulk recombination rate Rg is a simple reaction between vacancies and interstitials that assumes that any interstitial will recombine with any vacancy regardless of their charged states This assumption may overestimate the recombination rate The equation is expressed as Ry KCC Cr Cy 3 27 Silvaco 3 9 ATHENA User s Manual where K is the bulk combination coefficient and specified as 3 28 Ke KR Oexp SE kT where the parameters KR 0 and KR E are user definable in the INTERSTITIAL statement The interstitial trap rate Ry model was first introduced by Griffin 12 to explain some of the wide variety of diffusion coefficients extracted from different experimental conditions The Trap Equation which describes the evolution of the empty trap population in time is Tur K7 CerC E cr Cr Cer 3 29 1 e R where e Cris the total trap concentration e Kris the trap capture rate e Crris the empty trap concentration e Cyis the interstitial concentration e C isthe equilibrium interstitial concentration e e is the equilibrium empty trap to total trap ratio ewe ET Cr Both Kr and e are Arrhenius expressions that can be set in the TRAP statement with the total trap concentration Ry with the parameters shown in Table 3 4 Table 3 4 Parameters for interstitial traps En
247. addition to a distribution of point defects and is usable in a subsequent RTA diffusion step Damage is specified as a profile scaled to an implanted profile Independent vertical and lateral control of the scaled damage is definable CNET Diffusion Models A new series of models from CNET under the guidance of Dr Daniel Mathiot have been implanted and calibrated to better describe high dose effects during diffusion The series of five extra models include Impurity Defect pairing statistics static clustering percolation correlated interstitial amp vacancy mediated impurity diffusivities bimolecular recombination of defects through impurity states Temperature Dependent Fractional Interstitialcy The parameters for fractional interstitialcy Fi have been extended to include temperature dependence Fi 0 and Fi E If Fi is stated it will remain a fixed value Silvaco D 11 ATHENA User s Manual Indium Added as New Dopant Species The Indium dopant species has been included as it has shown promise as a good shallow junction forming alternative to Boron and BF2 implanted species Indium may further be passed though DEvEDIT and into ATLAS as part of the active net dopant calculation Gridding Capabilities Power Device Diffusion Model A new model for power device diffusion has been added This model will run around 4 times faster than the standard fermi model in SSUPREM4 enabling Athena to simulate larger power device structures
248. aims to provide you with a set of rules outlined indicating the correct model that can be used most of the time without you having a detailed knowledge of the physics involved The usual rules of model selection apply here The more complicated the model the greater the simulation time There is always a compromise between simulation accuracy and simulation time The following sections describe when to use the hierarchy of models so that the most complicated models are only used when you make a significant difference to the result 2 4 3 Choosing an Appropriate Model Using the Method Statement The hierarchy of diffusion and damage models available is broadly related to the maximum level of damage already in the semiconductor or the maximum level of damage that the next process step is likely to introduce at any particular time during the process flow The level of damage in the semiconductor at any one time is not a static quantity but will depend on when and how much damage was induced by a process step and how much annealing has occurred in subsequent thermal steps The range of models available to you can account for all of the above effects and allows accurate simulation of dopant diffusion if appropriate models have been chosen The choice of model or combination of models for any of the process steps described above is defined in the METHOD statement The METHOD statement serves a number of functions but in the context of defining damage models
249. ain specifies the type of file An intensity file initialized in this fashion is useful only for exposures that use the vertical propagation model N PUPIL also affects the accuracy of the aerial image calculation A higher N PUPIL value increases the number of source points by a factor 2 N PUPIL 1 squared and increases the accuracy and the computation time Note The Image Window and the Computational Window are not linked The computational window is automatically adjusted to include all mask features unless otherwise specified in the IMAGE command This means that the entire mask will be used in the image calculation You can use the Image Window to calculate a part of the entire image to increase the simulation speed You can override the selected image in the IMAGE command by specifying new window coordinates Silvaco 2 81 ATHENA User s Manual Aerial image intensity distributions can be added together by specifying MULT IMAGE on repetitions of the IMAGE command You can add any number of images together The first IMAGE statement shouldn t contain the boolean parameter MULT IMAGE because the preceding aerial images are erased from memory You can weigh of the aerial images by using the INTENSITY parameter on the ILLUMINATION command ONE DIM is a new parameter that has been added to the IMAGE command It allows you to calculate one dimensional aerial images This is used primarily for increasing speed in
250. al 2 8 3 MaskViews Interface This section describes an alternative to the manual specification of grid and etch steps described in Section 2 3 Creating a Device Structure Using ATHENA Defining Initial Rectangular Grid Using MaskViews An initial rectangular grid can also be defined by using Silvaco s IC layout editor MASKVIEWS MASKVIEWS is designed specifically for interfacing IC layout information with process and device simulators For more detailed information about MASKVIEWS and its interface with DECKBUILD see the VWF INTERACTIVE TOOLS USER S MANUAL VOL I This section gives several practical suggestions on how to prepare a good initial grid for ATHENA With MASKVIEWS you can omit ATHENA mesh definition statements because you can include the gridding information in the layout file When using MAsKVIEws to provide line information DECKBUILD will comment out existing line commands when it loads the MASKVIEws information Load the example 34 11 anex11 in from the MaskViews ATHENA section of the Deckbuild Examples Window See Figure 2 2 Then select MaskViews Starting MaskViews from the Tools menu of DECKBUILD to open MASKVIEWS Layout Files Popup Figure 2 47 Mask Views Layout Files Directory fexport main mishat athman Filter lay myvanexo 1 lay Co hd Filename mvyanexo1 lay Start MaskViews Figure 2 47 MaskViews Layout Files Popup Choose the anex11 1lay layout file from the scro
251. al is elevated but growth rate and oxide shape are not affected In all analytical models the initial silicon surface must be planar The ERFG model simulates the bird s beak oxide shape under nitride masks of different thicknesses 40 The ERFG model consists of two models ERF1 and ERF2 The ERF1 model describes the oxide growth under a thin nitride layer where the stress from the nitride mask layer is negligible ERF2 model describes the oxide growth when nitride layer thicknesses are large enough to cause stress in the oxide which can result in the oxide layer being pinched When ERFG is specified either the ERF1 or ERF2 model will be automatically selected based on the structure under consideration Both models are based on the error function shape of the oxide silicon and oxide ambient or oxide nitride interfaces 3 58 Silvaco SSUPREM4 Models Z Aerfc By C D 3 171 The A B C and D parameters are complex functions of several geometric parameters e initial thickness of oxide 1 and nitride lp e current thickness E of oxide given by the Deal Grove Model Equation 3 132 e the length of lateral oxidation under the nitride layer Lpp e and the lifting of the mask during oxidation H These functions are specified in the OXIDE statement All defaults are taken from 40 3 3 8 Recommendations for Successful Oxidation Simulations Achieving successful oxidation simulations can be a frustrating task for a novice user
252. al to which the parameters apply SILICON POLYSILICO TUNGSTEN TITANIUM PLATINUM COBALT WSIKX TISIX PTSIX COSIX and MATERIAL specify the second material to which parameters apply MTTYPE specifies the type metal or silicide of the user defined MATERIAL IMTTYPE specifies the type metal or silicide of the user defined MATERIAL ALPHA specifies the volume expansion ratio between MATERIAL and MATERIAL Examples The following example specifies the volume expansion between user defined material TiSi2 and standard material titanium SILICIDE MATERIAL TISI2 MITYPE SILICIDE MATERIAL TITANIUM ALPHA 0 4 6 102 Silvaco SOURCE 6 59 SOURCE SOURCE executes statements from the specified file Syntax SOURCE lt filename gt Description SOURCE reads statements from an input file Statements are read from the file until an end of file marker is found SOURCE is especially useful for executing a large group of statements SOURCE places the named file in the current input stream SOURCE statements can be nested up to the limit of open file descriptors system dependent Examples The following statement causes the contents of a file named test in to be included into the input stream SOURCE TEST IN Note To support the use of this function when running under the VWF AUTOMATION TOOLS place the file to be sourced i
253. alculated by OPTOLITH to a file called test str STRUCTURE OUTFILE TEST STR INTENSITY MASK The following statement mirrors the structure about its left boundary STRUCTURE MIRROR LEFT Silvaco 6 107 STRUCTURE ATHENA User s Manual Note The STRUCTURE command will only save all mesh and solution information It will not save any defined model or machine methods If you exit a simulator the middle of an input file you may need to manually parse the preceding METHOD and IMPURITY commands to reinitialize specified parameters This function is handled automatically when running under the VWF AUTOMATION TOOLS For more examples see INITIALIZE 6 108 Silvaco SYSTEM 6 64 SYSTEM SYSTEM allows execution of any UNIX C shell command within an input file Note The SYSTEM statement is executed by DECKBUILD and is fully documented in the VWF INTERACTIVE TOOLS USER S MANUAL VOL I Note The SYSTEM command must be enabled using an option on the DECKBUILD Main Control menu under Category Options Examples The following command will remove all files named test str before a DIFFUS the DUMP parameter is used system rm rf test str DIFFUSE DUMP 1 DUMP PREF test The SYSTEM command and UNIX commands are case sensitive E statement where UNIX commands can be concatenated on a single line using the semicolon operator For examp
254. ald ripening phenomena e The third step characterized by the supersaturation collapse is explained by the entire dissolution of the IC population due to the recombination at the surface It is clear that the CDD model alone dashed lines in Figure 3 5 cannot reproduce these three steps of the transient enhanced diffusion since the curves are monotonically decreasing Silvaco 3 39 ATHENA User s Manual Pelaz Experiment In this experiment a boron marker is deposited at a depth of 0 15 um To observe the boron diffusion Pelaz et al have performed a silicon implant to generate a high interstitial concentration at the surface Unlike the Cowern experiment the boron concentration is high enough to allow the formation of BICs Thus this experiment exhibits a particular effects of in the boron diffusion an immobilization and disactivation of the dopant at high concentrations even under the solid solubility limit Figure 3 6 demonstrates simulations of this particular experiment using various parts of the PLS model Tony Prot V28 14 A lt 2 gt File vew Plot Tools Prints Properties Help Pelaz s experiment As grown boron Silicon implant furnace annealing As grown Annealed at 800 C 35min Simulation As grown Simulation with full PLS model east Total Boron Active Boron Simulation with clasical IC model Simulation with classical model a 2 e Q Z g e O 4 4 4 4 0
255. ames of the Silvaco products ATHENA and ATLAS Silvaco Table of Contents Chapter 1 INTOGUCUON ai tvs wie deanna Cee ee eae ewe dee beens 1 1 tA Athena Ov rview isise i aee a eeewed tin as ba ease nl eee Seed eet 1 1 Tete Wings PHISIMANUAlt soiin eS capoeira rte ec hacer treed ere lait dat tate a AA tol 1 1 dNe2s R e olere Eeli o olo a reee en ee eh Se fa ate Oe ata et oi Oa eed aul ale aC gh 1 1 1 2 Athena Features and Capabilities 0 0 0 0 cece cece e eee eens 1 2 1 2 1 Using ATHENA With Other SILVACO Software 0 0 0 c cece eee eee eee 1 3 1 2 2 The Value Of Physically Based Simulation 0 00 cece cece ened 1 4 Chapter 2 Tutoriales cud oct ce itch Aes hat a aa weaned ee E E a E ees 2 1 2 1 Getting Started nc seei iener ea teats we AADO E ote DIDATE E ceases ARA EBA spene s 2 1 2 1 1 Running ATHENA Under DeckBuild aaaea 2 1 2 1 2 Loading And Running ATHENA Standard Examples 00 e cece eee eee eee es 2 2 2 2 Op ration Modes r a lak sees alan ee Mates sate theteate a nets Wee ciate estates 2 5 2 2 1 Interactive Mode With DeckBuild 00 e cece eee eens 2 5 2 2 2 Batch Mode With Deckbuild sce Sct ah canoes Kinks Perl Sas oie bute 8 De Sere hee eet 2 5 2 2 3 No Windows Batch Mode With Deckbuild 0 0 00 cece eee eee eens 2 5 2 2 4 Running ATHENA inside DeckBulld 20 3305 o00is serad erie bia beets Denia Peed eee dag ddeed 2 6 2 3 Creating a Device Structu
256. ameter MAX DAMAGE on the IMPLANT or MATERIAL statements also controls the rate at which the implanted material will amorphize e The silicide model has been enhanced to improve volume conservation during silicide calculations Parameters DSV 0 and DSV E have been added to the SILICIDE statement to control the dissolu tion of a contributing material during the silicide calculation e Improvements to the TWO DIM model and cylindrical coordinates to address bug fixes and model extensions have been included Silvaco D 21 ATHENA User s Manual D 14 3 ELITE Capabilities Reflow capabilities that allow spin on glass modeling with a physically based calculation that simul taneously calculates impurity diffusion are now included Reflow capability is now available with ELITE for individual materials by specifying the REFLOW parameter on the MATERIAL state ment Specifying the REFLOW parameter on the DIFFUSION statement invokes the reflow model The VISCOUS model should be selected on the METHOD statement prior to performing reflow The parameter GAMMA REFLOW lt n gt has been added to the MATERIAL statement to specify surface tension sigma for the reflow calculation When used in conjunction with either SSUPREM4 or FLASH the reflow capability allows simultaneous calculation of material flow and impurity diffu sion Monte Carlo deposit capabilities are now available as an optional functionality These allow physi cally based calculations tha
257. and C20 are described in Table 6 3 Coefficients for fifth seventh and ninth order aberrations must be entered in separate ABERRATION commands for each order Each of these parameters represents a particular aberration coefficient depending on the order specified by parameters FIFTH SEVENTH or NINTH Table 6 3 Aberration Coefficients Parameter Fifth Seventh Ninth C1 4C20 6C20 8C20 C2 2c40 4c40 6C40 C3 0C60 2C60 4c60 C4 5c11 0c80 2C80 Silvaco 6 9 ABERRATION ATHENA User s Manual Table 6 3 Aberration Coefficients Parameter Fifth Seventh Ninth C5 3C31 7C11 0C100 C6 1C51 5C31 9C11 C7 4C22 351 7C31 C8 2C42 1C71 5c51 c9 3033 6C22 3c71 C10 4C42 1C91 cii 2C62 8C22 C12 533 6C42 C13 3C53 4C62 C14 4c44 2C82 C15 7033 C16 5C53 C17 3C73 C18 6C44 c19 4c64 C20 555 Examples If high order aberrations are to be studied they must be entered on a separate command line for each order ABERRATION X FIELD 5 SPHERICAL 25 ABERRATION FIFTH C1 25 C2 5 ABERRATION SEVENTH Cl 3 C4 4 E For more examples see IMAGE ILLUMINATION PROJECTION ILLUM FILTER PUPIL FILTE LAYOUT z w 6 10 Silvaco ADAPT MESH 6 4 ADAPT MESH ADAPT MESH enables the adaptive meshing algorithm Syntax ADAPT MESH SMOOTH SMTH COU
258. and Image planes are totally independent There is no mesh in the object or reticle plane source condensor reticle projection aperture projection image plane lens stop lens Figure 5 2 The Generated Optical System 5 3 1 Discretization Errors The size of the window in the reticle plane is determined by the number of mesh points in the projector pupil the numerical aperture and by the chosen wavelength CW NP lambda NA 5 23 where e CW is acomputational or sampling window mask or image cell in the object or reticle plane e NP is the number of mesh points in the projector pupil e NA is the numerical aperture of the stepper e lambda is the chosen wavelength For an i line stepper with NA 0 54 the size of the sampling window is the square whose side length is equal to 6 8 um 10 0 365 0 54 No mask feature should exceed this dimension You can increase the size of the sampling window for this particular stepper to any size by increasing the number of mesh points in the projector pupil This will be done automatically to accommodate the mask and image windows that were specified Mask features cannot be placed outside of the sampling window As mentioned earlier the image mesh is totally independent of the mesh in the Fourier plane This allows you to arbitrarily specify the number and distance of image points 5 3 2 Mesh The size of the computational window is determined by Equation 5 23 and the position o
259. and anneals 2 10 2 Interface Mesh Control The Interface Mesh Control is used to control the mesh in the vicinity of a material interface This function allows you to add grid lines for example to run along under the gate of MOSFET at some distance from the Si Si02 interface The Interface Mesh Control is often useful for adding mesh as required by highly mesh dependent mobility models during a following device simulation It is also useful to be able to add mesh for better segregation modeling The ADAPT ADD I LINE n command controls the addition of a new mesh line Two materials are specified as parameters to the command defining an interface or a set of interfaces The mesh line is added to MATERIAL1 as follows ADAPT MESH ADD 1I LINE 0 001 MATERIAL1 MATERIAL2 For example in the case of adding an additional mesh line to the SILICON in the channel region of a MOSFET ADAPT MESH ADD I LINE 0 001 SILICON OXIDE The structural transition from 1D to 2D to create a base mesh is controlled by the BASE PAR parameters Figure 2 68 indicates the flow of events towards the formation of a base mesh and beyond in the case of MOSFET device Silvaco 2 89 ATHENA User s Manual Commands Action on Structure Method ID Adaptive Meshing Simulation Define Mesh Rules for ID to 2D Transition Switch from ID to 2D Simulation Adapt Par Modify 2D Adaptive Meshing Criteria Simulat
260. ant The implant creates some interstitials but also creates 311 defect clusters These clusters decay with time releasing point defects over an extended period of time This effect is particularly apparent at low temperatures Clearly then a key parameter for tuning RTA effects is the time constant for the dissolution of 311 clusters to interstitials This is controlled by the syntax CLUSTER SILICON TAU 311 0 lt val gt TAU 311 E lt val gt Measured data 128 shows that the enhanced diffusivity due to point defects extends over minutes at 800C Figure C 9 shows ATHENA results matched to the measured data in Figure C 10 of 128 In this case the value of TAU 311 0 is adjusted to show lower diffusion in the first 15 seconds than the FULL CPL model predicts For comparison a lower value of TAU 311 0 is used in Figure C 10 It is clear that this does not match the data in 128 as a significant part of the complete diffusion is in the first 15 seconds C 12 Silvaco Hints and Tips File View Ploty Tools Print Properties Help ATHENA RTA SIMULATION MATCHED TO EXPERIMENTAL DATA IN 3 20 F lt As Implanted RTA15 sec RTA1 min RTAS min furnace anneal 60 min Phos Cone cm 3 SILVACO International Figure C 9 RTA of a 5 0e13 phosphorus implant matched to experimental data in 128 File View v Ploty Tools Print
261. ant the units are cm sec and D E is the activation energy the units are eV CSTAR O and CSTAR E specify of the total equilibrium concentration of interstitials or vacancies in intrinsically doped conditions CSTAR O is the pre exponential constant the units are cm and CSTAR E is the activation energy the units are eV NEU 0 NEU E NEG 0 NEG E DNEG 0 DNEG E POS 0 POS E DPOS 0 and DPOS E specify the relative concentration of interstitials or vacancies in the various charge states neutral negative double negative positive double positive under intrinsic doping conditions All 0 parameters are unitless All E parameters are in eV 6 54 Silvaco INTERSTITIAL and VACANCY Bulk Defect Recombination Parameters KR 0 and KR E specify the interstitial or vacancy bulk recombination rate KR O is the pre exponential constant the units are cmsec and KR E is the activation energy in eV IVFACTOR and ITFACTOR specify I V Bimolecular recombination ratios in HIGH CONC model These parameters are valid only for the INTERSTITIAL statement KTRAP O and KTRAP E specify the interstitial trap reaction rate KTRAP 0O is the pre exponential constant the units are cm sec and KTRAP E is the activation energy in eV Note At present it is very difficult to extract exact values for these parameters The default values assume the trap reaction is limited by the interstitial concentr
262. anual Parameters Related to Numerics of Diffusion Oxidation IMPURITY specifies impurity for which one or several bound tolerance parameters will be applied during diffusion oxidation simulation see Section 6 2 10 Standard Impurities for the list of impurities INTERST VACANCY OXIDANT VELOCITY TRAPS PSI and PAC specifies type of solution for which one or several bound tolerance parameters will be applied during diffusion oxidation simulation REL ERR indicates the precision with which the impurity solution must be solved In general the actual error will be less than half of the indicated error The defaults are 0 01 for all impurities except the potential which is solved to 0 001 If this parameter is used an impurity should also be specified ABS ERR specifies the error tolerance absolute value For dopants the absolute error defaults to 1 0x10 For defects the absolute error defaults to 1 0x10 For the potential the error defaults to 1 0x10 6 If this parameter is used an impurity should also be specified FE RELERR and FE ABSERR specifies the relative error and absolute errors for the FERMI model TD RELERR and TD ABSERR specifies the relative and absolute errors for the TWO DIM model ST RELERR and ST ABSERR specifies the relative and absolute errors for the STEADY model FU RELERR and FU ABSERR specifies the relative and absolute errors for the FULL CPL model MIN FILL and MIN FRE Q specify a minimum fill
263. any implant depo or profile statement Now only newly added impurities are activated completely D 10 ATHENA Version 4 0 0 R Release Notes D 10 1 SSUPREM4 Diffusion Simulation Features Physical RTA Model A new TED model including the dynamic transient release of interstitial point defects has been added to SSUPREM4 lt 311 gt Clusters release Interstitials over time with a user defined time constant This model was derived from Dr Peter Griffin work at Stanford Dislocation Loop based point defect sink model A dislocation loop based interstitial sink model is now included for high dose RTA situation and may be used in conjunction with the lt 311 gt Cluster model This model was derived from the work of Dr Peter Griffin at Stanford University Point Defect Dopant Pair Recombination Capture Cross Section Control To account for high concentration effects extra terms have been added to the fully coupled diffusion model allowing for higher order dopant point defect dopant pair recombination Recombination may be controlled independently both in the bulk and as an extended surface recombination velocity This model was derived from the work of Dr Peter Griffin at Stanford University Extended Defects Extended defects may now be in introduced during Ion Implantation Both lt 311 gt Clusters and Dislocation Loops may be introduced during ion implantation along with an overlying amorphous region This damage may be introduced in
264. araea arera ea aea a e RAE EEA AREE aae 2 7 D FUTET NAITO MAAAR NE E EAE TA 2 4 2 7 6 5 athenamod 97 AAEE EET EET 2 6 Daage Amorphization Models eee ee ee gK 25 See also Implant Damage athenares a e dann in ti ane ppc ASONO fescetatstciy ne cevecsttn de bias ela tee teeooeeee tte 3 35 36 B BONN sete aaa io Geet 3 33 34 Bank Weiser Error Estimator cccccccccccccccecececcececcceceueuens 2 88 Phosphorus EPAPER E T E PRE E A T T setae 3 35 Basic Diffusion and Oxidation Models cccccccccccececeececceeeee 2 40 Deactivation ThreShold eeeseeessseeesseeessneeeeeneeeeeaes 3 18 19 Deal Grove Model scccescesseeeeeeeeseeseaeeeeeeteaeesseeeeaeeeaes 3 59 Silvaco Index 1 ATHENA User s Manual DeckBuild iv ntact Yee ieee het iis 2 1 2 6 100 6 109 Etch Mode Sareren eae eaaa neari aiaiai 3 92 4 12 19 Batch Modari a a eterna eos 2 5 ISOMOPIC i m unr aa Teh E 4 12 Interactive Mode cceccceseceeeeceeeesceeesseeeneseeeeeeeneneeeenens 2 5 Linear viet ie a ee al eit 4 18 No Windows Batch Mode ssceseseeeessereeeeneeeeeneeees 2 5 6 Monte Carosse Seaee tasted n aiana iedots aeons 4 14 19 Running ATHENA inside s an 2 6 Plasmatvesisoieth vindictive AR eet od 4 14 defect saiit T n a e act Ra Rage 3 1 PIE sssesseesseesteeseesssesneessesseeesecsnecsseeseesnsesseenscesaeanaes 4 12 13 Defect Diffusion Exposure Modulo 22 avse unveil ei ayavesaned 5 9 11 Time Step Control oi setensiceciene cet siet
265. arameters for the DEVELOP command are TIME in seconds STEPS and SUBSTEPS Silvaco 2 83 ATHENA User s Manual TIME is the total development time STEPS specifies the number of times the structure has to be regridded SUBSTEPS is the total number of times the development line has to be moved Each substep is performed for a time increment equal to TIME STEPS SUBSTEPS After each regridding of the structure you can dump out a standard structure file to show the progress of the development To do this specify the parameter DUMP 1 To name the structure file to be dumped specify DUMP PREFIX lt name gt and the structure will be created in the local directory with the name lt name gt str where is the current development time Post Development Bake A physically based reflow of the developed photoresist is available Specify it by using the BAKE command and the boolean parameter REFLOW along with TIME and TEMPERATURE 2 9 8 CD Extraction Smile Plots And Looping Procedures CDs are extracted from the structure using the function MAT1 MAT2 y This gives the horizontal intersection of material number 1 and material number 2 at the value y To extract a CD from a profile the following format is used GAS PHOTO 1 4 PHOTO GAS 1 4 This will give the CD at the horizontal line y 1 4 To generate swing curves use the FOREACH and
266. are implemented in order to achieve better compatibility with TSUPREM4 and TSUPREMS simulators E 1 General Syntax Capabilities e Added capability to specify that default values of some parameters correspond to those of TSUPREM4 The modified keyfile athenakey tma with some modified default values is introduced To run ATHENA with default parameters specified in athenakey tma file the syntax is go athena simflags tma For example TSUPREM4 defaults for TILT and ROTATION parameters in the IMPLANT statement are 0 while ATHENA uses 7 and 30 respectively e A plus character can be used as a line continuation sign instead of standard backslash e Boolean parameters can be set to false by preceding the parameter name with or character e The S character can be used to specify the comment line This should only be used at the begin ning of the line because the character can be used for substitution of parameters defined by SET or DEFINE statements of DECKBUILD e The maximum length of parameter names has extended from 12 to 16 characters some TSUPREM3 names are longer than 12 characters e The first character of a parameter name can be a numeral now E 2 Execution Control Capabilities Provided by Deckbuild The detailed description on these new functionalities will be published in the DECKBUILD manual Here we highlight only key features related to the compatibility issues E 2 1 DEFINE State
267. ary during the RTA diffusion and therefore affects the total diffusion of dopant into the single crystalline part of the emitter and the base width doping profile The second process parameter affects dopant pile up at the poly silicon silicon boundary and therefore the source doping concentration at the mono crystalline interface Once again this will affect the overall doping profile of the emitter in the mono crystalline region of the device 2 52 Silvaco Tutorial The third process parameter affects the velocity of transport of dopant across the polysilicon silicon boundary with similar effects to the parameters above You can use these parameters to tailor the emitter doping profile in the mono crystalline silicon region to match available measured data usually in the form of SIMS or capacitance information An accurate profile of dopant in the poly silicon part of the emitter is not too important if measured data concerning interfacial dopant concentrations is available This is because the work function of the poly emitter will be set in ATLAS by defining the poly emitter as an electrode All you need to calculate the correct work function at the poly silicon emitter is the interfacial doping concentration at the poly silicon silicon interface on the poly side of the junction See the Poly emitter work function Section on page 2 51 for setting the correct work function for the poly emitter Conclusions By using a logical c
268. as solid interface the surface of the silicon if exposed and solid solid interfaces have been strictly modelled within ATHENA Effects such as dopant loss from exposed silicon and dopant pile up at interfaces are simulated Silvaco 3 4 ATHENA User s Manual 3 1 1 Mathematical Description The mathematical definition of a diffusion model includes the following specifications for every diffusing species present e a Continuity Equation often called a Diffusion Equation e one or more flux terms e aset of boundary and interregional interface conditions In the case of impurity diffusion in semiconductors we need a set of equations for each dopant present and for each type of point defect if point defects are explicitly represented in the model Since dopants can only diffuse as participants in dopant defect pairs the dopant continuity equation is actually a continuity equation for defect dopant pairs The formulation of the continuity equation have a number of built in assumptions e Electronic processes take place on a time scale which is much smaller than the time scale of all other processes adiabatic approximation e The pairing reaction between dopants and defects is assumed to always be in equilibrium This may not be the case especially at a low temperature but would pose a much harder and more CPU intensive numerical problem to solve e Mobile dopants are electrically active and vice versa Models that explicit
269. at dictates the mesh should be mirrored about the x axis DEPTH STR and WIDTH STR specify the depth and width of the initial substrate structure dimension for use with the Process Adaptive Meshing algorithm Units are microns Example Starting from a file The following statement reads in a previously saved structure from the TEST STR file INITIALIZE INFILE TEST STR Example Using an GaAs Substrate The following statement creates GaAs substrate doped with Selenium concentration of 1x10 cem INITIALIZE GAAS C SELENIUM 1E15 For more examples see BOUNDARY LINE REGION STRUCTURE and BASE MESH Silvaco 6 53 INTERSTITIAL and VACANCY ATHENA User s Manual 6 31 INTERSTITIAL and VACANCY INTERSTITIAL specifies coefficients of interstitial diffusion recombination and generation VACANCY specifies coefficients of vacancy diffusion recombination and generation Note These two statements are almost equivalent Most parameters that exist in the INTERSTITIAL statement are also on the VACANCY statement INTERSTITIAL VACANCY ATERIAL D 0 lt n gt D E lt n gt CSTAR 0 lt n gt CSTAR E lt n gt NEU 0 lt n gt NEU E lt n gt NEG 0 lt n gt NEG E lt n gt DNEG 0 lt n gt DNEG E lt n gt POS 0 lt n gt POS E l
270. ate as a function of the points that are above point i For a flat surface K A S 1 Following the work of Warnock these three factors are calculated using the following set of equations The shadow factor is the one for flat surfaces But it is generally calculated basedon one or two points that shadow point i and is given by the equations below Az E encanto Saran a Azz gt 0 0 so S gt 1 Az is obtained by integration over the surrounding topography Light F Ag Secon cara 4 36 z 12 cosh GT FAC left In these equations ij 4 and i n refer to the two points that can possibly shadow point i The effect of these shadow points depends on the two parameters LENGTH FAC and HEIGHT FAC as shown in the equations The variable z is the vertical distance between the point i and the point ijoq i ignz The variable r is the horizontal distance between the point i and the point izeft iright The acceleration factor A is given by the equations below A is calculated for the two points that shadow point i In this manner multiple shadowing effects are taken into account through the term Ai left Airight This is the acceleration factor for the point s that shadow If point i shadows some other point j in the system it will increase A by a similar factor This increase is then passed on to Ajjeq Aten Auten 4 Bi d 148 4 37 ileft Aivignt Airigh 4i B U 16 4 38 The constant Bijop
271. ate gradient in the y direction log Logarithm log10 Logarithm base 10 mat1 mat2 Returns the y value of the interface between mat 1 and mat2 along a vertical slice at the given location mat1 mat2 Returns the x value of the interface between mat 1 and mat2 along a horizontal slice at the given location scales Scales the value given by the maximum value sqrt Square root xfn Takes y and z and finds a matching x yfn Takes x and z and finds a matching y afn Takes x and y and finds a matching z Examples PAR1 lt n gt PA PA PA R1 is a required numeric valued option PARI 4 0 EXP 2 0 8 62E 5 1173 0 R1 is a required numeric valued option assigned a real number expression PAR2 lt c gt R2 is an optional character variable For further examples of expressions see SET and EXTRACT in VWF INTERACTIVE TOOLS USER S MANUAL VOL I 6 1 5 Command Line Parsing ATHENA supports expressions on the command line For example DIFFUSE TIM E 10 60 TEMP 1000 Be careful when using parentheses as the precedence of arithmetic operators as in programming languages is not guaranteed in all cases Silvaco 6 3 ATHENA Statements List ATHENA User s Manual 6 2 ATHENA Statements List This chapter contains a complete description in alphabetical order of every statement used by any of the ATHENA products The follo
272. ate teed sneaks LEO bas san A DAS D 26 D 20 Additional SSUPREM4 Changes cccecee eee e eee eee eee eee ene eeeeeeeeeeaee D 27 D 20 1 Oxidation method defaults to CompreSS 00 c cece eee ene eens D 27 Appendix E SUPREM4 and TSUPREM3 Compatibility FeatureS ccceeeee eee eeeeees E 1 E T General Syntax Capabilities sssic cic pavatee tite te ete eee bee ewe Hed ee eet E 1 E 2 Execution Control Capabilities Provided by Deckbuild 0ce cee eee cece eee eee eee eens E 1 E 2 1 DEFINE Statement and Substitutions Capability 0 0 tenes E 1 E 2 2 IF ELSEIF ELSE IF END Capability eu iecee any heeta ee eid e le pheasant ed ey E 2 E 2 3 LOOP L END ASSIGN L MODIFY Capability nanana E 2 E 3 MESH Statement iii ericysiicssiirriss nrd adu nerona ENEA AT EE EEEE EE E 3 E 4 Using MASK statement with the parameter IN FILE and XLINES for Automatic grid generation in the horizontal direction s erisicsrirriserrererrs iorri Sa bee dks EON ea E 4 E 5 Using mask information with the EXPOSE MASK lt maskname gt statement 000eeeeeeeee E 5 E 6 Aliases and substitutions for some statements 00 cece eee eee e eee eee eens E 6 E 7 Changes in the INITIALIZE statement 00 cc cece eee e ete eee E 7 E 8 Changes in the DEPOSIT statement 00 ccc cece eee e eee eee eens E 7 E 9 Changes in the DIFFUSE statement 0 ccee cece cece eee eee eee eee eee eeeennenee E 8 E 10
273. atements are similar to those used for the ELITE deposition RATE DEPO and DEPOSIT and the ELITE etching RATE ETCH and ETCH The RATE POLISH statement sets up the parameters for a particular machine while the POLISH statement executes the actual polishing step using the machine 4 6 1 Hard Polish Model The Hard Polish Model 98 simulates the grinding down of the topography based on a rate calculated as a function of the pattern factor Pf of the surface The higher the pattern factor the lower the polishing rate Use the following formula R x y MAX HARD 1 Pf MIN HARD Pf 4 31 The hard polish model parameters are MAX HARD and MIN HARD Pf 1 corresponds to a flat surface Pf is calculated from the topography by the formula Pf ETN a 4 32 max where x y are points on the polished material surface y denotes both y and y _1 Y max is the highest point of the structure and AY is the rate effective height calculated by the previous rate multiplied by the current time step value For Figure 4 12 a pattern factor will be as follows AX AX x total Pf 4 33 The rate for points at height Y ay are equal to the R calculated in Equation 4 31 Points below Yax have a rate cause the structure to polish to the y coordinate Ymax AY Therefore the structure becomes more planar as shown in Figure 4 13 Silvaco 4 21 ATHENA User s Manual Figure 4 12 I
274. aterials e Universal tilt and rotation capability for both analytic and Monte Carlo calculations Oxidation e Compressible and viscous stress dependent models e Separate rate coefficients for silicon and polysilicon materials e HCL and pressure enhanced oxidation models e Impurity concentration dependent effects e Ability to simulate the oxidation of structures with deep trenches undercuts and ONO layers e Accurate models for the simultaneous oxidation and lifting of polysilicon regions Silicidation e Models for titanium tungsten cobalt and platinum silicides e Experimentally verified growth rates e Reactions and boundary motion on silicide metal and silicide silicon interfaces e Accurate material consumption model 1 2 1 Using ATHENA With Other SILVACO Software ATHENA is normally used in conjunction with the VWF INTERACTIVE TooLs These tools include DECKBUILD TONYPLOT DEVEDIT MASKVIEWS and OPTIMIZER DECKBUILD provides an interactive run time environment TONYPLOT supplies scientific visualization capabilities DEVEDIT is an interactive tool for structure and mesh specification and refinement and MASKVIEWS is an IC Layout Editor The OPTIMIZER supports black box optimization across multiple simulators For more information about VWF INTERACTIVE TOOLS see the VWF INTERACTIVE TOOLS USER S MANUAL VOLUMES 1 and 2 ATHENA is also frequently used in conjunction with the ATLAS device simulator
275. aterials e g an interstitial on a silicon oxide interface Interface conditions Between any two regions there must be some control on how any impurity species can exist in the vicinity of the interface For every such interface you must specify a Concentration Jump Condition and a Flux Jump Condition The Concentration Jump Condition accounts for discontinuities in particle concentrations across interfaces and encompasses particle transport across material interfaces due to different solid solubility ratios of the impurity species in the two materials The Flux Jump Condition enables the formulation of interface source and sink terms such as surface recombination particle injection and particle pile up at a moving interface For all species no flux boundary conditions are employed on the sides and at the bottom of the simulation structure This is hardwired into the software which means it is not user definable 3 4 Silvaco SSUPREM4 Models 3 1 2 The Fermi Model The Fermi Model assumes that point defect populations are in thermodynamical equilibrium and thus need no direct representation All effects of the point defects on dopant diffusion are built into the pair diffusivities The main advantage for using the Fermi Diffusion model is it will greatly improve the simulation speed since it does not directly represent point defects and only needs to simulate the diffusion of dopants Also the Fermi Model usually results in a
276. ation DAMALPHA specifies the interstitial recombination rate in the dislocation loop region The units are sec Interface Defect Generation and Recombination Parameters MATERIAL specify MATERIAL2 for setting generation and recombination parameters on the boundary between two materials see Section 6 2 9 Standard and User Defined Materials for the list of materials TIME INJ GROWTH INJ and RECOMB specify the type of reactions occurring at the specified interface The TIME INJ parameter means that a time dependent injection model should be chosen The GROWTH INJ parameter ties the injection to the interface growth velocity The RECOMB parameter indicates a finite surface recombination velocity KSURFE 0 KSURF E KRAT 0 KRAT E KPOW 0 and KPOW E specify the interstitial or vacancy surface recombination rate KSURF 0 is the pre exponential constant for surface recombination rate under inert conditions the units are cm sec KSURF is corresponding activation energy in eV KRAT 0 is the pre exponential constant for the growth rate dependent component of the surface recombination rate unitless KRAT E is the corresponding activation energy in eV KPOW 0 is the pre exponential constant of the power parameter in the surface recombination rate formula unitless and KPOW E is the corresponding activation energy in eV VMOLE THETA 0 THETA E GPOW 0 and GPOW E specify interstitial or vacancy generation parameters of the
277. ation allows the user to see effects due to lines perpendicular and parallel to the current cross section being studied but uses a two dimensional array of plane waves in the calculation The two dimensional mode requires much longer calculation time The one dimensional calculation uses D 22 Silvaco ATHENA Version History only lines which are perpendicular to the cross section This calculation uses only a one dimensional array of plane waves and is much faster When a two dimensional mask is defined only mask fea tures that are on the same level as the desired cross section are included in the calculation e The POSTBAKE statement has been replaced by the BAKE statement that performs either post exposure bake or post development bake TIME and TEMPERATURE parameters have been added to be used instead of diffusion length Associated photoactive component diffusivity parameters are also included in the RATE DEVELOP statement A new post development bake capability includes photoresist flow The REFLOW parameter on the BAKE statement invokes the material flow model e The BAKE statement includes the DUMP and DUMP PREFIX parameters that allow movies of bake processes to be created Setting DUMP 1 and DUMP PRE test will create a sequence of Sil vaco standard structure files that show the time evolution of the structure during the bake The files will be named test str where the indicates the time within the bake e The librar
278. ation and to select the desired diffusion and oxidation model complexity Appropriate defaults for the numerical parameters are included in the athenamod file so that you only need to specify the desired diffusion and oxidation model The numerical methods used in ATHENA for the solution of the diffusion equations are described in 7 Parameters Related to DIFFUSION models FERMI TWO DIM STEADY and FULL CPL specify the type of diffusion equations to be solved with particular regard to the point defect models see Chapter 3 SSUPREM4 Models Section 3 1 Diffusion Models The F ERMI parameter specifies the defects are assumed to be a function of the Fermi level only The TWO DIM parameter specifies that a full time dependent transient simulation should be performed The s EADY parameter specifies that the defects are assumed to be in a steady state The FULL CPL parameter specifies that full coupling between defects and dopants should be included The default is FI TWO DI alias for TWO DI M PD FUL M ERMI PD FERMI L is an alias for FULL CPL PD FERMI PD FULL is an alias for FULL CPL is an alias for FERMI is an alias for FI ERM PD TRANS is an alias for I PD TRANS is an Silvaco METHOD Note Chapter 2 Tutorial shows a complete description of the use of these diffusion models for typical applications PLS IC VC
279. ature is used if the temperature isn t specified in the ETCH or DEPOSIT statement The final stresses from the previous step is used as a initial condition for the subsequent step If the temperature is changed between the end of one step and the start of another the stress calculation with corresponding temperature ramp is automatically inserted If stresses are calculated using the STRESS statement or during oxidation with VISCOUS model or during REFLOW simulation then stresses previously calculated by the STRESS HIST model are ignored and new stresses are computed Silvaco 3 97 ATHENA User s Manual This page is intentionally left blank 3 98 Silvaco Chapter 4 ELITE Models 4 1 Overview The ELITE module of ATHENA allows the use of sophisticated models for deposition and etch processes These processes are modeled by defining a machine and invoking the machine to perform either deposit or etch ELITE also includes a model for material reflow ELITE can also be licensed with modules for Monte Carlo deposition Monte Carlo etching and Chemical Mechanical Polishing CMP In ELITE a number of default machines are defined so that specifying any process reasonably close to the standard is especially simple Process modifications or additions are easily implemented by changing or adding individual machines without affecting the remainder of the simulator For all models except Monte Carlo deposition a
280. atures of TONYPLOT for graphical display and analysis consult the TONyPLOT chapter of the VWF INTERACTIVE TOOLS USER S MANUAL VOL I 6 110 Silvaco TRAP 6 66 TRAP TRAP sets the coefficients of interstitial traps Syntax TRAP MATERIAL ENABLE TOTAL lt n gt FRAC 0 lt n gt FRAC E lt n gt Description This statement allows you to specify values for coefficients of the interstitial traps The statement allows coefficients to be specified for each of the materials ATHENA has default values only for silicon Polysilicon parameters default to those for silicon MATERIAL specifies the material for which the parameters apply see Section 6 2 9 Standard and User Defined Materials for the list of materials ENABLE indicates that traps should be enabled in the material specified TOTAL specifies the total number of traps in cm The default for silicon is 5 0x101 cm This value is appropriate for Czochralski silicon material FRAC 0 and FRAC E allows the specification of the equilibrium empty trap ratio Examples 017 The following statement turns on interstitial traps and sets the total to 5 0x10 and the fraction to a half TRAP SILICON TOTAL 5 0E17 FRAC 0 0 5 FRAC E 0 0 ENABLE Note The trap concentration depend upon the thermal history of the wafer starting material stress and temperature This history is not considered in the trap model in ATHENA
281. ave been included into ATHENA They are AlGaAs InGaAs SiGe and InP These materials are accessible via the INITIALIZE or DEPOSIT statement by specifying ALGAAS INGAAS SIGE or INP The fractional components of the elements can be entered via the parame ter C FRAC on either the INITIALIZE or DEPOSIT statements The DEPOSIT statement also allows a linearly graded variation in the fractional components by use of C FRAC as the fractional component of the first element ie for ALGAAS Al is the first component at the bottom of the deposit and C FINAL as the fractional component of the first element at the top of the deposit The fractional component of the second component i e for ALGAAS Ga is the second component is 1 C FRAC and 1 C FINAL These materials are also available on other statements such as STRETCH ETCH etc Ten more user materials were added to make a total of 20 user definable materials The parameters DONOR and ACCEPTOR have been added to the IMPURITY statement This allows an impurity to be specified as either donor or acceptor for a given material Active impurities are now part of the output file as well as chemical impurities Donors and accep tors are calculated from the active impurity concentration All impurity data can be entered via the IMPURITY statement The old statements BORON ARSENIC PHOSPHORUS and ANTIMONY can still be used as before as they are aliased to the IMPURITY statement Due to numerous additions
282. ave only regional and material information Click on the Plot menu button and the Display 2D Mesh popup will appear Select only the two left icons Mesh and Edges and the Initial Triangular Grid Figure 2 11 will appear in TonyPLoT Silvaco 2 15 ATHENA User s Manual TonyPlot 2 2 1 File vj i View F C Plot 7 Tools 9 Print 7 Properties 7 Help 7 ATHENA Initial Triangular Grid Microns SILVACO International 1994 f Figure 2 11 Initial Triangular Grid The grid in ATHENA consists of points connected to form a number of triangles Each point has one or more nodes associated with it A point within a material region has one node while a point which belongs to several regions has several nodes A node represents the solution e g doping concentration in a particular material region at the point For example a given node may represent solution values in silicon at a point with coordinates 0 0 0 0 an entirely different node may represent solution values in oxide at the same point 0 0 0 0 So the previous INIT statement creates the lt 100 gt silicon region of 1 0 um x 1 0 um size which is uniformly doped with boron concentration of 3e14 atom cm This simulation structure is ready for any process step e g implant diffusion Reactive Ion Etching
283. avelength 2 9 7 Structure Exposure Exposure post exposure bake and development each have separate statements EXPOSE BAKE and DEVELOP respectively In order to use these three statements some initial requirements must be met First an intensity cross section or Fourier Spectrum data must be available Second you must create a structure including photoresist using the techniques described in Section 2 3 Creating a Device Structure Using ATHENA This intensity cross section can come from three different places The first method is by running the imaging module prior to exposure This puts the intensity data array into memory The second method is by initializing with an intensity data array that has been stored in a standard structure file see Section 2 9 5 Imaging Control using the following command INITIALIZE INFILE STR INTENSITY The INTENSITY qualifier lets ATHENA know that this is an intensity file as opposed to a standard structure file After this command is entered the intensity data array will be placed in memory Wavelength will be stored in this file and can be changed only by re running the imaging module The third method of entering an intensity cross section is through a user data file The file should contain the wavelength the number of data points and the intensity and position of each point The first line of this file should contain the wavelength in mic
284. balt films on silicon J Appl Phys v 71 p 5892 1992 M A Nicolet and S S Lau in VLSI Handbook Ed Norman G Einspruch Academic Press p 430 1985 J Lindhard M Scharff and H E Schiott Range Concepts and Heavy Ion Ranges Kgl Dan Vid Selsk Mat fys Medd v 33 1963 R Smith Ed Atomic and Ion Collisions in Solids and at Surfaces Cambridge University Press 1997 D G Ashworth R Oven and B Mundin Representation of Ion Implantation Profiles be Pearson Frequency Distribution Curves J Phys D v 23 p 870 1990 A F Tasch An Improved Approach to Accurately Model Shallow B and BF2 Implants in Silicon J Electrochem Soc v 136 p 810 1989 K B Parab et al Analysis of Ultra Shallow Doping Profiles Obtained by Low Energy Ion Implantation J Vac Sci Technol v B14 p 260 1996 G A J Amaratunga K Sabine and A G R Evans The Modeling of Ion Implantation in a Three Layer Structure Using the Method of Dose Matching IEEE Trans Electron Dev v ED 32 p 1899 1985 A F Burenkov F F Komarov and M M Temkin Analytical Calculation of Ion Implantation through Mask Windows in Russian Microelektronika v 16 p 15 1987 A F Burenkov A G Kurganov and G G Konoplyanik Two Dimensional Local Ion Implantation Distribution in Russian Povekhnost Surface Sciences v 8 p 52 1989 J Lorenz W Kruger and A Barthel Simulation of the La
285. be implemented in future when more complete tables of lateral parameters will be generated using Monte Carlo simulations Silvaco 3 75 ATHENA User s Manual Specification of Implant Parameters in the Moments Statement As mentioned previously the analytical ion implantation simulations strongly depend on the input parameters moments ATHENA provides several ways of implant parameter specification They are look up tables user defined look up tables and the MOMENTS statement Two types of look up tables are currently provided with ATHENA The files containing the tables are in ASCII format and can be found in the lt install gt lib athena lt gt version gt common implant tables directory The first type are standard tables std tables file containing parameters for most ion material combinations used in ATHENA These are longitudinal parameters for the single Pearson distribution in the energy interval 10keV to 1MeV The energy interval is extended to 1keV 8MeV for B P and As in silicon silicon oxide polysilicon and silicon nitride These tables also include a limited set of parameters for the dual Pearson function only for B and BF2 in the energy interval 10 100keV tilt angle 7 rotation angle 30 and native oxide as well as with simple interpolation of the dose ratio parameter SR between different doses Parameters for the FULL LAT model are provided only for B implants in silicon They are based on the spatial mom
286. bility limit might refer to For example excess dopants could be participating in small clusters or larger precipitates Deactivation threshold would be a more proper designation for this limit and will be used throughout the rest of this section The notation cy will be used for the deactivation threshold Therefore for all the models described in this section the following points are assumed for each dopant type e Dopants in excess of the deactivation threshold are considered electrically inactive i e they do not contribute to the carrier populations 3 18 Silvaco SSUPREM4 Models e Additionally dopants in excess of the deactivation threshold are considered to be immobile i e they cannot diffuse Electrical Activation Model The Electrical Activation Model is used to calculate which level of dopant concentration the deactivation occurs For this purpose two different Electrical Activation Models are used for all diffusion models e An AsV Clustering Model for arsenic or similar models for other impurities e A Semi empirical Activation Model based on Table B 14 in Appendix B Default Coefficients for all other dopants These models can be activated in the IMPURITY statement by parameters CLUSTER ACT and SOL SOLUB respectively By default the CLUSTER ACT model is used only for As in silicon and polysilicon The AsV Clustering Model used in SSUPREM4 is based on the simple reaction As Vo o AsV
287. cal difficulties may occur during simulation For ramped thermal step a synonym T START can be used T FINAL specifies the final temperature for ramped thermal steps Synonym is T STOP T RATE specifies the ramp rate in C minute for ramped thermal steps Parameters to Define the Diffusion Ambient DRYO2 WETO2 INERY and NITROGEN specify the type of ambient during the diffusion step DRYO2 specifies that ambient is dry oxygen WETO2 specifies that ambient is wet oxygen NITROGEN specifies that ambient is inert INERT is a synonym for NI TROGEN HCL PC specifies the percentage of HCl in the oxidant gas stream PRESSURE specifies the partial pressure of the active species in atmospheres Units are atmospheres The default is 1 F 02 F H2 F H20 F N2 and F HCL specifies the relative flow rate of the components of oxygen hydrogen water nitrogen and HC1 in the ambient If these parameters are used the DRYO2 WETO2 NITROGEN set or HCL PC should not be specified Silvaco 6 25 DIFFUSE ATHENA User s Manual C IMPURITIES specifies concentration of the impurities in the ambient gas see Section 6 2 10 Standard Impurities for the list of impurity names that can be used e g I BORON Units are atoms cm You can define multiple impurity parameters for ambients with multiple impurities You can only use boron phosphorus and arsenic if you specify the advanced diffusion model
288. cally deposits a thin native oxide layer on all exposed silicon polysilicon surfaces at the beginning of oxidation steps The INITIAL parameter in the OXIDE statement determines the layer s thickness which has a default value of 20 A The two dimensional oxidation models in SSUPREM4 are based on the well known linear parabolic theory of Deal and Grove 30 Numerical aspects of the model implementation can be found in 31 Silicon oxidation is modeled by considering the following three processes 1 Oxidant e g H20 or Og is transported from the ambient gas into the SiO layer at the gas SiO interface 2 Oxidant is transported across the SiO layer until reaching the Si SiO interface 3 Oxidant arriving at the Si SiO interface reacts with silicon to form a new layer of SiO The transport of oxidant across the gas SiOg interface is given by F WC C ny 3 127 where e his the gas phase mass transport coefficient e C is the equilibrium oxidant concentration in SiO e Co is the oxidant concentration in SiO at the gas SiO interface e ng is a unit vector normal to the gas SiO interface pointing toward the silicon layer The equilibrium oxidant concentration in SiO is linearly related to the partial pressure of the oxidant P in the gas by Henry s law x C K P 3 128 where K is a constant Diffusion of oxidant molecules in the SiO is driven by a concentration gradient and is given by Fick
289. can control the grid inside deposited layer The grid distribution along normal direction is controlled by a number of divisions the user defined parameter DIVISIONS in a uniform vertical grid If non uniform vertical grid is used then specify the DY and YDY parameters in the DEPOSIT statement DY specifies nominal spacing and YDY specifies the position where to apply the nominal spacing The spacings further from the nominal position YDY increase or decrease according to geometrical series The coefficients of the geometrical series are calculated so that total number of the spacings will be equal to the DIVISIONS parameter When the conformal deposition algorithm fails to deposit next sublayer which happens when deposition occurs on the structure with narrow trenches or undercuts or both the current spacing divides into two and thinner layer is checked This spacing division algorithm is applied recursively because in these situations the number of sublayers divisions actually deposited can be higher than the number specified in DIVISIONS 3 6 3 Epitaxy Simulation SSUPREM4 models high temperature deposition of single crystal silicon through the EPITAXY statement This statement combines deposit and diffusion steps and parameters See Chapter 6 Statements Section 6 18 EPITAXY for more information Silvaco 3 91 ATHENA User s Manual 3 7 Etching Models Although etching is an integral process step in sili
290. caused by the built in electric field due to the dopant gradients e Both J and V are not considered to be in local equilibrium but can be annihilated by bimolecular recombination This feature of the CDD model performs annihilations between not only the free defects but also between the impurity defects pairs which play the role of recombination centers Therefore the J V recombination rate is strongly enhanced at high dopant concentration e It is now well established that transient enhanced diffusion TED is strongly correlated with the evolution of the self interstitial supersaturation governed by the nucleation and evolution during the high temperature anneal of a variety of extended defect structures such as the interstitials clusters Thus the predictive process modeling of the deep submicron MOSFET technologies requires the development of accurate diffusion models which take into account the full set of interactions between dopants and the point or extended defects clusters The PLS model coupled with the BCA implantation allows you to calculate the evolution of the clusters Therefore it is unneccessary to artificially add 311 or other clusters because the model automatically generates them and calculates their evolutions according to the Ostwald ripening theory e At concentrations near the solid solubility limit a dynamic clustering model is considered Immobile complexes As V or B Im are formed which result in decrease of t
291. center of the gate at x 0 0 for the initial grid Therefore polysilicon should be etched to the right from x 0 3 To do so select Right from the Geometrical type and set the Etch location to 0 3 This will give the following statement POLY DEFINITION The structure created by this You can obtain an arbitrary ETCH POLY RIGHT P1 X 0 3 ETCH statement is shown in the left hand plot of Figure 2 16 shape of geometrical etching by using the Any Shape button For example to make a tilted etch specify X and Y locations of four Arbitrary points as shown in Figure 2 17 The following four etch lines will be inserted into the input file POLY DEFINITION ETCH CONT X 0 4 Y 1 ETCH CONT X 2 Y 1 ETCH DONE X 2 Y 1 ETCH POLY START X 0 2 Y 1 2 20 Silvaco Tutorial TonyPlot 2 2 1 File v View Plot v Tools Print Properties Help ATHENA ATHENA Etch Right Arbitrary Etch amp R E amp B EA amp amp q k amp Ed amp oe eee coe FENE FERE FERS FENE FENE ERTE FERAI amp ENE ENEN ENEN FENE FERN ENEN ERNS FENE ENEN FNT Gon i Rian ag amp amp
292. cess in C REFLOW specifies that material flow should be calculated during the bake process Default is False TIME specifies the amount of time for the bake step in specified units Default is MINUTES HOURS MINUTES and SECONDS specify the units of the TIME parameter DUMP and DUMP PREFIX specify that a structure file be output at every DUMPth time step The files are readable with the STRUCTURE statement or can be displayed using ToNyPLOT The names will be of the form DUMP PREFIX lt time gt str where lt time gt is the current total time of the simulation Examples The BAKE command is entered with the user specified diffusion length for post exposure bake BAKE DIFF LENGTH 0 05 BAKE can also be entered with time and temperature parameters for post exposure bake BAKE TIME 45 TEMP 120 For photoresist reflow post development bake the above command is entered with the REFLOW parameter BAKE REFLOW TIME 45 TEMP 120 For more examples see DIFFUSE and RATE DEVELOP 6 14 Silvaco BASE MESH 6 7 BASE MESH BASE MESH specifies parameters of the base mesh used for initial grid generation Syntax BASE MESH SURF LY lt N gt SURF DY lt N gt ACTIVE LY lt N gt ACTIVE DY lt N gt EPI LY lt N gt EPI DY lt N gt SUB LY lt N gt SUB DY lt N gt BACK LY lt N gt BACK DY lt N gt
293. cifies a multiplier for chemical flux generated by the plasma etching machine Default is 1 0 MAX DEPOFL specifies a multiplier for deposition flux generated by the plasma etching machine Default is 1 0 Parameters used for Monte Carlo Plasma Etch Model ION TYPES specifies the number of different ions in etching plasma MC POLYMPT specifies the number of MC simulated polymer particles normalized to the volume of the ejected material MC RFLCTDIF specifies the reflection diffusiveness 1 corresponds to completely diffusive reflection 0 corresponds to ideal mirror reflection MC ETCHI specifies the etch rate parameter for the first type of ions unitless MC ETCH2 specifies the etch rate parameter for the second type of ions unitless MC ALB1 specifies the reflection parameter for the first type of ions unitless This coefficient can vary from 0 no reflection to 1 100 reflection MC ALB2 specifies the reflection parameter for the second type of ions which is unitless This coefficient can vary from 0 no reflection to 1 100 reflection MC PLM ALB specifies the reflection parameter for polymer particles which is unitless This coefficient can vary from 0 no reflection to 1 100 reflection MC NORM T1 specifies the plasma normal temperature for the first type of ions which is unitless MC NORM T2 specifies the plasma normal temperature for the second type of ions which is unitless MC LAT T1 specifies the plasma lateral
294. cifies the temperature of the substrate during implantation DIVERGENCE specifies the implant beamwidth in degrees When the BEAMWIDTH angle is specified the TILT angle is varied between TILT DIVERGENCE 2 0 Each ion will have an angle somewhere in this range decided by a random number generator Distribution of the ions is uniform across the defined angular distribution Correct specification of DIVERGENCE is generally required for accurate zero degree implant ranges Default is 1 degree IMPACT POINT specifies only in the Monte Carlo method that the ion beam enters the surface in the point with lateral coordinate x left IMPACT POINT L where left is the x coordinate of the left boundary of the structure and L is the length of the structure This parameter would be used for calculation of the point source 2D distribution and spacial moments using Monte Carlo method IONBEAMWIDTH specifies the ion beam width in nanometers It can only be used with the IMPACT POINT parameter SMOOTH specifies that a special Gaussian convolution smoothing to be applied to the Monte Carlo results SMOOTH multiplied by estimated standard deviation of the whole profile serves as the standard deviation for the Gaussian formula SAMPLING specifies that statistical sampling to be used in the Monte Carlo method DAMAGE specifies that damage formation should be calculated during Monte Carlo implant MISCUT TH and MISCUT
295. ckly the enhancement factor reach its maximum Note For exponentially varying solutions e g oxidation stress and dopant concentrations both C and ENH MINC are taken to be log base 10 of their respective value 6 90 Silvaco RATE ETCH 6 51 RATE ETCH RATE ETCH specifies the etch rate parameters for a machine which is used in a subsequent ETCH statement in ELITE Syntax RATE ETCH MACHINE lt c gt MATERIAL NAME RESIST lt n gt WET ETCH RIE PLASMA C PLASMA A H A A S U H U U S N M DIRECTIONAL lt n gt ISOTROPIC lt n gt CHEMICAL lt n gt DIVERGENCE lt n gt PRESSURE lt n gt TGAS lt n gt TION lt n gt VPDC lt n gt VPAC lt n gt LSHDC lt n gt LSHAC lt n gt FREQ lt n MGAS lt n gt MION lt n gt QI1O lt n gt QCHT lt n gt CHILD LANGM COLLISION LINEAR CONSTANT IONS ONLY NPARTICLES lt n gt ENERGY DIV lt n gt OUTF TABLE lt lt n gt gt OUTF ANGLE lt c gt ER LINEAR ER INHIB ER COVERAGE ER THERMAL K I lt n gt K F lt n gt K D lt n gt SPARAM lt n gt THETA lt n gt IONFLUX THR lt n gt MAX IONFLUX lt n gt MAX CHEMFL lt n gt MAX DEPOFL lt n gt ION TYPES lt n gt MC POLYMPT lt n gt MC RFLCTDIF lt n gt MC ETCH1 lt n gt MC ETCH2 lt n gt MC
296. con technology SSUPREM4 lacks a complete physical description of etching steps To circumvent this problem SSUPREM4 considers etching simulation as a purely geometrical problem Etching is simulated as a low temperature process Impurity redistribution is ignored during the etching process Etch steps are simulated using the ETCH statement in which the material to be etched and the geometrical shape of the etch region are specified It is not necessary that material to be etched be exposed or at the top surface of the structure There are five different ways to define an etch region 1 A polygonal region may be defined by specifying the x and y coordinate of each vertex in the poly gon Etching will confined to that polygon only A region to the left or right of a line segment may be defined by specifying the x and y coordinates of the end points of the line segment Etching will then proceed from the left or right of the line segment to the edge of the structure A region between the top boundary of the structure and a line obtained by translating exposed por tion of the top boundary down in the y direction may be defined by specifying the DRY parameter in the ETCH statement The THICKNESS parameter will determine the distance to etch in the y direction An extension of the DRY etch produces the etch region with slopped sidewalls and undercuts under the mask The UNDERCUT parameter specifies the top boundary s extens
297. cording to the diffusion equation with the diffusion constant D being independent of time concentration and location dM V DV 5 31 7 DVM M is the PAC concentration and is the PEB time For a more general discussion see 110 M is calculated by solving the two dimensional diffusion Equation 5 31 The diffusion length can be related to the bake time t and the diffusion coefficient D 2 2tD 5 32 For a PEB of 60 seconds at 125 C a diffusion length in the range of 0 04 lt o lt 0 06 microns would be appropriate PEB can also be specified with parameters temperature and time The diffusivity D is given by the equation D D exp Dg kT 5 33 You can specify the Dg and Dg parameters in the RATE DEVELOP command Reflective boundary conditions at the air resist interface and at the resist substrate interface must be incorporated to ensure that the total amount of dissolution inhibitor in the resist is conserved M x y t is extended into regions outside the resist by reflection at the planar interfaces A post development bake is also available It models a physically based reflow of the photoresist 5 12 Silvaco OPTOLITH Models 5 6 The Development Module The development model is based on the knowledge of the PAC distribution or dissolution inhibitor in the resist layer after exposure and post exposure bake In classical Novolac resists the dissolution inhibitor and the PAC are
298. ctivation parameters of B P As and SB in Germanium The experimental data from 132 and 133 were used for parameter estimations Silvaco D 1 ATHENA User s Manual D 1 2 Optolith Features i 2 Added the multi image capability to the proximity printing lithography module Improved MULT EXPOSE capability ATHENA now takes into account different values of Dill s C parameter for each wavelength in case of broadband illumination D 2 ATHENA Version 5 14 0 R Release Notes D 2 1 SSUPREM4 Features 1 D Do Introduced new parameters to control the trajectory visualization capability in Monte Carlo implant TRAJ FILE Specifies the name of the file in which ion trajectories calculated with the Monte Carlo BCA method are to be saved and N TRAJ which specifies the number of ion trajectories to be saved in the TRAJ FILE Improved statistics and consequently the effective accuracy of Monte Carlo BCA ion implantation simulation in 2D structures This is achieved by more accurate estimation of number of trajectories near the side edges of the simulation structure Improved stopping power model for 11 20 channel in SiC This is a rare event channel for standard wavers 0001 But it has some influence in case of angled implants into trenches parallel to the 11 20 plane Added capability to control triangle orientation of initial ATHENA grid New parameters TRI LEFT and TRI RIGHT are added to the LINE X s
299. cture Dislocation LOOpSiiiri i mpri atre a e aah ipa 3 1 ae E occccecccccccecccceccececcececceceececeececeececeececeeceeeee 2 26 28 DEIS 282 e N a E eh tae et 2s 3 2 Initial Substrate ccccccccccececccceeeeeenscceeeseessceeeeees 2 14 16 dopat eausa Pathe aknan EEE ad Heed caer a 3 1 Rectangular Grid scssssecssssssesesecesneecsnsseessneeesnnaes 2 8 14 Dos Loss Modelisasn inanin aR 3 7 Reducing Grid Points in Non Essential Areas See also Interface Trap using the Relax Parameter 2 22 25 Reflecting a Structure in the Y Plane E using the Mirror Parameter 2 25 26 Simple Film Depositions eeceeeeeeseesseeeeseeeereteaee 2 16 19 Electrical Deactivation and Clustering Models Simple Geometrical Etches ecceeeeeeseeeeeeeeeeeeeeee 2 20 22 Electrical Activation Model 3 19 20 Structure File for Plotting or Initializing an ATHENA Transient Activation MOdel cesesceeeesseeeeeneeeeeneeereeees 3 20 Input file for Further ProceSSing sseeeeesrereees 2 28 29 Epitaxy SIMUlAON asse iaa ai raat daad e eane vedanta 3 91 lon Implantation MOdEIS ceccscerecsrseeseeeeees 3 66 89 3 93 ERFG IGG ios adestatssdetiigtaienadtnubrwaeaciteten 3 58 Analytic sseeceesseseeseeseeseeteseeseeseeneeseseestereeneenssnesens 3 66 70 Monte Car orri rnea a othe eect N tices 3 77 Silvaco Index 2 Index Multi Layer ovs cccc cctcessitecteeteitacetecestepaaiei ebiattees Adeeise 3 70 72 Stopping
300. d by the SAMPLING parameter Note When TRAJ F ILE parameter is specified the Monte Carlo implant simulation will be performed on a single processor even if the multiple processors are available and parallel capability is specified by the P parameter in the GO ATHENA statement Silvaco 6 47 IMPLANT ATHENA User s Manual N TRAJ specifies the number of ion trajectories to be saved in the TRAJ FILE The default is minimum of N ION and 2000 Z1 specifies the atomic number of an inert ion used only for damage or amorphization of substrate No new impurity will be introduced into the structure except damage which will affect subsequent Monte Carlo implants M1 specifies the atomic weight of the inert ion with atomic weight of Z1 If M2 is not specified the atomic weight of the main isotope will be used Analytical Implant Example This example specifies that a 100keV implant of phosphorus to be done with a dose of 1 0 e14 and with a tilt angle of 15 to the surface normal The Pearson model is to be used to calculate the doping profile IMPLANT PHOSPH DOSE 1E14 ENERGY 100 TILT 15 SVDP Boron Implant Example This example shows the syntax for a zero tilt and 50keV boron implant through 5nm of screen oxide The oxide is defined by S OXIDE and this definition is independent of any actual oxide in the structure itself IMPLANT BORON DOSE 1E13 ENERGY 50 TILT 0 S OXIDE 0 005
301. d for energy range 80 100keV Only for 15 80keV Experimentally verified for 5 65keV For energy ranges 1 5keV and 65 80keV the same procedures is used for boron b c d Experimentally verified for 15 80keV Numerical extrapolation is outside this energy range e Experimentally verified for 5 180keV Interpolation between 5keV and UT MARLOWE calculated profile at 0 5keV If you choose a simulation outside the parameter ranges described in Table 3 7 ATHENA will not use the Dual Pearson Implant SVDP Models but will use the standard tables instead When using the Dual Pearson model remember the following e For implant energies below 15keV for boron BF and arsenic the simulation predicts the dopant profiles for implants into a bare silicon surface i e silicon wafer subjected to an HF etch less than two hours before implantation Low energy implant profiles at such low implant ener gies are found to be extremely sensitive to the presence of a thin 0 5 1 5nm native oxide layer or disordered silicon layer on the wafer surface 49 Remember this fact when using the model for the simulation of low energy ion implantation and when performing implantations e For implant energies between 10keV and 15keV the simulations are performed for boron BF and arsenic by using an interpolation between the Dual Pearson Model parameters at 15keV and the Dual Pearson Model parameters at 10keV The parameters at 15keV correspond to implanta
302. d gain throughout the full operating range In some cases the base current may be less affected in the very high and very low injection regions by changes in the surface recombination velocity and adding some scope to fine tuning the profile of the base current versus base emitter voltage curve It is important to define the poly emitter as an electrode so it can define the interfacial surface recombination velocity VSURFN and VSURFP using the CONTACT statement This is in contrast to the MOSFET calibration text where we strongly advise you not to define the polygate as an electrode Be sure not to get these two confused The parameter that activates the recombination velocity is SURF REC which is also in the CONTACT statement For example an NPN BUT statement would be CONTACT NAME emitter N POLYSILICON SURF REC VSURFP 1 5e5 A lower value of recombination velocity VSURFP will reduce the base current and increase the gain hfe The reverse is also true Silvaco 2 49 ATHENA User s Manual 2 6 3 Tuning the Collector Current All Regions Figure 2 37 shows the parameter that affects the collector current over the entire range is the intrinsic base resistance The base resistance is primarily determined by the dose of the base implant s An increase in the base implant dose will decrease the intrinsic base resistance and decrease the collector current in all injection regions In some cases however the collector current
303. d in the Physics Chapter of the ATLAS USErR s MANUAL VOL I There are three user definable parameters for the Klaassen band gap narrowing model The BGN E parameter has a linear dependency on doping concentration and has the default value of 6 92e 3 volts BGN c has a square root dependency with doping concentration and has the default value of 0 5 BGN N is the value of doping where band gap narrowing effectively starts to take effect and has a default value of 1 3e17 cm3 The equivalent default setting consequently should be written as MATERIAL BGN E 6 92e 3 BGN C 0 5 BGN N 1 3e17 You can alter these parameters to modify the current gain of the device in the medium injection regime For example reducing the linear parameter from 6 92e 3 to 6 5e 3 is sufficient to cause a significant increase in current gain in the medium injection region Although the bandgap narrowing parameters affect both collector and base currents the base current is affected to a greater degree The most sensitive plot to see the effect of small changes to bandgap narrowing is a plot of current gain versus log of collector current A reduction in bandgap narrowing will result in an increase in current gain in the medium current injection region 2 6 5 The Base Current Profile Low Injection This is one case where there is an interdependency on one parameter since the intrinsic base resistance not only affects the collector current in all regions s
304. d in the simulation MULT IMAGE specifies that the preceding and current images will be added You can add any number of images as long as the IMAGE command contains this boolean X CROSS and Z CROSS specify if the one dimensional image is parallel to the x axis or z axis respectively ONE DIM use a one dimensional image module that images a line drawn across a 2D layout This is the best method when the image will be used for subsequent Example EXPOS E statements This statement loads a mask named MASK SEC and specifies x resolution in the image window of DX 0 1 micrometers It then runs the imaging module IMAGE INFILE MASK SEC DX 0 1 For more information see ILLUMINATION PROJECTI ON IL LUM FILTER PUPIL FILTER ABERRATION LAYOUT and EXPOSE Silvaco IMPLANT 6 28 IMPLANT IMPLANT specifies an ion implantation process step Syntax IMPLANT GAUSS PEARSON FULL LAT MONTECARLO BCA CRYSTAL AMORPHOUS IMPURITY ENERGY lt n gt DOSE lt n gt FULL DOSE TILT lt n gt ROTATION lt n gt FULLROTATIO PLUS ONE DAM FACTOR lt n gt DAM MOD lt c gt PRINT MOM X DISCR lt n gt LAT RATIO1 LAT RATIO2 S OXIDE lt n gt ATCH DOSE RP SCALE MAX SCALE SCALE MOM ANY PEARSON N ION lt n gt MCSEED lt n gt TEMPERATURE lt n gt DIVERGENCE lt n gt TONBEAMWIDTH lt n gt
305. ded to include syntax ABOVE and BELOW to facilitate the truncation or planarization of structures for interfacing to device analysis or following point defect based diffusion ETCH ABOVE and ETCH BELOW both sustain one dimensional calculation and can be used in the inverse of the STRETCH operations described above The STRIP statement has been enhanced to include material specification This allows strip of any material If no material is specified STRIP removes all photoresists and BARRIER materials The IMPURITY statement has been added The IMPURITY statement allows the specification of parameters for the new impurities for FLASH and SSUPREM4 that have been introduced in this release The IMPURITY statement is intended to stop proliferation of multitudinous statements of the form PHOSPHORUS BORON ARSENIC etc as new impurities are added Parameters for boron for example can be specified with IMPURITY I BORON instead of the BORON statement The IMPURITY statement allows setting of atomic mass and atomic number using the AT MASS and AT NUMBER parameters respectively These parameters effect the Monte Carlo ion implant and allow user defined impurities for ion implant by redefining an existing impurity with the desired characteristics e The number of user definable materials has been increased to 10 e Shell statements such as QUIT and HELP have been made case insensitive D 20 Silvaco ATHENA Version History e A new algorithm for
306. developers Examples The following statement sets the routine output to include more information OPTION NORMAL Silvaco OXIDE ATHENA User s Manual 6 40 OXIDE OXIDE specifies coefficients for use during oxidation steps AMBIENT is a synonym for OXIDE Syntax OXIDE DRYO02 WETO2 ORIENT lt n gt IN L O0 lt n gt 1 BREAK lt n gt IN L E lt n gt LIN H 0 lt n gt LIN H E lt n gt L PDEP lt n gt PAR L 0 lt n gt PAR L E lt n gt PAR H 0 lt n gt PAR H E lt n gt P BREAK lt n gt ORI DEP ORI FAC lt n gt HCL PC lt n gt HCLT lt n gt HCLP lt n gt HCL PAR lt n gt HCL LIN lt n gt THINOX 0 lt n gt THINOX E lt n gt THINOX L lt n gt THINOX P lt n gt BAF DEP BAF EBK lt n gt BAF PE lt n gt BAF PPE lt n gt BAF NE lt n gt BAF NNE lt n gt BAF K0 lt n gt BAF KE lt n gt STRESS DEP VC lt n gt VR lt n gt VD lt n gt VT lt n gt DLIM lt n gt MATERIAL MATERIAL DIFF 0 lt n gt DIFF E lt n gt SEG 0 lt n gt SEG E lt n gt TRN 0 lt n gt TRN E lt n gt HENRY COEFF lt n gt THETA lt n gt ALPHA lt n gt MIN OXIDANT lt n gt INITIAL lt n gt SPLIT ANGLE lt n gt SPREAD lt n gt MASK EDGE lt n gt NIT THICK lt n gt
307. diffusivity and intrinsic carrier concentration through C Interpreter function specified by SIGECDF MOD and SIGECNI MOD parameters in the METHOD statement Silvaco 3 95 ATHENA User s Manual 3 9 3 Boron Transient Diffusion Suppression by Carbon Incorporation Models There are experimental indications 82 that interstitials diffuse slowly and tend to disappear or get trapped more intensively in SiGe layers with substitutional carbon These effects result in suppressing of the boron transient diffusion when carbon is incorporated into SiGe layer The following equations shows the diffusivity of interstitials Dz is controlled by the DCARBON E parameter DASiy_ yy Ge C DSi ep PEAREN E 3 243 l This model also introduces an additional sink for interstitials in the layers with high carbon concentration Intensity of the sink is proportional to the carbon concentration and is controlled by the recombination parameters KCARBON 0 and KCARBON E specified in the INTERSTITIAL SILICON statement This means that the following recombination term will appear on the right hand side of the interstitial transport equation Equation 3 24 _KCARBON Rcargoy KCARBON exp T 5 maea 3 244 Si 3 96 Silvaco SSUPREM4 Models 3 10 Stress Models ATHENA allows you to calculate stresses generated during semiconductor processing There are three ways to calculate stresses The first way is to calculate th
308. e during subsequent processing the polysilicon dopant can penetrate into the underlying silicon substrate This simulation artifact can cause threshold voltages to be very different than expected To rectify this simulation artifact you can control the number of grid layers added during the oxidation with the GRID OXIDE and GRIDINIT OXIDE parameters in the METHOD statement You should place this statement before the gate oxidation diffusion step Setting these parameters to a value which results in three or four grid layers in the gate oxide e g 15 angstroms for a 60 angstrom gate oxide thickness can alleviate this problem We suggest you to set back these parameters to the default values after the gate oxidation step Figure 3 17 shows a cross section of an NMOSFET with a highly doped phosphorus polysilicon gate The default grid spacing in the oxide is used in Figure 3 17 b while the grid spacing is adjusted properly for Figure 3 17 a By comparing these two figures it is obvious that phosphorus has penetrated through the gate oxide for Figure 3 17 b but does not penetrate through the gate oxide in Figure 3 17 a Silvaco 3 59 ATHENA User s Manual TonyPlot 2 6 9 Materials Materials Silicon Silicon SiO2 i SiO2 Polysilicon Polysilicon Aluminum Aluminum SILVACO International 1996 Figure 3 17 a MOSFET Structure with Proper Gridding in Gate Oxide b MOSFET Structure with Default Grid Spacing in Gate
309. e C 12 Comparison of doping profiles analytical extraction versus Monte Carlo Analytical implants are run instantaneously whereas Monte Carlo takes up to 30 minutes Silvaco C 15 ATHENA User s Manual This page is intentionally left blank C 16 Silvaco Appendix D ATHENA Version History This appendix lists the release notes in reverse chronological order for each ATHENA release The initial release of ATHENA incorporates the standalone capabilities of previously released versions of SSUPREM4 ELITE and other functionalities Version histories for SSUPREM4 are included here for reference D 1 ATHENA Version 5 16 0 R Release Notes D 1 1 SSUPREM4 Features 1 The Monte Carlo BCA implantation module has been multithreaded By default ATHENA runs on the maximum number of CPUs available online You can specify the number of CPUs to be used by the parameter P lt n gt in the athena command line as follows athena P 2 input in When running athena within DECKBUILD add the P 2 option to the simflags parameter of the Go statement The speedup achieved with the multi threading of the BCA module is close to linear for simulation of large number of trajectories N ION gt 50000 This almost optimal behavior is due to the inherent parallel structure of the BCA implantation module There are following limitations for using multi threading version of MC Implant Module If you specify the TRAJ FILE parame
310. e NA parameter on the EXPOSE statement is used only to specify that the vertical propagation model be used NA 0 0 The default is the large numerical aperture model Multiple exposure capability has been added to the EXPOSE statement Using the boolean parame ter MULT EXPOSE allows an arbitrary number of exposures to be simulated in the same resist Applications are multiple focal planes FLEX method and multiple wavelengths Multiple image capability has been added to the IMAGE statement Using the boolean parameter MULT IMAGE allows an arbitrary number of images to be superimposed in the same aerial image The application is for superposition of multiple images with different focal planes FLEX method A new parameter POWER MIN lt n gt in the EXPOSE statement has been introduced to control the extent of the exposure calculation This parameter controls the amount of loss to be considered in the calculation After reflection transmission and absorption the intensity may be so low as to be negligible POWER MIN sets the level below which the intensity will be ignored The imaging module now includes a one dimensional mode that allows the calculation of one dimen sional as opposed to two dimensional images The one dimensional image capability is invoked by specifying the ONE DIM parameter on the IMAGE statement The advantage of the ONE DIM mode is realized when using the calculated image in the exposure module The two dimensional cal cul
311. e a time step a point could jump over a slow region Errors in direction arise from non uniform rates along the string and from certain boundary conditions During each step perpendicularity to the front which is defined below is assumed to be constant in direction If two adjacent points have greatly differing rates however the quickly moving point cannot start turning towards the slower point until the end of the time step This mechanism tends to introduce relatively small errors in position because the error is roughly proportional to the cosine of the angle error Silvaco 4 3 ATHENA User s Manual 4 3 Deposition Models ELITE provides a set of deposition models that correspond to different physical deposition techniques Most of the models were first developed at UC Berkeley 84 85 86 and 87 and were originally implemented in the topography simulator SAMPLE 88 Any one of these models can be selected to define a machine for simulating processes on the structure In addition ELITE provides a conformal deposition capability that can be used to define initial structures In most integrated circuit processes at least one layer of interconnect is formed by depositing and patterning an Al or Al alloy film The trend toward lower temperature processing combined with the very steep edge profiles produced by anisotropic dry etching processes results in sharp step profiles which are difficult to cover with a uniform fi
312. e amount of fill is to be computed The DIAG parameter indicates that only the diagonal blocks should be factored in the matrix The KNOT parameter is inactive DIAG is the default parameter Although under certain conditions one dimensional stripes FULL FAC will perform better TRUNC DEF specifies that defect concentrations that become negative due to numerical difficulties be forced to a positive value 6 66 Silvaco METHOD Parameters Related to Timestep Control INIT TIME specifies the initial timestep value The default is 0 1 seconds PDINIT TIME specifies the initial time step for point defect diffusion Point defects are held fixed for the first timestep The default is 105 seconds T DEFECT specifies time in seconds for which point defect injection will be neglected during an oxidation The default is 5 seconds OXIDE GDT limits the timestep during oxidation to a fraction OXIDE GDT of the time required to oxidize the thickness of one grid layer GRID OXIDE The timestep may be limited by oxidation and by diffusion and the value of OXIDE GDT will limit the timestep if it is more stringent than the limits imposed by diffusion OXIDE GDT lt lt 1 is recommended to improve resolution of oxidizing diffusions The default is 0 25 REDO OXIDE saves time by not computing the oxide flow field every time the diffusion equation for impurities is solved The REDO OXIDE parameter specifies the percentage
313. e amount of runtime output has been set such that the default level provides appropriate infor mation for day to day use The level of output may be specified by the OPTION statement and either QUIET NORMAL VERBOSE or DEBUG parameters The default is NORMAL e The ECHO feature has been set to on by default This can be altered by specifying UNSET ECHO or SET ECHO to turn the echo off or on respectively e Command line continuation was supported The plus symbol at the beginning of a line indicates that it is a continuation of the previous line The at the end of a line indicates that the line follow ing it is a continuation The continuation symbol is now a space followed by a backslash character Silvaco D 27 ATHENA User s Manual at the end of the line to be continued e A smoothing algorithm has been incorporated into the mesh initialization calculation This guaran tees numerically desirable mesh characteristics for meshes with rapidly changing spacing e The deposition and epitaxy algorithm has been improved to be more robust and to provide more con sistent gridding The parameter MIN SPACE has been added to control the resulting grid e The parameter TOP LAYER has been added to the ETCH statement to indicate that only top layers of the etched material should be removed D 28 Silvaco Appendix E TSUPREM4 and TSUPREM3 Compatibility Features The following changes in ATHENA syntax and functionalities
314. e ateis 3 16 Resist exposure with accounting dose effect ese 5 11 DEPOSIT ien te ni a i eee 6 21 23 Resist exposure without accounting dose effect 0 5 11 Deposition Models sccceessecessseseeserereeeeeees 3 91 4 4 11 OUD rae a a Bente A AN 44 F GONPONMAN coe Sas EET BEPPE AE SOTIE EE E E E EAEE AEE 4 4 Flux Equations Conical CNET aena aa A as Aon es eine 3 26 28 KE r E E ae Flux EXPreSSiOn 0 scscseseseescseseceescsesecevevscsesevevsveessaveneeees 3 4 al ee VE eas A Tipe EE Me te 4 5 6 Flux Jump Eao ae EEE E E E 3 4 Epitaxy Simulation sessi einnsean EE 3 91 Free Point Defect Damage s sssssesssessseesseseeeseneesnee 3 87 Grid Conttol Saves civacen tetera RO 3 91 Fresnel diffraction ssceseeeeseseseseseetseeeteetenerenees 5 15 5 16 E EIA E E E A E 4 6 7 Fully Coupled Equations Monte CANO erapr aaee en nA nE EELEE A 4 10 11 CNET ts ha cist ev hate ter eet E AE AE EE 3 26 Planetary oiar a a a Pa eee 4 7 9 Fully Coupled Model Unidirectional sacair a oes 4 4 5 High Concentration Extension 0 c scsscsessesseseeseesesseeeeees 3 17 Deposition Wet Dry Etching RTA Diffusion Modelling ccccceeeeceeeeeeeeeeeeeeeeeeeaes 3 18 Defining ELITE Deposition Machines esseeeee 2 61 62 Defining ELITE Etch Machines G Modifying ATHENA ELITE Default Machines ee 2 60 Using A Specified Etch Machine ccccccssssssesceseteesceeeees 2 64 Generation Recombinati
315. e deposited layer where C INTERST specifies concentration at the bottom of the layer and F INTERST specifies concentration at the top of the layer C VACANCY specifies the concentration of vacancies in deposited layer Units are cm F VACANCY can only be specified together with C VACANCY This parameter generates the linearly graded vacancy concentration in the deposited layer where C VACANCY specifies concentration at the bottom of the layer and F VACANCY specifies concentration at the top of the layer Units are cm C FRACTION specifies the fractional component of the first element of a ternary compound to be deposited i e Al is the first component for AlGaAs The fractional component of the second component i e Ga is the second component for AlGaAs is 1 C FRACTION This parameter is valid for standard ternary materials AlGaAs and InGaAs or user defined ternary materials with the following standard names AlInAs InGaP GaSbP GaSbAs InAlAs InAsP GaAsP HgCdTe InGaN and AlGaN F FRACTION can only be specified together with C FRACTION This parameter generates the deposited layers with linearly graded fractional component where C FRACTION specifies the fractional component of the first element at the bottom of the layer and F FRACTION specifies the fractional component of the first element at the top of the layer This parameter is valid for standard ternary materials AlGaAs and InGa
316. e determined the fluxes of the polymer particles are calculated as follows As the result of ion flux interaction with the surface segment the polymer particles are generated The angular distribution of the polymer particles is uniform and the current density of these particles is determined by the etch model see Linear Etch Model on page 4 18 and the sum of the fluxes from incoming ions neutrals and from polymer particles ejected from other surface segments Obviously the latter flux needs to be pre calculated This flux is computed as follows First the configuration or geometrical factors are calculated These factors are the fractions of the number of particles ejected from one segment and absorbed by the other one These are calculated using the same trajectory tracing algorithms which are described above for the incident ions and neutrals with the only one difference starting points are not at the upper boundary of the simulation area but at the surface segments Then an iteration process is initialized At the first iteration only the incoming ion and neutral fluxes are used for calculation of the ejection rates from each surface segment Knowing the current densities of ejected particles and the configuration factors the polymer fluxes are calculated At subsequent iterations the polymer fluxes calculated at the previous iteration are used to update the etch and ejection rates The iterations are repeated until etch and ejectio
317. e entire grid in both directions Figure 2 20 the following lines will be inserted into the tutorial input file RELAX EVERYWHERE RELAX DIR X T DIR Y T TonyPlot 2 2 1 File vj View vj Plot yji Tools vj Print Properties 7 Help 7 ATHENA Spacer formation using dry etch foe ere RAR AR carat cpreritatrtetetetstet ty bebi ririt Hirti eer ea fy i ct U tite ni S iy o a CO A GD TD Micrans o k o n a i T 4 Silicon Sis Polysilicon o a gt i nternational 1994 Figure 2 19 Spacer Formation using Dry Etch The resultant grid is shown in the upper right corner of Figure 2 20 The total number of grid points is reduced from 708 to 388 When comparing with the grid before relaxation upper left corner of Figure 2 20 note that the grid within the oxide spacer and polygate has not changed This is due to three factors e the re
318. e grid must be improved First make a better grid in the y direction Usually it s necessary to get better resolution for the depth profile after the ion implantation step When adaptive gridding capability isn t used apply preliminary knowledge of the process you are going to simulate Suppose you want to perform a 60 keV boron implant so that the implant peak would be around 0 2 um It is reasonable to make a finer grid at this depth To achieve this simply add one more Y line by setting the Location to 0 2 and the Spacing to 0 02 The new rectangular grid Figure 2 6 will now appear Notice the number of points and triangles have increased to 231 and 400 respectively Silvaco 2 11 ATHENA User s Manual View Grid 241 points 400 triangles Figure 2 6 New Rectangular Grid The minimum spacing in the Y direction is at 0 2 um and the spacing gradually increases toward the bottom and the top of the structure Since the spacing at y 0 is still 0 1 only 3 grid lines lie between 0 and 0 2 um You may want to make a finer grid at the top of the structure To do this select the top line of the Y Location scrolling list change the spacing to 0 03 and press the Insert button The selected line will be replaced by Y LOC 0 00 SPAC 0 03 If you then press the View button there will be 8 grid lines between y 0 and y 0 2 Figure 2 7 View Grid r5 points 480 triangles Figure 2 7 Inserting New Grid Lines 2 12 Si
319. e has to be converted into SiGe material with a variable X composition to pass the device into ATLAS To do this type in the following STRUCTURE command example STRUCT OUTFILE HBT STR SIGE CONV This will save a structure call HBT str and converts the germanium dopant profile into the correct X composition SiGe material Silvaco 2 55 ATHENA User s Manual 2 8 Using Advanced Features of ATHENA 2 8 1 Structure Manipulation Tools Using the Structure FLIP Capability The Structure FLIP capability allows you to flip the structure in the x axis The STRUCT FLIP Y statement causes the structure to be vertically flipped This operation can be useful if some process steps e g etching deposition or implant take place from the backside of the wafer By using this statement you can flip a structure perform these steps and then flip it back Using the Stretch Capability In some cases a device characterization as a function of length is of interest For example the drain current characteristics depend strongly on the gate length The Stretch capability makes it possible to generate a number of MOSFET structures with different gate lengths from one ATHENA simulation The structure obtained so far in this tutorial See Figure 2 22 has a gate length of 0 6 u To increase the gate length to 1 5 use the STRETCH command To use this capability select Structure Stretch in the Commands menu and the ATHENA Stretch menu F
320. e or several materials can be specified at a time I IMPURITIES specify the impurities to be used for the grid adaptation see Section 6 2 10 Standard Impurities for the list of impurity names that can be used e g I BORON You can specify one or several impurities at a time I INTERST specifies that interstitials to be used for the grid adaptation I VACANCY specifies that vacancies to be used for the grid adaptation DISABLE specifies that the materials impurities combinations given are disabled to be effective on mesh adapting or smoothing Default is false MAX ERR specifies the maximum error allowable before adding points to the mesh unitless Error calculated above this value cause points to be added MIN ERR specifies the minimum error below which points can be deleted from the mesh unitless Error calculated below this value will remove points Both MAX ERR and MIN ERR are calculated using the Bank Weiser error estimator CONC MIN specifies the minimum impurity concentration below which adapting will stop Units are cm Default is 1 0 10 4em AREA MIN specifies the minimum triangle area below which adding points will stop Units are cm Default is 1 0 10 AREA MAX specifies the maximum triangle area below which deleting points will stop Units are cm Default is 1 0 10H EDGE MIN specifies the minimum edge length below which adding points will stop Units are cm Default is 1 0 10
321. e size of the initial structure and will trunc the RATIO BOX parameter is used to trade off mesh EPTH STR These parameters define MESH defined structure F ate the previous BAS Figures 2 74 and 2 75 show an example of this base mesh and of the subsequent 2D diffusion ha TonyPlot 2 5 4 File gt View 7 Ploty Tools Print Properties Help 7 ATHENA After 1 D Adaption Net Doping m3 21 a a 4 E a 3 a E SA c ESLL L A E N SEUS EEAS EL S EE o E O2 E 04 O05 06 OF 08 089 4 Distance along line SILVACO International 1996 xj TonyPlot 2 5 4 Filer View Plot Tools Print Properties 7 Help ATHENA Loss of Dopant Information Ra ARIAS i NAAAAAAAAN ARE Ae FAS bet SILVACO International 1996 al Figure 2 70 Mesh that is too coarse leads to Dopant Information Loss xj TonyPlot 2 5 4 File View Plot 7 Tools 7 Print Properties Help mem Tale CENTEA EnEn SEBE Fij TonyPlot V2 5 4 File 7 View Ploty Tools _Printv Properties v Help 7 ATHENA After 1 D Adaption gt Net Doping m3 2i 23 HAE ra gE dis HE Be Bpl Eh la i T T T T T T T T T 1 0 0 4 0 2 0 3 0 8 09 aI 04 06 0 6 Distance along line SILVACO International 1996 ATHENA Over Refinement 04
322. e stresses during viscous oxidation or viscous material reflow see Chapter 4 ELITE Models Section 4 5 Reflow Model The second way is to calculate the stresses due to thin film intrinsic stress or thermal mismatch using the STRESS statement at a certain moment of the processing sequence usually after thin film deposition or etching or both The third way is to follow stress history by specifying the STRESS HIST parameter in the METHOD statement In the cases of the second and third methods ATHENA performs a finite element analysis of the material structure solving the similar set of equations as in case of viscous oxidation Equations 3 144 3 146 The only difference are the thermal expansion and intrinsic terms added to the right hand side of Equation 3 144 T d POISS R 1 VISC E VISC 0 exp LCTE INTRIN SIG 3 245 1 2 POISS R KT i T3 The linear coefficient of the material thermal expansion LCTE can be specified as a function of temperature T in the MATERIAL statement The film intrinsic stress parameter INTRIN SIG is specified in the MATERIAL statement T and T are initial and final temperatures If the STRESS HIST method is specified ATHENA then calculates stresses when the simulation structure changes after etching deposition epitaxy and diffusion processes The temperature including ramp specified for current process step is used in the calculation The room temper
323. e through matter ions interact not only with the atoms from the lattice but also with the electrons Figure 3 22 shows the scattering geometry of two particles in the Laboratory Co ordinate System In the computational model it is assumed that ions from one deflection point to the next move along straight line segments these being the asymptotes of their paths At each collision ion loses energy through quasielastic scattering by a lattice atom and by an essentially separate electron energy loss part Silvaco 3 77 ATHENA User s Manual ion asymptote Y 4 path of ion r t initial location path of of lattice atom recoil recon asymptote Figure 3 22 The trajectories of the ion projectile and the lattice atom recoil The scattering angles of the projectile and the recoil are as follows tan 9 Afsin0 1 Afcos 6 3 218 tan 3 fsin 1 fcos 0 3 219 where f Jl Q E 3 220 Q is the energy lost by electron excitation A M M is the ratio of the mass of the target scattering atom to that of the projectile implanted ion O is the barycentric scattering angle calculated as follows o0 0 z 2pf l dr 3 221 R r g r where pP a AN g r D B N T 3 78 Silvaco SSUPREM4 Models where e pis the impact parameter e E AE 0 1 A is the relative kinetic energy Ey is the incident energy of the projectile e ris interatomic separation e V r is the interatom
324. e to Complete Figure 2 68 MOSFET Device Mesh Formation Flow The base mesh quality is important to allow a subsequent adaption in 2D The adjacent ratio of elements both in 1D and 2D relate directly to the smoothness of the final mesh quality The generation of a high quality adapted mesh starts with the BASE MESH command Here the 1D mesh is defined from where the final 2D mesh will evolve The BASE ESH command defines a 1D structure as a stack of up to five layers Five layers are used to define the five layers of a Bipolar device Each layer is described as having a thickness SURF LY ACTIVE LY EPI LY SUB LY and BACK LY and an associated mesh spacing per layer SURF DY ACTIVE DY EPI DY SUB DY and BACK DY The whole structure can also be offset in space with the point of origin determining the top left hand corner of the structure The OFFSET X and OFFSET Y parameters are used for this purpose An example of using the offset command might be defining the starting surface of an initial structure an epi thickness below the zero position That way the subsequent geometrical calculations are made easier Figure 2 69 indicates the relationship of the BASE MESH command to the initial 1D structure mesh 2 90 Silvaco Tutorial OFFSET Y SURFACE OF STRUCTURE SURF LY A DY ACTIVE LY x ACTIVE DY EPI LY EPI DY SUB LY y SUB DY BACK LY BA
325. e14 E X X 10 mins N2 str o 1um 10mins N2 str 1e13 E 4612 Se UL SLs fas 0 01 02 03 04 05 06 07 08 09 1 11 Boron Concentration cm3 Depth into surface um Loading file home derekk dk examples OXIDE 1 um 10mins N2 str OK SILVACO International 1996 Figure 2 27 Effect on boron diffusion profile when too small a substrate depth is used in the simulation Figure 2 27 shows the boron profiles for two identical anneals the only difference is the depth of the simulated substrate You ll see that a shallow modeled substrate always results in more total diffusion even though the substrate depth was greater than the total diffusion depth in both cases Modeling a deep substrate doesn t need to involve a huge number of extra mesh points since the mesh points can be placed quite far apart near the bottom of the substrate All that is required of the mesh points near the bottom of the substrate is that there be sufficient to model the gradient of interstitials in this region The number of additional mesh points can be further reduced in the X direction by the using several RELAX statements For normal small geometry MOSFET Bipolar processing a substrate depth of 20 um should be more than adequate This depth can be reduced by plotting the vertical interstitial profiles at various points in the process to find the maximum depth of interstitial diffusion There is little to be gained by reducin
326. ead any of the popular C language books such as 119 Additional information about the C Interpreter can be found in the SILVACO C INTERPRETER USER S MANUAL The function arguments of the C Interpreter functions are fixed in ATHENA Thus you need to make sure that the arguments and return values match those expected by ATHENA To help you a set of templates for functions available for the current release of ATHENA can be obtained by typing athena T filenam The filename is the name of the file where you want the template to be copied You can also obtain the C Interpreter templates by selecting Commands Templates in DECKBUILD The following example shows how to use the C Interpreter function get_damage_values to modify the default Plus One Model See Chapter 3 SSUPREM4 Models for interstitials generated during ion implantation A 2 Example Template for the C Interpreter function for defect formation during ion implantation y void get_damage_values input parameters int imp impurity index As 2 P 3 Sb 4 B 5 etc int mater material index Si 3 double x x coordinate in micron double y y coordinate in microns le double implanted_conc implanted concentration in 1 cm 3 double implanted_dam accumulated damage in eV cm 3 do not use without Monte Carlo BCA return parameters double I_val Interstitial concentration double V
327. eae E a EEEE hea ea eee hed a eens Bae dt D 8 D 6 3 Miscellaneous Features and Bug FixeS 2 00 e cece cece eee eee eee eee D 8 D 7 ATHENA Version 5 4 0 R Release NoteS 00 cece eee e cece e eee eee eee e ene D 8 DAESSURREM Renate eee earns tir Pen Ene Oat a aes tie niece eae Berar ery ene tee E etre D 8 xX Silvaco Table of Contents 7A PASC apa DES 2 5 ise at hh alate here nal ol ah he ge INU Ut el oh te a ot D 9 D7 SeOPTGLITH Capabilities 2X5 tee cook reat ee ead eh ee net cade ad ee en er D 9 Divas ELITE Capabilities na rrsan st actin aE Atal Awe E cot wha aR Late SA hao eats D 9 D 7 5 Miscellaneous Features and Bug FixeS 2c s et sade eed eee ee eed eed eee ede D 9 D 8 ATHENA Version 5 2 0 R Release NoteS cccceee eee eet eee eee e eee eens D 9 D 8 1 lon Implant BCA Model i 3 2c 24e0 beedebeads beady wee biwe nl pelted el aeteheedse cede D 9 D 8 2 Miscellaneous Features and Bug Fixes 0 00 cece eee eens D 10 D 9 ATHENA Version 4 5 0 R Release NoteS 0 ccc ee eee cere e eee eee eee e eee eeeeeee D 10 D 9 1SSUPREM4 i krisar wise ts wine ie T Waar E iE umn E e EA t duet ate whats t D 10 D 9 2 ELITE Capabilities nxiiseo bees dvevay aia wed EEEE E E E EAE EEE D 11 D 9 3 Generic ATHENA Capabilities 2a 902 canes ananuna aaeeea D 11 D 10 ATHENA Version 4 0 0 R Release Notes n u auuuuunnannnnnnnnnnnnnnrnunnnnnnrnnenrnrnrnnn D 11 DOSS SOU RMON oc onettek en Yt cued eine nna E ae e N
328. easured distribution moments The statistical technique uses the physically based Monte Carlo calculation of ion trajectories to calculate the final distribution of stopped particles 3 5 1 Analytic Implant Models ATHENA uses spatial moments to calculate ion implantation distributions This calculation method is based on range concepts from Range Concepts and Heavy Ion Ranges 45 in which an ion implantation profile is constructed from a previously prepared calculated or measured set of moments A 2D distribution could be essentially considered a convolution of a longitudinal along the implant direction 1D distribution and a transverse perpendicular to implant direction 1D distribution In the rest of this section we will first describe three 1D implant models and the method used to calculate 1D profiles in multi layered structures Then two models of transverse lateral distribution and a method of construction of 2D implant profiles will be outlined Finally three methods of implant parameter specification will be described Gaussian Implant Model There are several ways to construct 1D profiles The simplest way is using the Gaussian distribution which is specified by the GAUSS parameter in the IMPLANT statement 2 x R apa 3 178 an are 2 MAK LIAR where is the ion dose per square centimeter specified by the DOSE parameter R is the projected range AR is the projected range straggling or standard deviatio
329. eater than 900 and the device has been annealed for at least one minute following an implant where the dose is greater than 1e13 cm2 For a more accurate guideline see the Chapter 3 RTA Diffusion Modelling Table 3 6 shows the anneal temperature time combinations required for 95 of the clusters formed during high dose implants to dissolve Modeling these dopant defect clusters requires the fully coupled full cp1 and cluster damage cluster dam models Only when these clusters have dissolved can the two dim model be used without significant loss of simulation accuracy As a general rule we recommend that the method statement be changed to method two dim only after a diffusion time that is at least two or three times as long as the values quoted in the table Silvaco 2 31 ATHENA User s Manual If you wish to be certain of when it s safe to switch models the recommended procedure is to save a structure file at the point of interest load the file into ToNyPLoT and perform a 1D cutline Plot the clusters and interstitials If the cluster concentration is still visible it s too early to switch models For power devices where simulation time is at a premium the same method already described should be used But instead of using the cluster concentration as a guide of when to switch models the interstitial concentration should be used as the guide as to when to switch models one more time from the TWO DIM model to the basic F
330. ecifies the depth where the nominal spacing will be applied YDY is calculated relative to the top of the newly deposited layer Units are microns MIN DY specifies the minimum spacing in microns allowed between grid lines in the y direction in the new material The default is 0 001 microns 10 Angstroms Silvaco 6 21 DEPOSIT ATHENA User s Manual MIN SPACE specifies a minimum spacing between points on the surface of each sub layer Increasing this parameter will reduce the number of points on arced deposited surfaces Units are microns ARC SPACE is asynonym for this parameter Parameters Specific to Depositing Doped Layers C IMPURITIES specify the concentration of the impurity in the deposited layer in cm You can specify more than one of these parameters to define materials doped with multiple impurities F IMPURITIES can only be specified together with the corresponding C IMPURITY e g F BORON and C BORON This parameter generates the linearly graded concentration of the specified impurity in the deposited layer where C IMPURITY specifies concentration at the bottom of the layer and F IMPURITY specifies concentration at the top of the layer Units are cm C INTERST specifies the concentration of interstitials in deposited layer Units are cm F INTERST can only be specified together with C INTERST This parameter generates the linearly graded interstitial concentration in th
331. ect is illuminated by an element dx dzo of the effective source at xo Zg with its amplitude proportional to y x Zo the object spectrum a x z is then shifted by a corresponding amount In this instance the complex amplitude distribution on the entrance pupil sphere of the objective is Jo Z0 a x X0 Z Zo 5 12 The complex amplitude on the exit pupil reference sphere at E will be given by a x Z JN Xp Zo a x Xg Z Z fx z 5 13 In this equation f x z denotes the pupil function of the optical system If the system has an annular aperture where the central circular obstruction has the fractional radius e the pupil function has the form Vay aa 0 xX Z lt E f z t x Zz exp i k W x z x z lt l saa I x z a l 5 4 Silvaco OPTOLITH Models t x y is the pupil transmission which is usually set to one and W x z denotes the wave front aberration For an entirely circular aperture s becomes zero Note that the approach taken here is somewhat similar to the one used in the investigations on phase contrast microscopy 105 The function W x z gives the optical path difference between the real wave front and the exit pupil reference sphere Commonly the wave front aberration is expanded into a power series 103 giving 2 2 2n Waz Y Wyma E ert 20 z 5 15 Lm n for a particular position x z in the exit pupil y and denote the fractional coordinates of the image field The va
332. ections the right most and left most will be taken These conditions are only true if gas is specified If gas is not specified it returns the x intersection for y in the same manner as MATI MAT2 X A bug in the RELAX capability has been repaired This makes RELAX function more completely and makes it remove triangles for cases where they were left in the past D 16 SSUPREM4 Version 6 0 e Version 6 0 of SSUPREM4 incorporates a number of new models as well as convenience features SSUPREM4 now includes the first available two dimensional silicide model The DEPOSIT ETCH and model statements now include materials TUNGSTEN TITANIUM PLATINUM WSIX TISIX and PTSIX Silicidation can also be performed using user defined materials for other metal systems e The silicide model parameters can be specified in a number of model statements and in the METHOD statement e DEPOSIT EPITAXY and DIFFUSION now allow specification of multiple impurities The multiple impurity deposition capability is exhibited in an example of BPSG type material e The DIFFUSION statement now allows simultaneous oxide growth and impurity predeposition This allows physically based modeling of processes such POCL deposition e One remaining area of concern for modeling such processes is that impurity diffusion in highly doped oxide type materials such as BPSG or PSG will tend to be faster The impurity diffusion coef ficient must typically be adjusted in order to model such
333. ed as a function of temperature and time for lightly doped substrates annealed at atmospheric pressure with no chlorine content in the ambient The parameters in Equation 3 167 are specified for the appropriate oxidant species using the OXIDE statement The pressure dependence and chlorine dependence are described in the following sections Pressure Dependence The effects of pressure on the kinetics of the silicon oxidation process have been studied by Razouk et al 36 for pyrogenic steam and Lie et al 39 for dry oxygen The parabolic rate varies with pressure because of its dependence on the oxidant equilibrium concentration in the oxide C which is directly proportional to the partial pressure of the oxidizing gas The following relation is used to model this dependency P PDEP Bp P 3 168 Here P is the partial pressure of the oxidizing gas in atmospheres and P PDEP is specified on the OXIDE statement See Figure 3 13 for a plot of SiO thickness as a function of time and pressure Chlorine Dependence It has been observed that additions of chlorine during thermal oxidation also affect the parabolic rate constant One possible explanation is that as chlorine enters the oxide film it tends to cause the SiO lattice to become strained which increases the oxidant diffusivity 35 Chlorine concentration dependence on the parabolic oxidation rate is modeled in a similar manner to that of the linear rate constant Gi
334. ed for oxidations on lightly doped substrates annealed at atmospheric pressure with no chlorine content in the ambient The parameters appearing in Equation 3 157 are specified in the OXIDE statement The remaining factors in Equation 3 156 are described in the following sections Silvaco 3 51 ATHENA User s Manual Orientation Dependence The silicon substrate orientation is known to affect the oxidation kinetics 84 35 The influence of orientation on the linear rate constant is modeled as B A in Equation 3 156 The orientation dependencies for lt 100 gt and lt 110 gt orientations are modeled by appropriate reduction factors and B A ri for lt 111 gt substrates is unity Figure 3 12 shows the silicon dioxide thickness dependence as a function of the substrate orientation for several oxidation temperatures TonyPlot 2 6 9 r Files View Plot Tools gt Print Properties v Help X 11 ls Silicon O S lt 110 Silicon lt 100 Silicon Silicon Dioxide Thickness Angstroms Oxidation Time Minutes SILVACO International 1996 Figure 3 12 Silicon Dioxide Thickness versus Time for Different Substrate Orientations and Temperatures 3 52 Silvaco SSUPREM4 Models Pressure Dependence High pressure silicon oxidation allows one to grow relatively thick SiO films while keeping the temperature low so that dopant redistribution is reduced 36 The pressure depende
335. ed implants It is applicable to a wide energy range from 1 keV to few MeV It includes damage accumulation model which allows accurate simulation of dose dependency effect Several improvements are made in analytical implant models Improved handling of wrong user defined or tabulated combinations of skewness and kurtosis for longitudinal profiles The values are corrected to provide legitimate bell shaped profiles The cor rected values could be checked by using the parameter PRINT MOM in the IMPLANT statement Calculations of cluster and dislocation bands from implant profiles parameters MIN CLUSTER MAX CLUSTER MIN LOOP and MAX LOOP are fixed and available for both analytical and Monte Carlo methods A new parameter FULL DOSE has been added If it is set to TRUE the adjusted full dose for the angled implant will be applied A more accurate integration of non Gaussian lateral distribution functions is implemeted User specified models for implant damage lt 311 gt clusters and dislocation loops can be controlled through a C Interpreter file The name of the file is specified in the parameter DAMAGEMOD FN in the MOMENTS statement Diffusion Simulation Features A new numerical scheme for diffusion calculations the Implicit Linear Finite Element Method ILFEM is implemented The ILFEM uses a new internal data structure an advanced spatial discretization scheme an extremely fast and robust linear solve
336. ed with the input file structure files setfiles for TONYPLOT and layout files for MAsKVIEWS will be copied into your current directory 9 9 Silvaco Tutorial Once the input file is in the Deckbuild Text Subwindow bottom panel of the window press the Run button in the Main Deckbuild window or follow the special instructions in the Deckbuild Examples Window to run the input file Most of the ATHENA examples contain preset calls to the graphical postprocessing tool TONYPLoT One or more plots will appear while the selected example is running If you are not familiar with DECKBUILD use a simple example to learn the basic DECKBUILD features and capabilities For more information see the VWF INTERACTIVE TOOLS USER S MANUAL VOL I This will assist you in working through the rest of the tutorial The details of these functions are described in the VWF INTERACTIVE TOOLS USER S MANUAL VOL I Deckbuild Examples Return to index 17 LASER Laser Diode Application Examples 16 THERMAL Thermal Distribution Application Examples 19 SEU Single Event Upset Application Examples 20 DIODE Diode Application Examples 21 INTERCONNECT Interconnect Parasitics Application Examples 22 ATHENA _SSUPREMA4 2 D Process Simulation 23 ATHEMA_ELITE 2 D Topography Simulation 24 4ATHENA_OPTOLITH Optical Lithography Simulation 25 ATHENA FLASH Compound Semiconductor Process Simulation Exampl 26 ATHENA CALIBRATION Proces
337. ee the previous section Figure 2 37 however also has an effect on the base current in the low injection region For a small range of implant doses around the optimum the base doping concentration will also affect the position of the knee or the rate or both of fall off of the base current in the low injection operating region of the device This is most noticeable as a loss of current gain in the low injection region for the alternative standard plot of current gain versus collector current An increase in the base implant reduces the intrinsic resistance and typically increases the base current in the low injection region resulting in a decrease in current gain for very low currents A similar effect to increasing the base doping is observed if the base doping is kept constant but the overall doping is reduced in the mono crystalline silicon region of the emitter You can tune the doping profile in the mono crystalline region of the emitter using three parameters in ATHENA The main physical effect of these ATHENA parameters is to change the doping profile of the emitter in the mono crystalline silicon These process parameters are as follows e The total interstitial concentration in the poly emitter e The dopant segregation effects in the poly emitter e The dopant velocity across the silicon polysilicon boundary The first process parameter will affect how quickly the dopant in an implanted poly emitter reaches the silicon polysilicon bound
338. efined Materials for the list of materials Default is SILICON I IMPURITY specifies an impurities to be used for the dislocation loop scaling see Section 6 2 10 Standard Impurities for the list of impurity names which can be used e g I BORON MIN LOOP CO and MAX LOOP CO define the upper and lower bounds of the dopant concentrations where the loops are placed Dislocation Loop Generation Example The following example switches on the loop model and then places loops in the position where indium concentrations lie between 1e16 and 1e15 cm METHOD I LOOP SINK DISLOC LOOP MIN LOOP CO le15 MAX LOOP CO le1l6 I INDIUM SILICON IMPLANT INDIUM DOSE 1e15 ENERGY 45 For more examples see METHOD CLUSTER INTERSTI HW IAL VACANCY DIFFUSE and IMPLANT 6 28 Silvaco ELECTRODE 6 17 ELECTRODE ECTRODE defines electrodes and names for ATLAS or other device simulation 7 re Syntax ELECTRODE NAME lt c gt X lt n gt Y lt n gt BACKSIDE LEFT RIGHT Description This statement defines a whole material region as an electrode NAME gives a name to the electrode that can be plotted or referenced in TONYPLOT or ATLAS X specifies the horizontal location or x coordinate of the region which will be defined as an electrode Y specifies the vertical location or y coordinate of the electr
339. egions to be adapted on This may be one or several materials at a time The default materials include SILICON OXIDE POLYSILICON ete e Boolean I BORON ILARSENIC specify impurities to be adapted on This may be one or several impurities at a time The available impurities include ILBORON ILARSENIC ILPHOSPHORUS I ANTIMONY I INTERST I VACANCY etc e Boolean DISABLE specifies that materials impurities combinations given are disabled to be effec tive on mesh adapting or smoothing e Float MAX ERR specifies the maximum error allowable before adding points to the mesh unitless Error calculated above this value causes points to be added e Float MIN ERR specifies the minimum error below which points may be deleted from the mesh unitless Error calculated below this value causes points to be removed Both MAX ERR and MIN ERR are calculated using the Bank Weiser error estimator which is defined as 2V Ci ERE T D 1 where h is the average of the edge lengths associated with node i Ci is the impurity concentration at node i e Float CONC MIN specifies the minimum impurity concentration below which adapting will stop units 1 0 cm3 e Float AREA MIN specifies the minimum triangle area below which adding points will stop units cm2 e Float AREA MAX specifies the maximum triangle area below which deleting points will stop units cm2 e Float EDGE MIN specifies the minimum edge length below which adding points
340. el includes dynamic processes of the transformation from crystalline to amorphous state as ion implantation proceeds Each pseudo projectile in the simulation represents a portion of the real dose where N is the number of projectiles aoa 2 3 235 N The deposited energy is accounted for each grid point of the target and accumulated with the number of projectiles As the implantation proceeds deposited energy increases and the crystalline structure gradually turns into an amorphous structure This is quantified by the Amorphization Probability Function as follows feat ap 5e 3 236 Cc Here AE r is the energy deposited per unit volume at the grid point r and Eis the critical energy density which represents the deposition energy per unit volume needed to amorphize the structure in the relevant volume Silvaco 3 81 ATHENA User s Manual It is defined as 3 237 E T oy E T Eu elre B 00 where is activation energy k is Boltzmann s constant and T is the temperature at and above which the infinite dose is required for crystalline to amorphous transition Some experimental values for E E are given by F L Vook 65 In the BCA module the value fr 0 6 corresponds to a fully amorphized state and any additional energy deposited at point r does not contribute to the amorphization process Implantation Geometry Figure 3 23 shows the orientation of the ion beam relative to the crystallographic
341. ellaneous Features and Bug Fixes 1 10 11 12 13 14 15 16 17 18 It is now possible to use clust trans model when impurities other than B P As and Sb present in the structure Also the model can be used in Polysilicon Solid solubility tables are extended down to T 600C Also solid solubility in polysilicon is set equal to that in silicon Added a NEUTRAL type impurity as an alternative to DONOR ACCEPTOR in the IMPURITY statement For example I SILICON is considering as DONOR in GaAs but should be NEUTRAL in Si As the result Si atoms implanted in order to preamorphized silicon crystal would not affect diffusion of other impurities and will not contribute into the net concentration New impurities Nitrogen and Oxygen are added to all relevant statements Also impurity Fluorine is now available in all statements Standard material GERMANIUM is added Fixed a bug in initial gridding The fix makes sure that the distances betweem vertical grid lines remain constant if the SPACING parameters are equal at the adjacent LINE statements Add several aliases to command names to achieve better syntax compatibility with TSUPREM4 DIFFUSION for DIFFUSE LOADFILE for INITIALIZE from a structure file SAVEFILE for STRUCTURE and AMBIENT for OXIDE Made SILICON to be a default material in the STRETCH statement Removed obsolete parameters LABEL and TITLE from the PRINT 1D statement
342. els these parameters control whether the crystalline lattice structure is considered or not Parameters Applicable for All Implant Models IMPURITY specifies the impurity to be implanted see Section 6 2 10 Standard Impurities for the list of impurities BF2 is also available ENERGY specifies the implant energy in keV DOSE specifies the dose of the implant Dose is calculated in a plane normal to the implant direction The units are in cm FULL DOSE specifies that the implanted dose is adjusted to compensate for the tilt angle This type of dose specification is often used for high tilt implants Adjusted Dose DOSE cos TILT Silvaco 6 45 IMPLANT ATHENA User s Manual TILT specifies the tilt with respect to the vertical of the implantation ion beam The units are degrees The default is 7 ROTATION specifies the angle of rotation of the implant relative to the plane of the simulation The units are degrees The default is 30 FULLROTATION specifies that the implant be performed at all rotation angles PLUS ONE synonyms are UNIT DAMAGE and D PLUS and DAM FACTOR synonym is D SCALE specify the implant damage calculation UNIT DAMAGE specifies that the interstitial profile should be a scaled version of the doping profile from the implant DAM FACTOR specifies the scaling factor to be used for the UNIT DAMAGE model At a depth the interstitial concentration from the UNIT DAMAGE model will equal to
343. ent calculations in amorphous silicon as in 58 59 The auxiliary file wserimp in the lt install gt lib athena lt version gt common directory provides a template for specifying implant parameters in the format of standard tables The second type of look up tables are SVDP tables described in the Dual Pearson Model on page 3 68 The format of these tables is much more flexible than the format of the standard tables It also allows parameters for lateral distribution to be added easily The SVDP tables are used by default If no moments are found ATHENA will search through standard tables If it cannot find parameters for a specified energy ion material combination a warning message is issued which will tell you a very small projected range and straggling will be used in simulation for this combination The message will also suggest that you use the Monte Carlo method in order to find the right moments This is the sequence of ATHENA actions in the case when no MOMENTS statement precede the current IMPLANT statement The MOMENTS statement serves to control the moment parameters tables to be used in subsequent IMPLANT statements If you specify the STD_TABLES parameter ATHENA will skip searching through SVDP tables and proceed directly to the standard tables If you specify the USER_STDT or USER_SVDPT parameter then the user defined file specified with the USER TABLE lt filename gt parameter will be used as the first choice Of course if
344. ent descriptions Gas flow specification Gas flow can now be explicitly specified during diffusion calculations This functionality supports the use of mixed ambients and is described in the DIFFUSION state ment description RELAX statement added for improved gridding A new statement RELAX has been added to allow the removal of excess grid points at any time during the simulation This greatly enhances efficiency by allowing free manipulation of the grid Improved MaskViews interface The interface to MASKVIEWS now can be invoked interactively during SSUPREM4 simulation This interface has also been improved to provide for automatic grid generation that is tied to layout information This interface and capability are demonstrated in the first standard example ELECTRODE statement The name and position of electrodes in a SSUPREM4 structure can now be defined using the ELECTRODE statement This information is incorporated in the MASTER structure file format and can be read transparently by SPISCES 2B D 20 Additional SSUPREM4 Changes D 20 1 Oxidation method defaults to compress e The HCL PC parameter has been added to the diffusion statement to allow the inclusion of HCl e The readability of the online help facility has been improved and additional comments have been added e These can be accessed by specifying HELP or HELP lt statement name gt in interactive mode e The initial HELP statement list has been alphabetized e Th
345. ent out all existing line statements and will automatically run line statements generated by MASKVIEWS For example the following output will appear in the DECKBUILD Text Subwindow if the default sec 1 generated for the CMOS Inverter is loaded ATHENA gt LINE X LOC 0 000 SPAC 0 100 TAG LEFT ATHENA gt LINE X LOC 0 300 SPAC 0 100 ATHENA gt LINE X LOC 0 500 SPAC 0 100 ATHENA gt LINE X LOC 0 600 SPAC 0 100 ATHENA gt LINE X LOC 0 800 SPAC 0 050 ATHENA gt LINE X LOC 1 100 SPAC 0 150 ATHENA gt LINE X LOC 1 500 SPAC 0 150 ATHENA gt LINE X LOC 1 800 SPAC 0 050 ATHENA gt LINE X LOC 2 000 SPAC 0 100 ATHENA gt LINE X LOC 2 100 SPAC 0 100 ATHENA gt LINE X LOC 2 300 SPAC 0 100 ATHENA gt LINE X LOC 2 600 SPAC 0 100 TAG RIGHT ATHENA gt LINE Y LOC 0 00 SPAC 0 03 TAG TOP ATHENA gt LINE Y LOC 0 20 SPAC 0 02 ATHENA gt LINE Y LOC 1 00 SPAC 0 10 TAG BOTTOM 2 70 Silvaco Tutorial Using MaskViews for Generating Masks in ATHENA The dry etching capability of ATHENA and the physical etching capability of ATHENA ELITE can be used in conjunction with the mask generating capability provided by DECKBUILD and MASKVIEws A cutline loaded into DECKBUILD has information on the x location of the photomask edges You should specify the sequence of mask creation and stripping steps in the ATHENA input file This ca
346. entration independent and therefore only the first term of this equation is non zero for these impurities It appears that diffusion of Ge in GaAs is proportional to the second power of n n The diffusivity for acceptors is the following 2 Dy H Dit t Day 2 3 240 acceptor n l Different terms are dominant for different acceptors Carbon diffusivity is considered as concentration independent Be and Mg diffusivities are proportional to p n while diffusivity of Zn is proportional to p n It s important to know that some dopants in compound semiconductors are amphoteric and can be either donor or acceptors under certain conditions This means you can use the DONOR and ACCEPTOR parameters in the IMPURITY statement to specify the type of dopant Boundary and interface condition for impurities in compound semiconductors are specified using the transport velocity parameters TRN 0 and TRN E and the segregation coefficients SEG 0 and SEG E The impurity activation in compound semiconductors is calculated using solid solubility model with default value for solid solubility limit for all impurities set at 101 cm Silvaco 3 93 ATHENA User s Manual 3 8 2 Implantation Models Ion implantation models for compound semiconductors are essentially the same as those for silicon The Pearson analytical approximation uses look up tables derived from experiments 79 and calculations 59 The
347. ep This allows the mesh to conform to the dopant after a time step The difference between the dopant contours and the change in the mesh density distribution will thus be limited to the difference of dopant profiles between time steps This difference is substantially smaller than that over total diffusion time Thus mesh adaption can allow more accuracy and minimize the mesh density for the dopant representation at any given time A Simple Example GO ATHENA LINE X LOC 0 00 SPAC 0 1 LINE X LOC 2 00 SPAC 0 1 LINE X LOC 0 00 SPAC 0 1 INIT SILICON C ARSENIC 10E14 DIFE IME 50 TEMP 950 DRYO2 DEPOSIT POLY LEFT PL X 1 2 CH POLY LEFT PL X 1 2 RUCT OUTF MOS_0 STR ERFORM ADAPTIVE MESHING FOR BOTH IMPLANT AND DIFFUSION Ti e u W U E D Z w Q prel O Z DOSE 1 0E13 ENGERY 15 PERSON TILT 0 UCT OUTF MOS_1 STR OSIT OXIDE THICK 35 DIV 6 H OXIDE THICK 35 LANT BORON DOSE 1 0E14 ENGERY 15 PERSON TILT 0 OS_2 STR EMP 1000 TIME 30 F MOS_3 STR HO WH ie Ss 0 QU wt NOUONHE ot C Q Q G Hy ll BS C Q E CH QUIT LISTING 1 A SIMPLE EXAMPLE OF IMPLANT ADAPTIVE MESHING This simple example creates a LDD MOS device structure The initial simple mesh is specified with the four LINE statements This initial me
348. ependencies 62 The local inelastic energy losses are based on the model proposed by Firsov 63 In this model the estimation of the electronic energy loss per collision is based on an assumption of a quasi classical picture of the electrons i e the average energy of excitation of electron shells and electron distribution and motion according to the Thomas Fermi model of the atom In this quasi classical picture the transfer of energy AE from the ion to the atom is due to the passage of electrons from one particle to the other Thus resulting in a change of the momentum of the ion proportional to its velocity v and a rising of a retarding force acting on the ion When ions move away from the atom despite being trapped by ions electrons will return to the atom There is no transfer of momentum calculated back because the electrons fail in higher energy levels The energy loss in the Firsov s Model is calculated as follows 5 3 05973 x Z ap 0205973 X Z Z3 E M oy AR 1 0 31 Z Z R where e Rp is their distance of closest approach in A which is approximately equal to the impact parameter in case of small angle collisions e E is the energy of the moving atom the ion in eV M is its mass in a m u In a binary collision the scattering angles are affected by the inelastic energy loss AE see Equation 3 228 through the parameter f The non local electronic energy losses are based on the model
349. ependent Viscous Model 3 3 4 Linear Rate Constant For short oxidation times and low oxidation temperatures the oxide growth is linearly related to the oxidation time The interface processes oxidant transport across the gas SiO interface and oxidant reaction at the Si SiO interface are the determining factor in describing the growth kinetics In this regime the oxide thickness can be approximated as si a 3 153 where B A is called the linear rate constant and is obtained by dividing Equation 3 135 and Equation 3 134 resulting in the following equation 7 c t 1 3 154 1 The equilibrium oxidant concentration in the oxide C is defined by Equation 3 128 K in Equation 3 128 is specified by the HENRY COEF parameter in the OXIDE statement The gas phase mass transport coefficient h is given by the following Arrhenius relation h TRN O exp TENE k T 3 155 3 50 Silvaco SSUPREM4 Models where the TRN 0 and TRN E parameters are specified in the OXIDE statement The interface reaction rate constant k is determined from Equation 3 154 and experimentally determined values of B A The linear rate constant is composed of several dependencies including orientation pressure chlorine additions and doping effects 9 00 O82 sr os B A is given by LIN L derp HILL T lt L BREAK G _ 3 157 LINH Desp LIN H E T gt L BREAK kT which is the linear rate constant determin
350. er the loop 2 84 Silvaco Tutorial To generate SMILE plots focus exposure latitude curves you need to require a double loop The input language used for a typical double loop is shown below PRINTF ATHENA gt SMILE PRINTF 24 3 3 gt SMILE PRINTF DEFOCUS gt SMILE PRINTF CDS gt SMILE PRINTF DOSE gt SMILE F F I I OREACH I 200 TO 300 STEP 25 OREACH J 1 5 TO 1 5 STEP 0 5 NITIALIZE INFILE ANOPEX12 STR1 AGE DEFOCUS J WIN X LO 5 WIN X HI 5 WIN Z LO 0 WIN Z HI 0 CLEAR EXPOSE DOSE I BAKE DIFF LENGTH 0 05 STRUCT OUTFILE ANOPEX12 J 1 STR DEVELOP MACK TIME 45 STEPS 5 SUBSTEPS 10 STRUCTURE OUTFILE ANOPEX12 J I STR3 PRINTF J ZZZ GAS 1 4 GASS ZZZ 1 4 I gt SMILE END In this smile plot example exposure DOSE is varied in the outer loop and DEFOCUS is varied in the inner loop The output is written to a file called SMILE The difference between the smile plot and the swing plot is that smile plots must distinguish between several types of data To do so a third column called Group is added see the ToNYPLoT chapter in the VWF INTERACTIVE TOOLS USER S MANUAL VOL I The final PRINTF statement prints DEFOCUS J CDs and DOSE I To display the plot outside of DECKBUILD enter the tonyplot da SMILE command and the plot will appear In the
351. er transform plane CIRCLE SQUARE GAUSSIAN and ANTIGAUSS defines or changes the shape of the exit pupil of the projection system The shape of the pupil must be entered as a character string GAMMA defines or changes the GAMMA value for GAUSSIAN and ANTIGAUSS pupil transmittance GAMMA is a parameter that defines the truncation of the GAUSSIAN by the pupil In the limit of GAMMA gt 0 the pupil transmittance will be uniform IN RADIUS and OUT RADIUS defines or changes the intensity transmittance and phase transmittance of an annular zone inside the exit pupil or either the illumination or the projection system This qualifier is used to simulate spatial filtering techniques IN RADIUS and OUT RADIUS are used to define an annular zone in the exit pupil having the pupil transmittance equal to TRANSMIT and producing the phase angle equal to PHASE Radius values are specified in fractions of unity and phase is specified in degrees Note that the annular zones should not overlap The outer radius of an inner zone must be smaller than the inner radius of an outer zone The shape of the annular zone is specified by the shape parameter above The maximum radius is one PHASE specifies the phase shift in degrees produced by the pupil filter 180 lt PHASE lt 180 TRANSMIT specifies the pupil transmittance caused by the pupil filter CLEAR FIL resets the projection filter list Examples This set of commands defines a square aperture in the
352. eraction with a free interstitial according to the following reactions 0 0 ks 0 0 Ks i Pr P ol Ip f gt R te DAT E Da Sog ky k k Silvaco 3 31 ATHENA User s Manual where according to 19 I B Agn r_ Dy Ey EKn K 4A0RyDyexp Ee no o exp 5 E 3 110 Here Ref represents the effective capture radius The elementary jump length is equal to the inter atomic distance ag 2 35 A 0 is the number of dissociating sites E is the self interstitial formation energy defined in Equation 3 76 and En is the formation energy per interstitial for clusters of size n The number of reactions taken into account is specified in the ic mod file The value of effective energy barrier Ag in Equation 3 110 can be represented as the sum of two components tot Ag max byt Eoo 3 111 I eq The first of the two terms represents the change in free energy associated to the change in chemical potential when an interstitial jumps from the supersaturated phase to the cluster The second term is the formation energy per interstitials This parameter is a function of the size and of the crystallographic structure of the cluster The values of the formation energy per interstitial are specified in the ic mod file The first nine parameters corresponding to the formation energy per interstitial for clusters of size 2 through size 10 are defined in 20 For larger clusters the energy of formation per interst
353. erial But for structures containing many non planar layers material regions and for the cases which have not been studied yet experimentally requires more sophisticated simulation models The most flexible and universal approach to simulate ion implantation in non standard conditions is the 3 76 Silvaco SSUPREM4 Models Monte Carlo Technique This approach allows calculation of implantation profiles in an arbitrary structure with accuracy comparable to the accuracy of analytical models for a single layer structure ATHENA contains two models for Monte Carlo simulation of ion implantation Amorphous Material Model and Crystalline Material Model Both of them are based on the Binary Collision Approximation BCA and apply different approximations to the material structure and ion propagation through it Nature of the Physical problem A beam of fast ions energy range approximately 50 eV amu to 100 keV amu entering crystalline or amorphous solid is slowed down and scattered due to nuclear collisions and electronic interaction Along its path an individual projectile may create fast recoil atoms that can initiate collision cascades of moving target atoms These can either leave the surface be sputtered or deposited on a site different from their original one Together with the projectiles being deposited in the substrate this results in local compositional changes damage creation and finally amorphization of the target Depending on t
354. erial is included within th mesh If you do not include REGION statement between the LINE statement and the INITIALIZI statement you can define the material on the INITIALIZE statement amp FI Gl MATERIAL specifies the material in a region see Section 6 2 9 Standard and User Defined Materials for the list of materials XLO YLO XHI and YHI specifies the bounds of the region rectangle The value lt string gt should be one of the tags created in a preceding LINE statement Examples The following REGION statement specifies silicon as the material for the entire mesh LINE X LOC 0 SPA 1 TAG LEFT LINE X LOC 1 SPA 0 1 LINE X LOC 2 SPA 1 TAG RIGHT LINE Y LOC 0 SPA 0 02 TAG SURF LINE Y LOC 1 SPA 0 1 TAG BACK REGION SILICON XLO LEFT XHI RIGHT YLO SURF YHI BACK INIT Note If you do not use REGION statement and no material appears on the INIT statement then ATHENA assumes Silicon is the starting material If you do not specify enough regions to describe the material at every part of the grid it may not be detected until the execution of a subsequent command For more examples see INITIALIZE 6 96 Silvaco RELAX 6 54 RELAX RELAX loosens the grid in an ATHENA mesh Syntax RELAX MATERIAL X MIN lt n gt X MAX lt n gt Y MIN lt n gt Y MAX lt n gt DIR X DIR Y SURFACE DX SU
355. es eee 3 1 3 1 Diffusion MODGIS croait doh earcareaele ai wa onal oe Ba ntinad nanan NE CEE Lape n ea tie 3 1 3 1 1 Mathematical Description 352 lance Papas eaea wee Mota Ae eee ny ace as eta ie 3 2 SA2 URC s PON Model ras i Season kama a one heat oe cian A rater oh ts iene AAA Nera ETA act N ek 3 5 3 1 3 Impurity Segregation Model n crete ne ei hele Neuadd de Bio Bul Sih sie ei ote haat 3 6 3 1 4 The Two Dimensional Model sat cthedesad bout E eet eee Pein ee og tet Mile Sia tort Ste 3 7 3 1 5 The Fully Coupled Model nunana nnana 3 16 3 1 6 Electrical Deactivation and Clustering Models 00 e cece eee ene ees 3 18 3 1 7 Grain based Polysilicon Diffusion Mod6l 23 cics4s Sees eee eel ge cheat eer es ee A 3 21 3 2 Advanced Diffusion ModelS is ices ica tenet siwet sian dene e een Weed t be dae ewe awd ee aai 3 23 3 2 1 Classical Model of Dopant Diffusion CDD 0 cece teenies 3 24 3 2 2 Solid Solubility Model tse key eens pee ela Aes eps ig Sees ree 3 31 3 2 3 Interstitials Clusters Model IC sc evens ce ancea aanndear ae noes ee ee shane edar ee ne ceabensad 3 31 3 2 4 Vacancy Cluster Model VC ii teeiteeerieeedeeesae ers deibta bee webiia edinelieeniies 3 33 3 2 5 Electrical Deactivation and Clustering Models DDC 00 cence eens 3 34 3 20 Typical EXaMpleS cates ccc kee ee ee ete EEEE we tee RA ee nsw A alga Week a 3 36 3 2 Oxidation Models Geni at sacle dy cate teumtnaie ete EET aia me
356. escribed Note that the Distance parameter is equivalent to the location parameter in ATHENA Also the Add button is equivalent to the Insert button of the ATHENA Mesh Define Menu Then press Return after entering the Distance or the Spacing values If the Distance and Spacing are set as shown in Figure 2 49 the grid will be the same in the Y direction as the grid produced using the ATHENA Mesh Define Menu 2 66 Silvaco Tutorial Maskviews Yertical grid control Distance 0 00 Space 0 03 Distance 0 20 Space 0 02 Distance 1 00 Space 0 10 Distance 9 2 0 00 H 10 00 Spacing 0 02 0 00 k 4 00 Figure 2 49 Vertical Grid Control Popup MASKVIEWS also controls the initial ATHENA grid in the X direction MASKVIEWS generates ATHENA line statements for each mask edge on valid layers crossed by a cutline The grid spacing and the validation of layers can be set by the ATHENA Grid Control Menu See Figure 2 50 To open this menu select Grid gt X from the Define menu Maskviews ATHENA grid control Current layer POLY first poly defi Offset from point 0 00 Valid layers Spacing at edge 0 05 Grid inside edge Distance 0 30 Spacing 0 15 Grid outside edge Distance 0 20 Spacing 0 10 Figure 2 50 ATHENA Grid Control Menu Figure 2 50 shows the line locations and spacings preset for the POLY layer This set of parameters means that for each POLY edge crossed by a cutline three line statements are to be in
357. ese statements manipulate the geometry or attributes of the structure or create output files e ADAPT MESH enables the adaptive meshing algorithm e ADAPT PAR specifies adaptive meshing parameters e BASE PAR defines adjacent mesh characteristics of an automated base mesh e ELECTRODE names electrode regions e GRID MODEL defines a template file containing adaptive meshing commands e PROFILE causes ATHENA to read in an ASCII file of depth and doping data e RELAX loosens the grid within a user specified area e STRETCH allows changes in structure geometry by stretching at a horizontal or vertical line e STRUCTURE writes the mesh and solution information into a file This is the main output statement for generating program data to be plotted 6 4 Silvaco ATHENA Statements List 6 2 3 Simulation Statements These statements apply physically based models for processing operations to the structure BAKE performs post exposure or post development photoresist bake DEPOSIT deposits a material layer DEVELOP performs photoresist development DIFFUSE performs a time temperature step on the wafer and calculates oxidation and diffusion of impurities EPITAXY performs high temperature silicon epitaxial growth ETCH performs a geometric or machine type etch on the structure EXPOSE models photoresist exposure IMAGE calculates a 2D or 1D aerial image IMPLANT models ion implantation POLISH simulates chemical mechanical polishin
358. esist called SECRETX RATE DEVELOP NAME RESIST TEST E1L DILL 1 E2 DILL 0 5 E3 DILL 003 For more examples see EXPOSE BAKE and DEVELOP Silvaco 6 89 RATE DOPE ATHENA User s Manual 6 50 RATE DOPE RATE DOPE specifies the enhancement parameters for dopant enhanced etching in ELITE Syntax RATE DOPE MACHINE lt c gt MATERIAL I IMPURITY ENH MAX lt n gt ENH SCALE lt n gt ENH MINC lt n gt Description This statement is used to define dopant enhanced etching and can be applied to an etch machine defined using the RATE ETCH statement Note Dopant enhanced etching is not applicable to MC PLASMA etch model MACHINE specifies the machine name for which the dopant enhanced model to be applied MATERIAL specifies material in which the dopant enhanced model to be used see Section 6 2 9 Standard and User Defined Materials for the list of materials IMPURITY INTERST VACANCY SXX SYY and SXY specify impurity or other solution which concentration is used in the dopant enhanced etching model see Section 6 2 10 Standard Impurities for the list of impurities ENH MAX specifies the maximum enhancement ENH MINC specifies the solution value below which enhancement decays ENH SCALE specifies the spread of the enhancement over solution values i e how qui
359. et of lines close to the silicon silicon dioxide interface For this to work the existing grid spacing at the interface must be greater than 0 005 microns Note that since the lines are added only between existing mesh lines and the interface the lines must be specified in this order i e getting closer to the surface ADAPT MESH SILICON OXIDE ADD 1I LINE 0 005 ADAPT MESH SILICON OXIDE ADD I LINE 0 001 ADAPT MESH SILICON OXIDE ADD 1I LINE 0 0005 ADAPT MESH SILICON OXIDE ADD I LINE 0 0001 For more examples see ADAPT PAR Silvaco 6 11 ADAPT PAR ATHENA User s Manual 6 5 ADAPT PAR ADAPT PAR specifies adaptive meshing parameters Syntax ADAPT PAR ATERIALS I IMPURITIES I INTERST I VACANCY DISABLE MAX ERR lt n gt MIN ERR lt n gt CONC MIN lt n gt AREA MIN lt n gt AREA MAX lt n gt EDGE MIN lt n gt EDGE MAX lt n gt MIN ADD lt n gt AX POINT lt n gt MAX LOOP lt n gt IMPL SMOOTH DIFF SMOOTH IMPL SUB DOSE ERR lt n gt DOSE MIN lt n gt DIFF LENGTH lt n gt ANISOTROPIC Description ADAPT PAR specifies parameters used during adaptive meshing enabled by the ADAPT MESH statement MATERIALS specify standard materials or user specified material regions in which mesh adaptation takes place see Section 6 2 9 Standard and User Defined Materials for the list of materials On
360. eters The mask to be imaged will already be defined either by a MASKVIEWS sec file or by the LAYOUT command If a MASKVIEWS sec file is used the IMAGE command will be of this form IMAGE INFILE sec If the mask is defined using the LAYOUT command the mask features will be stored in memory and the only required input related to mask features is the OPAQUE CLEAR specification OPAQUE specifies the background intensity transmittance to be zero CLEAR specifies the background intensity transmittance to be one OPAQUE is the default setting OPAQUE and CLEAR cannot be used with an input file from MASKVIEWS The Image Window not the Computational Window is specified with the parameters WIN X LOW WIN Z LOW WIN X HI and WIN Z HI These parameters define the minimum and maximum range of the x and z values see Figure 2 65 The aerial image is then calculated only inside this window This allows for faster computation when you only want a cross section If you want a simple cross section set the window parameters for z WIN Z2 LO WIN Z HI to the same value for a cross section parallel to the x axis This value WIN Z LOW WIN 2Z HI gives the location of the cross section Z A WIN Z HIGH Image WIN X LOW WIN X HIGH gt xX Window WIN Z LOW Figure 2 65 The Image Window can be Placed Anywhere in the XZ Window The resolution in the image window can be controlled by tw
361. eviations in performing optical proximity correction Thus it must be used together with the OPC parameter OPC specifies the normalized intensity level for OPC evaluation An image file SEC will be generated for this particular intensity level and is to be used by MASKVIEWS FLIP Y indicates that the structure should be flipped around the x axis This is used to invert structures for backside processing MIRROR LEFT RIGHT TOP and BOTTOM mirrors the grid about its left or right top or bottom boundary respectively This is useful for turning half of a MOSFET simulationstructure into full structure for subsequent ATLAS simulation The default reflection is about the right axis INTENSITY specifies the aerial image intensity distribution to be saved in the output file MASK specifies layout mask information to be saved in the output file REMOVE GAS specifies that the gas region is to be removed from the output structure Currently the overlaying gas region is automatically added to the structure for Monte Carlo etch and BCA implant simulations SIGE CONV converts the layer of silicon that is highly doped with Ge into a Sij_ Ge layer so it can be used in ATLAS TWO DIM specifies that the structure to be transformed into 2D if it s still 1D Examples The following statement writes the current structure to a file called test str STRUCTURE OUTFILE TEST STR The following statement saves an aerial image and masks c
362. extra Si SiO interface recombination reactions In the bulk extra terms for point defect recombination apply at high concentrations where statistically a high level of dopant defect pairing is prevalent D D K IIFACTOR tit r K 3 56 r D Dy r 5 D D K IVFACTOR t K 3 57 r D D The extra model parameters calibrate the ratio of effective capture cross sections of dopant defect defect to defect defect recombination mechanisms The IIFACTOR and IVFACTOR parameters can be set in the INTERSTITAL statement as follows INTERSTITIAL SILICON IVFACTOR lt n gt IIFACTOR lt n gt Silvaco 3 17 ATHENA User s Manual RTA Diffusion Modelling SSUPREM4 has the capability to model rapid thermal annealing RTA processes within the framework of existing diffusion models i e the two dimensional model and the fully coupled model Since RTA is a short time thermal cycle involving steep temperature ramping to high temperatures Transient enhanced diffusion TED will dominate whenever a significant amount of lattice damage is prevalent Because the amount of dopant diffusion is intimately coupled to the evolution of the point defect populations you can calibrate these models to RTA conditions by tuning the point defect related parameters The ratio of interstitial damage in the form of 311 clusters to that in the form of free interstitials and the characteristic time for dissolution of interstitial cluster
363. f Spacial Moments RANGE RP specifies the projected range Units are microns STD DEV DRP specifies the standard deviation Units are microns GAMMA SKEWNESS specifies the third moment Default is 0 0 KURTOSIS specifies the fourth moment Default is 3 01 LSTD DEV LDRP specifies the lateral standard deviation Units are microns SKEWKXY specifies the mixed third moment KURTXY specifies the lateral mixed fourth moment KURTT specifies the lateral fourth moment Default is 3 0 SRANGE SRP specifies the projected range for second Pearson Units are microns 6 70 Silvaco MOMENTS SSTD DEV SDRP specifies the standard deviation for second Pearson Units are microns SGAMMA SSKEW specifies the third moment for second Pearson function Default is 0 0 SKURTOSIS specifies the fourth moment for second Pearson function Default is 3 01 LSSTD DEV LSDRP specifies the lateral standard deviation for second Pearson Units are microns SSKEWXY specifies the mixed third moment for second Pearson Default is 0 0 SKURTXY specifies the mixed fourth moment for second Pearson Default is 0 0 SKURTT specifies the lateral fourth moment for second Pearson Default is 3 0 DRATIO specifies the dose ratio R in the double Pearson function Default is 0 9 Reset Parameter IGNORE_MOM specifies that all previous MOM Examples ENTS statements will be ignored The MOMENTS statement is used to define user moments through a convenient command language
364. f all the PAC has been decomposed R1 KIM must be expressed in microns sec R2 KIM corresponds to the dissolution rate of the unexposed resist material R2 KIM must be expressed in microns sec R3 KIM corresponds to the dissolution sensitivity of the resist material R3 KIM is dimensionless R4 KIM corresponds to a specific depth into the resist film for surface retardation effects R4 KIM must be specified in microns R5 KIM describes extraordinary 6 88 Silvaco RATE DEVELOP retardation effects R5 KIM is dimensionless positive and less than one R6 KIM describes extraordinary retardation effects R6 KIM is dimensionless positive and less than one R7 KIM describes extraordinary retardation effects R7 KIM is dimensionless positive and less than one R8 KIM describes extraordinary retardation effects R8 KIM is dimensionless positive and less than one R9 KIM describes extraordinary retardation effects R9 KIM is dimensionless positive and less than one R10 KIM describes extraordinary retardation effects R10 KIM is dimensionless positive and less than one CO EIB C1 EIB C2 EIB and C3 EIB are the parameters for the Eib development model DIX 0 and DIX E specify pre exponential constant in cm sec and activation energy in eV for diffusion of photoactive compound that are used in the post exposure bake Examples The following statement defines the Dill development parameters for a user defined r
365. f simplifying simulation of mask deposition over highly non flat structures A region to be etched may be any area not containing a mask on a clear field layer or any area containing a mask on a dark field area You can specify this in the Field section in the MaskViews Layers Popup In the case of the POLY mask and cutline in Figure 2 56 the barrier layer will be etched to the left of x 0 8 and to the right to x 1 8 The following echo output will appear in the Deckbuild Text Subwindow as the result of defining of the POLY mask HENA gt DEFINING POLY MASK HENA gt ASK NAME POLY BARRIER THICK 0 10 UCT OUTFILE HISTORY 9 H BARRIER START X 0 100 Y 20 H CONT X 0 100 Y 20 H CONT X 0 800 Y 20 DONE X 0 800 Y 20 UCT OUTFILE HISTORY 10 H BARRIER START X 1 800 Y 20 H CONT X 1 800 Y 20 H CONT X 2 800 Y 20 H DONE X 2 800 Y 20 UCT OUTFILE HISTORY 11 O pp DD DDD DD Db YI Zz S Vv v QARAW WY I D z A D V n vs If the Reverse Mask box is checked in the ATHENA Photo popup the following lines will be inserted into the input file DEFINING POLY MASK MASK NAME POLY REVERSE and the effect of the field attribute is reversed i e the barrier area will be etched between x 0 8 and x 1 8 When the mask is defined the ATHENA dry etch capab
366. f simulation is difficult because it needs to take into account a large number of phenomena including strong defect recombination at the surface and the fact that TED duration is function of the implant setup and subsequent diffusion duration Despite of these difficulties Figure 3 8 shows a very good fit with experimental annealed data which proves the excellent quality of the model Nevertheless as presented in the literature the TED effect decreases at very low implant energy and high temperature annealing This explain why the simulations done using CDD and IC models and only CDD model sometimes give reasonable agreement with experiments Tony Piot VZET LA 2 Fie View Plot Tools Printe Properties Help Boron Diffusion after implantation Low energy implant RTA As implanted B 2keV 1E14enr Annealed at 1000 C 10sec Simulation As implanted with SVDP Simulation with full PLS model Simulation with classical IC model Simulation with classical model b VE 2 Q 4 E 8 fe O 0 01 0 02 0 03 od d Depth um Cikk to place P changes alignment o drag to get leader SILVACO Intematknal 2003 Figure 3 8 Simulation of boron diffusion at 1000 C during 10 s after an implantation at 2 keV with a dose of 1 10 4cm Experimental data are from 28 Experiment with arsenic implanted at 2 keV The PLS model is not only design for the boron diffusion but also for other common dopan
367. f the effective source The condenser system is assumed to be diffraction limited that is free of aberrations Residual aberrations of the illuminator do have an appreciable influence on the final image for Koehler type illumination systems as shown by Tsujiuchi 104 5 2 Silvaco OPTOLITH Models Figure 5 1 shows a schematic diagram of a generalized optical system The actual source and the condenser system are replaced by the equivalent effective source having an irradiance distribution of g xo Zo The effective source for the object plane U is taken to lie in the exit pupil reference sphere of the condenser lens This means that directing from arbitrary points xg zg on the effective source plane waves propagate towards the object plane U having irradiance values of y xg Zo source condensor reticle plane Imaging system Image plane U U Figure 5 1 Schematic Diagram of a Generalized Optical System The reduced coordinates 103 on the object plane are defined as follows EE pea 5 3 A v Fon sina y 5 4 where amp and n are the Cartesian coordinates of the object plane 27 is the absolute value of the wave vector and n sina is equal to the numerical aperture NA of the imaging system Primed quantities indicate the corresponding coordinates and angles in the image space of the projection system The fractional coordinates on the object pupil spheres are defined as follows x g 5 5 h z il 5 6 h where
368. f the mask points By positioning this window so that the mask cell in the object plane is covered multiple image cells can be calculated Silvaco 5 7 ATHENA User s Manual 5 3 3 Computation Time To increase computation first use a very coarse mesh for screening type simulations and then refine the mesh as you approach specific points of interest Computation time is linearly dependent on the number of source points which is determined by the coherence factor 5 8 Silvaco OPTOLITH Models 5 4 The Exposure Module The Exposure Module computes the intensity distribution in the photoresist through the numerical solution of the Helmholtz equation Equation 5 24 VE Kn E 0 5 24 Here E is the electric field n x y is the complex refractive index of media and k is the wave number The specific solution of this general equation is determined by a set of boundary conditions Generally the target substrate coated with the resist film consists of an arbitrary number of different materials with n in Equation 5 24 and interfaces between material regions Also material refractive index can depend on the absorbed dose i e n can vary with the exposure time According to the electromagnetic theory any field distribution can be represented by unique set of plane waves Generally such a set has an infinite number of terms But the main contributions are the incident wave and reflections from interfaces Therefore
369. f the radiation This is acceptable if all convergence angles are small According to Watrasiewicz 100 who experimentally investigated the limiting numerical aperture the breakdown of the scalar theory occurs at angles of convergence greater than 30 which corresponds to a numerical aperture of 0 5 Similar results were published by Richards and Wolf 101 who used theoretical calculations to investigate the electromagnetic field near the focus produced by an aplanatic system working at a high convergence angle They also found appreciable departures from scalar theory for convergence angles larger than 30 Since the convergence angles are calculated in air we can assume that the accuracy of this model is even better inside the photoresist where angles are reduced in accordance with Snell s law Consequently it can be stated that the scalar diffraction theory gives a reliable limit for imaging system numerical apertures of 0 5 The approach used for calculating the image irradiance distribution is based on the work of Hopkins 102 and 103 which showed the partially coherent illumination of the object structure can be simulated by the incoherently illuminated exit pupil of the condenser The exit pupil serves as an effective source which produces the same degree of coherence in the illuminated object plane as the actual condenser system The degree of coherence in the object plane is therefore determined by the shape and angular size o
370. face It has been also demonstrated that the fundamentals of the point defect properties are critically important in accurate prediction of device behavior For example the reverse short channel effect The recombination flux of silicon self interstitials at a nonoxidizing interface T is given by the formula ame all Tie z0 ve 3 105 tes 7 q q where Ly and L are called the recombination length at the surface These parameters can be used for adjusting the recombination rate at the surface and are specified in the defect mod file 3 30 Silvaco SSUPREM4 Models In the case of dopant implantation an exodiffusion may occur during the thermal treatment Therefore you can write the dopant flux at the surface as o AV 3 106 tot oA Jav ip Jir r where caz and ogy are the exodiffusion coefficients for each dopant defect pair and are defined in dopand mod files These parameters are defined by the following Arrhenius functions E _ 0 OAI Oy O4 exXp 0 E oar ar Oy ot yexp f 3 107 kT 3 2 2 Solid Solubility Model The model for precipitation is assumed to be a constant solid solubility cut off This means that all solute atoms above the solid solubility level will form a precipitate almost instantaneously This model is activated by adding the SS parameter to the METHOD PLS statement The rate equation for the solid solubility model can be formulated as follows 23 DA A A T for A gt
371. ffusion Model is identical to the Two Dimensional Model Be familar with that model before reading any further The one important difference is that the diffusion of the defects is now influenced by the diffusion of the dopants by the addition of the joint pair fluxes to the flux terms in the governing equation of the defects Therefore there is a true two way interaction between the diffusion of dopants and the diffusion of point defects which gives this model its name The fully coupled model is slightly more CPU intensive than the two dimensional model but encompasses the capability of reproducing certain important aspects of semiconductor processing such as the Emitter Push Effect in the case of phosphorus diffusion From a physical viewpoint however this original fully coupled model suffers from the shortcoming of not explicitly representing pairs and the consequential lack of a subdivision of defects and dopants into paired and non paired fractions Therefore this model cannot reproduce the saturation of the dopant diffusivity that is believed to occur at very high damage concentration due to a total pairing of dopants In other words the model relies on the dilute approximation i e the assumption that the concentration of pairs is much smaller than both the dopant and the defect concentrations To use the Fully Coupled Model specify parameter FULL CPL inthe METHOD statement 3 16 Silvaco SSUPREM4 Models The Fully Coupled M
372. fusion and flux equations for vacancies are largely similar to the interstitial equations described above OC i V J R Here Jy is the flux of vacancies and Rpg is the bulk recombination rate The Vacancy Flux Expression is 3 47 3 48 C Sy DyCy lt Cy which correctly accounts for the effect of an electric field on the charged portion of the vacancies by C taking the gradient of the normalized concentration TA The term Dy is the diffusivity of free vV vacancies not to be confused with the pair diffusivity Day which was mentioned in Section 3 1 2 The Fermi Model The vacancy diffusivity is set according to the following equation D D vexp P2 3 49 where the D 0 and D E parameters are set on the VACANCY statement 3 14 Silvaco SSUPREM4 Models The bulk recombination rate Rp is a simple reaction between vacancies and interstitials that assumes that any interstitial is equally likely to recombine with any vacancy regardless of their charged states This assumption may overestimate the recombination rate but is a commonly applied assumption The equation is expressed as x Ry RCC CFC 3 50 where K is the bulk combination coefficient and is specified as K KR Oexp ARE 3 51 where the parameters KR 0 and KR E are user definable in the INTERSTITIAL statement This is the same equation for bulk recombination as described earlier for interstitial bulk recombination
373. g energy for each Boron Interstitial Cluster and can be specified and modified in the boron mod file Phosphorus For phosphorus the similar situation can be considered 22 Therefore phosphorus atoms can form clusters with self interstitials These clusters have been experimentally observed and are called phosphorus interstitials clusters PIC The following reactions are taken into consideration kp I RA P I SA PIAT gt Pio 3 123 kpot Pl These and additional reactions and their parameters can be specified in the phosphorus mod file Arsenic The case of arsenic is a little bit different Arsenic migrates through the vacancy and interstitial mechanisms with roughly the same proportion It is well known that arsenic can form clusters with vacancies of type As V Therefore the DDC model needs to take into account the following reactions hich hs y zr As AsV As AsV As V 7483V V 3 124 k asv Asg V Silvaco 3 35 ATHENA User s Manual Kev As3V AsV AS4V V 3 125 AsgV The system of equations for DDC model is a ot w at ear V J GRy GRy GRyy_p GRaio V Jy GRy GRyay GR4ay_y GRayo V JIy GRy GR _y GRue aye 3 126 E V Jay GRay GRyy_ 1 GRave OA Ss VS 4y t GRyy GR yy GR y_ 1 GRarc GRavos OAIC ar GRiic OAVC Ba GRiyo where GRayc GRajc are generation recombination terms caused by
374. g in the ELITE module STRESS computes the thermal elastic stresses STRIP removes photoresist or another user specified material 6 2 4 Model Statements These statements are used to change model parameters and coefficients The parameters are described in the statement descriptions When starting up ATHENA executes the model statements in the file named athenamod located in the SILVACO lib athena subdirectory corresponding to the version number and system type of ATHENA that you are running This file contains the default parameters for most model statements ABERRATION defines aberration parameters of the optical projection system CLUSTER specifies parameters of 311 cluster model ILLUMINATION describes the photolithographic illuminating system ILLUM FILTER defines filters used in the illumination source for photolithography IMPURITY sets the coefficients of impurity kinetics INTERSTITIAL sets the coefficients of interstitial kinetics LAYOUT describes the mask reticle for imaging MATERIAL sets the coefficients of various materials METHOD sets the numerical options or models for solving the equations MOMENTS specifies moments for Pearson implant model OPTICAL specifies the coefficients of reflection and refraction OXIDE specifies oxidation coefficients PROJECTION defines the photolithographic projection system PUPIL FILTER defines filters in the pupil plane RATE DEPO specifies deposition rates for machine type deposits
375. g the depth of simulation however if the combination of large grid spacing is deep in the substrate and the RELAX statement is used appropriately Silvaco 2 33 ATHENA User s Manual Simulating lon Implantation Ion implantation is the main method used to introduce doping impurities into semiconductor device structures Adequate simulation of the ion implantation process is very important because modern technologies employ small critical dimensions CDs and shallow doping profiles high doses tilted implants and other advanced methods The IMPLANT statement can be set by using the ATHENA Implant Menu Figure 2 28 To open this menu select Process gt Implant in the Commands menu Impurity Dose ions cm2 Energy Ke Model Tilt degrees Rotation degrees Continual rotation Material type Point defects Deckbuild AT Phosphorus HENA Implant Arsenic Bf2 Antimony Silicon Zinc Selenium Beryllium Magnesium Aluminum Galliurn Carbon 4 0 1 0 j 3 9 Exp E 13 EQ O 500 0 50 9 ie o s 180 O crystaline Damage farter to amp Comment gt Channel implant WRITE Figure 2 28 ATHEN A Implant Menu The following gives the minimum set of parameters that should be specified e Name of implant impurity e g boron e Implant dose using the slider for the pre exponential value e g 4 0 and the Exp menu for t
376. gh a process called Ostwald ripening But here they re considered as existing immediately after the implantation The cluster release rate follows a simple exponential decay in time specified by O 317 1 t Rin fe 4 exp 4 3 33 where f x is the as implanted profile of 311 clusters and t is an Arrhenius type temperature dependent time constant calculated from 3 34 r TAU 311 0exp TAU3ILE kT where the TAU 311 0 and TAU 311 E parameters can be specified in the CLUSTER statement The profile f x of the 311 clusters is created from a previous IMPLANT statement For more information see Section 3 5 5 Ion Implantation Damage To activate this model a previous IMPLANT statement has to introduce 311 clusters with the CLUSTER DAM flag in the METHOD statement For example METHOD CLUSTER DA CLUSTER BORON MIN CLUSTER 1 0E17 MAX CLUSTER 1 0E19 CLUST FAC 1 4 TAU 311 0 8 33e 16 TAU 311 E 3 6 SILICON IMPLANT BORON ENERGY 100 DOSE 1E15 Here the METHOD statement switches the model on and the CLUSTER optional statement decides the location and scaling of the 311 cluster profile In this example clusters are present in the regions of the substrate where the chemical boron concentration is between 1 0e17 cm and 1 0e19 cm which are scaled by a factor of 1 4 relative to the boron concentration Notice that the activation energy for TAU
377. gh temperature linear steam oxidation rate for silicon the following syntax can be used OXIDE SILICON WET LIN H 0 lt real gt LIN H E lt real gt whereas for polysilicon the syntax is OXIDE POLY WET LIN H 0 lt real gt LIN H E lt real gt C 2 Silvaco Hints and Tips Question When simulating a structure with a heavily doped polysilicon gate unreasonably high concentration of the impurity is sometimes observed at silicon oxide interface under the gate Is it possible to avoid this situation Answer The impurity transport through oxide is controlled by the impurity diffusion coefficients within oxide and the impurity transport coefficients at the poly oxide and oxide silicon boundaries Not all of these coefficients are well characterized If you know that for your process the impurity diffusion through oxide is negligible you may prevent the impurity transport from polysilicon through oxide into the substrate by specifying zero transport coefficient as follows lt IMPURITY NAME gt POLY OXIDE TRN 0 0 0 If the impurity concentration at the gate oxide silicon interface is measured you can use the measured value for tuning the TRN 0 parameter Question In which cases should the viscous oxidation model with stress dependence be used Which parameters should be tuned to match experimental shape of the grown oxide Answer The viscous stress dependent model is described in C
378. gt ST ABSERR lt n gt GAUSS DIAG wu LOOP SINK POL MODEI SIGE C M VE OCITY T D RELERR lt n LEU CG RELERR lt n BACK lt METHOD selects numerical methods and models for diffusion and oxidation Y DIFF IN TEMP lt RAPS PSTI n gt PAC gt TD ABSERR lt gt FU ABSERR lt n gt BLK KNOT FUL 1 FAC PDINIT TIME REDO OXID Col wu E lt n gt ERE ERF1 C ERFG LIFT POLY LIF GRI GRI GLOO OXIDE OX OBFIX lt n gt FILL ADAPT DEPO SMOOTH ETCH SMOOTH STRESS HIST Description ERF2 COMPR ESS I DE FEC lt n gt D OXIDE lt n gt D SITLICI lt n gt P IMAX lt n gt BRARLY lt n gt OXIDE OXID I GRI GLOOP EMIN lt n gt LATE lt n gt P ERIMET T VISCOUS E LIFT NITRID GRIDINIT OX lt n gt DINIT SI lt n gt RBDF FORM OX THR GLOOP EMAX OXIDE ER lt n gt REL TRUNC DEF lt n ULA ESH lt n gt lt n gt lt n gt DIFF SMOOTH STEP SMOOTH n gt n gt gt ITLIM lt n gt SKIP SIL This statement is used to set flags to select the various mathematical algorithms that will be used to produce the simul
379. gt DVP E lt n gt DVPP 0 lt n gt DVPP E lt n gt SOL SOLUB CLUSTER ACT CTN 0 lt n gt CTN E lt n gt CTP 0 lt n gt CTP E lt n gt SS CLEAR SS TEMP lt n gt SS CONC lt n gt ACT FACTOR lt n gt TRACT 0 lt n gt TRACT E lt n gt TRACT MIN lt n gt MATERIAL SEG 0 lt n gt SEG E lt n gt TRN 0 lt n gt TRN E lt n gt TRNDL O lt n gt TRNDL E lt n gt PD DIX 0 lt lt n gt gt PD DIX E lt n gt PD EFACT lt n gt PD SEG E lt n gt PD TAU lt n gt PD SEGSITES lt n gt PD GROWTH 0 lt n gt PD GROWTH E lt n gt PD CRATIO lt n gt PD SEG GBSI lt n gt Description This statement allows to specify coefficients of impurity diffusion transport segregation and so on Generic Parameters I IMPURITY is the name of impurity which parameters to be specified see Section 6 2 10 Standard Impurities for the list of impurities DONOR ACCEPTOR and NEUTRAL specify the type of the impurity in the given material Default is NEUTRAL MATERIAL specify the material in which the impurity parameters apply as well as MATERIAL1 for the segregation and transport parameters on the boundary between two materials see Section 6 2 9 Standard and User Defined Materials for the list of materials AT NUMBER and AT MASS specify the atomic number and atomic mass of the i
380. h general set dose 1e13 set energy 70 set ion boron Set non standard material set mat tisi define substrate material init mat First use MC method and save the structure file including moments implant ion dose dose energy energy monte n ion 20000 print mom struct outf tmpfile str EXTRACT AND RE USE THE MOMENTS THIS MIGHT BE IN A SEPARATE ATHENA RUN H extract analytical moments stored in structure file i extract init infile tmpfile str extract name rp param RP extract name drp param DRP extract name skew parame SKEW extract name kurt param KURT Use them in the moments statement moments mat i ion dose dose energy energy i rp rp drp drp skewn skew kurtos kurt Now analytical implant can be used implant ion dose dose energy energy print mom any pearson F 1 ATHENA started ATHENA Figure C 11 Syntax for extracting implant parameters from a Monte Carlo simulation C 14 Silvaco Hints and Tips File View Ploty Tools Print Properties gt Help ATHENA 70 ke Boron Implant into Titanium Silicide i Sy p Ry 3 F J N 5 A N e TAA i x N J g R F 16 ie i 3 Monte Carlo Simulation i Pearson Using Extracted Moments 15 T T T T T T T T _ T 7 ia T T 0 0 1 0 3 0 4 SILVACO International Figur
381. h is the radius of the pupil The fractional coordinates of the exit pupil of the condenser are given by Xo 5 7 Q Is Silvaco 5 3 ATHENA User s Manual Z Zy 5 8 oO In these equations ng sind 5 9 n sind where ag and a are angular semi apertures of the condenser and the objective respectively ng and n are the refractive indices in the image space of the illuminator and the object space of the imaging system usually both are set to one The ratio o is the radius of the effective source referred to the aperture of the objective and governs the degree of spatial coherence in the object plane The limits o gt 0 and o correspond respectively to coherent and incoherent illumination The object is taken to be infinitely thin Therefore a complex amplitude transmission function can describe the object which gives the change in magnitude and phase produced on the radiation passing through it The object has the complex transmission A u v Its real part is given by H A u v 1 in transparent areas 5 10 0 in opaque areas The complex amplitude of the Fraunhofer diffraction pattern on the entrance pupil reference sphere at E of the imaging system is given apart from a constant factor by a x z Effa v exp i ux vz dudv 5 11 which is the inverse Fourier transform of the complex amplitude transmission of the object If not stated otherwise integration ranges from to to If the obj
382. hanges to a grid if obtuse triangles would result from the mesh relaxation Consequently RELAX will typically only work on meshes that were initially defined using LINE statements in ATHENA For other structures you can use DEVEDIT For more examples see VWF INTERACTIVE TOOLS USER S MANUAL VOLUMES I AND II Silvaco 6 97 SELECT ATHENA User s Manual 6 55 SELECT SE ar ECT selects the variable for printing using the PRINT 1D statement Note This command has been superseded for use with PRINT 1D by the EXTRACT command See VWF INTERACTIVE TOOLS USER S MANUAL VOL I Syntax SELECT Z lt c gt TEMPERATURE lt n gt Description SELECT specifies the variable that will be printed by the PRINT 1D statement You can only use one variable at a time Each SELECT statement overrides any previous statements w Z is set equal to the selected variable The operators all work as standard algebraic operators would Z can be set to any of the vector variables shown on the next page Table 6 4 Select Operator Variables Vector Variables Description ANTIMONY antimony concentration ARSENIC arsenic concentration BORON boron concentration CI STAR equilibrium interstitial concentration CV STAR equilibrium vacancies concentration DOPING net active concentration ELE
383. hapter 3 SSUPREM4 Models Viscous Model There are also two examples in the ATHENA SSUPREM4 section of the Deckbuild Examples Window See Figure 2 2 that demonstrate the use of the model for LOCOS and SWAMI isolation processes The following considerations should be kept in mind when using this model 1 The stress dependent viscous oxidation model is an extremely time consuming simulation method Therefore it should be used only when it is absolutely necessary and alternative approaches fail 2 Typical cases for use of the model are those where a kinked oxide surface is observed and when the simulated bird s beak is longer than the measured one 3 In some cases the alternative compress method with increased Young s modulus for nitride could give a reasonable shape see the ATHENA SSUPREM4 example in the Online Help Section 4 The grid for the stress dependent viscous oxidation should be as simple as possible but it cannot be too coarse in the direction of oxidant diffusion x direction in the case of simple LOCOS 5 The higher than default relative error for oxidation rate should be chosen to allow faster convergence For example METHOD OXIDE REL 0 01 6 The main parameter for tuning the model is nitride viscosity which is specified in the NITRIDE statement MATERIAL NITRIDE VISC 0 lt real gt The higher the nitride viscosity the stronger the stress dependence It is important to know that ni
384. hat can be accessed by the MATERIAL statement Parameter VISC O VISC E VISC X YOUNG M POISS R OXIDE 1 99x1077 5 292 0 499 8 3x1011 0 2 wet OXIDE gasio 7 405 0 499 8 3x1011 02 dry NITRIDE 1 8x1015 0 0 499 1 0x1024 0 3 SILICON 1x103 0 0 499 1 7x1012 0 28 POLY 5x1011 0 0 499 1 7x1022 0 28 OXYNI 5x1012 0 0 499 3 89x1012 0 3 B 4 Silvaco Default Coefficients B 1 11 Linear Coefficients Of Thermal Expansion These parameters can be accessed by specifying the LCTE parameter in the MATERIAL statement Table B 9 Linear Coefficients of Thermal Expansion Parameter Value SILICON CTE 3 052e 6 2 6 206e 10 T 293 OXIDE CTE 1 206e 7 2 2 543e 10 T 293 ALUMINUM CTE 2 438e 5 2 6 660e 9 T 293 NITRIDE CTE 3 0e 6 POLY CTE 3 052e 6 2 6 206e 10 T 293 B 1 12 Volume Expansion Ratio The volume expansion ratio ALPHA can be set in the OXIDE statement Table B 10 Volume Expansion Ratio Ratio Value silicon oxide unitless 0 44 poly oxide unitless 0 44 Other combinations unitless 1 00 B 2 Impurity Diffusion Coefficients Table B 11 Impurity Diffusion Coefficients Parameter Antimony Arsenic Boron Phosphorus Silicon 121 DIX 0 cm2 s 0 214 8 0 0 037 3 85 DIX E eV 3 65 4 05 3 46 3
385. hat it also can be integrated in the close form through the incomplete gamma function Selection of transversal distribution function is subjective because it is based on comparison with the lateral cross section of the 2D Monte Carlo distributions which cause accuracy to diminish further away from its maximum The analysis of 56 based on the BCA simulation see Section 3 5 4 Monte Carlo Implants showed that when By S25 2 8 which usually happens for random part of the 2D distribution or for amorphous implants the Pearson type II function slightly underestimates concentrations obtained in the BCA calculations while the MGF slightly overestimates these concentrations Therefore it was decided to use in ATHENA an average between the Pearson type II and the MGF for all B lt 3 When P 3 both functions reduce to standard Gaussian Finally in the case of higher values of lateral kurtosis it was found 56 that the MGF appears to be a better approximation so it is used in ATHENA It is very difficult to find B x as was done for a x already mentioned because the spatial moments of fifth and sixth order would be needed to build analytical functions for B x Therefore ATHENA uses constant By the KURTT and SKURTT parameters for the first and second Pearson functions correspondingly when you specify the FULL LAT model in the IMPLANT statement The generic approximations 56 for ox instead of Equation 3 210 and for B x will
386. he exponent e g 12 e Implant energy in KeV e g 60 e Tilt angle in degrees e g 7 e Rotation angle in degrees e g 30 All other parameters can use their default values Press the Write button and the following statement will appear in the input file CHANNEL IMPLANT IMPLANT BORON DOSE 4 0 CRYSTAL E12 ENERGY 60 P EARSON TILT 7 ROTATION 30 Silvaco Tutorial All of the parameters in the statement above are self explanatory except CRYSTALLINE The CRYSTALLINE parameter indicates that for all analytical models the range statistics extracted for a single silicon crystal will be applied when available If AMORPHOUS is selected the range parameters measured in pre amorphized silicon will be used when available The CRYSTALLINE parameter also has another meaning for the Monte Carlo or BCA implant models It invokes the Crystalline Material Model which takes channeling into account Note that the latter model is much slower 5 10 times than the Amorphous Material Model The Crystalline Material Model is the default model for BCA or Monte Carlo simulation For a detailed description of ion implant model selection see Chapter 3 SSUPREM4 Models Section 3 5 Ion Implantation Models You can specify tilt and rotation angles of the ion beam Positive tilt angles correspond to the ion beam coming from the top left Specifying the rotation angle makes sen
387. he crystal orientation or the direction of the beam or both the implanted projectiles and the damage created by them has different spatial distribution With even more higher fluency these phenomena will cause collisional mixing in a layered substances changes of the surface composition due to preferential sputtering and the establishment of a stationary range profile of the implanted ions Method of Solution The paths of the individual moving particles and their collisions are modeled by means of the binary collision approximation for a crystalline polycrystalline and amorphous substance using a screened Coulomb potential for the nuclear collisions and a combination of local and non local free electron gas approximation for the electronic energy loss For each nuclear collision the impact parameter and the Azimuthal Deflection Angle are determined according to the crystal structure using its translational symmetry For amorphous materials the impact parameter and the azimuthal deflection angle are determined from random numbers A proper scaling is chosen so that each incident projectile pseudo projectile represents an interval of implantation dose Subsequent to the termination of each pseudo projectile and its associated collision cascades the local concentrations of the implanted species created vacancies and interstitials are calculated according to the density of the matrix Nuclear Stopping As mentioned before during their passag
388. he device is to be stretched about a specified location If device characterization as a function of length is of interest the stretch function will save massive amounts of CPU time in generating multiple gate length structures The stretch capability is also useful for power devices MATERIAL specifies material that defines the stretch region see Section 6 2 9 Standard and User Defined Materials for the list of materials Default is SILICON LENGTH specifies the final value in microns to which the specified material region is stretched Alternatively you can specify X VAL using STRETCH VAL to specify the position of a vertical cut line and the distance to be stretched respectively The grid spacing within the stretched region is defined either by spacing or by division X VAL and Y VAL specifies the horizontal or vertical position in microns at which stretch occurs LENGTH overrides the STRETCH VAL X VAL and Y VAL parameters If LENGTH is specified the cut line stretch location defaults to the center of the specified material The default material is polysilicon SPACING specifies the grid spacing within the stretched region Units are microns DIVISION specifies the number of grid divisions within the stretched region SNAP indicates that X VAL should snap change value or locate to the nearest grid point before stretching SNAP is recommended to minimize the potential for obtuse triangle generation SNAP
389. he effective diffusivity and increase of the inactive dopant concentration These complexes are not assumed to be in local equilibrium with the other species e When the dopant concentration exceeds a few 102 cm the dopant vacancy pairs can no longer be considered as isolated entities because the vacancies can interact with more than one dopant atom 3 2 1 Classical Model of Dopant Diffusion CDD The basic idea of the model is isolated substitutional dopant atoms A are immobile The dopant diffusion occurs only through the migration of dopant self interstitial AJ and of dopant vacancy AV pairs Moreover in this enhanced model local equilibrium is not assumed between the pairs and their components unlike the original CNET model 5 All possible charge states of the free defects and of the pairs have been considered and their relatives concentrations depending on the local Fermi level position To turn on the CDD model specify PLS parameter in the METHOD statement All physical parameters of the model can be modified in the mod files To specify the location of these files use the B MOD P MOD AS MOD IC MOD and VI MOD parameters in the DIFFUSE statement By default all these files are located in the SSILVACO lib athena lt version_number gt common pls directory Charge States Point Defects The result of diffusion studies in metals and ionic crystals have led to the establishment of several basic atomic diffus
390. he final profile shows the reflow shoulders and the proximity effects seen following a 10 minute reflow heat cycle at 950 C xj TonyPlot 2 4 1 File v View Plot Tools Print Properties Help 7 BEFORE REFLOW HEAT CYCLE Microns T T N Sn 0 4 8 12 16 20 24 AFTER REFLOW HEAT CYCLE 0 4 8 12 16 20 24 Loading file tmp_mnt main striker andys dev cmp history25 str OK SILVACO International 1995 Figure C 5 Reflow of a via array Silvaco C 9 ATHENA User s Manual Question How can the reverse short channel effect RSCE in MOSFETs be simulated using ATHENA and ATLAS How can the physical effect behind RSCE be tuned Answer RSCE in MOSFETs is where the threshold voltage increases with decreasing channel length At very short channel lengths the normal short channel effect takes over and the threshold voltage decreases The cause of the increasing threshold voltage is a non uniform enhancement of diffusion of the channel implant laterally along the MOS channel This non uniformity arises from the extra point defects generated in the source and drain areas of the MOSFET The source of these point defects is most commonly the damage caused by the heavy n and LDD implants Other possible causes that can be modeled in ATHENA are oxidation or silicidation of the source and drain area The amount of implant damage from the source drain implants is controlled using the DAM FAC
391. he following statement prints the selected value at x equal to one micron between the top of the mesh and the 3 0 micron point PRINT 1D X VAL 1 0 X MAX 3 0 The following prints the selected variable along the silicon side of the silicon oxide interface PRINT 1D SILICON OXIDE For more examples see SELECT and PRINTE Silvaco 6 79 PRINTF ATHENA User s Manual 6 43 PRINTF PRINTF is a string printer and desk calculator Note Functions of this statement have been replaced by the EXTRACT statement Description The ECHO statement merely prints the string given to it This is useful for placing comments in an output file The statement attempts to parse the string to a legal real number if possible It has a regular expression parser built in This allows Examples The following command will send the string Athena 1 ECHO to be used as a desk calculator m Is My Favorite Process Simulator to standard output ECHO Athena Is My Favorite Process Simulator The following command will print 4096 ECHO ECHO 25354 The following command will print 8 373 which is the solution to the arithmetic expression 15 0 12 0 EXP 4 0 2 0 6 0 Silvaco PROFILE 6 44 PROFILE PROFILE reads a 1D doping profile into ATHENA Syntax PROFILE INFILE lt c gt MASTER IMPURITY INTERST VACANCY CLUSTER DAM DIS LOOP L
392. he grid spacing can increase rapidly in spacing away from the oxide silicon interface Figure 2 36 shows the effects of changing the mesh spacing at the interface on the simulated drain current You can see from this figure that too coarse of a mesh always results in too high of a current simulated 2 46 Silvaco Tutorial TonyPlot 2 6 10 A e File 7 View Plot Tools Print Properties Help TE GUESS CT Ges siya T EET Effect of Y Direction Grid Spacing on the l V Curve Spacing Key Files in Angstroms 1000A log 500A log 100A log 50A log 20A log 10A log 5A log 2A log 1A log x x Drain Current A um gt a Ligh aa mL tp Hie mae a mL nll amr aa num LST pms 0 0 5 1 1 5 2 2 5 3 Gate Voltage V Loading file home derekk manuals 1A log OK SILVACO International 1996 Figure 2 36 The effect of changing the mesh spacing at the interface on the simulated drain current If contact resistance is a problem then include it in the CONTACT statement The resistance added to the CONTACT statement should be the measured resistance per contact divided by the number of contacts on each individual electrode Obviously for D C measurements the resistance on the gate contact will have no effect on the results since no current flows in this direction Checking the Predictive Powers of Tuned Process Parameters If the process simulat
393. he polymer material generated as a mixture of incoming ions with etched sputtered molecules of substrate material Also the module has interface to the C Interpreter which allows simulation of several other processes such as wet etch and deposition ion milling and sputtering deposition of various materials The Monte Carlo etch module was successfully used by Toshiba researchers for simulations of reactive ion etching of narrow deep trenches in oxide 96 Simulation of Incoming lons and Neutrals Direct modeling of the plasma sheath is not included into this release and will be added later It is assumed that ions and neutrals fluxes leaving plasma sheath are represented by bimaxwell velocity distribution function along the direction determined by user specified incident angle L L K Uj vy I exp 2L 24 4 16 Ty T where V is the ion velocity component parallel to the incident direction V is the ion velocity component perpendicular to the incident direction I ion or neutral current density specified by parameters MC ION CU1 or MC ION CU2 in the RATE ETCH statement T is the dimensionless parallel temperature specified by parameters MC NORM T1 or MC NORM T2 T is the dimensionless lateral temperature specified by parameters MC LAT T1 or MC LAT T2 The incident angles are specified by the Mc ANGLE1 and MC ANGLE2 parameters Calculation of lon and Neutral Fluxes During each time step the simulati
394. he projected range and range straggling in the layer are normalized according to the probability for the ion to penetrating into the layer The only available measure of the probability is the portion of the dose accumulated in the specific layer Therefore the corrected projected range Rp and range straggling AR in the i layer are calculated as follows Silvaco 3 71 ATHENA User s Manual f 2 39 k Ric OR gt Rp 3 201 k 1 i l venii k Ric OR gt Rp 3 202 k 1 where i l go 4 gt F r 3 203 k 1 You can use the SCALE MOM parameter together with any of three depth matching methods 3 5 3 Creating Two Dimensional Implant Profiles Convolution Method ATHENA calculates 2D implant profiles using a convolution method described as follows First it calculates the implantation direction within the simulation plane using the TILT O and ROTATION angle parameters specified in the IMPLANT statement O is the angle between the ion beam direction and y axis is the angle between ion beam direction and the simulation plane For example 0 and 9 gt 0 correspond to an ion beam parallel to the simulation plane and directed toward the lower right corner of the simulation area The case of 90 and O gt 0 correspond to an ion beam in the plane perpendicular to the simulation plane and directed from behind the simulation plane The effective implantation angle in the simulation plane
395. her standard silicides PTSIX and WSIX as well as for user defined Silicides the TiSi2 growth rates are used The silicide growth rates can be modified by varying parameters D 0 and D E for silicon or interstitial diffusivity in silicide D which are specified in the INTERSTITIAL statement Defaults are D 0 1 96 and D E 1 81 for all silicides Silicide formation usually leads to a large volume decrease The ratio between consumed volumes of silicon and metal and resultant volume of silicide are specified by ALPHA parameters in the SILICIDE statement The default values for the ALPHA parameter are taken from 44 3 64 Silvaco SSUPREM4 Models The 2D movement of growth interfaces and volume change cause the viscous flow of the silicide layer This silicide flow is modeled analogously to the compress model of oxidation where the equations solved are VEVE 3 175 V y p 3 176 u E 3 177 FLD where e V is the velocity e P is the pressure e u is the viscosity e v is Poisson s ratio e E is Youngs modulus The parameters v and E are specified using the POISS R and YOUNG M parameters in the MATERIAL statement Silvaco 3 65 ATHENA User s Manual 3 5 lon Implantation Models ATHENA uses analytical and statistical techniques to model ion implantation By default the analytic models are used Analytical models are based on the reconstruction of implant profiles from the calculated or m
396. hosphorus Silicon oxide SEG 0 30 0 30 0 1126 0 30 0 unitless SEG E eV 0 0 0 0 0 91 0 0 Poly oxide SEG 0 300 30 0 1126 0 30 0 unitless SEG eV 0 0 0 0 0 91 0 0 Other Impurities and Pairs of Materials SEG eV 0 0 B 6 Silvaco Default Coefficients B 4 Interface Transport Coefficients Table B 13 Interface Transport Coefficients Parameter Antimony Arsenic Boron Phosphorus Silicon gas 123 TR unitless 2 5x1073 LS 27 9 1 5 R eV 1 04 12 99 2 48 1 99 Poly gas TRN unitless 2 5x1073 1 5 2439 15 R eV 1 04 LHII 2 48 1 99 Other Impurities and Pairs of Materials TR unitless 1 55 x 1077 R eV 0 0 B 5 Solid Solubility In Silicon Solubility can be modified for a particular temperature using the SS T in each of the impurity statements EMP and SS CONC parameters Table B 14 Solid Solubility in Silicon 124 125 Temperature Boron Phosphorus Antimony C cm cm cm 800 3 4499x1019 2 3000e 825 0 4 1291x1019 850 0 4 9027x1019 2 7943x102 875 0 5 7777x1019 900 0 6 7615x1019 3 1585x102 3 0000x1019 925 0 7 8610x1019 950 0 9 0832x1019 975 0 1 0435x107 1000 101922x102 303981x107 4 0000x1019 1025 103552x107 1050 1 5331x107 1075 1 7263x102 1100 1 9356x102 3 7943x107 4 8000x10 1125
397. hree operating regions is dominated by a different physical phenomenon Therefore successful modeling of a BJT involves matching both the base and collector currents in each of the three general operating regions making a total of six areas for calibration The derived parameter hfe is also a good parameter to monitor since this is sensitive to errors in the ratio of collector to base current The following text suggests an approach and describes which of the six regions are effected by each change The general technique is to calibrate the parameters that have the greatest effect on device performance in all regions first and then to move on to more subtle phenomenon that effect certain parts of the base or collector currents or both In general matching the collector current for all injection regions is less problematic than matching the base current at the extremes of the injection regions Consequently there are more sections on tailoring these parts of the curve The text is divided into the following sections Tuning Base and Collector Currents All Regions Tuning the Base Current All Regions Tuning the Collector Current All Regions The Base Current Profile Medium Injection OURS OO NO at The Base Current Profile Low Injection 6 Conclusions If you follow this order there should be a reasonable correlation between measured and simulated data Most of the tuning parameters however have some degree of interdepende
398. ic potential e R is defined from equation g R 0 In ATHENA the intersections of the incoming and outgoing asymptotes are evaluated with the hard core approximation of the time integral x ptan 0 2 3 222 x 0 3 223 Interatomic Potential ATHENA uses two body screened Coulomb potentials with a screening function which is a numerical fit to the solution given by Firsov 60 It also preserves the same analytic form as for the isolated atom AA 2 V r area 3 224 where Z and Zg are the atomic numbers of the two atoms and dq is the screening length defined by 1 3 dy 0 88534 pZ 3 225 where Z is an average atomic number of the two atoms calculated as a _ 1 Burs Ae a M 3 226 The main drawback of these two body potentials is their relatively slow decay as r The screening parameter a 0 is often regarded as an adjustable parameter for each two body combination which can be matched either to self consistent field calculations or to experimental data ATHENA uses the screening function in the form 4 X p3 a exp b x 3 227 i where a and b are taken from 61 Silvaco 3 79 ATHENA User s Manual Electronic Stopping Electronic stopping used in the simulation consists of two essentially separate mechanisms for inelastic energy losses local and non local These two types of electronic stopping are quite different in nature and behavior they have different energy and spatial d
399. id or not This obstacle can be overcome by simulating ion fluxes and by setting the etch rate to zero if the flux on the surface segment is less than some small threshold value Surface Movement A sophisticated string algorithm is used to move all segments according to the rates positive or negative calculated at each time step If the rate is negative the surface moves outside and the area is filled with redeposited material by default polymer If the rate is positive the surface moves inwards and the area is filled with vacuum Silvaco 4 19 ATHENA User s Manual 4 5 Reflow Model A two dimensional viscous reflow capability is included in ELITE The vitreous silica e g oxide BPSG are modeled as the viscous incompressible fluids which are dynamically deformed under the driving force of surface tension The finite element method is used to solve the creeping flow equations for the chosen materials With a 7 node triangle element as the basic discretization unit arbitrarily shaped 2D regions and surface curvatures are automatically described Using the built in user defined material capability you can simulate multiple material combinations The flow equation solver can be coupled with impurity diffusion to simulate the impurity redistribution and oxide growth The reflow is invoked by setting the reflow flag in the DIFFUSE statement and by setting the REFLOW flag in the MATERIAL statement to choose a specific material
400. id thermal annealing e Models simultaneous material reflow and impurity diffusion e Impurity diffusion in polysilicon accounting for grain and grain boundary components Epitaxy e 2D epitaxy simulation including auto doping Etch e Extensive geometric etch capability e Wet etching with isotropic profile advance e RIE model that combines isotropic and directional etch components e Microloading effects e Angle dependence of etchant source e Default etch machine definitions e Monte Carlo plasma etching e Dopant enhanced etching Silvaco Introduction Table 1 1 Athena Features and Capabilities Features Capabilities Exposure e Model is based on the Beam Propagation Method simulating reflections and diffraction effects in non planar structures with capability to take into account local modification of material optical properties the absorbed dose e Defocus and large numerical aperture effects Imaging e Two dimensional large numerical aperture aerial image formation e Up to 9th order imaging system aberrations e Extensive source and pupil plane filtering for enhanced aerial images e Full phase shift and transmittance variation mask capabilities Implantation e Experimentally verified Pearson and dual Pearson analytical models e Extended low energy and high energy implant parameter tables e Binary Collision Approximation Monte Carlo calculations for crystalline and amorphous m
401. ied independently from the depth standard deviation The LSTD DEV parameters can be specified in seriously improved MOMENTS statement or in user defined tables see below Also they are added into the standard look up table for a few ion material combinations Simplified control of the lateral distribution could be achieved by using LAT RATIO parameter in the IMPLANT statement Generic Pearson Distribution To achieve better compatibility to several other implant simulation programs e g UT at Austin deviations from standard Pearson IV distribution formula could be allowed using new ANY PEARSON parameter It means that kurtosis fourth moment could be slightly smaller than the critical kurtosis of the Pearson IV formula Range Parameters are Eliminated from the IMPLANT statement This capability has become obsolete after complete implementation of the MOMENTS statement The capability was very limited because it could be used only for unimaterial structures New PRINT MOM parameter of the IMPLANT statement Tells ATHENA to printout range parameters used for all ion material combinations for specified energy and dose It also refers user to a source where these parameters are taken from standard tables user specified tables or the MOMENTS statement In the case of Monte Carlo simulation PRINT MOM prints spatial moments calculated from the Monte Carlo based profile Improved Control of Moments Selection The selection of implan
402. ields When modelling the actual diffusion process there are additional effects to consider such as impurity clustering activation and how interfaces are treated Fundamentals of the models described in this section could be found in 5 6 and 7 Note In the following sections the terms impurity and dopant shall be used interchangeably although an impurity doesn t necessarily have to be a dopant Also the term defect shall mean the same as point defect unless otherwise indicated in the context Diffusion of dopants and point defects in SSUPREM4 is described by a number of user specifiable models The three most basic models are the following e The Fermi diffusion model e The two dimensional diffusion model e The fully coupled diffusion model The models are natural extensions of each other in the sense that the Fermi model is included in the two dimensional model which is included in the fully coupled model The two significant differences between them are the way point defects are represented and treated throughout the simulation and how the specific dopant diffusivities are formulated The selection of which model to use will depend upon the existence or the generation of point defects during the diffusion process and the dopant concentrations within the silicon Careful reading of the following sections is critical to understanding which model to use All three models rely on the concept of Pair Diffusion which says that a
403. ier e g A min You can select one of seven unit specifiers from the menu e Deposition rate e g 1000 This parameter is specified in the user selected units The SIGMA DEP parameter is optional and defaults to 0 2 The SMOOTH WIN and SMOOTH STEP parameters provides an alternative to a complete reflow calculation It allows a geometric averaging over a window of width SMOOTH WIN microns that is performed over a number of steps SMOOTH STEP These parameters perform a post deposition smoothing that effectively emulates a reflow process The wider smoothing window produces a more intensive surface redistribution of the deposit material The default number of smoothing operations 1 is adequate for most applications One or several model specific parameters are attributed to each model For example only the ANGLE1 parameter is required for the unidirectional model Table 2 4 indicates which parameters are required for each model The Machine Type section of the ATHENA Rate Deposit Menu includes only those parameters that are relevant to the selected model Each parameter has a default value which will be inserted in the input file If some of the parameters are undefined the simulation may be invalid or may produce unpredictable results If the ATHENA Rate Deposit menu is set as shown in Figure 2 44 the following RATE will be inserted into the input file EPO statement 0 RATE DEPO MACHINE TEST01 ALU
404. ies linear coefficient for Ge content dependency formula of intrinsic carrier concentration for Boron diffusion model in SiGe SiGeC EAFACT SIC specifies linear coefficient for Ge content dependency formula of intrinsic carrier concentration for Boron diffusion model in SiGe SiGeC Example The following statement defines some properties of a material called BPSG The material is composed of silicon oxygen boron and phosphorus with fraction composition 0 3 0 6 0 05 and 0 05 respectively Monte Carlo Implants could be performed into this material based on this definition MATERIAL MATERIAL BPSG AT NUM 1 14 AT NUM 2 8 AT NUM 3 5 AT NUM 4 15 AT MASS 1 28 086 AT MASS 2 16 AT MASS 3 10 8 AT MASS 4 31 ABUND 1 3 ABUND 2 6 ABUND 3 05 ABUND 4 05 For more information see OXIDE STRESS and DIFFUSE Silvaco 6 63 METHOD ATHENA User s Manual 6 36 METHOD Syntax FR PURI HOD RMI TWO PLS LUSTER DAM LUS iGECDF MOD DIM Ic ve RANS ST DDC SS HIGH CONC DOSE SiG LOSS 1 ERROR lt n gt E RELERR lt n gt RELERR lt n gt MIN TIME STR ERROR INIT TI OXID FILL E lt n gt E GDT lt n gt ABS F FREQ lt n gt NEWTON EADY FULL CPL ECNI MOD Y INTERST VACANCY OXIDAN ERROR lt n gt E ABSERR lt n
405. ification Wi Total number of grid layers fe Nominal grid spacing um 0 02 _ ra Grid spacing location ym Minimum grid sparing dumb Wi Minimum edge spacing pm Impurity concentrations ato mcm3 Antimony Arse pie Sergiy Phosphorus Silica fine Se ban begun feryillani Magne slit Adaggpn Ea igaes Galligan Carbon chramh iph heruian ian Composition fractions initial ramus Hien fracti Final compositios fraction Comment Poly deposition Figure 2 13 Impurity Section of the ATHENA Deposit Menu Click on the Phosphorus checkbox and set the doping level e g 5 0x10 using the slider and the Exp menu You can set a non uniform grid in the deposited layer by changing the Nominal grid spacing and the Grid spacing location parameters To create a finer grid at the polysilicon surface set the total number of grid layers to 10 the Nominal grid spacing to 0 02 um and the Grid spacing location to 0 0 at the surface Then click on the Write button and the following deposition statement will be written in the input file as DEPOSIT POLY THICK 0 5 C PHOSPHOR 5 0 DY 0 02 YDY 0 0 MIN SPACING 0 E19 DIVISIONS 10 001 2 18 Silvaco Tutorial Use the Cont button to continue the ATHENA simulation This will create the three layer structure shown in the lef
406. ify the Total number of grid layers in the deposited material region If you set this number to 10 it will insert the following ATHENA ELITE DEPOSIT statement USING DEFAULT DEPOSIT MACHINE PE4450 DEPOSIT MACHINE PE4450 TIME 2 0 MINUTES DIVISIONS 10 You can specify impurity concentrations in the deposited region in the Impurity concentration section of the ATHENA Deposit Menu by clicking on the Impurities box Silvaco 2 59 ATHENA User s Manual HS Deckbuild ATHENA Deposit Type Conformal Machine Display Impurities PARAMETERS TO RUN THE DEFINED MACHINE Machine name PE4450 Time of run 2 0 E minutes Grid specification Wi Total number of grid layers Nominelarid spacing dumk 3 Grid suacing lacefion Gam Minimum arid seacing um Wining edge seacine pmi Composition fractions O jaltialramuesition fraction 3 O Fipai campaesikipn fraction Monte Carlo Parameters O Rluinibed of partichgs Comment Using default deposit machine PE4450 Figure 2 43 ATHENA Deposit Menu with Machine Section Modifying ATHENA ELITE Default Machines The file athenamod defines PE4450 as follows Notice that a is used to concatenate or continue a long input line RATE DEPO MACHIN E PE4450 ALUM INU U M SIGMA DEP 35 HEMISPH ANGLE1 72 ANGLE2 70 E D EP RAT
407. ify the column entries with the parameter HCLT which is an array of numerical values surrounded by double quotes and separated by spaces or commas Specify the dependence of B A with the parameter HCL LIN which is an array of numerical values surrounded by double quotes and separated by spaces or commas The number of entries in HCL LIN must be the product of the number of entries in HCLP and HCLT Specify the dependence of B with the parameter HCL PAR which is an array of numerical values surrounded by double quotes and separated by spaces or commas The number of entries in HCL PAR must be the product of the number of entries in HCLP and HCLT BAE DEP BAF EBK BAF PE BAF PPE BAF NE BAF NNE BAF KO and BAF KE relates to the doping dependence of the oxidation rate The doping dependence is activated when BAF DEP is true default MATERIAL1 must be specified with these parameters only SILICON and POLYSILICON make sense here STRESS DEP VC VR VD VT and DLIM controls the stress dependence of oxidation which is only calculated under the VISCOUS model STRESS DEP turns on the dependence VC is the activation volume of viscosity VR is the activation volume of the reaction rate with respect to normal stress VT is the activation volume of the reaction rate with respect to tangential stress VD is the activation volume of oxidant diffusion with respect to pressure DLIM is the maximum increase of diffusion permitted under tensile stress
408. igure 2 39 will appear iy Deckbuild ATHENA Stretch Stretch From Material Center Snap to Grid ii Stretch Length ym 1 5000 01 e k 2750 Stretch fron K pesiiien dam mou hae nS cone RECO ri Grid Divisions O Grii pacing dumb GJ cAi Stretch Target Material Polysilicon Comment Stretch to 1 5 microns Figure 2 39 ATHENA Stretch Menu Then select Stretch Polysilicon upper left hand corner Next set Stretch Length to 1 5u and choose 10 as the number of Grid Divisions Then press the Write button and the following command will appear in the input file STRETCH TO 1 5 MICRONS STRETCH LENGTH 1 5 POLY SNAP DIVISION 10 As a result the polygate will be stretched from its initial length of 0 6 u left plot in Figure 2 40 to 1 5u right plot in Figure 2 40 Ten additional vertical grid lines will be inserted in the center of the gate area The LENGTH parameter of the STRETCH command can serve as a split parameter for the Virtual Wafer Fab Split Experimentation capability For more information about this capability see the VWF AUTOMATION AND PRODUCTION TOOLS USER S MANUAL 2 56 Silvaco Tutorial Another use of the Stretch capability is in the simulation of large power device structures where active areas are uniform everywhere except in close proximity to the mask edges and are separated from each other by long non active or isolation regions You can simulate a shrunken struct
409. ility can be used to etch the specified thickness of a material not covered by the mask After the dry etch is complete strip the mask by clicking the Strip Mask button in the ATHENA Photo popup A typical mask definition fragment should appear as follows POLY DEFINITION ASK NAME POLY CH POLY THICK 0 5 TRIP E n If the cutline from Figure 2 51 is loaded this will give the structure shown in upper plot of Figure 2 57 If the reverse parameter is added the structure will appear as shown in the lower plot of Figure 2 57 2 72 Silvaco Tutorial TonyPlot 2 1 beta File vj View vj Plot yji Tools vj Print Properties Help 7 Le Microns Microns ATHENA Nermal Magik ona 04 3S eS ee So o Q Dp Q m G o 6 po A IU Reverse Masik SILVACO International 1993 Figure 2 57 Using Mask Capability for POLY Definition Silvaco 2 73 ATHENA User s Manual 2 9 Using ATHENA OPTOLITH 2 9 1 Overview ATHENA OPTOLITH is designed as an optical lithography tool integrated into a complete process framework Specific functions of ATHENA OPTOLITH include 2D aerial image formation 2D photoresist exposure and development post exposure bake and post processing cap
410. ill be grown only over silicon while Poly Si will be grown elsewhere Silvaco D 7 ATHENA User s Manual Stress Simulation Feature The Stress History Model has been added This model is specified by using the STRESS HIST parameter in the METHOD statement The default is FALSE If this method is specified ATHENA then calculates stresses when the structure changes after etching deposition epitaxy and diffusion processes D 6 2 ELITE Capabilities 1 The capability to specify direction of incident ion beams for Monte Carlo plasma etch module is activated The MC ANGLE1 and MC ANGLE2 parameters have been introduced The old limit of 6000 on the number of nodes in the structure allowed during reflow simulation is removed D 6 3 Miscellaneous Features and Bug Fixes 1 a oP amp Capability to transform 1D structure into 2D structure when writing standard structure file in the STRUCTURE statement has been added When the TWO DIM parameter is specified and the current simulation structure is 1D it will be transformed into 2D before saving in the specified str file User defined materials with the names corresponding to SILVACO standard materials are now saved in Standard Structure Files as standard materials so that they will be recognized by ATLAS DEVEDIT and other SILVACO tools The ELECTRODE statement now recognizes the regions with material names specified as metals in the list below
411. ill be used in place of the build in function A 2 Silvaco Appendix B Default Coefficients This appendix contains the list of impurity and material default coefficients default model parameters and other parameters used in ATHENA calculations Most of these coefficients are initialized in the athenamod file The file athenamod will appear when you select Commands Models in DECKBUILD while ATHENA is the current simulator Almost all of these coefficients can be modified to match measured results You should check the contents of athenamod for updates to default values that may be more current than those shown in the following lists B 1 Oxidation Rate Coefficients B 1 1 Dry Ambient For lt 111 gt Orientation These parameters are from the bibliography reference 34 Table B 1 Parabolic and Linear Rate Constants for Dry Ambient Parameter Value PAR H 0 ym min 12 8667 PAR H E eV 1 23 P BREAK C 0 LIN H O pm min 1 038x10 LIN H E eV 2 0 1 BREAK C 0 B 1 2 Wet Ambient for lt 111 gt Orientation These parameters are from the bibliography reference 36 Table B 2 Parabolic and Linear Rate Constants for Wet Ambient Parameter Value PAR L O um min 283 333 PAR L E eV 1 17 PAR H O um min 20 PAR H E eV 0 78 P BREAK C 950 Silvaco B 1 ATHENA User s Manual
412. in a given time frame Adaptive Meshing During Ion Implantation and Diffusion A series of important improvements are now available in SSUPREM4 in the area of automated adaptive meshing Improvements include efficient 1D adaption and a new basemesh generation routine during the auto transition to a 2D structure 2D adaption employs a new smoothing capability Time stepping control also allows greater versatility Templates for a range of technology are supplied to more automatically generate the mesh Implant Simulation Features Advanced 2D Implant Distribution Model Analytical 2D distribution model which takes into account depth dependence of lateral standard deviation is implemented It is invoked using parameter FULL LAT in the IMPLANT statement In order to use this advanced model the following additional spatial moments should be furnished LSTD DEV LGAMMA and LKURTOSIS Corresponding parameters could be specified for the second Pearson distribution in the case of double Pearson model All above mentioned new parameters can be specified in seriously improved MOMENTS statement or in user defined tables see below Also they are added into the standard look up table for a few ion material combinations Flexible Control of Lateral Distribution More accurate and flexible modeling is implemented also in the case of simple lateral implant distribution with constant lateral standard deviation The lateral standard deviation now can be specif
413. in films to reduce the surface energy associated with areas of high curvature A prediction of the trends in local film density can be achieved Plot with discs can be obtained using the parameter OUTFILE lt filename gt in the DEPOSIT statement Figure 4 7 shows the vapor flux distribution arriving can be defined using the ANGLE1 parameter describing the angle measured between the vertical from the source and the wafer normal To use multiple steps for both MONTE1 and MONTE2 models set the DIVISION parameter in the DEPOSIT statement The number of incoming particles can be defined by the N PARTICLE parameter in the DEPOSIT statement 4 3 9 Custom Deposition Models ELITE implements two slightly different custom deposition models In both models the angle is d Silvaco 4 11 ATHENA User s Manual 4 4 Etch Models ELITE provides a set of etch models that correspond to different physical etching techniques 93 94 and 95 Any one of these models can be selected to define a machine that can then be invoked to perform processing on the structure In addition ELITE provides a primitive etching capability that can be used to define initial structures 4 4 1 Isotropic Etch Model To use the model specify the WET ETCH parameter in the RATE ETCH statement In wet etching and simple plasma etching the substrate is immersed in a fluid liquid or gas which chemically reacts with the exposed surface In wet
414. in iste carte eh anew a eee ep aie oe state 6 42 viii Silvaco Table of Contents 6 27 IMAGE ss edite se toed eee line AAA a ated oat onl etc et eae eg ee 6 43 6 28 IMPLANT i a a ck cate 2S cates are A bats ch Gn ed teeters aie any aisha Sache alm heal ome pace AS 6 45 6 29 IMPURITY soisi Ses cried aaa Soin te whee he tot aie ae Setwrane Rens a Oe eee ewe BaN A 6 49 6 30 INITIALIZE SoS AS fOr wake hate re aa he ck Wet E a ca ee Eien Lah Kale Wi tl 6 52 6 31 INTERSTITIAL and VACANCY i ic 0cc 20 ecie raga niieoe ewe eaa a AAE acids aw 6 54 6 32 LAYOUT scoccccriereadvecersedpreatig onl pesa Peeee Pau SiGe DoE AAE ERA eles Diit 6 57 6 33 LINE x Ser a aie cua Rectan a aa a inte reed aA vil cia Heelan beeen Tees 6 59 O34 MASK r c ewan retl deat Poteet E T Dieta cent Re nemen A deen meee Meet 6 61 6 35 MATERIAL ree rtan ive ori een toys oe eid xcanae elds Stan eae etite ved er aE A N 6 62 6 36 METHOD 5 00 Serenata n Er EnA DLEA ETUE EAA ET Case ontsesd ew areewebatas ass 6 64 6 37 MOMENTS ok irota a eaae a a a a eaa a keia ha Enia 6 70 6 38 OPTICAL saa AAE A ATREA ECAA AA 6 72 039 OP TON a E US eet a A E A AN a A AOE 6 73 6 40 OXIDE ri ig sie eect tee seem ie edge A a Ma GARAE Oa 6 74 6 41 POLISH ieoi a Eiaa Ea AE A Gn EEEE A nd Et AAE ET owe aA a 6 78 6 42 PRINTI D yorini a a a hs E wa wc E See ae ee E wee eee 6 79 6 43 PRINTE ee a aa ea r a a a aaa a a a eA EA E aa tol aa ate kate S 6 80 6 44 PROFILE p irene a a a a a a a a S 6
415. ine or change the intensity transmittance and phase transmittance of an annular zone inside the exit pupil of the illumination system This qualifier is used to simulate spatial filtering techniques IN RADIUS and OUT RADIUS are used to define an annular zone in the exit pupil having the pupil transmittance equal to TRANSMIT and producing the phase angle equal to PHASE Radius values are specified in fractions of unity and phase is specified in degrees Note that the annular zones should not overlap The outer radius of an inner zone must be smaller than the inner radius of an outer zone PHASE specifies the phase shift in degrees produced by the illumination source filter 180 lt PHASE lt 180 TRANSMIT specifies the intensity transmittance produced by the illumination filter 0 lt TRANSMIT lt 1 CLEAR FIL resets the illumination source filter list Example The following example defines a SHRINC illumination source where the quadruple circular illumination sources are located at 45 to the x axis with the center at a radius of 0 2 from the origin and a circle radius of SIGMA 0 2 ILLUM FILTER SHRINC RADIUS 2 SIGMA 2 ANGLE 45 Silvaco 6 41 ILLUMINATION ATHENA User s Manual 6 26 ILLUMINATION ILLUMINATION specifies the basic illumination parameters in OPTOLITH Syntax ILLUMINATION I LINE G LINE H LINE KRF LAS X TILT lt n gt Z TILT lt n gt INT
416. ined in the METHOD statement would invalidate the remainder of the following section Calibrating a bipolar process flow entails matching the two parameters base current and collector current versus base emitter voltage to measure results throughout the full operating range of the device By implication the current gain of the device Ic Ib will also be matched All of the following paragraphs refer to the standard plot of collector and base currents measured against the base emitter voltage Vbe unless it s specifically stated otherwise This standard IV graph is usually referred to as the Gummel Plot Another way of plotting the same information in a different format that can prove useful is a plot of current gain hfe versus the log of the collector current This graph however is a derivation of the same information that makes it less clear as to which current is increasing or decreasing for each change Therefore a less useful graph when it comes to understanding exactly what is happening to the collector and base currents The full operating range of a bipolar junction transistor BJT consists of three general regions defined by the current density injected into the base These three operating regions are usually described as low medium and high current injection regimes The medium injection region is the most important part of the curve to model correctly as this represents the typical operating condition of the BJT Each of the t
417. ing and surface diffusion effects SIGMA DEP parameter This new model is called USER DATA 2 The necessary parameters are in an ASCII input file of the same form as the USER DATA 1 model Angle and deposition rate are the input values in the file where the deposition rate is taken as a rel ative deposition rate and the overall deposition rate is determined by the DEP RATE parameter SUBSTEPS has been added to the DEPOSIT statement This parameter controls the number of steps made for each division of the deposit This parameter is very important in terms of shadowing effects as these effects are calculated every time there is a change in SUBSTEPS or DIVISIONS In general the larger the number of SUBSTEPS the more accurate the calculation However a large number of SUBSTEPS also increases calculation time SUBSTEPS 1 is useful for the USER DATA 2 model if there are a large number of points in the ASCII input file This will speed up depositions made with this model and will not affect the accuracy of the shadowing as shadowing effects are calculated for each point in the ASCII input file The default value for SUBSTEPS is 8 Silvaco D 19 ATHENA User s Manual D 13 3 FLASH Capabilities For the new materials AlGaAs InGaAs SiGe and InP implantation and diffusion models were enabled Currently diffusion in AlGaAs InGaAs and InP have the same parameters as GaAs as specified in the model file SiGe uses the parameters for Si again as s
418. intrinsic pair diffusivity terms Pair charge states beyond two are unlikely to occur which is why they have been omitted Also for most dopants it is seldom that more than three of the terms above are non vanishing Table 3 2 Table of intrinsic pair diffusivities for different pair types Pair Charge State A seal Activation Energy AV x DVX 0 DVX E AV DVM 0 DVM E AV DVMM 0 DVMM E AV DVP 0 DVP E AV DVPP 0 DVPP E Silvaco 3 5 ATHENA User s Manual Table 3 2 Table of intrinsic pair diffusivities for different pair types Pair Charge State aaa Activation Energy AI x DIX 0 DIX E AI DIM 0 DIM E AI DIMM 0 DIMM E AI DIP 0 DIP E AI DIPP O DIPP E Note Since the point defect populations are by definition assumed to be in equilibrium in the Fermi model there are no separate continuity or boundary condition equations for these species Additionally neither the vacancy concentration C nor the interstitial concentration Cy appear explicitly in Equations 3 9 3 10 or 3 11 3 1 3 Impurity Segregation Model In multilayer structures dopant segregation across material interfaces must be considered Such interfaces can represent either a solid solid interface or a gas solid interface the surface Interface segregation is modeled empirically by a first order kinetic model for the interregional flux S C F Sigh hee 3 12 2 where
419. ion Etching in complicated structures latch up etc memory allocation and freeing bugs eliminated D 24 Silvaco ATHENA Version History Syntax Changes REGRID and layout interface related syntax for INITIAL statement has been removed D 17 SSUPREM4 Version 5 1 4 Version 5 1 of SSUPREM4 incorporates a number of new models as well as convenience features and numerous bug fixes Eliminated a bug in the PRINT 1D statement for structures including BARRIER material The memory requirements for SSUPREM4 were reduced dramatically through a change to the maxi mum number of materials and regions allowed in a simulation Boundary conditions bug fixes eliminated some difficulties during TWO DIM diffusions D 18 SSUPREM4 Version 5 1 Version 5 1 of SSUPREM4 incorporates a number of new models as well as convenience features and numerous bug fixes PREDICT2 Feature Incorporation As part of an ongoing collaboration with the Microelectron ics Center of North Carolina Version 5 1 of SSUPREM4 is coupled with initial model implementa tion of PREDICT2 The models in PREDICT2 are the most accurate available for high concentration diffusion Rapid Thermal Processing RTP and Transient Enhanced Diffusion TED The use of these models is described in the DIFFUSION and METHOD statement descriptions and in the Ref erence Manual DeckBuild example facility added A set of standard examples for SSUPREM4 and other SIL VACO simulator
420. ion has been correctly tuned the process and device simulators should have predictive powers To check the validity of the tuning process use a new set of electrical data that was not used during the tuning process For example a good alternative set of data is to check the threshold voltage versus gate length for a non zero voltage applied to the MOSFET body contact Conclusion Using just one set of easily obtained measured electrical data namely a plot of threshold voltage versus gate length you can obtain most of the tuning parameters required for accurate process simulation The other most important piece of data required is an accurate measurement of the gate oxide thickness which is routinely measured in any instance You have been given specific advice as to which process and device models to use for each process in order to get the best results out of the simulation software In particular the correct use of models for the implantation and diffusion processes is stressed as this has a dramatic effect on MOSFET characteristics especially as anneal times and device dimensions decrease Silvaco 2 47 ATHENA User s Manual 2 6 Calibrating ATHENA for a Typical Bipolar Process Flow As with MOS calibration text we assume you are familiar with the mechanics of making an input file and using the correct methods and models see Section 2 4 Choosing Models In SSUPREM4 For example incorrect selection of diffusion models def
421. ion mechanisms These mechanisms dominate the interpretation of silicon diffusion experiments with the exception that in silicon there is a very wide energy range available to the Fermi level Therefore a given lattice defect can appear in a variety of ionized states The fundamental principles of thermodynamic predict that such defects will exist in equilibrium at all temperatures above 0 K because the presence of such defects minimizes the free energy of the crystal The entities Vee and i are the equilibrium defect concentrations for vacancies and silicon self interstitials in 3 24 Silvaco SSUPREM4 Models their neutral charge state The weight factors y and 6 account for the different charge states for distribution of point defects under extrinsic conditions Each y and 6 is assumed to be temperature dependent through Arrhenius expressions For point defects V or J five various charge states are considered E MAP MAY n 3 73 where s is one of the charge states 2 1 0 1 or 2 All parameters y and 6 are specified in the charge state statements of the defect mod file With this consideration the equilibrium concentrations V and I are estimated as eq 0 ay 1 0 gay 2 i ae 74 i leq gt n Veg ye n 3 Under intrinsic conditions i e n p n the equilibrium concentration can be simply written as 1 A 5 Py ae yo 3 75 l l Equilibrium concentration for vacancies and silicon self interstitials a
422. ion of the etch region under the mask The ANGLE parameter defines the slopes of sidewalls of the region The bottom line of the etch region is defined by vertical translation of the top boundary with undercut taken into account The ANGLE less than 90 results in trenches narrowing to the bottom The ANGLE greater than 90 produces retrograde sidewalls The UNDERCUT length is measured along the boundary line between etched material and masking layer In a special case when the etched material layer is sandwiched between two other layers the THICKNESS parameter is ignored and UNDERCUT is applied to both the upper and lower boundaries of the etched material layer All regions of a particular material may be etched by specifying the ALL parameter of the ETCH statement When a region is defined in one of the first three ways By default all materials in the defined region will be etched Specifying a material in the ETCH statement limits etching to only that material within the defined region For a complete description of physically based etch models see Chapter 4 ELITE Models ELITE is a complete 2D topography simulator included in the ATHENA framework Silvaco SSUPREM4 Models 3 8 Compound Semiconductor Simulation ATHENA allows you to simulate basic technological processes in compound semiconductors The set of standard compound materials includes GaAs AlGaAs InGaAs and InP Additional user
423. ion of the light intensity over the resist surface The real resist exposure with the light absorption into resist reflections from materials interfaces and following photo chemical resist modifications are considered the same way as for the projection printing Silvaco 5 15 ATHENA User s Manual The common wave equation is Verve EE E PRO 5 45 Ox Oy or where E is the electromagnetic field and k is the wave vector Because the normal incidence on the mask is considered and propagation of light from the mask to resist surface is to be calculated it is convenient to represent E r A r exp ikz 5 46 where A is amplitude of electromagnetic field and z is direction along the light propagation Substitution from Equation 5 46 into Equation 5 45 gives GA GA ZA 2 0 5 47 ao y o Obviously A is modified with z weakly Therefore you can neglect the third term in Equation 5 47 As the result Equation 5 47 can be rewritten as a ey 0 5 48 ae Oy Now Equation 5 48 looks as a diffusion equation with complex diffusion coefficient where z replaces diffusion time It means that the propagation along the z direction can be formally considered as a diffusion of complex amplitude A in the x y plane with a complex diffusion coefficient The known 6 solution of traditional diffusion equation ft 14 rn 0 5 49 aw ow r is as follows 1 y 4 2 fo exp A 5 50 The substitution in Equation 5 48 results in Ag
424. ion step when point defects remain at their equilibrium values FERMI and when point defects are allowed to obtain non equilibrium values TWO DIM It is evident from Figure 3 19 a that boron diffusion is enhanced for the TWO DIM case The corresponding interstitial concentrations are shown in Figure 3 19 b The interstitial concentration is above the equilibrium interstitial concentration for the TWO DIM case thus allowing oxidation enhanced diffusion to be observed but remains at equilibrium for the FERMI case TonyPlot 2 6 9 i f f 5 oO 5 5 fr Interstitial Concentration cm 3 02 04 closer ae ee LE ts Z SILVACO International 1996 Figure 3 19 a Boron Concentration Versus Depth b Corresponding Interstitial Concentration Versus Depth You can also have a diffusion retardation effect during thermal oxidation For dopants diffusing primarily through a vacancy mechanism you can reduce their diffusivities during oxidation because of the recombination of vacancies with injected interstitials at the SiOv Silicon interface Figure 3 20 shows an example of this phenomenon Figure 3 20 a shows the resulting antimony concentration profiles after an oxidation step where the FERMI and TWO DIM models were used In contrast to boron Figure 3 19 a the resulting antimony concentration profile is shallower for the TWO DIM case when compared to the FERMI case Figure 3 20 b shows the reduced vacancy concentra
425. ion were not independently accessible from those for silicon oxidation Experiments have shown that polysil icon oxidation can be significantly different from silicon oxidation All coefficients for oxidation are now accessible independently for silicon and polysilicon oxidation By default the rates for polysili con and silicon oxidation are the same Geometric Mode Added The capability to specify at initialization that a simulation is to be per formed without impurities has been added as a parameter on the INITIALIZE statement This specifies the so called geometric mode that describes all material layers but produces no impurity information speeding up SSUPREM4 execution time immensely Silvaco D 25 ATHENA User s Manual Coarse Grid Mode Added The parameter SPACE MULT has been added to the INITIALIZE statement to globally manipulate the initial grid specification for SSUPREM4 Setting the value of SPACE MULT to a value greater than one will increase the effective value of each of the spacing parameters on preceding LINE statements This gives a quick way to globally reduce the grid den sity in a SSUPREM4 simulation for reduced simulation time for preliminary analyses Full Rotation Capability Added To IMPLANT Statement Full rotation for implant can now be specified on the IMPLANT statement Specifying the FULLROTAT parameter will perform implantation at the specified tilt angle from all rotation angles as would occur with a rota
426. is set to true by default Stretch Examples The following statement will stretch a device about the center of its polysilicon region This device can have been a MOSFET with a polysilicon gate 1 micron long The STRETCH command creates a 1 8 micron long MOSFET in this case STRETCH LENGTH 1 8 The following example will stretch an oxide isolation structure from the x position of 2 3 microns by a value of 1 3 microns The stretched region contains 14 grid spaces This case can be useful for generating large isolation regions that take too long to simulate numerically STRETCH OXIDE X VAL 2 3 DIVISIONS 14 STRETCH VAL 1 3 Note The stretch function may not be valid or physically correct in the case of very short initial structures e g with RSCE effect in MOSFETs The location selected for stretching should correspond exactly to a grid line for best results It will provide best grid quality if the stretch location does not touch areas in which the grid has been relaxed The STRETCH command often results in grid failure for complex structures and is not recommended for complex topographies DEVEDIT provides a superior stretch feature for these cases Silvaco 6 105 STRIP ATHENA User s Manual 6 62 STRIP STRIP removes all photoresist and barrier materials Syntax STRIP MATERIAL Description MATERIAL specifies the material to be stripped see Section 6 2 9 Standard and User Defined Material
427. istribution in both cases with and without taking into account dose to n effect is specified in the EXPOSE statement as NUM REFL The maximum dose that corresponds to completely exposed resist is specified with DOSE parameter The examples below show how to use the OPTICAL and EXPOSE statement to specify parameters for the exposure module Example Resist exposure with accounting dose effect OPTICAL NAME RESIST CURREN_RESIST I LINE REFRAC REAL 1 4 REFRAC IMAG 0 02 DELTA REAL 0 2 DELTA IMAG 0 01 EXPOSE DOSE 200 NUM REFL 5 Example2 Resist exposure without accounting dose effect OPTICAL NAME RESIST CURREN_RESIST I LINE REFRAC REAL 1 4 REFRAC IMAG 0 02 EXPOSE DOSE 200 NUM REFL 5 Silvaco 5 11 ATHENA User s Manual 5 5 Photoresist Bake Module Post Exposure Baking PEB of the photoresist has been demonstrated to dramatically reduce standing wave fringes of the developed resist image resulting from optical interference of monochromatic illumination This effect is generally accepted to be a result of bulk diffusion of the PAC and photo reaction products The simple physical model which is adopted here to describe the PEB is that just one chemical constituent of the resist diffuses This constituent is generally assumed to be PAC or the dissolution inhibitor which diffuses ac
428. itial is calculated by the following empirical formula 1 B En Alog 1 oF J Ee a 3 112 The empirical parameters of the model A B C and Boe are specified in the ic mod file Thus the system of coupled equations for the IC model can be written as N oe V J GRy GR 47 2GRic2 GR icin n 3 ot a Ve Jy GRiy GRyy tot OAI 2 ao Vid jp GR gy GR is ate t eav z V Jiy CRay GR ys OA oe GRy GRay GRyy_ GRy _ p gt a GRicm CR oaa pp n 2 3 N 3 32 Silvaco SSUPREM4 Models where the generation recombination term es am ae 3 114 n n nn CRicmn a It is important to notice that the IC model is completely independent of the dopants involved in the process 3 2 4 Vacancy Cluster Model VC In the VC model a cluster containing m vacancies V evolves to a cluster of size m 1 by interaction with a free vacancy according to the following reactions K K Y Vvsyvs W Payv zi Vn rere 3 115 ki k E where V E RE EV Elmh _ _ K 4R Dy Pes o enco E SD a Me 3 116 Here Re represents the effective capture radius and the elementary jump length is equal to the inter atomic distance 0 is the number of dissociating sites Hy is the vacancy formation energy defined in Equation 3 76 and Em is the formation energy per vacancy for clusters of size m To take into account the interactions VC with interstitials the following re
429. l 2 1 Getting Started This chapter is to help you start using ATHENA by providing a step by step tutorial centered on a typical process simulation sequence It explains how ATHENA uses the VWF INTERACTIVE TOOLS i e DECKBUILD TONYPLOT MASKVIEWS DEVEDIT and OPTIMIZER These tools make ATHENA easier to use and they provide visualization and interface capabilities with other Silvaco tools This tutorial assumes that you are familiar with the basic features of the VWF INTERACTIVE TooLs For more information about these tools see VWF INTERACTIVE TOOLS USER S MANUAL This chapter begins by explaining how to start ATHENA and continues with tutorials on how to use the program We recommend that you read Section 2 3 Creating a Device Structure Using ATHENA before you move on to the section appropriate to the particular tool you will be using This section explains how to start ATHENA how to load and run standard examples and how to use the ATHENA online help facility The following explanations assume that ATHENA has been properly installed See the SILVACO INSTALLATION MIGRATION AND TROUBLESHOOTING GUIDE if you encounter installation difficulties 2 1 1 Running ATHENA Under DeckBuild DECKBUILD is an interactive graphic environment that is used for the following purposes e Generating input files for process or device simulation or both e Running simulations interactively e Interfacing between different simulators e Invoki
430. l describing dose loss at silicon oxide interface Units for TRNDL 0 are cm sec units for TRNDL E are eV Polysilicon Diffusion Model Parameters PD DIX 0 and PD DIX E specify impurity diffusivity along grain boundaries PD DIX 0 is the pre exponential factor of grain boundary diffusivity units are cm sec PD DIX E is the activation energy for grain boundary diffusivity units are eV PD EFACT specifies entropy factor of grain boundary segregation coefficient PD SEG E specifies the activation energy of grain boundary segregation coefficient Units are eV PD TAU specifies the grain boundary time constant Units are seconds PD SEGSITES specifies density of segregation sites at grain boundary Units are sites cm PD GROWTH 0 specifies the grain growth rate pre exponential coefficient Units are eV cm sec PD GROWTH E specifies the grain growth rate activation energy Units are eV PD CRATIO specifies initial ratio between impurity concentration in grain boundaries and total concentration PD SEG GBSI specifies the factor which controls segregation between polysilicon grain boundaries and Silicon Examples The following statement changes the neutral interstitial diffusivity component of phosphorous in silicon IMPURITY I PHOSPHORUS SILICON DIX 0 3 85 DIX E 3 85 The following statement changes the segregation parameters at the silicon silicon dioxide interface The concentration of phosphorous in silicon will be
431. l sub micron CMOS process ELITE models might be required for e Trench isolation e Spacer formation e Reflow of oxides over non planar surfaces e Metal to active area contact cuts e Metal deposition over step e Inter metal dielectric formation In general ELITE should be used for any etch process with a degree of isotropy since perfectly anisotropic etches can be handled geometrically in SSUPREM4 For deposition processes ELITE is appropriate when the deposition is significantly non conformal C 6 Silvaco Hints and Tips Many topography simulators exist but interfacing them to process simulation programs such as SSUPREM4 has traditionally been a problem Without the tight integration of ATHENA the interface has traditionally been one way for example creating a non planar topography such as a trench and then using the surface to create the initial structure for a SSUPREM4 simulation In ATHENA the bi directional interface between topography and process simulation is completely automatic and transparent to the user Figure C 3 shows this interface used to form a self aligned trench isolation for a sub micron CMOS process The initial part of the simulation uses SSUPREM4 to set up a LOCOS oxidation next to a nitride spacer ELITE is then used to remove the nitride and etch a trench into the silicon SSUPREM4 is used to oxidize the trench sidewalls Then the ELITE deposition models are used to fill the trench with oxide Fi
432. laces in the name of the file so as to be able to movie diffusion effects during the initial short time steps Simulated Structure can be Truncated from a Side or from the Bottom by using NOEXPOSE parameter in the ETCH LEFT RIGHT or ETCH BELOW statements Alternative Model Files With ATHENA V4 0 0 R users may now select alternative model files using the modfile command option The option argument names the alternative model file ATHENA should use during the simulation ATHENA V4 0 0 R is shipped with a new updated model named smod96a This file contains improved model parameters and its use is recommended Silvaco D 13 ATHENA User s Manual D 11 ELITE Monte Carlo Plasma Etching A new monte carlo plasma simulation function is available to calculate the angular energy distribution of ions emitted from a RIE machines dark space sheath Shadowing is calculated and etch rates over complex topograhical surfaces result Sputtering efficiency as a function of angle is also controllable Doping Concentration Dependent Etch Rate A doping level etch rate enhancement factor allows user control over the relative etch rates of doped materials This function is unique to the mesh based ATHENA product and can not be treated with a simple string based tool Stress Dependent Etch Rate Etch rates may be enhanced as a function of material stress Oxidation induced stress creates defectivity in materials that will increase the local etch rates
433. late the image in the case of contact printing The GAP parameter enables the model and specifies the mask to wafer gap D 7 4 ELITE Capabilities The ELITE etching algorithm has been improved Now if the ETCH RATE parameter for a material specified in the MACHINE statement is equal to zero or is not specified all regions for this material it will not be changed during all etch process steps which utilize this machine D 7 5 Miscellaneous Features and Bug Fixes 1 The C interpreter capability in the DIFFUSE statement has been removed This capability will be re implemented and expanded using a newer more flexible and extensive SILVACO C INTERPRETER The standard tables for BF2 implants were extended down to 1 keV Fixed a bug in Pearson VI function which occasionally resulted in a non physical tail in case of high energy implants in photoresist The flip triangle procedure after etching deposit and epitaxy steps has been removed The bug in saving and loading standard structure files after using the POLY DIFF model has been fixed The bug in separation of floating and substrate oxidizable regions has been fixed This bug use to distort the substrate in some structures obtained from DevEdit D 8 ATHENA Version 5 2 0 R Release Notes D 8 1 lon Implant BCA Model 1 aon Fw S Considerable speed up for 2D simulations Profile smoothing capability is available after BCA implant The PRINT MOM parameter now w
434. late them through the experimentally known intrinsic diffusivities D4 and the interstitialcy component fz The p and pP y parameters are defined as an Arrhenius functions Finally the flux equations for instance for boron point defect pairs are defined as follows xe EF B J J D 4 x 3 91 Bi 2 Br BI pst s1 s Kyr Y s isk A BENN l ABT pf s ae x 2 Sy fs P S Sark 2 3 S J fot ee ae p st 51 Se a 3 92 Bv n l aBy Bys Ap K 2 73 P BV Ni S where s s 1 The equations for pairs formed by donor impurities are completely symmetrical to the equations above 3 28 Silvaco SSUPREM4 Models Generation Recombination Terms The generation recombination terms GRy in Equation 3 82 describe the evolution of particular reactions For example GR 4y represents the formation of the dopant vacancy A VED AV 3 93 kK AV Formation of Pairs In the model the formation of dopant defect pairs is taken into account by simulating the following reactions Pi 3 94 vi A P oa kar Ky StV osm 3 95 ky where k and k are the reaction kinetic constants for each reaction They are define as K 42RD k K o K 3 96 Ky 40RD Kay Kay K p 3 97 where K4 Kay are defined by Equation 3 81 and R is the silicon lattice constant Thus the generation recombination terms in Equation 3 82 are as follows GR KA P oR GAL 3 98 GRyy K
435. lated Diffusion parameters for epitaxial silicon however are considered the same as for single crystal silicon The epitaxy process is defined in the ATHENA Epitaxy Menu Figure 2 32 To open this menu select Process Epitaxy in the Commands menu The ATHENA Epitaxy Menu consists of five sections e The Time temperature section selects temperature step parameters in the same way as in the DIFFUSE statement e The Thickness rate section selects either the total thickness of the epitaxial layer or the deposit rate in microns minute In the latter case the total thickness will be determined by the rate and time e The Grid Specification section specifies the vertical grid structure within the grown epitaxial layer All grid parameters are equivalent to those of the ATHENA Deposit Menu See Figure 2 12 2 40 Silvaco Tutorial The Ambient section is where the gas pressure can be modified to the value used in the Epitaxial Chamber The Impurity Concentrations section specifies the growing epitaxial layer in the same way as in the DIFFUSE statement All parameters in the last three groups are optional If the parameters of an epitaxial step are set exactly as shown in Figure 2 32 the following statement will appear in the input file EPI LAYER EPITAXY TIME 30 TEMP 900 T FINAL 1000 THICKNESS 5 DIVISIONS 20 DY 0 05 YDY 0 00 Note The diffusion during the epitaxy process will use the Diff
436. lation Features 1 2 3 Three types of silicon carbide materials are added SIC_6H SIC_4H and SIC_3C BCA implantation model for the silicon carbide materials is implemented BCA implantation model for two superconductor materials Ba2YCu307 and Ba2NdCu307 is implemented Silicide Simulation Features 1 2 Two or more metal silicide pairs can be simulated simultaneously The volume reduction effect is now specified by two volume ratio parameters ALPHA for silicide metal and silicide silicon or polysilicon Cobalt and CoSix materials are added User defined metals and silicides specified by parameters MATTYPE and MATTYPE in the SILICIDE statement will be recognized as electrodes in the ELECTRODE statement Silvaco D 5 ATHENA User s Manual Etch and Deposition Features 1 It is now possible to simulate deposition or epitaxy of the layers with linearly graded impurity or point defect content New parameters F BORON F PHOSPHORUS F INTERST F VACANCY are added to the DEPOSIT and EPITAXY statements D 5 2 ELITE Capabilities 1 Fixed dopant enhanced etching in ELITE It is applicable for all etching machines except Monte Carlo Plasma Etch All impurities are now explicitly specified in the RATE DOPE statement D 5 3 OPTOLITH Capabilities 1 Time units parameters SECONDS MINUTES and HOURS are added to the BAKE statement D 5 4 Misc
437. lation and Measurement of Density Variation in Mo Films Sputter Deposited Over Oxide Steps J Vac Sci Technol v A8 p 1593 1990 93 S F Meier Etching Simulation Of Nonplanar Layers M S Thesis UC Berkeley 1987 94 J L Reynolds A R Neureuther and W G Oldham Simulation of Dry Etched Line Etched Profiles J Vac Sci Technol v 16 p 1772 1979 95 A R Neureuther C Y Liu and C H Ting Modeling Ion Milling J Vac Sci Technol p 1167 1979 96 S Takagi K Iyanagi S Onoue T Shinmura and M Fujino Topography Simulation of Reactive Ion Etching Combined with Plasma Simulation Sheath Model and Surface Reaction Model Japanese J Appl Phys v 41 no 6A p 3947 2002 97 P Sutardja Y Shacham Diamand and W G Oldham Two Dimensional Simulation of Glass Reflow And Silicon Oxidation VLSI Technology Technical Digest p 39 1986 98 P A Burke Semi Empirical Modeling of SiO2 Chemical Mechanical Polishing Planarization Proc VMIC Conf p 379 1991 99 J Warnock A Two Dimensional Process Model for Chemimechanical Polish Planarization J Electrochem Soc v 138 p 2398 1991 100 B M Watraslewicz Image Formation in Microscopy at High Numerical Aperture Optical Acta v 12 p 167 1965 101 B Richards E Wolf Electromagnetic Diffraction in Optical Systems II Structure of the Image Field In An Aplanatic System Proc Phys Soc v A 253 p 358 1959
438. lation is projected on which is what finally goes into ATHENA structure The orientation of the simulation plane is specified by the ROT SUB parameter in the INITIALIZE statement By default the simulation plane is oriented along equivalent lt 101 gt direction ROT SUB 45 In summary the laboratory coordinate system used in the BCA implant simulation is right hand sided Y is depth X is the other co ordinate and Z is from the observer The azimuth angle is measured in the X Z plane ROT SUB is relative to X Simulation plane is X Y The simulation plane where TONYPLOT displays results and ATHENA calculates is always parallel to the major flat which is specified by ROT SUB 90 lt ROT SUB lt 90 The implant calculation in bulk is 3D Ray tracing for BCA i e calculation of ion impacts scattering from walls and re implantation is 2 5D In other words structure is infinitely extended in the third dimension along Z All simulation results doping damage are projected on the simulation plane and appropriately scaled ROTATION is measured from the major flat and in ATHENA case from the simulation plane because it is coupled with the wafer s major flat For silicon and other crystalline materials you can think of TILT ROTATION as always relative to the simulation plane giving the same shadowing effects while ROT SUB defines which direction the simulation plane will slice the crystal structure through in
439. lation of phosphorus predeposition at 900 C during 1 hour with various surface concentration Squares pluses and crosses are experimental data from 26 As our simulations prove the PLS model accurately reproduces the experimental profiles features Particularly the simulated profiles exhibit the enhanced tail with more or less pronounced inflexion in the surface region This inflexion is the result of the strong coupling between the defect gradients and the dopant 3 38 Silvaco SSUPREM4 Models Cowern s Experiment To validate the IC model we first try to compare our model prediction with the experimental data obtained by Cowern et al 20 Briefly described this experiment consists of observing the diffusivity of two boron marker layers after a silicon implantation at 40 keV to a dose of 2 10 atoms cm simulated in the left hand side from Figure 3 5 From these observations Cowern et al have estimated the diffusion enhancement and the evolution of supersaturation with time On the right hand side of Figure 3 5 the PLS simulation of the supersaturation evolution is shown File View Pite Tools Print Properties Help e bw rint Properties Help al Boron Diffusion in Cowern experiment i COWERN EXPERIMENT ele Complete simulation Evolution of the seff interstitial supersaturation 19 zj Experimental Data at Experimental data at 600
440. lax algorithm works only with rectangular base grid e it never eliminates grid lines adjacent to a region boundary e the relaxed area should be at least five by five grid points Silvaco 2 23 Within silicon each second horizontal line is eliminated The lower part of each second vertical line is also eliminated This happens because the algorithm doesn t allow the formation of obtuse triangles ATHENA User s Manual in _direction SILVACO International 1994 4 Microns Microns Relax entire grid Relaw lower holf ont NANI BUSI T 4 Lal gt 2 gt EEEE AANA es NARAANANAANS ARERR ERASERS CS PARAS IAS RANMA RASA aD Sa S NaS A T SE Tea aN E S EEES PERE EREET RRR RERE E ERTS RAMAN AAAS AALAND Ea ANAS AAAS ogg 8 8 dg ag re ebruotures Microns Micrans Before relox Relox lower half of the File vj i View F C Plot 7 Tools 7 Print 7 Properties 7 Help 7 CUS IA BUSS y Silvaco d leave 0 3 ion an irect 1 00 Y MIN T DIR YSE F which prevents elimination in Y 0 3 remains intact see the plot in directions happens only below y 0 3 1 00 DIR X 0 3 choose Selected and set all four boundaries in the LAX statement 0 3 Y MAX F am CTION RELAX X MIN 0 00 X MAX F ama T the
441. le to run a third party program that reads and writes Silvaco format files with the fixed and output str STRUCTURE OUTF mysave str names input str system mv mysave str input str source myprog exe mv output str myrestart str INIT INF myrestart str The UNIX re direct symbol gt is not supported by the system command The UNIX echo and sed syntax can be used instead to output values or variables to a given filename For example to save the extracted value of the variable myvariable to the file called myfile system echo S myvariable sed n w myfile Silvaco 6 109 TONYPLOT ATHENA User s Manual 6 65 TONYPLOT Tonyplot starts the graphical post processor TONYPLOT Note The Tonyplot statement is executed by DECKBUILD which is fully documented in the VWF INTERACTIVE TOOLS USER S MANUAL VOL I Examples All graphics in ATHENA is performed by saving a file and then loading the file into ToNyPLor The command tonyplot causes ATHENA to automatically save a file and plot it in TonyPLoT The ToNYPLOT window will appear displaying the material boundaries Use the Plot Display menu to see more graphics options The following command will display the myfile str file tonypl lot st myfile str The following command will overlay the results of myfilel str and myfile2 str tonypl Note For do lot overlay myfilel str myfile2 str cumentation of the extensive fe
442. lf interstitials and vacancies m Parameters D Dy E and Ey are defined in the defect mod file Finally taking into account that np n and s ea 9 _ a I 3 86 the flux equations for I and V are defined as follows f A za ot ots An f Jo I D P f 3 87 kY SA Pe a y D D SOC me ci 3 88 Dopant Defects Pairs The diffusion of dopant defect pairs implies on the fact that diffusivities Di and Di are calculated from the basic parameters for the pairs defined by Equation 3 77 through 3 81 as follows i ees q Ves 2 we q 3 Di fDi YP Ae D U fpD4 20 yay 3 89 where fy is the interstitialcy component under intrinsic conditions and D4 is the intrinsic diffusivity of the impurity A The fy and D4 parameters are defined for each dopant in the corresponding dopant mod file Silvaco 3 27 ATHENA User s Manual It should be emphasized that several relationships exist between this various parameters Therefore you can decrease the number of free fitting parameters From physical point of view it can be safely assumed the ratio between the diffusivities of the various charge states must be equal Consequently we have D Dus D gt De Al AI AV AV AY _AV 3 90 ie Pip F Piy AI AI AV AV where s 1 for donors and s 1 for acceptors These diffusivities are free parameters when the various coupling parameters in Equation 3 81 are known Therefore you can calcu
443. lity factor WETO2 and DRYO2 specify whether the viscosity parameters are for wet or dry oxidation These parameters are valid only if the specified material is oxide YOUNG M specifies the Young s modulus for the material in dyne cm This parameter is used in stress calculations and also with the compress model for oxidation Also see METHOD LIFT POLY POISS R specifies the Poisson s ratio for the material This parameter is used in stress calculations LCTE specifies the linear coefficient of thermal expansion as a function of temperature T It is expressed as a fraction rather than a percentage INTRIN SIG specifies the initial uniform stress state of a material such as a thin film of nitride deposited on the substrate It can be specified as a function of temperature by using an expression and the variable T in K 6 62 Silvaco MATERIAL OXIDIZABLE specifies that the material could be oxidized If you set OXIDI ZABLE to TRUE then all oxidation related parameters for this material will be set equal to those specified earlier for Silicon You can specify different values in the OXIDE statement Parameters Related to Material Characteristics used by Monte Carlo Implant DENSITY specifies the density of the material in g em COMPONENTS specifies number of atomic components in the material AT NUM 1 AT NUM 2 AT NUM 3 and AT NUM 4 specify the atomic numbers of the constituent
444. lling list and press the Start MaskViews button The MASKVIEWS window will then appear as shown in Figure 2 48 Silvaco 2 65 ATHENA User s Manual Maskviews 2 6 5 mvanex01 lay dir export main mishat athman Files v Define v Edit Options write file Properties ATHENA z 5 te Layers WELL AAD VIA RES EN EN PC fC CC ESE ele minaman ajoaan proncmmAam P 1993 SILYACO International Figure 2 48 MaskViews Window This section will describe how to modify a grid specification for ATHENA First set the grid in the Y direction by selecting Grid Y from the Define menu Figure 2 49 shows the Vertical Grid Control popup will appear You can add modify and delete the lines for ATHENA initial rectangular grid exactly the same way as using the ATHENA Mesh Define menu from DECKBUILD as previously d
445. llowing approach is realized Non linear dependence of n on dose is defined through the PAC concentration as i N unexposed t GULL Mpycl x 5 30 Here Nynexposed is the complex refraction index of the unexposed resist An Nexnosed unexposed 18 the difference between values of n for completely exposed and unexposed resists Current intensity distribution is calculated after each simulation of direct propagation and all the reflections from interfaces with BPM Then current Mpac and n x y are calculated using Equation 5 29 and Equation 5 30 respectively for each point of the resist volume The new values of n x y are used during next recursion of the field and intensity simulations Thus the resulting intensity distribution is obtained as an accumulation of intermediate results You can specify the optical properties of the simulated material in the OPTICAL statement You can specify the refraction and absorption indices for unexposed resist and or for any other material using 5 10 Silvaco OPTOLITH Models the REFRAC REAL and REFRAC IMAG parameters respectively To specify the difference of the refraction index for the completely exposed resist from the unexposed one use the DELTA REAL and DELTA IMAG parameters If this difference isn t specified the effect of intensity on the resist refraction index will not be taken into account during the simulation The number of recursions to obtain the intensity d
446. llowing are the lists of those SILVACO standard materials that can be used in ATHENA as previously described e Semiconductors Fictive GaAs AlInAs AlAs Alpha Si 1 Alpha Si 2 Alpha Si 3 Alpha Si 4 AlxGal_xAs_x 0 25 AlxGal_xAs x 0 5 AlxGal_xAs_ x 0 75 InxGal_xAs_ x 0 50 Unstr InxGal_xAs_x_0 33 Str GaAs InxGal_xAs_ x_0 75 Str InP AlxIn1 xAs_ x 0 50 Diamond AIP AlSb GaSb GaP InSb InAs ZnS ZnSe ZnTe CdS CdSe CdTe HgS HgTe PbSe PbTe SnTe ScN GaN AIN InN BeTe InGaP GaSbP GaSbAs InAlAs InAsP GaAsP HgCdTe CdZnTe InGaAsP AlGaAsP AlGaAsSb SiN Si CulnGaSe InGaN AlGaN InAlGaN InGaNAs InGaNP AlGaNAs AlGaNP AlInNAs AlInNP InAlGaAs InAlGaP InAlAsP Penta cene Alq3 TPD PPV and Organic e Insulators Sapphire Vacuum TEOS BSG BPSG PMMA SOG Polyimide Cooling package material Ambient Air Insulator Polymer and ITO e Metals and Silicides Gold Silver AlSi Palladium Molibdinum Lead Iron Tantalum AISiTi AlSiCu TiW Copper Tin Nickel WSix NiSix TaSix PdSix MoSix ZrSix AlSix Conductor Contact Ba2YCu307 and Ba2NdCu307 The generic name MATERIAL specifies the second material in those statements which specify parameters related to the boundary between two materials 6 2 10 Standard Impurities Different impurities can be specified as parameters in various statements The generic name IMPURITY appeared in a sta
447. llustration of the Hard Polish Model Structure before Planarization al le Dxi gt Ymax Total X Distance zl Figure 4 13 Illustration of Hard Polish Model Structure after Planarization A total amount of AY is always removed at each time step in the above fashion You can mix the hard polish model with the soft polish model and isotropic etch component by specifying the ISOTROPIC parameter of the RATE POLISH statement 4 22 Silvaco ELITE Models 4 6 2 Soft Polish Model The soft polish model is based on the work of J Warnock 99 It has four parameters SOFT LENGTH FAC HEIGHT FAC and KINETIC FAC SOFT is the polish rate on a flat surface LENGTH FAC is the horizontal deformation scale in microns It is a measure of the polishing pad s flexibility It describes the distance at which shadowing will be felt by a tall feature HEIGHT FAC is the vertical deformation scale in microns This measures how much the polishing pad will deform with respect to the height of the feature KINETIC FAC increase the vertical polish rate as the surface becomes more vertical The following formula gives the polishing rate K A P S L 4 34 K is the kinetic factor or horizontal component of the polish removal rate at point i A is the accelerating factor of point i and is large for points that are higher and shadow other points S is the shadow factor and decreases the polish r
448. lly Coupled Equations The point defects dopant defect pairs and active dopant evolve according to the following continuity equations ot s V J GR GR GRyy_p ot i V Jy GR y GRyy GR4r1 y 3 82 ON V J4 GR4 GR a Vda GRay AI Vs t uy A V Jiy GRay GRay_p OA GR GRyy GRay_1 GRay_y Here X is the total concentration of the species X and A is active dopant concentration in substitutional position J represents the flux of species X and GRy is a generation recombination term corresponding to the reactions that contain X Flux Equations Point Defects As the migration rate of a given species may depend on its charge state two distributions must take into account in the flux equation The first one is the Fickian term and the second is the Nerstian term Thus the flux can be written for a given species X with the charge state s Ip Dy ax 5x22 3 83 3 26 Silvaco SSUPREM4 Models For point defects the PLS model makes the assumption that the diffusivity is independent of the charge state at high temperature This implies that for vacancy and silicon self interstitial at any charge state s the following is always true D Ds Dy Dy 3 84 Moreover it is assumed that diffusivities at high temperature follow the simple Arrhenius law Dy De exp 2L 3 85 D p exp H IT ia where E and Ep represent respectively the migration energy for se
449. lm of metal 4 3 1 Conformal Deposition You can perform conformal deposition by specifying a material to deposit a thickness and a number of vertical grid spacings on the DEPOSIT statement The conformal deposition model produces unity step coverage 4 3 2 CVD Deposition To use this model specify the cvD parameter in the RATE DEPO statement as well as the material type the deposition rate DEP RATE and step coverage STEP COV The local deposition rate R x y for the cvp model is given by R x y DEP RATE 1 STEP COV cos 0 STEP COV 4 1 where 0 is the angle between the surface segment and the horizontal 4 3 3 Unidirectional Deposition To specify this model specify the UNIDIRECT parameter in the RATE DEPO statement As shown in Figure 4 2 the region of the substrate not shadowed sees the arrival of the vapor streams in one direction only The growth rate of the deposited film in the shadowed region is equal to zero According to these assumptions growth rate on the substrate R x y can be expressed as R x y 0 if point x y is shadowed 4 2 R x y Csinoi Csino j 4 3 where e isthe angle between the y axis and the direction of the vapor stream e iandj are the unit vectors in the x and y direction respectively e Cis the growth rate of an unshadowed surface normal to the vapor stream e Angle o is specified as ANGLE1 on the RATE DEPO command
450. ls This implementation removes the use of the rectangu lar grid that is utilized in other versions of SSUPREM4 and that is frequently responsible for large memory requirements during implant calculations This model can be invoked by specifying the D 26 Silvaco ATHENA Version History TILT parameter in the IMPLANT statement e Speed enhancements for diffusion and oxidation calculations Speed enhancements have been incorporated that provide an overall speed improvement by a factor of two for typical diffusion calculations Monte Carlo ion implant model Version 5 0 of SSUPREM4 introduces a fast Monte Carlo ion implant calculation This calculation is very general and because of significant developments in modeling and computational techniques is from 10 to 100 times as fast as similar calculations from other sources The model includes the following effects Implant angle tilt and rotation Substrate damage and damage temperature dependence Reflected Ions Physical modeling of penetration through multi layer structures This model can be invoked by specifying the MONTECAR parameter on the IMPLANT statement Non uniform grid capability A non uniform grid can now be specified in the vertical direction for either deposit or epitaxy process steps This is especially useful for modeling epitaxial processes This capability can be invoked by specifying the DY and YDY parameters as described in the EPIT AXY and DEPOSIT statem
451. lt n gt ABUND 2 lt n gt ABUND 3 lt n gt ABUND 4 lt n gt REFLOW GAMMA REFLO lt n gt NO FLIP NIFACT SIGE lt n gt EAFACT SIGE lt n gt NIFACT SIC EAFACT SIC Description MATERIAL specify the material for which all parameters apply see Section 6 2 9 Standard and User Defined Materials for the list of materials Parameters Related to Impurity Diffusion NI 0 NI E and NI POW specify parameters of the intrinsic electron concentration as a function of temperature NI 0 is the preexponential constant in the intrinsic electron concentration formula NI E is the corresponding activation energy NI POW is the unitless power constant These parameters are used only in diffusion calculation and not in EXTRACT electrical calculations EPS specifies the relative dielectric permittivity of the material This value is used to calculate electric field in semiconductors during diffusion simulation This value isn t used in EXTRACT electrical calculations E FIELD specifies that electric field term will be included in the impurity diffusion equations for this material The default is true Parameters Related to Material Stress and Viscosity VISC 0 VISC E and VISC X specify the material viscosity parameters VISC 0 is the pre exponential coefficient in gcm s VISC E is the activation energy in eV VISC X is the incompressibi
452. lues of l m and n describe the order of aberrations while the coefficients W l m n determine the magnitude of the aberrations The aberration coefficients up to the ninth order of aberration are specified in the ABERRATION statement For third order aberrations 1 m and n take the values 1 0 m 0 n 2 spherical aberration 1 0 m 1 n 1 coma 1 0 m 2 n 0 astigmatism l 1 m 0 n 1 field curvature l 1 m 1 n 0 distortion 1 0 m 0 n 1 defocus where isoplanatism is assumed for the particular section of the image field for which the irradiance distribution is calculated The coefficient Wo can be determined from hoot D Woo1 g sna 5 16 where refers to the distance of the defocused image plane to Gaussian image plane The resulting amplitude in the image plane due to a wave coming from the point xo Zo of the effective source is A XxXo Zou 5y NX Zo fa X X9 Z Zg exp i ux vZ dxdz 5 17 where u v refers to a point in the image plane The irradiance distribution associated with the illuminating wave of the effective source will then be represented by 2 dI Xo 29 U5 v A Xo Zo sus v dxodzo 5 18 Since by definition the effective source is equivalent to a self luminous source the total irradiance at u v can be obtained by integrating over the entire source Iusyv fia Xo Zou 5 V N dxo dzo 5 19 y where indicates the
453. lvaco Tutorial To improve the initial grid in the x direction consider two things First make sure that a good 2D profile resolution is specified under the mask edges Second make sure the vertical grid lines are placed along future mask edges To build half of a 0 6 um MOS structure with the center of the gate at x 0 there must be an additional X line at x 0 3 and spacing at this line must be small enough to obtain good lateral resolution of source drain implants To add these items return to the X direction specification in the Mesh Define menu and insert an additional X line at x 0 3 with spacing 0 02 After this final insertion and adding any desired Comment information the Mesh Define menu should appear as shown in Figure 2 8 The grid will have 525 points and 960 triangles see Figure 2 9 Deckbuild ATHENA Mesh Define Direction Location s loc 0 00 spac 0 10 s l c 0 3 spac 0 02 s loc 1 spac 0 1 Insert Delete View Co hd Location 0 00 0 00 H 16 00 Spacing 0 10 0 00 m 1 00 Comment Non uniform grid Figure 2 8 ATHENA Mesh Define Menu View Grid 525 i triangles Figure 2 9 Redefined Grid Silvaco 2 13 ATHENA User s Manual Finally write the Mesh Define information to the file by pressing the Write button A set of lines like these will appear GO ATHENA NON UNIFORM GRID LINE X LOC 0 00 SPAC 0 1 LINE X LOC 0 3 SPAC 0 02 LINE X LOC
454. ly take pair populations into account have been implemented by various research groups But all these models suffer from the lack of established experimental data such as binding energies or pairing coefficients for which reason the predictability of these models is questionable The lack of data especially for the energy levels of the different charge states of the point defects in the band gap at typical diffusion temperatures poses a serious gap in our knowledge Some of these energy levels have been measured in low temperature experiments such as DLTS deep level transient spectroscopy but no one knows how these levels adjust themselves relative to the band edges when the band gap narrows as a function of increasing temperature Van Vechten 8 has theoretically argued that the acceptor states 0 and and the donor states and 0 of the mono vacancy follow the conduction band edge with increasing temperature Mathiot 5 however chooses to scale the positions of the energy levels relative to the band edges with the size of the band gap In addition to the models described above which are all specific for dopants and point defects in silicon there is a smaller number of hard coded models that are used for other materials such as oxide or poly In the sections that follow we apply standard notation used in the literature for dopants point defects interstitials and vacancies and the different charge states as shown in Table
455. m a complex network The texture and morphology of the grain structure depends on the deposition conditions and on subsequent thermal treatment during which recrystallization can occur Impurities inside the grain will diffuse differently than those in the grain boundaries Dopant will also transport through grain and grain boundary interfaces A model for impurity diffusion in polysilicon outlined in 15 16 and 17 is incorporated in SSUPREM4 To use polysilicon diffusion specify the POLY DIFF parameter in the METHOD statement Most of parameters which control the model have a prefix PD and are specified in the IMPURITY statement In this model the concentration of each impurity C is split into two components namely the concentration within grain interior C and concentration in the grain boundary C The impurity diffusion within grain interior is simulated by the standard model used for crystalline silicon see Equations 3 2 to 3 7 ar D v z c ivn G ot l n 1 3 66 where diffusivity of impurity i within grains D is calculated exactly as in Equations 3 10 and 3 11 The diffusion in the grain boundary is assumed to be constant and very rapid gb OCR b gb l where DE b is diffusivity of impurity i along the grain boundaries b D PD DIX 0 gt exp ees 3 68 kT The PD DIX 0 and PD DIX E parameters are specified in the IMPURITY statement The last term G in Equations 3 66 and 3 67 controls impurit
456. m empirical modeling The goal of empirical modeling is to obtain analytic formulae that approximate existing data with accuracy and minimum complexity Empirical models provide efficient approximation and interpolation Empirical models however doesn t provide insight predictive capabilities or capture theoretical knowledge Physically based simulation is an alternative to experiments as a source of data Empirical modeling can provide compact representations of data from either source Physically based simulation has become very important for two reasons One it s almost always much quicker and cheaper than performing experiments Two it provides information that is difficult or impossible to measure Physically based simulation has two drawbacks you must incorporated are that all the relevant physics and chemistry into a simulator and numerical procedures and you must be implemented to solve the associated equations But these tasks have been taken care of for ATHENA users Physically based process simulation tools users must specify the problem to be simulated ATHENA users specify the problem by defining the following e The initial geometry of the structure to be simulated e The sequence of process steps e g implantation etching diffusion exposure that are to be sim ulated e The physical models to be used The subsequent chapters of this manual describe how to perform these steps Silvaco Chapter 2 Tutoria
457. m the fully coupled model The main physical points taken into account in the models are the following e Dopant diffusion of all species is assisted by both vacancies V and self interstitials Z These point defects exist in various charge states and their relative concentrations depend on the local Fermi level position i e on the local dopant concentration e Both J and V have strong binding energies with the dopant atoms and consequently the diffusing species are dopant defect pairs the isolated substitutional dopants are immobile These impurity defect pairs in their various charge states are not assumed to be in local equilibrium with the free substitutional dopant atoms and the free defects In the PLS model at high dopant concentrations Silvaco 3 23 ATHENA User s Manual the concentrations of these pairs are not considered to be negligible with respect to the substitutional active dopant concentration Therefore the pair concentrations are explicitly taken into account to compute the total dopant concentration and the Fermi level position i e carrier concentration Consequently a partial self compensation takes place at high doping concentrations which contributes to the differences between total and active concentrations and affects the variations of the extrinsic diffusivities as a function of the total doping e The flux of each diffusing species dopant defect pairs and free defects includes the drift terms
458. mation as follows deposit an oxide of a specified thickness e g 0 2um and then etch the same thickness again CLEAN GATE OXIDE ETCH OXIDE DRY THICK 0 03 SPACER DEPOSITION DEPOSIT OXIDE THICK 0 2 DIVISIONS 8 SPACER ETCHING ETCH OXIDE DRY THICK 0 23 The dry etching step etches the specified material in the region between the top exposed boundary of the structure and a line obtained by translating the boundary line down in the Y direction The etch distance is specified by the THICK parameter Figure 2 19 shows the resulting spacer Reducing Grid Points in Non Essential Areas using the Relax Parameter The previous sections demonstrate that the quality of the grid is extremely important for ATHENA simulation The rectangular based grid generated by the INITIALIZE or DEPOSIT statements may remain intact in those areas not involved in the process steps affecting the grid e g etching or oxidation The Grid Relax capability allows the spacing to be increased in such areas at any point during the simulation This capability is useful for two reasons First the initial small spacings are propagated throughout the structure For example the fine grid in the X direction shown in Figure 2 9 may be needed only in the upper portion of the structure where doping occurs Eliminating some grid lines and points in the lower portion of the structure will not affect the accuracy of implant and diffusion
459. may be affected a little in the very high injection region giving scope for fine tuning the profile of collector current versus base emitter voltage File View T Plot Tools T Print Properties Help ee Effect of Base Doping Concentration on the Gummel Plot Comparison between 5e17 cm3 and 1e18 cm3 o 5017 A 1018 0 6 0 7 0 8 Base Voltage V Loading file home derekk dk examples BIPOLAR atlas BJT standard 1e18 OK SILVACO International 1996 Figure 2 37 Effect of base doping profile on low injection base current in BUT If the pinched or intrinsic base sheet resistance is a measured parameter the simplest way to match measured and simulated data is to make slight changes to the base implant dose so that the simulated dose is not outside the expected error in actual implanted dose in conjunction with the error in percentage activation In some designs where the base contact is close to the collector contact or the base contact is the substrate or is generally wide the collector current can also influence all current injection regions by specifying a surface recombination velocity at the base contact For a typical design with a buried n collector and surface contacts the surface recombination velocity at the base contact may have little affect on the collector current 2 50 Silvaco Tutorial 2 6 4 The Base Current Profile Medium Injection In ATLAS there are two maj
460. me and improve statistical quality of simulated profiles ATHENA implements a three dimensional rare event algorithm An implantation profile can differ significantly in concentration values across implantation depth Low concentrations in the profile are due to low probability of implanted species rare events to reach that point in space Therefore the number of cascades simulated to get good statistics profile depends on the desired number of orders of magnitude of accuracy Even in real experiments depending on device size implant distributions below some threshold concentration value could exhibit significant statistical noise The algorithm uses trajectory splitting to achieve increased occurrence of the rare target events by generating several independent sub trajectories from less rare events The original idea 67 68 and 69 was first developed into a refined simulation technique by Villi n Altamirano et al 70 They call their version of this approach Restart The basic idea is to identify subspaces from where it is more likely to reach the target subspace where the rare event occurs Each time these subspaces are reached current events sequence are split in a number of replicas all continuing from the splitting state The number of rare events will then increase depending on the number of restart thresholds defined and the number of replicas generated The trajectory splitting algorithm naturally fits into the problem of Monte Ca
461. me of run 1 0 minutes TUNING PARAMETERS Wi Surface Grid Multiplier 1 0 9 1 O gE fact i dt max 0 25 0 00 0 50 Comment Figure 2 46 ATHENA Parameters to Run the Define Machine Etch Menu Section The Machine Name TESTO2 the time units e g minutes and the Time of run e g 1 0 must be specified There are also two tuning parameters that control time stepping during the etch process To improve the smoothness of the etch surface decrease the maximum time step parameter DT MAX from its default value of 10 percent of the specified Time of Run value If you set the ATHENA Etch Menu as shown in Figure 2 46 the following ETCH statement will appear in the input file when you press the WRITE button 1 MINUTE ETCHING USING TESTO2 ETCH MACHINE ETCH MACHINE TESTO02 TIME 1 0 MINUTES DT MAX 0 25 A new parameter DX MULT will be added to the ETCH statement to allow enhanced discretization during individual ELITE Etch steps Increasing the value of DX MULT from its default value of 1 0 will result in larger surface segments and a reduced discretization Decreasing DX MULT will result in better discretization in both space and time during the etch calculation Reducing the value of this parameter allows realistic modeling of wet etches that previously were poorly resolved Use the DX MULT is preferable to the use of DT MAX 2 64 Silvaco Tutori
462. ment and Substitutions Capability DEFINE statement specifies strings for substitution in the following input statements until the UNDEF INE statement is encountered The following DEFINE statements and corresponding substitu tions are allowed DEFINE dconditions temp 1000 time 10 dry DIFFUSE dconditions DEFINE t1 5 0 DEFINE t2 10 DIFFUSE temp 900 time t1 t2 or DIFFUSE temp 900 time Stl St2 Silvaco E 1 ATHENA User s Manual or DIFFUSE temp 900 time t1 t2 In case when you have to redefine a string you should use the DEFINE statement E 2 2 IF ELSEIF ELSE IF END Capability This allows you to perform segments of input deck depending on conditions set in the IF COND condition or IFELSE COND condition For example the following sequence extracts the gate oxide thickness If it is greater than required 100 A then the extra oxide thickness is etched Otherwise the lacking oxide thickness is deposited extract name gateox thickness material Si0 2 mat occno 1 x val 0 5 extract name gateoxdiff 1 e 5 Sgateox 100 0 IF cond S gateoxdiff gt 0 0 etch oxide thick gateoxdiff ELSE deposit oxide thick Sgateoxdiff IF END E 2 3 LOOP L END ASSIGN L MODIFY Capability LOOP and L END statements defines the beginning and end of an inpu
463. meter will be shifted so its center is in the point 0 0 the origin of coordinates for the computational window WIN X LOW WIN X HIGH WIN Z LOW and WIN Z HIGH set the minimum and maximum x and z values that define the image window If unspecified default values from the mask file will be used The units are microns DX specifies the mesh resolution for the image window in x If DX is not specified X POINTS and Z POINTS will be used The units are microns DZ specifies the mesh resolution for the image window in z The default is DZ DX The units are microns X POINTS and Z POINTS are the number of x and z coordinate points in the image window respectively These parameters are used only if DX is not specified Default value is 10 for both coordinates N PUPIL defines or changes the number of mesh points in the projector s exit pupil used in imaging simulations The value of N PUPIL sets the number of mesh points along the exit pupil s radius Larger values provide better accuracy The default setting should be adequate for accuracy N PUPIL also sets the size of the mask or image cell for imaging simulations Finally N PUPIL affects the Silvaco 6 43 IMAGE ATHENA User s Manual discretization of the source This means that if a very fine source discretization is required N PUPIL should be set to a larger value Note that computation time grows linearly with the number of pupil mesh points and source points use
464. mi Energy is measured as being 0 leV from the conduction band edge the work function of the poly emitter set in the CONTACT statement should be set to 4 17 0 1 4 27V i TonyPlot V2 6 12 A File 7 View Ploty Tools Print Properties 7 Help 7 E Effect of Changing Poly Emitter Work Function on Current Gain versus Log Collector Current En EEE nh gt X WF 4 17eV O WF 4 276V gt sd micelles lias 2 3 tg b SG o ie pe 3 o d gt 36 42 a9 40 4 Ff GS Sw wo A a Collector Current A um Click to place P changes alignment or drag to get leader SILVACO International 1996 Figure 2 38 Effect of emitter contact work function on bipolar gain Silvaco 2 51 ATHENA User s Manual Bandgap Narrowing Effects If the BIPOLAR parameter is stipulated in the MODELS statement in ATLAS bandgap narrowing is included automatically The inclusion of bandgap narrowing in the MODELS statement is strongly advised since this phenomenon has a significant effect on the current gain of the device But to validate the default Klaassen bandgap narrowing model you should also use the Klaassen mobility model Use the additional keyword KLA in the MODELS statement to activate this model For example MODELS BIPOLAR KLA The parameters in the Klaassen bandgap narrowing model are user definable in the MATERIAL statement and describe
465. models e The MATERIAL statement which specifies some basic parameters for all materials e The SILICIDE statement which specifies silicidation coefficients Table 2 3 shows the basic diffusion and oxidation models Table 2 3 Basic Diffusion and Oxidation Models Process Model Assumption Recommendation Diffuse Fermi Default Defect in equilibrium For undamaged substrates in inert ambients two dim Transient defect diffu during oxidation and after medium sion dose implant e g OED full cpl Defect and impurity Post high dose implant amp co diffusion binding energy model effects but execution time is high Oxidation Vertical Planar 1D oxidation only should never be used Compress Default Non planar with lin 2D oxidation e g birds beak ear flow Viscous Non planar with non 2D oxidation e g birds beak with thick Elastic linear flow SigN4 however execution time is higher For a detailed description of all diffusion and oxidation models see Chapter 3 SSUPREM4 Models Sections 3 1 Diffusion Models and 3 3 Oxidation Models 2 4 8 Simulating the Epitaxy Process ATHENA SSUPREM4 can simulate a high temperature silicon epitaxial processes The epitaxy process is considered as a combination of deposit and diffuse processes Therefore processes such as autodoping from a highly doped buried layer into a lightly doped epitaxial layer can be simu
466. modes and has several statements to control it This section describes the adaptive meshing related statements and how to use them Table 2 6 list these statements Table 2 6 Summary of Adaptive Meshing Control Parameter Description METHOD Switches the various automated adaption modes on and off ADAPT MESH Invokes a stand alone adaption of the mesh at a specific point ADAPT PAR Control both the stand alone adaption and the automatic adaption meshing criteria GRID MODEL Describes an external template file containing mesh related statements specific to a general technology or device BASE MESH Defines the 1D starting point of a mesh for an adaptive mesh based simulation BASE PAR Specifies the adaption criteria for the base mesh only The Mechanics of the Base Mesh Formation ATHENA uses adaptive meshing in both 1D and 2D modes ADAPT PAR parameters control both these modes The concept of the Base Mesh however needs to be described A typical simulation e g a MOS is simulated in 1D initially and then switched to 2D during mid process flow perhaps at the Poly Gate definition process step Here the mesh is extruded from 1D to 2D and the result is the base mesh The Base Mesh then forms the basis and is the starting point for 2D Adaptive Meshing The mesh quality of this base mesh is important for success of future adaption for example during source drain implants
467. mplex implantation geometries could lead to large deviations behavior of the system thus overbiasing and underestimating the relevant statistics 3 5 5 lon Implantation Damage Ion implantation induced crystal damage can play an important role in the various mechanisms related to diffusion and oxidation ATHENA includes several different types of damage formation which can be used in a subsequent diffusion calculation Implantation induced damage results from cascades of atomic collisions If these collisions cascades are dense it may result in the crystal lattice becoming locally amorphized Accurate simulation of collision cascades with simultaneous estimation of generating various types of point defects clusters and spatial defects can be done only in elaborated Binary Collision Approximation BCA or Molecular Dynamics MD simulators Such simulations are usually time consuming and impractical within general purpose process simulators Generally the amount of damage and distribution of defects associated with it depend on the energy species and dose of implanted ions ATHENA includes several simple models that link various types of defect distributions with ion implantation distributions calculated using any of the models described in previous sections The following types of defects can be estimated e Interstitial profiles e Vacancy profiles e 311 Clusters e Dislocation Loops You can describe the damage types to the simulator
468. mpurity respectively This parameters are used in Monte Carlo BCA implant calculations Diffusion Parameters The units for all pre exponential diffusion constants are cm sec while the units for activation energies are eV DIX 0 and DIX E specify the diffusion coefficient for the impurity diffusing with neutral interstitials DIX 0 is the pre exponential constant and DIX E is the activation energy Silvaco 6 49 IMPURITY ATHENA User s Manual DIP O and DIP E specify the diffusion coefficient for the impurity diffusing with single positive interstitials DIP 0 is the pre exponential constant DIP E is the activation energy DIPP O and DIPP E specify the diffusion coefficients for the impurity diffusing with double positive interstitials DIM 0 and DIME specify the diffusion coefficient for the impurity diffusing with single negative interstitials DIM 0 is the pre exponential constant DIM E is the activation DIMM 0 and DIMM E specify the impurity diffusing with doubly negative interstitials DIMM 0 is the pre exponential constant DIMM E is the activation energy DVX 0 and DVX E specify the impurity diffusing with neutral vacancies DVX 0 is the pre exponential constant DVX E is the activation energy DVM 0 and DVM E specify the impurity diffusing with single negative vacancies DVM 0 is the pre exponential constant DVM E is the activation energy DVMM 0 and DVMM E specify the impurity diff
469. n Pearson Implant Model Generally the Gaussian distribution is inadequate because real profiles are asymmetrical in most cases The simplest and most widely approved method for calculation of asymmetrical ion implantation profiles is the Pearson distribution particularly the Pearson IV function ATHENA uses this function to obtain longitudinal implantation profiles The Pearson function refers to a family of distribution curves that result as a consequence of solving the following differential equation ea _ x a f x gt 3 179 x bo b 5x box in which f x is the frequency function The constants a by b and bg are related to the moments of f x by AR B 3 jae 3 180 2 2 AR 48 E A Lim a Ie 3 181 by a 3 182 3 66 Silvaco SSUPREM4 Models ote 3 183 2 A where A 10B 12y 18 y and B are the skewness and kurtosis respectively These Pearson distribution parameters are directly related to the four moments u p Ly Hy lg of the distribution f x e _ _ 3 _ 4 Rp Rp Hi is given by o0 a xflx dx 3 185 oe i fees i m ORR Mende 1 23 4 3 186 00 The forms of the solution of the Pearson Differential Equation depend upon the nature of the roots in the equation b tb x b x 0 There are various shapes of the Pearson curves You can find the complete classification of various Pearson curves found in Atomic and Ion Collision in Solids and at Surfaces 46 Obviously o
470. n decrease if necessary for more accuracy DT MAX is used with ELITE type polish calculations By default the upper limit for the micro timestep DT MAX is one tenth of the total etch time specified This is a good compromise between calculation accuracy and calculation time But sometimes it is useful to adapt this value to the specific simulation problem Allowing the time steps to become greater gives a higher simulation speed but the accuracy may suffer For smaller time steps the simulation speed will decrease but the accuracy may be greater Examples The following statements illustrate running the chemical mechanical polish module A RATE POLISH statement sets the values for the polish model and must precede the POLISH statement RATE POLISH OXIDE MACHINE cmp u s MAX HARD 0 15 MIN HARD 0 03 ISOTROPIC 0 001 POLISH MACHINE cmp TIME 5 MIN For more examples see RATE POLISH and ETCH 6 78 Silvaco PRINT 1D 6 42 PRINT 1D PRINT 1D prints values along a one dimensional cross section or an material interface Note Use of this statement is not recommended All functions are available using the EXTRACT command within DECKBUILD Syntax PRINT 1D X VALUE lt n gt Y VALUE lt n gt MATERIAL MATERIAL ARCLENGTH LAYERS X MIN lt n gt X MAX lt n gt FORMAT lt c gt Description This command prints the values along cross sections through the device You
471. n Condensed Matter Phys Rev B v 25 p 5631 1982 F L Vook Defects in Semiconductors p 60 1972 Silvaco BIB 3 ATHENA User s Manual 66 67 68 69 70 71 72 73 T4 75 76 77 78 79 80 81 82 83 84 85 86 87 I R Chakarov and R P Webb CRYSTAL Binary Collision Simulation of Atomic Collision and Damage Buildup in Crystalline Silicon Radiation Effects v 130 131 p 447 1994 H Kahn and T E Harris Estimation of Particle Transmission by Random Sampling National Bureau of Standards Applied Mathematics Series v 12 p 27 1951 B A Bayes Statistical Techniques for Simulation Models The Australian Computer Journal v 2 p 190 1975 J M Hemmersley and D C Handscomb Monte Carlo Methods Methuen and Co Ltd London 1964 M Villi n Altamirano and J Villi n Altamirano RESTART A Method for Accelerating Rare Event Simulations Proc 13th Int Teletraffic Congress ITC 18 Queueing Performance and Control in ATM P C Cohen J W Ed North Holland Copenhagen Denmark p 71 1991 A Phillips and P J Price Monte Carlo Calculations on Hot Electron Energy Tails Appl Phys Lett v 30 1977 S H Yang D Lim S Morris and A F Tasch A More Efficient Approach for Monte Carlo Simulation of Deeply Channeled Implanted Profiles in Single Crystal Silicon Proc NUPAD p 97 1994
472. n be done by selecting the Tools gt Maskviews gt Cutfiles option from DECKBUILD when ATHENA is active which will open the MASKVIEWS Cutline Popup Names of all available mask layers are in Figure 2 55 When you select a name e g POLY from the list press the Apply Mask button and the following lines will appear in the input file DEFINING POLY MASK MASK NAME POLY Deckbuild ATHENA Photo Edit layer Label POLY T E Name Insert layer 7 i Reverse Mask He Delete layer Comment Defining POLY mask Apply mas Strip mask Figure 2 55 ATHENA Photo Popup During runtime DECKBUILD converts the MASK statement into a DEPOSIT statement followed by a series of ETCH statements The mask thickness and material type are defined in the MaskViews Layers Popup Figure 2 56 in the Define menu of MASKVIEWws Maskviews Layers Current layer Label POLY Name first poly def Field Electrodes yj Mis alignments x 0 0 y 0 0 Delta CD 0 00 thickness 0 1 Add Delete Figure 2 56 MaskViews Layers Menu Silvaco ATHENA User s Manual Two types of mask material are available Photoresist and Barrier The real thickness of a photoresist layer should be specified because it can be used as a mask for implantation Barrier is a fictitious material It is impenetrable for any implants and can serve only as a masking material This material is implemented in ATHENA for the purpose o
473. n easier numerical problem due to the avoidance of numerical stiffness But since point defects are not directly simulated the Fermi model cannot deal with certain process conditions in which the defect populations are not in equilibrium such as in wet oxidation where Oxidation Enhanced Diffusion OED is important emitter base diffusions and wherever implantation results in an initial high level of implant damage To use the Fermi Model specify FERMI partameter in he METHOD statement In the Fermi Model each dopant obeys a continuity equation of the form E ch_ Vv Dix VC 2 C 2 ae X 1V where Cop is the chemical impurity concentration Z is the particle charge 1 for donors and 1 for acceptors Day and D4 are the joint contributions to the dopant diffusivity from dopant vacancy and dopant interstitial pairs in different charge states 5 C4 is the mobile impurity concentration and is the electric field The terms Day and Dy depend on both the position of the Fermi level as well as temperature and are expressed as ie 1 2 1 ERNS DTZ D Dix 4 Diy 4 Diy 4 EDs 4 3 10 n n n n i 1 l n l where the temperature dependency is embedded in the intrinsic pair diffusivities which are specified by Arrhenius expressions of the type E D Dye Dea 3 11 Table 3 2 shows the names of the pre exponential factors D 0 and activation energies D E for each of the charge states c of the various
474. n linear effect of the intensity distribution on the local optical properties of the resist material The third reason is because it provides a good accuracy to run time ratio The BPM is used to solve the Helmholtz equation for electromagnetic field inside the structure During the simulation the field distribution is formed as the superposition of incident light with all the reflections from all elements of the resist substrate interface and secondary reflection s from the upper resist surface 106 shows the formal descriptions of the BPM Papers 107 108 and 109 also describe some applications using BPM In this model the Helmholtz equation Equation 5 24 for the electric field E in the media with complex refractive index n x y is solved in two main stages The first stage the diffraction over a small spatial step along the propagation is calculated Thus obtaining the new field amplitude distribution without absorption taken into account Then the actual field distribution is computed as a product of this amplitude distribution and the distribution of the complex absorption over the step Let the wave propagate along the y axis Silvaco 5 9 ATHENA User s Manual We find the solution as a quasi plane wave E A x y exp inky with a slowly varied amplitude A A is then modified with y which is slower than phase term inky In this case the Fourier image of current distribution A in the plane y yq is defined as foll
475. n rates converge Silvaco 4 17 ATHENA User s Manual Calculation of Rates The second stage involves calculation of the etching rates as well as ejection and redeposition rates of the polymer particles During each time step the two processes simultaneously take place on each surface segment The first is redeposition of the polymer with the rate equal to the polymer flux The second is etching by incoming ions and neutrals The combination of these two processes can be treated as deposition of a virtual polymer layer with subsequent etching of the two layer structure If the etch rate of polymer by incoming ions and neutrals is less than the polymer deposition rate the result is the redeposition of a polymer layer on the surface If the etch rate of polymer by incoming ions and neutrals is larger than the polymer deposition rate the result is actual etch of the underlying material Linear Etch Model In the case of the linear model the etch rate ER m of each material m is calculated as ER m S EP m 1 Vans 4 23 n where n is the number of plasma ion types specified by the parameter I10N TYPES n could be equal to 1 or 2 EP m i is the etch parameter for material m and ion type i specified by parameters MC ETCH1 and MC ETCH2 V_ is the ion velocity as calculated in Equation 4 22 abs If calculated ER polymer is less than the polymer flux redeposition rate PF the actual etch rate and ER is negative which
476. nally a planarization etch is performed ATHENASSUPREMS ATHENAGSUPREMS ATHENAELITE NITRIDE SPACER FORMATION LOCAL GRIDATION NITRIDE SPACER REMOVAL Migone Mirona ATHENAALITE ESUPREMS ATHENAELITE ATHENAELITE TRENCH ETCH and CRIDATION TRENCH REFLL TRENCH PLANARIZATION Figure C 3 Simulation of self aligned trench isolation process using the ELITE and SSUPREM4 modules of ATHENA SSUPREM4 is used for the LOCOS and trench oxidation ELITE is used for the trench etch and refill The interface between SSUPREM4 and ELITE is completely automatic and transparent to the user The syntax needed to access the ELITE models can be found using the Deckbuild Command Menus The main parameters are RATE ETCH MACHINE lt name gt to set up parameters for the etch machine and ETCH MACHINE lt name gt TIME lt value gt to run that machine for a given time Analogous commands exist for depositions One key parameter for users of ELITE is DX MULT lt value gt on the ETCH statement This parameter sets the ratio between the grid spacing used by SSUPREM4 and the surface accuracy used by ELITE The default is 1 0 Lower DX MULT values will improve the accuracy and smoothness of etch shapes at the expense of some additional CPU time Silvaco C 7 ATHENA User s Manual Question Can dopant diffusion be modeled simultaneously with the material reflow Answer An extremely important feature of ATHENA is that simulation of t
477. nation model is enabled The IIFACTOR and IVFACTOR parameters on the INTERSTITIAL command are used when METHOD HIGH CONC is enabled I LOOP SINK specifies that a dislocation loop band can be specified during a subsequent implant and that the loops may behave as an interstitial sink during diffusion The DISLOC LOOP command is used to set parameters for this model POLY DIFF specifies that the two stream polysilicon diffusion model should be used To operate accurately set this model before the deposition of the polysilicon material See Section 6 29 IMPURITY and Chapter 3 SSUPREM4 Models Section 3 1 7 Grain based Polysilicon Diffusion Model for more information CLUST TRANS enables the Transient Activation Model DOSE LOSS specifies that Interface Trap Model for dose loss at Silicon Oxide Interface is enabled MODEL SIGEC enables special B diffusion model in SiGeC SiGeC SIGECDF MOD specifies the name of the C Interpreter file for boron diffusivity model in SiGe SIGECNI MOD specifies the name of the C Interpreter file instrinic carrier concentration model used in boron diffusion mode in SiGe SiGeC MIN TEMP specifies the minimum temperature for which impurity diffusion is considered At temperatures below MIN TEMP the impurities are considered immobile The default is 700 C With caution you can set this parameter to a lower value for certain diffusion steps Silvaco 6 65 METHOD ATHENA User s M
478. nce in the linear rate constant is given by a _ geet anes where L PDEP is specified on the OXIDE statement for each oxidant and P is the partial pressure of the oxidizing gas Figure 3 13 shows the silicon dioxide thickness versus time with PRESSURE as a parameter TonyPlot V2 6 9 File 7 View Plot Tools Print Properties 7 Help C 3 E E E 5 H Oxidation Conditions Dry02 900 C lt 100 gt Si 0 HCI 2 Oxidation Time Minutes SILVACO International 1996 Figure 3 13 Silicon Dioxide Thickness versus Time with Pressure as a Parameter Silvaco 3 53 ATHENA User s Manual Chlorine Dependence The addition of chlorine to the oxidation system results in better passivation and higher oxide dielectric strength 30 35 For a dry oxygen ambient chlorine introduction gives rise to a higher oxidation rate It has been suggested 35 that chlorine reacts with O to produce HO and Cl as products The oxidation rate is higher in HO ambients than in O ambients because equilibrium concentration of H320 in the oxide is higher A look up table approach is implemented to model the increase in the linear rate constant in Equation 3 156 though the B A c term The table gives an enhancement factor to the linear rate constant as a function of chlorine percentage and temperature The default values for chlorine dependence are included in Appendix B Default Coefficients The effect
479. ncy to the extent of which is also device design specific Therefore some degree of iteration of the tuning parameters is to be expected When tuning bipolar transistors there is a greater emphasis to accessing tuning parameters by using the device simulator ATLAS compared to optimizing MOSFETs where most tuning parameters are process related A powerful combination is the tuning of a BiCMOS process where you can use the MOSFET part of the process flow to tune the process parameters while using the Bipolar part of the flow to tune ATLAS This technique should yield a high degree of predictability in the results 2 48 Silvaco Tutorial Tuning the process simulator parameters in ATHENA are mainly required to model effects such as the implantation induced defect enhanced diffusion responsible for the Emitter Push Effect which is essential to obtain the correct depth of the base collector junction The correct process modeling of the out diffusion of dopant from the poly emitter into the mono crystalline substrate is also critical to obtaining well matched IV curves Another critical process modeling area is the base implant because it is essential to match measured and modeled base resistance for correct modeling of the collector current These and other issues are discussed in these sections 2 6 1 Tuning Base and Collector Currents All Regions The most important parameter to model the general level of base and collector currents i
480. nd 3 19 is similar to the Fermi Diffusion Model except for two elements The additional term Cy Cy has been added to model the enhancement or retardation of diffusion due to non equilibrium point defect concentrations The term fy takes into account the knowledge that some impurities diffuse more by interstitials than by vacancies or vice versa Although this dependency is of a phenomenological character it seems reasonable and is the one used by most diffusion simulators to account for the diffusion enhancement of dopants during oxidation enhanced diffusion OED or transient enhanced diffusion TED Remember the point defect equilibrium concentrations are temperature as well as Fermi level dependent and can be calculated from the following expressions 2 1 2 n n n n neu ne dneg pos 2 dpos i i n ni ni ni O E EO S 3 22 neu neg dneg pos dpos x Cy i represents the equilibrium defect concentrations of interstitials and vacancies under intrinsic conditions and the weight factors neu neg dneg pos and dpos account for the distribution of defects of different charge states under intrinsic conditions All of these are assumed to be temperature dependent through Arrhenius expressions of the following type where Cy 3 23 neu NEU 0exp MEU E kT where the pre exponential factors and activation energies in this case NEU 0 and NEU E can be specified in the VACANCY and INTERSTIT
481. nd Monte Carlo Etching ELITE uses a string algorithm to describe topographical changes that occur during deposition and etching processes This chapter describes the models and techniques used in ELITE and the command language used to access model parameters Silvaco 4 1 ATHENA User s Manual 4 2 String Algorithm The ELITE simulation regime consists of a set of triangles that hold information on the materials that are being simulated The string algorithm treats each of these interfaces as a set of segments that move in response to a particular process calculation As microfabrication technology becomes more complex modeling each step of the manufacturing process is increasingly important for predicting the performance of the technology Etching is a step that is universal in microfabrication It may take place as the dissolution of a photoresist by an organic solvent the etching of an oxide by an alkali or the plasma etching of an electron resist Whatever its physical details the etching process can in many cases be modeled as a surface etching phenomenon Etching simulation starts from an initial profile that moves through a medium in which the speed of etching propagation can be a function of position and other variables that determine the final profile Two major assumptions limit the generality of the string algorithm in ELITE First the pattern to be etched is uniform in one dimension so the problem can be solved using only tw
482. ne Material Monte Carlo The crystalline model used in ATHENA is based on the program CRYSTAL described elsewhere 66 In order to calculate the rest distribution of the projectiles ATHENA simulates atomic collisions in crystalline targets using the Binary Collision Approximation BCA The algorithm follows out the sequence of an energetic atomic projectiles ions launched from an external beam into a target The targets may have many material regions each with its own crystal structure crystalline or amorphous with many kinds of atoms The slowing down of the projectiles is followed until they either leave the target or their energy falls below some predefined cut off energy The crystal model is invoked with the MONTE parameter in the IMPLANT statement ATHENA will then choose which model to use depending on the predefined crystal structure of the material Specifying CRYSTALLINE has no affect on the implantation and the BCA parameter is just a synonym for MONTE You can manipulate the implantation module to consider all materials amorphous by adding the AMORPHOUS parameter in the IMPLANT statement At that moment the materials with predefined crystal structure are Si Ge GaAs SiGe InP as well as three types of silicon carbides 8C SiC 4H SiC and 6H SiC and two types of superconductors Ba2YCu307 and Ba2NdCu807 All remaining materials in ATHENA are considered amorphous Statistical Sampling In order to reduce calculation ti
483. nformation appears in the standard output or in the DeckBuild Text Subwindow unless it is re directed into CPUFILE LOG enables logging of CPU usage when true and disables CPU logging when false The default is true CPUFILE specifies a name of the file to which CPU log is written The default is the standard output Examples The following example enables ATHENA to gather CPU statistics and store it in the file timeusage out CPULOG LOG CPUGFILE timeusage out Note The accuracy of time statistics depends on the computer and operating system It is usually around 0 01 sec 6 20 Silvaco DEPOSIT 6 13 DEPOSIT DEPOSIT deposits a layer of specified material DEPOSITION is a synonym for this statement Note Unless the ELITE module is used all deposition steps in ATHENA are 100 conformal This means deposition on all surfaces with a step coverage of 1 0 Syntax DEPOSIT MATERIAL NAME RESIST lt c gt THICKNESS lt n gt SI_TO_POLY TEMPERATURE lt n gt DIVISIONS lt n gt DY lt n gt YDY lt n gt MIN DY lt n gt MIN SPACE lt n gt C IMPURITIES lt n gt F IMPURITIES lt n C INTERST lt n gt F INTERST lt n gt C VACANCY lt n gt F VACANCY lt n gt C FRACTION lt n gt F FRACTION lt n gt MACHINE lt c gt TIME lt n gt HOURS MINUTES SECONDS N PARTICLE lt n gt OUTFILE lt c gt SUBSTEPS lt n gt VOID
484. ng other VWF Interactive tools To start ATHENA under DECKBUILD in interactive mode enter the following UNIX command deckbuild an After a short delay the Main Deckbuild Window See Figure 2 1 will appear The lower text window of this window will contain the ATHENA logo and version number a list of available modules and a command prompt ATHENA is now ready to run To become familiar with the mechanics of running ATHENA under DECKBUILD load and run some of the ATHENA standard examples The method described here is the recommended procedure for starting the program There are other methods and modes of running ATHENA Section 2 2 Operation Modes or Section 2 8 Using Advanced Features of ATHENA will describe these methods and modes Silvaco 2 1 ATHENA User s Manual Deckbuild 3 5 3 Beta manualin dir tmp_mnt main lucky stacy manual i File 7 View v CEdit 7 Find 7 Main Control v Commands 7 i Tools 7 co he 0 00 spac 0 10 0 30 spac 0 02 1 00 spac 0 10 go athena Joc 0 00 spac 0 03 Joc 0 20 spac 0 02 Joc 1 00 spac 0 10 silicon boron 3 0e14 orientation 100 two d struct outfile manual_ 1 str tonyplot manual _ 1 str set manual_ 1 set deposit oxide thick 0 02 divisions 2 struct outfile manual_ 2 str deposit poly thick 0 50 c phosphor 5 0e19 divisions 10 dy 0 02 ydy 0 0 4 min spac 0 001 struct outfile manual_ 3 str init infilesmanual_ 2 str deposit
485. nitialization to convert a complete input file to the appropriate fast mode and back to normal operation Question It is known that Silvaco s device simulator ATLAS allows the simulation of device structures with cylin drical symmetry Does ATHENA support the grid with cylindrical symmetry Silvaco C 1 ATHENA User s Manual Answer Yes you can specify the cylindrical coordinate system in the INITIALIZE statement choose Cylindri cal in the ATHENA Mesh Initialize Menu The axis of symmetry is always at x 0 Question In some cases the grid within oxide generated during the oxidation step is very coarse Does this affect accurate estimation of dopant segregation Does the shape of the oxide region depend on the quality of internal grid Is it possible to control the grid during oxidation Answer The thickness of grid layers during oxidation is controllable Two parameters of the METHOD statement affect the oxide grid GRID OXIDE and GRIDINIT OXIDE GRID OXIDE specifies the maximum grid layer thickness in microns GRIDINIT OXIDE specifies the maximum thickness of the very first grid layer generated in the growing oxide For both parameters the default is 0 1 microns These defaults are reasonable for simulation of thick 0 6 1 0 field oxide growth But for thinner oxides these parameters should be decreased For example if an 0 025p gate oxide is growing it is a good idea to set GRIDINIT OXIDE to 0 005 and GR
486. nly bell shaped curves are applicable to ion implantation profiles It is readily shown by Ashworth Oven and Mundin 47 that fx has a maximum when bg b x box lt 0 You can reformulate this as the following relation between B and y oko 5 9f 16 i 2567 oy 3 E eee 3 187 2 50 y 2 with the additional constraint that y lt 50 Only Pearson type IV has a single maximum at x a R and monotonic decay to zero on both sides of the distribution Therefore Pearson type IV is usually used for ion implantation profiles it is the solution of Equation 3 178 when the following conditions are satisfied 2 2 3 2 z 39y 48 6 y 4 B tne Ra S30 3 188 2 32 7 Silvaco 3 67 ATHENA User s Manual This gives the following formula for Pearson IV distribution 1 b i 4 22 bg 2by x R b fix K bg b x R b x R op Se ay e 3 189 j Hoet 4b b3 b qo 87 where K is defined by the constraint o0 Ande 7 3 190 0 In the narrow area of 8 7 plane where Pearson IV type criterion Equation 3 188 is not satisfied while bell shaped profile criterion Equation 3 187 holds ATHENA by default uses other than type IV Pearson functions These functions are bell shaped but they are not specified over the whole interval Usually this doesn t affect the quality of calculated profiles because the limits of these functions are situated far from their maximums If you want to use
487. nough so that any curve that developed would be well defined i e there should be some maximum angle between adjacent segments perhaps 0 1 radians This criterion however led to a proliferation of segments in regions where the front was either expanding or contracting The algorithms in ELITE attempt to maintain approximately equal segment lengths This results in position errors of about one half segment length The error can be reduced by decreasing the average segment length with a proportional increase in computation time 4 2 Silvaco ELITE Models exact solution Etchant local normal to the etch front Material being etched Figure 4 1 String Model approximation to the Etch Front For the most cases of interest the etch rate varies with position This leads to some errors in the position and in the direction of each point on the string Errors in position arise from the use of a rather simple integration algorithm The local rate at the start of each time step is assumed to be constant throughout the step This can easily lead to position errors as large as the distance covered in one step Consider for example an etch front in a photoresist approaching an unetchable substrate A point which is barely outside the substrate at the start of the time step will advance into the substrate at the rate associated with the resist Thin layers of alternating fast and slow etch rates could spawn errors in position With too larg
488. nreliable and unpredictable e Any concentrations of dopant initialized in ATHENA will be overwritten if a PROFILE statement is used to load a SSF file Silvaco PROJECTION 6 45 PROJECTION PROJECTION specifies the basic optical projection parameters for OPTOLITH Syntax PROJECTION NA lt n gt FLARI Description E lt n gt This statement specifies the numerical aperture NA the defocus distance and the possible flare in the optical or resist systems NA is the numerical aperture of the optical projection system FLARE is the amount of flare for the particular imaging problem FLARE must be expressed in percentages Examples The following statement sets the numerical aperture and flare value for the projection system For more examples ABERRATION PROJECTION NA 5 FLARE 2 see IMAGE ILLUMINATION ILLUM FILTER PUPIL FILTER LAYOUT and Silvaco PUPIL FILTER ATHENA User s Manual 6 46 PUPIL FILTER PUPIL FILTER specifies the projection pupil type and filtering for OPTOLITH Syntax PUPIL FILTER CIRCLE SQUARE GAUSSIAN ANTIGAUSS GAMMA lt n gt IN RADIUS lt n gt OUT RADIUS lt n gt PHASE lt n gt TRANSMIT lt n gt CLEAR FIL Description This command allows you to specify four different pupil types and allows spatial filtering in the Fouri
489. nte beats Males E waen ee 3 44 3 3 1 Numerical Oxidation Models 22s 2nekcths deat ved cuwe eed de peda yee eee te ee ees 3 46 3 9 2 COMpless MOE 2 8 5 cute we Noe OS ea eh a ee reeeo a T neno 3 47 Side VISCOUS Modet ar r ce andi tet ned AE E aA Gnd mu E A EEA NMEA Ru Ed A Aa nas 3 48 3 3 4 Linear Rate Constant ss say teads eee Renee ewer eit eie ls eecer testy 3 50 3 3 5 Parabolic Rate COnstats cac cudacr eien EREE EEE teat EA E SE TERE Red ee 3 57 3 3 6 Mixed Ambient Oxidation n nuanua n ananena 3 58 3 3 7 Analytical Oxidation Model nuanse a du aid p40 eee teele dee Seca date at aed eee y 3 58 3 3 8 Recommendations for Successful Oxidation Simulations 0 0 c cece eee 3 59 3 4 Silicidation Model jcs226 2 240 ses den 8 tea Ghee eelediew ec SEA eee ee ed dete eee andes 3 64 3 5 lon implantation MOG el S cus 522 ce ase wwii e a a a bre Aenea ae ata aaa Se 3 66 3 5 T Analyticimplant Model Ssa5 25885 atin ited here ead Dawe aes A hee SAU seem ee SeaT eee 3 66 3 5 2 M lti Layer Implants sas 2c cen cc ae Ary bene Ale ah he BN oh Ae ed steko 3 70 3 5 3 Creating Two Dimensional Implant Profiles 0 2 0 0 ccc cece eee eee eens 3 72 3 5 4 Monte Carlo Implants esto area a eater ete Bink eee ORs Ie ek hte e 3 76 3 5 5 lon lmpl ntation Damages ine net Gi ts nnana e a eA a a a aaa eee 3 87 yi Silvaco Table of Contents 3 5 6 Stopping Powers in Amorphous Materials and Range Validation
490. nterpolates into a table of experimental e g temperature and deactivation threshold data pairs and finds a concentration independent deactivation threshold C th that corresponds to the current simulation temperature You can set these pairs in the IMPURITY statement by assigning values to the parameters SS TEMP and SS CONC The temperature should be specified in Celsius Silvaco 3 19 ATHENA User s Manual 2 A logarithmic concentration dependency is incorporated by setting the final deactivation threshold to the value th b th act act th c ad Cae ETUS 1 b Cact Cact 3 62 act th Cact 7 Cact lt Cact where the parameter b must be in the range of 0 8 1 0 Parameter b can be specified as ACT FACTOR in the IMPURITY statement The effect of Equation 3 62 is to produce a rounding in the top of the active profile that slightly follows the form of the chemical profile Transient Activation Model The Transient Activation Model assumes that dopants after an implant are inactive A certain time is required before the dopants become active After an ion is implanted into silicon this model assumes that all dopants are inactive and may not be activated immediately but become gradually active instead The Transient Activation Model simulates this behavior and applies it to activating the implant dopants The following equation for the active concentration C4 is solved ggl Chem C 3 63
491. nto a directory directly visible to the simulation run regardless where the simulator is executing Silvaco 6 103 STRESS ATHENA User s Manual 6 60 STRESS STRESS calculates elastic stresses Syntax STRESS TEMP 1 lt n gt TEMP2 lt n gt NEL lt n gt Description This command calculates stresses due to thin film intrinsic stress or thermal mismatch TEMP1 and TEMP2 are the initial and final temperatures in C for calculating thermal mismatch stresses NEL is the number of nodes per triangle to use Currently only 6 or 7 are allowed 6 nodes are faster than 7 and usually gives adequate results Default is 6 Examples The following calculates the stresses in the substrate and film arising from a nitride layer which has an intrinsic stress of 1 4 x10 4 dynes cm when deposited uniformly MATERIAL NITRIDE INTRIN SIG 1 4E10 STRESS The following calculates thermal mismatch stress in the whole structure as the result of a temperature change from 1000 to 100 C STRESS TEMP1 1000 TEMP2 100 For more examples see MATERIAL 6 104 Silvaco STRETCH 6 61 STRETCH STRETCH stretches structures about a specified location Syntax STRETCH MATERIAL lt c gt LENGTH lt n gt X VAL lt n gt Y VAL lt n gt STRETCH VAL lt n gt SPACING lt n gt DIVISION lt n gt SNAP Description This statement specifies that t
492. ny sense because ATHENA uses Biersack Brandt Kitagawa stopping model where sqrt E dependency doesn t exist explicitly Moments are Calculated during Monte Carlo Implant Simulation All spatial moments are integrated during Monte Carlo calculations and then can be printed out when PRINT MOM parameter is specified BEAMWIDTH Capability for Monte Carlo Implant now works properly for any number of trajectories It used to wrongly estimate random angle Boundary Conditions PERIODIC and REFLECT now work properly even in the case of 1D simulation SMOOTH Capability now works in all cases used to fail for several combinations of other parameters Oxidation Oxidation Threshold Model Oxidation only occurs for oxidant concentration above some critical value Miscellaneous Features Solid Solubility Tables Extended Boron Solid Solubility Tables have been extended down to 700 minimum temperature New PD Time Stepping Control The initial time step may not be set independently for point defects to dopant This allows greater flexibility to study events occurring during the initial time of an RTA time cycle specifically when employing a new TED diffusion model Equilibrium Point Defects Concentration The equilibrium point defects concentrations Ci and Cv are now output into the SSF file These may now be visualized in TONYPLOT Dump filename extended the files dumped during a diffusion now include three extra decimal p
493. o SR The dual Pearson model will be used only when all nine parameters are present see the Specification of Implant Parameters in the Moments Statement on page 3 76 and the AMORPHOUS parameter is not specified in the IMPLANT statement the default is CRYSTAL Otherwise the single Pearson formula will be used 3 68 Silvaco SSUPREM4 Models SIMS Verified Dual Pearson SVDP Model By default ATHENA uses SIMS Verified Dual Pearson SVDP implant models These are based on the tables from the University of Texas at Austin These tables contain dual Pearson moments for B BF2 P and As extracted from high quality implantation experiments are also conducted by the University of Texas at Austin Table 3 7 show these implantation tables contain dose energy tilt rotation angle and screen oxide thickness dependence Table 3 7 Range of Validity of the SVDP Model in ATHENA lons Energy keV Dose cm Tilt Angle Rotation Angle Screen Oxide A B 1 100 1013 8x1015 0 10 0 360 native oxide 500 BF 1 80 1018 8x1015 0 10 0 360 native oxide P 12 2004 1013 8x1015 0 10 0 360 native oxide As 1 200 1013 8x1015 0 10 0 360 native oxide a Experimentally verified for 5 80keV For energy range 1 5keV an interpolation between 5keV and 0 5keV calculated with UT MARLOWE is used an extrapolation is use
494. o different sets of parameters The first set is DX and DZ DX and DZ are the resolution in micrometers for the x and z directions The second set X POINTS and Z POINTS is based on the number of points in each direction The resolution will be the length of the side of the image window divided by the number of points To study the defocus of the aerial image use the DEFOCUS parameter DEFOCUS uses units of micrometers 2 80 Silvaco Tutorial The N PUPIL parameter specifies the computational window If N PUPIL is not specified it is automatically calculated to a size that encompasses all mask features In these cases you can set the computational window manually using the following formula Neue e yy 2 1 lambda where length is the intercept coordinate for the x and z axes of a square centered at the origin that delimits the Computational Window as shown in Figure 2 66 Computational gt X Window Figure 2 66 The Computational Window is Always Centered at the Origin Once you calculate the image you can store it in a standard structure file by using the STRUCTURE OUTFILE STR INTENSITY command The INTENSITY modifier identifies the file to be different than a standard structure This file can later be initialized into memory and used without running the imaging module To initialize an intensity file type INITIALIZE INFILE STR INTENSITY The intensity modifier ag
495. o dimensions For most microfabrication problems the important cases involve the cross sections of lines so this model is directly applicable In certain other cases such as round holes the symmetry of some cross sections is such that the algorithm is still valid The second major assumption is that the etch rate is a scalar function of position and is independent of the direction of local etch front motion and the history of the front In some real situations this does not hold PMMA for instance has been found to have a gel region at the resist solvent interface during development so the etch rate is a function of the history of the adjacent regions as well as of the exposure Another case where the second assumption does not hold is in the so called preferential etching where etching proceeds more quickly along certain crystal directions making the etch anisotropic The algorithm described here is known as a string algorithm 83 The etch front is simulated by a series of points joined by straight line segments forming a string During each time increment each point advances perpendicularly to the local etch front as in Figure 4 1 A major portion of the algorithm adjusts the number of segments to keep them approximately equal in length Other subroutines input the data and output the etch front Choosing suitable criteria for segment length was a major problem in developing the algorithm It seemed that segments must be short e
496. ode being defined If no value of Y is specified the top of the structure is assumed BACKSIDE specifies that a flat zero height electrode will be placed on the bottom of the simulation structure This is the one exception to whole regions being defined as electrodes If a metal region is present on the bottom of the structure this parameter will not be used and the XY coordinates used instead BOTTOM is a synonym for this parameter LEFT specifies that the top left region of the structure will be defined as an electrode RIGHT specifies that the top right region of the structure will be defined as an electrode Note The ELECTRODE statement recognizes the regions made of polysilicon standard metals see Section 6 2 9 Standard and User Defined Materials for the list of standard metals or user defined materials with the following standard names Gold Silver AlSi Palladium Molybdenum Lead Iron Tantalum AISITi AlSiCu TiW Copper Tin Nickel NiSix TaSix PaSix MoSix ZrSix AlSix Conductor Contact Note ATLAS contains syntax that makes use of the common electrical names for highly preferred terminals These are anode cathode emitter base collector gate source drain bulk and substrate Metal Region Electrode Definition Example The following gives the name source to the metal or polysilicon region at location x 1 micron on the top of the current structure ELECTRODE X 1 0 NAME SOURCE Sub
497. odel 5 5 c 2 Sew ee ed las bh te See aoe eee eae BS 5 14 5 6 4 Trefonas Development Model s 2 se5chs atcosaccioesatty atsexliasan shan wpe pata wp s awe 5 14 5 6 5 Hirai s Development Modelos oacren dt sete bh ten date whet Sa ee are th ee ok io tere uy 5 14 5 7 Proximity PUM G ieee Sse a ark wee tsar ascot ware une a ra EEEa ran tang rience nnn A AAS 5 15 5 7 1 General Description of Proximity Lithography 00 cece cece eee eee eens 5 15 5 7 2 Theory of Proximity Printing sis teiare aiieaaahiies Star ee tWS seek eeu twee ea aebe east 5 15 5 7 3 Simulation Method i seriy deai eee eb eieade EA pis tele whe ewe eats eee EE 5 17 Chapter 6 StatemeniS sirin cake a nin sie awae Gans E kau Mave kava taeda ete eae care 6 1 BT OVGIVIGW 32 chore ata tertiles Richins Ba ara ea a ak enlaces ata et ea Cone euskal edie Mend erent Sra 6 1 Dold DDI OVIANONS is tans thn S Geo a Rae Se SPOS cece ae SY is So eet a E ot 6 1 6 1 2 Continuation LINCS ia S cst Sean aed ook ot wane te ata np ere Mita o e edge ierkty Sitters Sea one tanaka 6 2 Be BE COMIMGNIS en UE Se ak ee EER ee ta eRe oe ir Bo aed Se ae 6 2 6 1 4 General Syntax Description os tr eee iain ee ee ute ah ee hed ee eo ea ohh 6 2 6 1 5 Command Line Parsing 33 i evis ceed ete bawds teded eevee ee ledid bowie ede sadbewise aris ees 6 3 6 2 ATHENA Statements Listi itero eink ales peewee aoe eee eds eee 6 4 6 2 1 Structure and Grid Initialization Statements 0 2 0
498. odel specify parameter TWO DIM in the METHOD statement The Two Dimensional Model is based on the Fermi Model so read the Fermi Model description before continuing The major difference between the Fermi Model and the Two Dimensional Model is the direct representation and evolution of non equilibrium point defect populations Therefore there are three different sets of governing diffusion equations one for dopants one for point defect interstitials and one of point defect vacancies In addition you also need to take into account the 311 cluster formation and dissolution bulk and interface recombination and the generation of point defects through oxidation Each of these are described in detail in the following sections Dopants The continuity equation for dopants is Vedax 3 18 Silvaco 3 7 ATHENA User s Manual C C E Jax SP ax de Aae an Cx Cx where Cy and Cy are the actual concentrations of interstitials and vacancies and C and C y are the equilibrium concentrations The fy factor is an empirical defect factor which for interstitials is assumed to be temperature dependent through the following Arrhenius expression FLE fi FI oexp E 3 20 where the FI 0 and FI E parameters can be specified in the IMPURITY statement The value of f is clamped to a number between 0 and 1 The equivalent term for vacancies is calculated according to fy 1 f 3 21 The formulation of equation Equations 3 18 a
499. odel establishes a two way coupling between the diffusion of dopants and point defects respectively by adding the joint dopant defect pair fluxes to the flux terms of the defect equations which then become a Z Cy gt i V Jy S J R 3 54 4 F 7 z A c Ac Ae a VP i Sek Rr Rin 3 55 A c A C where summations run over all dopants and pair charge states The rest of the Fully Coupled Model Equations are identical to those in the Two Dimensional Model described in the previous section The effect of the correction terms only displays itself at very high dopant or high implant damage conditions or both where the Fermi level enhancement and point defect supersaturation will increase the dopant diffusivities significantly High Concentration Extension to the Fully Coupled Model This extension to the fully coupled model takes into account additional higher order defect dopant defect pairing the extra point defect recombination mechanisms This model was developed at Stanford University 14 to include higher order dopant defect interactions in the cases where the number of dopant defect pairs are significant This is the case for high dopant concentration in silicon It is activated by the following command METHOD HIGH CONC FULL CPL It is an extension of the basic fully coupled model and may only be used in conjunction with the METHOD FULL CPL command This model includes two extra bulk recombination reactions and two
500. of ion trajectories 1000 1 H 1000000 Relative smoothing 0 25 0 00 j 050 Comment Channel Implant WRITE Figure 2 29 ATHENA Implant Window 2 36 Silvaco Tutorial Simulating Diffusion Simulation of thermal process steps is a focal point of SSUPREM4 The hierarchy of diffusion and oxidation models is described in this chapter and in Chapter 3 SSUPREM4 Models Sections 3 1 Diffusion Models and 3 3 Oxidation Models This section will demonstrate how to set different parameters and models of diffusion oxidation and silicidation The last process will take place only if at least one refractory metal or silicide layer is present in the structure The parameters and models of a diffusion oxidation step can be prepared from the ATHENA Diffuse Menu Figure 2 30 Deckbuild ATHENA Diffuse Display Time Temp Ambient Impurities Models settings Time tem perature Time minutes 0 jH 500 Temp Temperature C 500 1300 Conctant End temperature Ch fo 5 13 Temperature rate C7 minh eE Rate Ambient Ambient Dry 02 wet O2 Nitrogen Gas Flow vf Gas pressure atm 1 00 0 00 10 00 vf HCL 2 0 10 Comment LOCOS WRITE _ Properties Figure 2 30 ATHENA Diffuse Menu To open this menu select Process Diffuse in the Deckbuild Commands menu The Diffuse menu has four sections Only the Time Temperature and Ambient fields appea
501. of the Deckbuild Window 2 2 2 Batch Mode With Deckbuild To use DECKBUILD in a non interactive or batch mode add the run parameter to the command that invokes DECKBUILD A prepared command file is required for running in batch mode We advise you to save the run time output to a file because error messages in the run time output will be lost when the batch job completes deckbuild run an lt input filename gt outfile lt output filename gt Using this command requires a local X Windows system to be running The job runs inside a DECKBUILD icon on the terminal and quits automatically when the ATHENA simulation is complete You can also run DECKBUILD using a remote display For example deckbuild run an lt input file gt outfile lt output file gt display lt hostname gt 0 0 2 2 3 No Windows Batch Mode With Deckbuild The ascii parameter is required for completely non X Windows operation of DECKBUILD For example deckbuild run ascii an lt input filename gt outfile lt output filename gt This command directs DECKBUILD to run the ATHENA simulation without any display of the DECKBUILD window or icon This is useful for remote execution without an X windows emulator or for replacing UNIX based ATHENA runs within framework programs Silvaco 2 5 ATHENA User s Manual When using batch mode use the UNIX command suffix to detach the job from the current command shell To run a remote ATHENA simulation under DEC
502. of the substrate which is dependent on the carousel rotation speed and the efficiency of the cooling system Note Ifthe process has been correctly modeled the device simulation will also be accurate if appropriate models have been chosen If a simulated device exhibits electrical characteristics that are totally inaccurate you may have done something wrong in the process simulation Do not make the mistake of changing well known default values in the simulators to make a curve fit one set of results because this will lead to poor predictive behavior Try and find the cause of a discrepancy 2 5 1 Input Information It may seem obvious but must be emphasized that an accurate process flow is vital for simulation accuracy especially for Rapid Thermal Anneals see Section 2 4 6 Simulating Rapid Thermal Anneals RTA Notes for details Other process information required is an accurate cross section of the oxide spacer Modeling the spacer profile accurately ensures the lateral damage distribution due to the subsequent source drain implants is correctly modeled Turning to electrical data the most important device electrical data is a plot of threshold voltage versus gate length for the NMOS devices Figure 2 33 shows typical plots of threshold voltage versus gate length In this figure the RTA anneal temperature and times were varied to show the various profiles that can be expected A more typical plot is represented by the 1000 C RTA pr
503. ofile showing a peak value around 1 2 microns with a tail off for longer or shorter gate lengths 2 42 Silvaco Tutorial TonyPlot V2 6 10 A m Files view Ploty Tools Print Properties v Help gt ajg EFFECT OF ANNEAL TEMPERATURE ON THRESHOLD VOLTAGE Data from VtL log T 850C 8 mins T 900C 4 mins T 950C 2 mins T 1000C 1 min T 1050C 30 secs T 1100C 15 secs 0 Gate Length um SILVACO International 1996 Figure 2 33 A plot of Threshold Voltage vs Gate Length for NMOS devices Gate oxide thickness measurements are also required Be careful here if oxide thickness is measured with capacitance voltage C V methods since quantum effects in very thin oxides less than 5nm can lead to inaccuracies because the actual location of the peak concentration of the accumulation charge is not at the interface as classic physics predicts but a short distance into the silicon Use the QUANTUM model in ATLAS to match accumulation capacitance with oxide thickness for very thin oxides Other useful electrical input information is data that won t be used now but later for the calibration process itself testing the predictive nature of the simulation Typical device characteristics used for predictive testing includes threshold voltage versus gate length measurements for a non zero substrate bias 2 5 2 Tuning Oxidation Parameters During oxidation interstitials are injected into
504. oint defects diffusing 30mm into the substrate The lateral diffusion length of point defects should be of a similar order Many parameters can be used to tune the fully coupled diffusion model The most effective for RSCE is the surface recombination of the interstitials KSURF 0 Figure C 7 shows threshold voltage versus channel length as a function of KSURF 0 for a fixed DAM FACT High values of KSURF 0 show no RSCE effect while lower values show strong increases in threshold at lengths around 1 0 micron Tuning RSCE using DAM FACT and KSURF 0 is possible using ATHENA ATLAS and VWF Users should note that both these parameters will affect process simulation results such as source drain junction depth Figure C 8 shows a graph of junction depth of an arsenic implant after a fixed diffusion as a function of DAM FACT and KSURF 0 For a given measured result for junction depth it is clear there are a whole set of DAM FACT and KSURF 0 combinations that can produce the correct answer However the effect of each combination that matches a junction depth is not the same on RSCE C 10 Silvaco Hints and Tips VWF graphics RSCE2 1dxjkS Control Print Figure C 6 Threshold voltage vs gate length for various values of implant damage VWE graphics RSCE2 ex3 Control Print Figure C 7 Threshold voltage vs channel length as a function of KSURF 0 for fixed DAM FACT VWF graphics RSCE2 1dxjkS
505. oltage roll off rate matches that of the measured data PMOS Tuning PMOS devices are a special case since the boron doped Source Drain implants overall tend to absorb interstitials rather than emit them The reverse short channel effect in buried channel PMOS devices can be caused by high angle implants If high angle implants are used the reverse short channel effect can be tuned using the LAT RATIO1 parameter in the IMPLANT statement 2 5 5 Related Issues on using the Device Simulator ATLAS for MOS Process Tuning It should now be known that calibrating an ATHENA process file involves using the device simulator ATLAS to a significant extent Hence it s imperative that the use of the device simulator doesn t create additional errors rendering the process calibration results invalid Fortunately the device physics involved in simulating the conditions required to extract a threshold voltage are not demanding The drain voltage required to extract a threshold voltage is only 50 100mV so effects such as impact ionization can be neglected The field perpendicular to the gate is also relatively low around the threshold voltage so field effects in this direction will do little effect We recommend however using at least the models SRH and CVT during the calculation Other parameters for silicon are sufficiently well known for silicon to the point that the results from the device simulator are reliable The first important point is to ensure
506. ombination of tuning parameters available in both the process simulator ATHENA and the device simulator ATLAS and with the influence of each parameter you can get a good match for bipolar transistors for most device designs Since it is usually less problematic to match the collector current for all levels of applied base emitter voltage compared to the matching of base current you will probably find that more time is spent trying to match the base current for very small and very large values of applied base emitter voltage You should however spend a good amount of time on making sure that the correct process models are used in the process flow to reduce the overall uncertainty as to which parameters require calibration Silvaco 2 53 ATHENA User s Manual 2 7 Using ATHENA for Simulating SiGe Process The recommended method for simulating SiGe process is to treat germanium as a dopant in silicon rather than depositing the SiGe material This is because it allows boron diffusivities to be germanium concentration dependant by using the model sige parameter in the METHOD statement You can also define SiGe related parameters in the MATERIAL statement The example below show a typical set of statements for depositing a 200A thick trapezoid profile SiGe base region in an HBT may be as follows METHOD FULL CPL MIN TEMP 600 MODEL SIGE MATERIAL SILICON NIFACT SIGE 100 EAFACT SIGE 1 5 NO FLIP DEPOSIT SILIC
507. oms Default is 40 MION specifies the atomic mass of the plasma ions Default is 40 QIO specifies the momentum transfer cross section Units are m Default is 1 7e 19 QCHT specifies the charge exchange cross section Units are m Default is 2 1e 19 CHILD LANG COLLISION LINEAR and CONSTANT specify a model used in calculation of the voltage drop in the plasma sheath Default is LINEAR IONS ONLY specifies that neutrals to be ignored in plasma simulation Default is false NPARTICLES specifies number of particles used for Monte Carlo calculation of the ion flux coming from plasma Default is 10 000 ENERGY DIV specifies number of energy divisions used for calculation of the plasma ion flux Default is 50 OUTE TABLE specifies the name of an output file in which complete table of simulated plasma ions and neutral distributions is saved The table cannot be loaded using ToNyPLoT The meanings of the columns in the table are e i index of energy from 0 to Nrow ENERGY DIV 1 where ENERGY DIV is the number of energy divisions specified in the RATE ETCH statement default is 50 e k index of angle from 0 to Ncol 14 The interval 0 90 degrees is divided into 15 intervals and each of them are divided into 4 sub intervals each of 1 5 degrees wide e Cergy lt n gt number of ions in each sub intervals n 0 3 e Cergyn lt n gt number of neutrals in each sub intervals n 0 3
508. on North Holland New York 1981 F Lau et al A Model for Phosphorus Segregation at the Silicon Silicon Dioxide Interface Appl Phys A v 49 p 671 1989 Y S Oh D Ward A Calibrated Model for Trapping of Implanted Dopants at Material Interface During Annealing IEDM Tech Digest p 509 1998 P B Griffin and J D Plummer Process Physics Determining 2 D Impurity Profiles in VLSI Devices IEDM Tech Digest p 522 1986 S M Hu On Interstitial and Vacancy Concentration in Presence of Injection J Appl Phys v 57 p 1069 1985 S Crowder Processing Physics in SOI Material Ph D Thesis Department of Electrical Engineering Stanford University 1995 B J Mulvaney W B Richardson and T L Crandle PEPPER A Process Simulator for VLSI IEEE Trans on Computer Aided Design v 8 p 336 1989 L Mei M River Y Kwart and R W Dutton Grain Growth Mechanism in Polysilicon Proc 4th Intern Symp on Silicon Materials and Technology v 81 p 1007 1981 L Mei and R W Dutton A Process Simulation Model For Multilayer Structures Involving Polycrystalline Silicon IEEE Trans Electron Devices v ED 29 p 1726 1982 F Boucard Dopant diusion modelling in silicon for shallow junctions processing Ph d thesis Louis Pasteur University September 2003 C Ortiz and D Mathiot A new kinetic model for the nucleation and growth of selfinterstitialclusters in silicon Mat Res Soc S
509. on This model introduces an additional sink for interstitials in the layers with high carbon concentration Additional parameter to control diffusion of interstitial in SiGeC region DCARBON E is added to the INTERSTITIAL statement It allows to decrease interstitial diffusivity in SiGeC and indirectly supresses transient boron diffusion in this region C Interpreter functions are now available for Boron diffusion model in SiGeC It is now possible to include non equilibrium interstitials into epitaxially grown or simply deposited silicon layer Handling of impurity activation models has been improved Now the type of activation model can be specified for each impurity material combination in the IMPURITY statement The SOL SOLUB and CLUSTER ACT parameters haven been added to the IMPURITY statement The TWO DIM and FULL CPL models can be used for all semiconductor materials There are no verified default parameters for vacancies interestitials and traps in materials other than Si but user can specify those parameters for any semiconductor material POLYDIFF model is completely rewritten The following new names for the model parameters are specified in the IMPURITY statement PD DIX 0 PD DIX E PD EFACT PD SEG E PD TAU PD SEGSITES PD SEG GBSI PD CRATIO PD GROWTH 0 and PD GROWTH E Use help impurity in the ATHENA command line to find a short description of these parameters Implant Simu
510. on Lines Since it may be necessary for a statement line to contain more than 256 characters ATHENA allows you to specify continuation lines If a statement line ends with a backslash the next line will be interpreted as a continuation of the previous line 6 1 3 Comments Comments are indicated by the COMMENT statement or a number sign All characters on a line which follow a comment indicator COMMENT or will not be processed by ATHENA The comment symbol is not supported anymore The should be avoided for use as a character in strings since it is used as part of shell capabilities included in DECKBUILD 6 1 4 General Syntax Description An ATHENA statement is a sequence of words starting with a statement name and followed by some or all of the statement s parameters This manual describes the syntax for each statement in the following way STATEMENT NAME DESCRIPTION OF PARAMETER 1 DESCRIPTION OF PARAMETER 2 Parameters are described in the following form PARAM lt n gt a real valued parameter PARAM lt c gt a string valued parameter PARAM a Boolean parameter Boolean parameters are those that recognize the Boolean values TRUE and FALSE as valid values In ATHENA Boolean parameter values are automatically set to true if the name of the Boolean parameter appears by itself in a statement A Boolean parameter can be set to false using the syn
511. on Models In the previous section an introduction to one dimensional oxidation modeling was presented This section describes the two dimensional numerical oxidation models implemented in SSUPREM4 The numerical oxidation models build on the Deal Grove oxidation theory and provide the capability to simulate arbitrary two dimensional structures The numerical oxidation models require solving the oxidant diffusion equation at incremental time steps at discrete grid points in the growing SiOz layer The oxidant diffusion equation is given by H V F 3 138 where e Cis the oxidant concentration in SiO e tis the oxidation time e F is the oxidant flux Equation 3 138 is solved by substituting Equation 3 129 for F and defining appropriate boundary conditions at material interfaces with SiO At the gas SiO interface Equation 3 127 describes the interface transport flux of oxidant molecules accounting for the boundary condition at that interface The boundary condition at the Silicon or Polysilicon SiO interface is described by Equation 3 130 The flux at boundaries between SiO and other materials in the simulation structure is set to zero By solving Equation 3 138 the oxidant concentration is determined at each grid point in the SiO layer The SiO growth rate or Si SiO interface velocity V is determined at each point along the interface by combining Equations 3 130 and 3 132 resulting in the following kCun vy U 3 139
512. on Terms Development Models CNET aiid ies denon en eee hae ie 3 29 30 Dille as Ses eat ne nee eed aed a 5 13 Grain based Polysilicon Diffusion Models s ssesseo 3 21 22 PEA ca Kes nahi Nl Gell anh ha 5 14 TWO SUCAM sees eeseeseeseeeeseesteneesestesensenneeneseeseeneenennenen 3 21 KIN hy ndi Pinks Aas Mere aA Sas ies 5 13 14 Grid Control ai iente eaten eaaa tA AM ee 2 19 6 68 MACK i ster oar E E dee dedicnea satis tes edeb th toed ne oben derek 5 14 Gulimmel Plots ieena hts becca te tae Bd Ate idle 2 48 WWOTOMAS APET E EAE 5 14 Development MOdUules eessceesseeeeesnteeeneeeeeneeeeees 5 13 14 H Devige eee tka Ns CAE TAR AW ae alien a eae Helmholtz equation i te ae a t e eaa 5 9 Dirtusion Equation intao ies 3 3 4 Huygens diffraction approximation z era Aniak 5 15 Diffusion Models ccccccccccccsseseseceeeeeeeeeeaeeeees 3 1 22 3 92 Electrical Deactivation and Clustering Models 3 18 20 l P E 3 5 6 Fully Coupled Model s senesnensenenserernsrnrnernererrernerene 3 16 18 Imaging Mod le snittet cane ceiatectin tea sanue 5 2 6 Impurity Segregation ssssscsssecssssseeeneeeesneeesneeeennees 3 6 7 HPU 2222 ieee crass eased tein don a es 3 1 Sr A EA N TN gt Gy A Impurity Segregation MOEl sccccsecseeecsetecseeseteeeenees 3 6 7 l ATAA AA SE I S a eo ey E A Interiace Traprain eean aeeoea DE eana 3 7 Dislocation Loop Based Enhanced Bulk Recombination 3 1 Initial Stru
513. on consists of the three stages These stages are as follows 1 Calculation of ion neutral and polymer fluxes 2 Calculation of etch polymer ejection and redeposition rates 3 Surface movement On the first stage the fluxes of incoming and reflected ions and neutrals are calculated on the each segment of the surface Computation of the ion fluxes is done by tracing the user defined number of particles Figure 4 11 model a Each particle is generated at random positions on top of the simulation area with normal and lateral velocities randomly determined from the bimaxwell distribution function Equation 4 1 Then each particle trajectory is traced until the ion is either absorbed by the surface or back scattered out of the simulation area Silvaco 4 15 ATHENA User s Manual simulation limit incoming ion Q reflected id rand b C Figure 4 11 Diagram of Plasma Flux algorithm a including zoom in of ion reflection models a amp b The interaction of the ion with material surface is governed by two factors The first is the reflection coefficient P which is specified by the Mc ALB1 and Mc ALB2 parameters for two types of plasma particles The second is Mc PLM ALB for polymer particles and roughness of the surface R which is specified by the Mc RFLCTDIF parameter Both factors depend on the surface material and the type of ion Reflection coefficient is the probability of the particle to be reflected from the su
514. onding high temperature numbers L PDEP is the exponent of the pressure dependence The value given is taken to apply to lt 111 gt orientation and later adjusted by ORI FAC according to the substrate orientation present 6 74 Silvaco OXIDE PAR L O PAR L E PAR H 0 PAR H E PBREAK and P PDEP specifies the parabolic rate coefficients B ORI FAC is the ratio of B A on the specified orientation to the orientation ORI DEP specifies whether the local orientation at each point on the surface should be used to calculate B A The default is true If it is false the substrate orientation is used at all points THINOX 0 THINOX E and THINOX L specifies coefficients for the thin oxide model proposed by Massoud 14 THINOX 0 is the pre exponential factor in microns min THINOX E is the activation energy in eV and THINOX Lis the characteristic length in microns THINOX P is the thin oxide model pressure dependence HCL PC HCLT HCLP HCL PAR and HCL LIN is where the numerical parameter HCL PC is the percentage of HCl in the gas stream It defaults to 0 The HCl dependence of the linear and parabolic coefficients is obtained from a look up table specified in the model file The table rows are indexed by HCl percentage Specify the row entries with the parameter HCLP which is an array of numerical values surrounded by double quotes and separated by spaces or commas The columns are indexed by temperature Spec
515. one oxidant e g O and H30 the partial pressure of each oxidant is used to calculate C for each species From C k and Deg for each oxidant species are calculated in a similar manner as that described in the pairs sections respectively Equation 3 138 is solved for each oxidant to obtain each oxidant s concentration distribution in the growing SiO The contributions of each oxidizing species to the Si SiO interface velocity is calculated with the following equation am kj Cima ij 3 170 Ni where ANT 3 139 has been used and j corresponds to the j oxidant gas The flow equations are also calculated for a mixed ambient where both O gt and H3O exist and COMPRESS or VISCOUS has been specified on the METHOD statement The stress dependence of Dey and k is a function of the composition of dry or wet oxide which depends on oxidation history Mixed ambient oxidation simulations take longer to solve than simple ambient equations 3 3 7 Analytical Oxidation Model You can use the analytical oxidation models to simulate a limited set of simple structures Possible structures include a silicon substrate with an oxide layer deposited or grown on it Since you can only specify the mask at the left part of a simulated structure oxidation will only occur to the right of the mask edge Analytical methods do not account for any real material layer located to the right of the specified mask edge As the oxide layer thickens the materi
516. only Pearson IV distribution set the ANY PEARSON parameter to FALSE In all cases when B and y do not satisfy one of the mentioned criteria ATHENA will automatically increase B up to the value that satisfies the criterion used In the standard Pearson model the longitudinal dopant concentration is proportional to the ion dose 9 C x Of x 3 191 This single Pearson approach method has been proved to give an adequate solution for many ion substrate energy dose combinations But there are many cases when the channeling effects make the Single Pearson Method inadequate Dual Pearson Model To extend applicability of the analytical approach toward profiles heavily affected by channeling Al Tasch 48 suggests the dual or Double Pearson Method With this method the implant concentration is calculated as a linear combination of two Pearson functions Cx f E pf 3 192 where the dose is represented by each Pearson function fj 9 x f1 x and fo x are both normalized each with its own set of moments The first Pearson function represents the random scattering part around the peak of the profile and the second function represents the channeling tail region Equation 3 191 can be restated as C x PRA U Pf x 3 193 where is the total implantation dose and I 2 I To use dual Pearson distribution supply nine parameters four moments for each Pearson function with the dose rati
517. ons exceed the limits of 20000 points or 1000 horizontal or vertical points the program gives an error message and exits e Non integer specification of the DIVISIONS parameter on the DEPOSIT statement is now allowed This allows parameterized gridding e The INITIALIZE statement now accepts material specifications This allows the specification of an initial grid for any material using only LINE and INITIALIZE statements TAG parameters for boundary definition do not need to be specified REGION and BOUNDARY statements are not needed and for most commonly used boundary conditions are set up by default e Improved grid refinements following oxidation deposition silicidation etching or other grid mov ing steps Silvaco D 23 ATHENA User s Manual e This update includes a new parser function MAT1 MAT2 Y that will return the x intersection point between materials mat1 and matz2 for the y value given to the function The other parser func tion MATI MAT2 X returns a y intersection point given x However the two functions are very different The former allows the intersection point with gas to be found specifically for the applica tion of extracting critical dimensions CDs for photolithography applications The latter will not handle gas material In the case of extraction of cds a special format is used PRINTF GAS PHOTO Y PHOTO GAS Y This is the right intersection the left intersection If there are more than two inters
518. onte Carlo models are removed 4 Improved Kinchin Pease Model for interstitials and vacancies generated during BCA implant is implemented 5 AMORPHOUS parameter is now applicable to the BCA simulation D 8 Silvaco ATHENA Version History D 7 2 FLASH Capabilities FLASH diffusion model now completely corresponds to FERMI Model for silicon Generic diffusivity formulae are now used for all available dopants instead of specific terms for each dopant The DIPP 0O and DIPP E parameters for diffusivity with doubly positive defects are added to the IMPURITY statement The electric field effect on diffusion in GaAs materials is fixed Now the type of impurity donor acceptor specified in the IMPURITY statement is properly taken into account Germanium is set as n type dopant in GaAs and all appropriate parameters are added Equilibrium interstitial and vacancy concentrations in compound semiconductors are now available in the structure file and TONYPLOT D 7 3 OPTOLITH Capabilities New model for simulation of resist exposure process is implemented instead of old one which used ray tracing algorithm It is based on the Beam Propagation Method see Chapter 5 OPTOLITH Models Section 5 4 The Exposure Module The main advantage of the new method is its capability to take into account dose dependency of the local optical properties refraction index of the photoresist We ve added a new model which allows you to simu
519. opography effects such as reflow in ELITE can be combined with in wafer simulation of dopant diffusion or oxidation in SSUPREM4 A previous Hints and Tips column April 1995 showed how ATHENA can simulate individual process steps from SSUPREM4 and ELITE with seamless integration In this case the ELITE and SSUPREM4 simulation is done on the same process step The reflow heat cycle will also trigger diffusion of the dopants in the silicon including transient enhanced diffusion effects where appropriate A single DIFFUSE statement with the REFLOW parameter can both produce reflow and dopant diffusion Figure C 4 shows an example of a 0 5mm contact cut to an arsenic diffusion During the reflow cycle at 875 C the edges of the contact cut are flowed while the arsenic is diffusing TonyPlot 2 4 1 File 7 View Plot Tools Print Properties gt Help F BEFORE REFLOW HEAT CYCLE 1 z Net Doping cm3 0 8 21 20 4 19 06 S 18 1 z 16 2 504 A 02 oo T o2 z se 04 ss i a EES ESN Te POPAS IETS lee al o 04 08 anpa 16 2 24 AFTER REFLOW HEAT CYCLE 1 Net Doping em3 08 21 19 0 6 18 I 17 16 2 0 4 2 A I 62 o2 als fee Stee oa 1 r a WSA ARE E T TO Se et et 0 04 08 aa 16 2 24 Loading file tmp_mnt main striker andys dev cm
520. or parameters that have a significant affect on the base current in the medium injection regime These parameters are the Poly emitter Work Function and the Bandgap Narrowing Effect These parameters are described below Poly emitter work function If the poly emitter is described as N POLYSILICON in the CONTACT statement for an NPN device as already described the Poly emitter Work Function is then set to 4 17V and is correct for saturation doped n polysilicon But if the poly emitter is not saturation doped the work function will differ from this ideal and have a pronounced affect on the base current and current gain in the medium injection regime as shown in Figure 2 38 The work function of the poly gate can vary from 4 17V for n poly silicon to 4 17V E for p polysilicon depending on the position of the Fermi Energy Changing the work function of the poly emitter by just 0 1V from 4 17V to 4 27V can often reduce the current gain in half in the medium injection regime so it s very important to assign the correct value The CONTACT statement below assigns a work function of 4 27eV to the poly emitter while keeping the other parameters the same as before CONTACT NAME emitter SURF REC VSURFP 1 5e5 WORKFUN 4 27 The poly emitter work function can be calculated by measuring the position of the Fermi Energy at the poly silicon silicon interface relative to the conduction band and adding this value to 4 17V For example if the Fer
521. or three dimensional targets which is partly possible because the Monte Carlo method treats an explicit sequence of collisions so the target composition can change on arbitrary boundaries in space and time The rest of the distribution is built up from a vast number of ion trajectories and the statistical precision of which depends directly on this number JN As the ion penetrates a solid it undergoes a sequence of collisions with the target atoms until it comes to rest A simplified model of this interactions is a sequence of instantaneous binary nuclear collisions separated by straight line segments free flight path lengths over which the ion experiences continuous non local electronic energy loss The collisions are separated i e the state of an ion after a collision depends solely on the state of the ion before the collision 3 84 Silvaco SSUPREM4 Models The model assumes that the arrangement of the target atoms is totally randomized after each collision i e the target has no structure and no memory As a result a sequence of collisions is described by randomly selecting the location of the next collision partner relative to the pre flight location and velocity direction of the ion This means that this model cannot simulate the anomalous tail penetration observed for implanted ions into aligned single crystal targets The model adequately describes the ion penetration into multilayer non planar structures Crystalli
522. orks for BCA simulation Improved damage model and electronic stopping Now the value of Implant Damage is in atomic density per cm Improved BCA model for indium and germanium implants Silvaco D 9 ATHENA User s Manual D 8 2 Miscellaneous Features and Bug Fixes 1 aoa PR WN The memory problem that use to result in failure during multiple implant steps with FULLROTATION has been fixed Several problems related to switching from 1D to 2D simulation have been fixed Missing donor acceptor concentrations after BCA implant have been added 311 cluster distribution after BCA implant has been added Wrong oxide thickness after several consequent viscous oxidation steps has been fixed Reading of some DEVEDIT structures into ATHENA have been fixed The license for SILICIDE material model is now checked only when silicidation process starts This allows to have structures with deposited silicide materials without having the license D 9 ATHENA Version 4 5 0 R Release Notes D 9 1 SSUPREM4 Implant Simulation Features 1 New Binary Collision Approximation Module for Monte Carlo type simulation of ion implantation in amorphous and crystalline materials is implemented The parameter BCA is used to turn on this model BCA and MONTE are mutually exclusive This module is much more accurate than previous Monte Carlo implementations It is able to accurately calculate implant profiles in difficult cases of well channel
523. osition parameters and the machine name for one of ten deposition models available in ELITE MACHINE specifies the machine name for the RATE DEPO statement MATERIAL specifies material to be deposited by the deposit machine see Section 6 2 9 Standard and User Defined Materials for the list of materials NAME RESIST specifies the name of photoresist to be deposited CONICAL CVD PLANETARY UNIDIRECT DUALDIRECT HEMISPHERIC MONTE1 MONTE2 CUSTOM1 and CUSTOM2 specify a particular model for the machine definition DEP RATE specifies the deposition rate used by the models CONICAL CVD UNIDIREC DUALDIREC HEMISPHE PLANETAR MONTE1 and MONTE2 DEP RATE is a rate multiplier for the CUSTOM1 and CUSTOM2 models INFILE specifies the name of a file containing angle and deposition rate information for the CUSTOM model A H A M A S U H U M U S and N M specify that the deposition rate DEP RATE is in Angstroms per hour Angstroms per minute Angstroms per second microns per hour microns per minute microns per second and nanometers per minute respectively Default is A S STEP COV specifies the step coverage used by the model CVD ANGLE1 specifies the angle parameter used by the models HEMISPHE DUALDIREC and PLANETAR CONICAL UNIDIREC ANGLE2Z specifies the angle parameter used by the models DUALDIREC PLANETAR and HEMISPHE
524. ow of the material is to be performed during the diffusion step Predeposition Example The following statement specifies a 1000 30 minute boron pre deposition DIFFUSE TIME 30 TEMP 1000 C BORON 1 0E20 Oxidation Example The following statement instructs the simulator to grow oxide for 30 minutes in a dry oxygen ambient DIFFUSE TIME 30 TEMP 1000 DRYO2 Gas Flow Example The following command performs diffusion with a mixed ambient with relative components of oxygen hydrogen and HC1 of 10 10 and 0 1 respectively DIFFUSE TIME 10 TEMP 1000 F 02 10 F H2 10 F HC1 1 Hydrogen and Oxygen are combined in a ratio 2 1 to form the ambient WETO2 Any excess hydrogen is considered inert Any excess oxygen is considered as the ambient DRYO2 Since the total pressure of the gas flow is defined or defaults to one atmosphere the partial pressure of WETO2 will be reduced if any excess hydrogen or oxygen is present File Output Example The following commands perform diffusion in dry oxygen ambient for 30 minutes at 1000 C After every second timestep a structure file is written with a name prefix TEST Following the diffusion the TONYPLOT statement plots each timestep output file in a manner suitable for creating a diffusion movie 6 26 Silvaco DIFFUSE A SYSTEM command is used to execute a UNIX command prior to the diffusion step to remove all TEST str files from previous run
525. own in Figure 3 18 TonyPlot V2 6 9 E p vy View v Plots Tools gt Print Properties gt Help 7 _ gt X Default Oo S improved i f 5 z E 5 Depth Microns SILVACO International 1996 Figure 3 18 Comparison of Arsenic Profiles in Silicon with Default Grid Spacing and Improved Grid Spacing in the Growing SiO layer Oxidation Enhanced Diffusion OED Oxidation Retarded Diffusion ORD During silicon thermal oxidation some of the dopant in silicon gets incorporated into the growing SiO layer and some remains in silicon where it diffuses As oxidation proceeds silicon lattice atoms become interstitial interstitials are injected into silicon at the Si SiO interface as oxygen molecules are incorporated into the lattice to form SiO Due to the injection of interstitial defects during oxidation you can enhance dopant diffusivities To properly simulate this effect you must include the creation and movement of point defects vacancies and interstitials in the simulation By specifying TWO DIMin the METHOD statement before the oxidation step non equilibrium point defect concentrations including injection and recombination at the Si SiO interface are included in the simulation For more information on point defect diffusion kinetics see Section 3 1 Diffusion Models Silvaco 3 61 ATHENA User s Manual Note Figure 3 19 compares the boron concentration profiles after an oxidat
526. ows F k AG Vo exp ik x dx 5 25 After propagating over a small step Ay each component of F4 obtains additional phase shift 2 2 corresponding to the value of k yee yk k Thus the amplitude distribution at new y without accounting for absorption can be written as follows 5 26 A x Yo Ay Fu exp ik Ay exp ik x dk Since actual material optical properties differ from the properties of vacuum the field at the new plane is computed simply as follows E x y Ay A x y exp ik n x x 1 Ay T The algorithm is repeated recursively step by step over all simulation domains The same calculations are then applied to reflections from all interface segments The current intensity distribution is calculated from the field distribution as 2 I x y E x y 5 28 During the exposure the resist structure is modified So the dissolution inhibitor is converted to the photo reaction product The initial normalized concentration of photoactive compound PAC is defined by local intensity magnitude as Mpac exp Cl x y 5 29 where I x y is the current intensity distribution and C is Dill s C parameter Accordingly the optical properties of the resist complex refraction index n which includes both refractive and absorption indices are modified too The capability to take into account is the effect of dose intensity on the refraction index n which is implemented into the module and the fo
527. p history1 1 str OK SILVACO International 1995 Figure C 4 Simulation of simultaneous dopant diffusion and glass reflow in ATHENA Question How can dielectric reflow be modeled Which calibration parameters are important for tuning the reflow C 8 Silvaco Hints and Tips Answer ATHENA contains a model for the reflow of materials as part of the ELITE module The model treats the dielectric material i e SiO2 BPSG as an incompressible viscous fluid The material is then deformed under the driving force of the surface tension of the topography The calculation of the changing topography of the material then proceeds according to the applied time and temperature The reflow model for a given material is enabled by setting the REFLOW parameter on a MATERIAL statement In addition the parameter REFLOW should be given on a DIFFUSE statements corresponding to the flow heat cycle The following is typical syntax MATERIAL OXIDE VISC 0 1 862E 20 GAMMA REFLO 1E3 REFLOW DIFF TIME time TEMP temp REFLOW This example syntax also includes two of the most useful tuning parameters VISC 0 sets the viscosity of the oxide GAMMA REFLO sets the surface tension factor for the flow calculation Figure C 5 shows the results of an example of reflow calculation with ATHENA The initial structure has a set of 1 micron contacts with a 2 micron pitch after the anisotropic contact etch T
528. pecified in the model file The analytic implant capabilities for these materials are as follows SiGe uses Si moments tables where they are available AlGaAs InGaAs and InP use moments tables for GaAs where they are available The Monte Carlo implant capabilities are as follows SiGe uses the Si crystal lattice AlGaAs and InGaAs use the GaAs crystal lattice InP uses its own crystal lattice Carbon has been added as a dopant for GaAs with diffusion coefficients and implant tables borrowed from Beryllium until better data is found D 13 4 OPTOLITH Capabilities Problem with annular sources for exposure has been fixed D 13 5 Known Bugs GPLOT visualization plots do not work when remotely displaying on Solaris 2 4 D 14 ATHENA Version 2 0 Version 2 0 of ATHENA incorporates a number of new models as well as convenience features The FLASH module is now available as a component of ATHENA ATHENA now includes a Monte Carlo based deposit algorithm and a reflow calculation D 14 1 ATHENA Capabilities ATHENA Framework capabilities have been enhanced by the inclusion of some helpful geometric manipulations Namely e The STRETCH statement has been extended to allow vertical stretches to easily extend structures for device analysis or point defect based diffusion calculations The parameter Y VAL on the STRETCH statement specifies the vertical position in the structure at which the stretch will occur e The ETCH statement has been exten
529. pond to a mask name contained in the layout file invoked using DECKBUILD The mask names are case sensitive and cannot be abbreviated REVERSE specifies that the mask polarity should be reversed or that negative type photoresist should be modeled DELTA specifies an offset in mask size The offset corresponds to a change in CD critical dimension of the mask Each edge of the mask is moved by a distance DELTA to enlarge or contract the mask feature Examples The following statement deposits photoresist on the top of the simulation structure and etches it with the pattern prescribed by the MaskViEws layout The layout file must be specified using the MASKVIEWS interface as described in the VWF INTERACTIVE TOOLS USER S MANUAL VOL II or in Chapter 2 Tutorial MASK NAME CONT For more examples see STRIP Silvaco 6 61 MATERIAL ATHENA User s Manual 6 35 MATERIAL MATERIAL sets the coefficients for materials Syntax MATERIAL MATERIAL NI 0 lt n gt NI E lt n gt NI POW lt n gt EPS lt n gt E FIELD VISC 0 lt n gt VISC E lt n gt VISC X lt n gt WETO2 DRYO2 YOUNG M lt n gt POISS R lt n gt LCTE lt c gt INTRIN SIG lt n gt OXIDIZABLE DENSITY lt n gt COMPONENTS AT NUM 1 lt n gt AT NUM 2 lt n gt AT NUM 3 lt n gt AT NUM 4 lt n gt AT MASS 1 lt n gt AT MASS 2 lt n gt AT MASS 3 lt n gt AT MASS 4 lt n gt ABUND 1
530. projection pupil that is opaque over a square annular region PUPIL FILTER SQUARE PUPIL FILTER IN RADIUS 1 OUT RADIUS 2 PHASE 0 TRANSMIT 0 For more examples see IMAGE ILLUMINATION PROJECTION ILLUM FILTER LAYOUT and ABERRATION 6 84 Silvaco QUIT 6 47 QUIT QUIT terminates execution of ATHENA The EX QUIT statement Syntax QUIT Description IT STOP and BYE a statements are synonyms of the All statements after a QUIT statement will not be checked or executed Silvaco RATE DEPO ATHENA User s Manual 6 48 RATE DEPO RATE DEPO specifies the deposit rates of a machine which is used in a subsequent DEPOSIT statement Syntax RATE DEPO MACHINE lt c gt MATERIAL NAME RESIST lt c gt CONICAL CVD PLANETAR UNIDIRECT DUALDIRECT HEMISPHERIC MONTE1 MONTE2 CUSTOM1 CUSTOM2 DEP RATE lt n gt INFILE lt c gt A H A MJA S U S U M U H IN M STEP COV lt n gt ANGLE1 lt n gt ANGLE2 lt n gt ANGLE3 lt n gt C AXIS lt n gt P AXIS lt n gt DIST PL lt n gt SIGMA DEP lt n gt SIGMA 0 SIGMA E SMOOTH WIN lt n gt SMOOTH STEP lt n gt MCSEED lt n gt STICK COEF lt n gt pan Description This statement is used to define dep
531. ques A common syntax and user interface however allow you to switch easily from one module to another Some specifics of the proximity lithography and a brief theory used in the proximity printing module are described below 5 7 1 General Description of Proximity Lithography Proximity lithography is used to print images of masks without expensive projection systems It has internal resolution limits but still applicable for relatively large objects Figure 5 3 shows the optical scheme of proximity lithography Figure 5 3 Scheme of proximity optical system Light illuminating the mask creates a diffraction image in resist film placed on some distance gap from the mask Due to diffraction effects in the gap the image in the resist film is a distorted mask image Distortion depends on the gap size and on the radiation wavelength and size of printed features The reliability and longevity of the mask also depends on the gap Therefore identification of the optimal printing conditions is a very important technology design task 5 7 2 Theory of Proximity Printing A formal description of diffraction on a gap can be found in any classical book on physical optics The Fresnel diffraction is applicable here Usually the Fresnel diffraction integral is obtained by applying Huygens diffraction approximation It can be derived from the first principles by applying specific limitation to common wave equation The goal is to obtain distribut
532. r and an object oriented hierarchical representation of the impurity and defect transport models The ILFEM module solves impurity and defect transport equations much faster than previous SSUPREM4 solvers It also has better convergency The following diffusion models are currently implemented within the ILFEM module FERMI TWO DIM FULL CPL 311 CLUSTERS and HIGH CONC It also handles all corresponding boundary conditions including impurity segregation defect generation and recombination models D 10 Silvaco ATHENA Version History To activate ILFEM use METHOD ILFE To disable ILFEM use METHOD ILFEM f The ILFEM module is currently applicable to the following e impurities boron phosphorus arsenic antimony and indium e materials silicon polysilicon oxide nitride and aluminum D 9 2 ELITE Capabilities A new Monte Carlo Etch Module is implemented The main application of this module is the simulation of plasma or ion assisted etching The module can take into acccount the redeposition of the polymer material generated as a mixture of incoming ions with etched sputtered atoms and molecules of substrate material C Interpreter can be used for introduction of user defined etch and ejection rate models D 9 3 Generic ATHENA Capabilities Active concentration calculations are improved Previously all existing impurities in the structure were set to completely active after
533. r bisector of adjacent segments For A the process is anisotropic yielding vertical sidewalls see Figure 4 9 Figure 4 10 illustrates the regions of significance for each component in the RIE model The shadowing effect is accounted for by the riso component in the shadowed area Incident Flux Initial line of action r gir COS Q Advance due to fair r dir Figure 4 9 Point Advance due to Directional Influence Incident Flux Mask Mask l Tiso A A r Tigo r Figo Fir COS Q p liso Tar Figure 4 10 Regions of Significance of rgir and risc Silvaco 4 13 ATHENA User s Manual 4 4 3 Dopant Enhanced Etching Dopant enhanced etching is a feature included in ELITE and allows the etch rate at any point on the surface to be changed depending on the value of any solution variable present The etch rate at any point is then given by the formula ER 1 enh ERM 4 14 where ER is the enhancement or retardation due the presence of particular dopant All impurities as well as interstitials vacancies and stress solutions S Szy and Syy can be specified in the model This enhancement is calculated using the formula enh 0 5enH max tanh ENH SCALE S ENH MINC 1 4 15 where ENH MAX is the maximum value of enhancement or retardation ENH MINC gives the solution value below which enhancement decays and ENH S
534. r called POW FACTOR has been added as the power of the energy ratio energy ratio 1000 current ion energy linitial ion energy of the ion The default value of POW FACTOR 0 5 or is the square root of the energy ratio These parameters apply to both the CRYSTAL and AMORPH implants The Hobler electronic stopping model and its parameters were originally for Boron in Si crystal implants The Hobler model is used by default for Boron in Silicon It can also be used for Si with any impurity by specifying HOBLER on the IMPLANT statement The Hobler parameters and their default values are PMAX HOBLER 2 35 XNL HOBLER 0 4 and FHOBLER 0 8 PRE FAC TOR can also be used with the HOBLER model C Interpreter Capabilities The C Interpreter has been integrated into ATHENA The first models accessible by the C Interpreter are for the phosphorus arsenic antimony boron interstitial and vacancy diffusion coefficients The latter two are only applicable for the advanced diffusion models The file name for model substitution is set on the DIFFUSE statement with the string parameter P DIF COEF lt filename gt This syntax is valid for all of the above with the string parameters being P DIF COEF AS DIF COEF SB DIF COEF B DIF COEF I DIF COEF and V DIF COEF for phosphorus arsenic antimony boron interstitial and vacancy diffusion coefficients respectively The segregation calculation can also be accessed by the C Interpreter for phosphorus arsenic
535. r initially The Impurities and Models fields appear only when the corresponding check boxes are selected The minimum set of diffusion step parameters is as follows e Time e g 60 minutes e Temperature e g 1100 Celsius e Gas pressure 1 atmosphere is default The following input file statements will appear DRIVE IN DIFFUSE TIME 60 TEMP 1100 NITRO PRESS 1 00 Silvaco 2 37 ATHENA User s Manual If you choose the Ramped box and End Temperature or Temperature rate a ramped temperature thermal step is simulated The temperature rate is a variable by default but it can be set to a specific constant temperature rate by selecting Constant in the Rate box If the End Temperature is set to 1000 the following lines appear RAMPING DOWN DIFFUSE TIME 60 TEMP 1100 T FINAL 1000 NITRO PRESS 1 00 The same pull down menu used for inert diffusions is also used for oxidations described in the Simulating Oxidation Section on page 2 39 But since there are special considerations for inert diffusions which come under the category of Rapid Thermal Anneals RTA the special notes pertaining to this specific set of conditions are described in the next section These notes are very important for accurate simulation of high temperature short duration anneals We recommend that you read these notes before attempting to write the RTA section of the input file 2 4 6 Simulating Rapid Thermal Anneals R
536. rabolic approximation for depth dependent f will be used if the FULL LAT parameter is used in the IMPLANT statement and when mixed spatial moments SKEWXY parameter and Yay xy KURTXY parameter are non zeros In the case of the Dual Pearson longitudinal function the mixed spatial moments for the second Pearson SSKEWXY and SKURTXY can be also specified The values of spatial moments are not yet included in the default moments tables and should be specified in the MOMENTS statement see the Specification of Implant Parameters in the Moments Statement on page 3 76 Non Gaussian Lateral Distribution Functions Detailed Monte Carlo simulations 55 and 56 also show that in most cases transversal distribution function f is not Gaussian In other words the transversal kurtosis By is calculated as o0 P fina y y dxdy 3 217 0 and is not always equal to 3 0 and also depends on depth Several non Gaussian transversal distribution functions were examined in 46 Their conclusions were as follows The symmetrical Pearson functions type II for S3 and type VII when B gt 3 are acceptable providing an agreement with amorphous Monte Carlo simulations and have computational advantage because they can be integrated over x in a closed form through incomplete beta functions 57 Another good alternative for transversal distribution function is the Modified Gaussian Function MGF suggested in 55 It is shown in 57 t
537. rare event algorithm uses the integrated dose as a criterion when to split 73 ATHENA uses the same criterion to determine the splitting depths Dose integration is carried out along the radius vectors of ions co ordinates thus roughly taking into consideration the three dimensionality of the ion distribution Threshold s tates depths Initial impact Figure 3 25 Restarting Collision Events by splitting at m thresholds Due to the discrete nature of collision cascades the number of sub trajectories created at each split depth should be an integer number greater or equal to two Suppose T is the event at each threshold state i e this is the event of ion passing through a split depth Suppose also the probability of an ion being in state T _ to reach the state T is p P T T _ Then the recommended number of replications at each threshold a split depth is R 1 P This relation gives the link between the number of replications at each split and the criterion to identify the threshold states i e the split depths If R 2 then the number of ions passing through split depth will be twice smaller the number of particles passing through split depth In ATHENA the criterion to determine the split depths is the integrated dose along the radius vectors of stopped particles d i e split depths d7 dy d3 and so on will be at doses 0 50 0 75 0 8750 and so on where is the total retained implant dose
538. re Using ATHENA 00 cece eee eee eee eee eee eee etn e en neee 2 7 2 3 1 Procedure Overview iii vecse inet avenladlsw eae wbebeekebewiywce ages teks edioweeisd 2 7 23 2 ATHENA INOUUOUIDUL s ieren saarna ce sae ents tense tae or eee ee ete hte E E E RE 2 7 2 3 3 Creating An Initial Structure nnan nn ede geks pete eno wela ee Wa awie ete yeesy 2 8 2 4 Choosing Models In SSUPREM4 0c cece cece eee ene 2 30 2 4 1 Implantation Oxidation RTA Diffusion and Epitaxy 00 cece eect eens 2 30 2 4 2 The Reason for Multiple Models for Each ProceSS 1 0 c eee eect tenes 2 30 2 4 3 Choosing an Appropriate Model Using the Method Statement 000 cece eens 2 30 2 4 4 Changing the Method Statement During the Process Flow ccc eee eee eee eee ee 2 31 2 4 5 Modelling the Correct Substrate Depth 0 ccc eee eee 2 32 2 4 6 Simulating Rapid Thermal Anneals RTA Notes 0 cece eee e eet eee eee 2 38 2 4 7 Simulating Oxidation n s ehaince ts cee arn ax yaks Beene aus A estan ne Ue conte tap aehittens ak St 2 39 2 4 8 Simulating the Epitaxy Process sc vi45 Seeds phe eendy iia weeks Peli Pe eee es 2 40 2 5 Calibrating ATHENA for a Typical MOSFET FIOW 0 0 c cece cece eee e eee eee eens 2 42 2 5 1 Input Information vs Peta eoe aa ladle kala wea bei Hews eee eee bewis bedlageeds 2 42 2 5 2 Tuning Oxidation Parameters ic 2e ze es nre tea te eceteee ae om dewee da
539. re defined as a simple Arrhenius functions EF eq _ Er eq_ 0 F L K exp 74 V Ky e iT 3 76 where E and E represent respectively the formation energy for I and V and K K are coefficients These parameters are specified in the defect mod file Dopant defect Pairs Using the same assumption as for point defects the dopant defect pairs are defined by the charge states of 1 0 1 For boron pairs we have s _ C0 ae ea 2 3 77 ee Cre ae Par rawr l 243 nag 2 ey pe BI K s 1 44 mongey pD i k 6 BV re The BV pairs exist however with positive 1 and neutral 0 charge states only Moreover it is well established BI pairs exhibit a negative U behavior and neutral state is unstable Therefore the three charge states BI BI and BI are considered Silvaco 3 25 ATHENA User s Manual For donor impurities the three charge states for dopant interstitial pairs are also considered Though only AV and AV are known to exist E centers f K 9 E Ap a n Al nO Al X sop 3 79 nr AI K 32 33 1 j K AI K s t 54 ay n Jawo nO cn cal z 6 oe 3 80 s AV K lt 1 53 i AV 7 5 AV The Ky and K y parameters for both donors and acceptors represent the pairing coefficients for the dopant interstitial and dopant vacancy pairs and are defined by following formula x Eat _ x Egy 3 81 Ko Ky exp iT Ko Ko exp 7 Fu
540. rface Roughness determines how the ion is reflected If R 0 the reflection is specular Figure 4 11 model b If R 1 the reflection is random with uniform angular distribution Figure 4 11 model c In general the velocity v efl of the ion after a collision with a surface segment could be presented as follows Vrefl O ion is absorbed ifx gt Prefi 4 17 Vig Vp CER RV eg Rif x gt Prefi 4 18 where Soy is the ion velocity after specular reflection 4 16 Silvaco ELITE Models V anq is the ion velocity after random reflection e xis arandom number Vpl Wrandl lv where v is the velocity of incident ion Each absorbed ion is used to compute the incoming flux F at the surface segment The following characteristics describe the flux IN ise E Nabs TON raj 4 19 where Norm is the normalized number of absorbed particles e Nps is the number of absorbed particles e N traj is the number of trajectories specified by the MC PARTS1 and MC PARTS2 parameter for each type of plasma particles and by the MC POLYMPT parameter for polymer particles e normalized normal Views and tangential Vv velocity components of the absorbed particle ands k before the encounter with the surface I Viabs N Vi 4 20 traj Ne I v v 4 21 labs Dl Nabs e normalized kinetic energy of absorbed particles 2 I 2 ee 4 22 raj N Calculation of Polymer Fluxes After ion and neutral fluxes ar
541. rimental data from 25 Although not corresponding to the modern deep sub micron technologies this simulation represents the high dopant concentration features that reveal the complex couplings between dopant and point defects Therefore it is considered as a meaningful basic test for any advanced diffusion models Although this model has been developed for advanced silicon technologies it still can be used as the standard diffusion model for any diffusion step For example in the case of buried layer formation the TED phenomena become irrelevant Therefore you can use only CDD part of the PLS model while ignoring IC and DDC models Phosphorus To illustrate the improvements that result from the CDD model we show simulations of phosphorus predeposition profiles at high and intermediate concentrations The simulation results are compared to the SIMS data of Yoshida and Matsumoto This experiment represents high dopant concentration features that reveal the complex couplings between dopants and point defects Therefore it is considered as a meaningful test for advanced diffusion models Silvaco 3 37 ATHENA User s Manual Ej TonyPlot Y2 8 18 A Files View v Plot Tools Print Properties gt Help gt PLS Model Phosphorus Predeposition at 900C for 1 hour ite z 3 5 a I A Click to place P changes alignment or drag to get leader SILVACO International 2004 Figure 3 4 Simu
542. ring like nature of the pad surface KINETIC FAC is the multiplier which increases the vertical component of the hori zontal polish rate on sloped surfaces KINETIC FAC increases the vertical polish rate as the surface becomes more vertical An isotropical rate component is also available on the RATE POLISH statement via the ISOTRO PIC parameter The two polish models HARD and SOFT can be used together or separately The isotropic compo nent can be added to either polish model The polish is initiated by the POLISH statement The syn tax of the POLISH statement is very similar to the ETCH statement for machine etches Temperature dependence has been added to the surface diffusion model for ELITE deposits The RATE DEPO statement now includes SIGMA 0 and SIGMA E for this model The dependence is SIGMA DEP SIGMA 0 EXP SIGMA E KT Temperature is entered on the DEPOSIT statement The string advance algorithm and the diffusion algorithm have been modified to give a more realis tic movement The WET RIE etch capabilities of ELITE have been converted from a string based algorithm to a mesh based algorithm This gives greater accuracy when etching near boundaries The CUSTOM deposit has been renamed to USER DATA 1 CUSTOM remains as an alias for this deposit model A new user deposit model was created that allows the same form of input file as USER DATA 1 but also contains all of the functionality of the UNIDIREC model including shadow
543. rity concentration in substrate material see Section 6 2 10 Standard Impurities for the list of impurities Multiple parameters can be used to define compensated doping in the substrate material RESISTIVITY specifies the resistivity of the initial substrate material If RESISITVITY is specified the impurity concentration specified by C parameter will be ignored and calculated from the resistivity vs concentration tables which are available only for boron phosphorus arsenic and antimony The units are Q cm C INTERST specifies the uniform interstitial concentration in substrate material Units are cm C VACANCY specifies the uniform vacancy concentration in substrate material Units are cm BORON PHOSPHORUS ARSENIC ANTIMONY specify the type of impurity when initial doping is defined by RESISTIVITY parameter 6 52 Silvaco INITIALIZE NO IMPURITY specifies that the calculation be performed without impurities No impurities will be introduced during the simulation This speeds calculation and allows quick analysis of oxidation deposit and etch results Parameters Related to Dimensionality of Simulation ONE D TWO D AUTO set whether the run will be in 1D 2D or the dimensionality automatically determined from the process flow AUTO is the default If ONE D is used to select a 1D calculation The calculation will be performed at a location indicated by the X LOCAT parameter TWO D
544. rlo simulation of stopping and ranges i e ion implantation A similar method was first used by Phillips and Price to simulate hot electron transport 71 Yang et al were the first ones to apply the rare event algorithm was applied to simulation of transport phenomena of ions in matter 72 Then Beardmore et al significantly refined this algorithm 73 A brief but comprehensive review of trajectory splitting methods used in modelling of ion implantation is given in 74 With the rare event trajectory splitting technique the speed up is due to changes in the statistical behavior so that rare events are provoked to occur more often The rare event algorithm in ATHENA achieves this by identifying subspaces from where it is more likely to observe given collision event followed by making replicas of the cascade sequences that reach these subspaces Silvaco 3 85 ATHENA User s Manual Figure 3 25 illustrates the trajectory splitting and restart of events replicas as a new threshold is reached When applying splitting to collision cascades or any other specific system the two things that need to be determined are when to split and how many sub trajectories to create when splitting There are different criteria that can be used to obtain the threshold states when splitting need to occur For example Bohmayr et al use a trajectory split method based on checking the local dopant concentration at certain points 75 Beardmore et al
545. rode Type Specified Position AUTO Name source Wi X Position 0 9 0 00 k 16 00 OD YPaestien 200 it mo pRiN Figure 2 24 ATHENA Electrode Menu The following backside electrode statement will appear in the input file ELECTRODE NAME BACK BACKSIDE If an electrode name is not specified DECKBUILD issues the error message NO ELECTRODE NAME SPECIFIED and the command is not written to the input file If an incorrect position for electrode is specified for example ELECTRODE NAME JUNK X 0 6 ATHENA will output the following warning message Cannot find the electrode for this structure Electrode statement ignored and ignores the statement Saving a Structure File for Plotting or Initializing an ATHENA Input file for Further Processing As mentioned in the Standard Structure File Format Section on page 2 8 the DECKBUILD history function saves structure files after each process step In many cases however you need to save and initialize structures independently There are several reason why it s needed to save and initialize structures independently The first reason is because the stack for the history files is limited 50 by default The second reason is because it is usually undesirable to keep dozens of history files on disc each of which occupy hundreds of Kbytes after the DECKBUILD session ends The third reason is because users often
546. roduced during ion implantation scaled to the dopant and within two user defined concentration thresholds For example you can scale clusters to 1 4 times the dopant concentration but only exists between the dopant concentrations of 1e19 and 1e17 cm This allows a scalable approach where clusters will follow implanted dopant as energies and doses vary see Figure 3 26 The following syntax to both switch on and control the cluster model damage scaling METHOD FULL CPL CLUSTER DAM CLUSTER CLUST FACT 1 4 MIN CLUST lel7 MAX CLUST 1lel19 PHOS See RTA Diffusion Modelling on page 3 18 on how to use the 311 clusters during RTA ae La LUST EACT DSP IS acim plunted Figure 3 26 Cluster Damage Control Dislocation Loops Model Dislocation loops can also be scaled to the as implanted dopant profile Loops are introduced as a simple static band to act as an interstitial sink Here interstitials will be recombined at an enhanced rate according to Rate damalpha C C 3 238 3 88 Silvaco SSUPREM4 Models Here Cy is interstitial concentration and C is equilibrium interstitial concentration Loops are placed in a band scaled to dopant concentration with the following command before implantation DISLOC LOOP MIN LOOP 1e16 MAX LOOP 1le18 PHOSPHORUS The recombination rate within the loop band is controlled as follows INTERSTITIAL SILICON DAMALPHA 1e8 C Interpreter Model The C Inte
547. rometers The second line should contain the number of points The following line should contain the position and then the intensity of the first point on the same line This should be repeated for each point This input file is read in the EXPOSE command using the following format EXPOSE INFILE EXP 2 82 Silvaco Tutorial Once the intensity array is initialized or when the Fourier spectrum data is in memory through the IMAGE command you can expose a structure if it exists in memory and if it has photoresist as its top layer s You can either create the structure in the input file or initialize it as described in Section 2 3 Creating a Device Structure Using ATHENA The EXPOSE command has many parameters that control the accuracy and speed of the exposure simulation as well as related imaging parameters The following parameters control simulation speed and accuracy and are unnecessary for a preliminary simulation FLATNESS NUM REFL FRONT REFL BACK REFL ALL MATS The most important of these parameters is the FLATNESS parameter If FLATNESS is set equal to zero the algorithm uses the entire grid for the calculation and may lengthen the simulation time The remaining parameters refer to the image to be exposed Both TE and TM modes are available in exposure but they must be performed separately Select TE by adding the PERPENDICULAR parameter to the EXPOS
548. ropriate to set 700 C temperature limit since for most models the default diffusion parameters are not well known at lower temperatures Numerical rounding bug is fixed in geometrical calculation for very flat triangles during oxidation Improved triangulation during oxidation which reduced probability of creating extremely small triangles The parameter TWO DIMin the STRUCTURE statement now always forces 1D to 2D transformation of the current structure Before it was applied this happened only when structure was written into the outfile D 4 Silvaco ATHENA Version History 14 Increased number of material regions up to 1000 which allows you to create a super lattice structures consist of hundreds of layers D 4 3 ELITE 1 Etch rate retardation can be specified in the RATE DOPE statement D 5 ATHENA Version 5 8 0 R Release Notes D 5 1 SSUPREM4 Diffusion Simulation Features 1 2 3 10 A complete set of Advanced Diffusion Models is implemented see Section 3 2 The earlier implemted CNET model and all related parameters are removed The Boron diffusion model in SiGe is extended to include effect of diffusion suppression by carbon incorporation Additional model for suppression of boron transient diffusion in SiGeC is implemented There are experimental indications that interstitials tend to disappear or get trapped more intensively in SiGe layer with substitutional carb
549. rpreter capabilities shown in Appendix A C Interpreter allows you to extend control over the damage formation models described The template for the implant damage model function is also shown in Appendix A The function is introduced by setting the DAM MOD parameter in the IMPLANT statement The user defined damage model introduced in the function will be used only within the current IMPLANT statement All subsequent implants will use the default damage models 3 5 6 Stopping Powers in Amorphous Materials and Range Validation Stopping powers in amorphous materials have been validated against available experiments Figure 3 27 shows a validation of boron and phosphorus ranges in amorphous silicon where compiled experimental data are taken from 61 10000 E 1000 wn oO D g pej oO 100 o 10 10 100 1000 10000 Energy keV Figure 3 27 Comparison of Monte Carlo simulated project ranges lines and measured ranges dots for Boron and Phosphorus in Silicon Experiments are from 61 Silvaco 3 89 ATHENA User s Manual The solid lines were calculated with ATHENA s Monte Carlo Module The spread of the experimental points in Figure 3 27 is typical and cannot be avoided For example systematic errors due to the depth calibrations and memory effects in SIMS measurements if accounted improperly would yield less accurate usually longer ranges Therefore the Monte Carlo module in ATHENA is calibrated
550. rsal function f y See Equation 3 207 fany AEVO 3 207 This approximation is used in ATHENA by default Obviously the function f y must be symmetrical and have a bell shape Gaussian Lateral Distribution Function The traditional selection for this function is a Gaussian ATHENA uses the Gaussian approximation unless the transversal kurtosis By KURTT in the MOMENTS statement is specified to be different from its default value of 3 0 In this case Equation 3 206 can be easily integrated into the following equation 1 t a 2 ti a 2 C x y f d erfe erfe 3 208 Pei IASA l J2AY J2AY where AY is the transversal lateral standard deviation defined from OO AY fot yyy dedy 3 209 00 Silvaco 3 73 ATHENA User s Manual Specification of Lateral Standard Deviation You can specify Lateral Standard Deviation LSTD DEV or LDRP together with other moments in the MOMENTS statement see the Specification of Implant Parameters in the Moments Statement on page 3 76 You can also control it with the LAT RATIO1 parameter in the IMPLANT statement LAT RATIO1 is the ratio between AY and AR p which is equal to 1 0 by default This means that if the lateral standard deviation and LAT RATIO1 are not specified it will be equal to projected range straggling AR p In the case of dual Pearson model for longitudinal profile corresponding parameters SLSTD DEV or SLDRP and LAT RAT
551. ructure information e Interfacing with device simulators e Using different VWF INTERACTIVE TOOLS These operations are relevant to all individual ATHENA process simulators This part of the tutorial should help you if you re new to each of the process simulators The three sections of the tutorial SSUPREM4 ELITE and OPTOLITH are devoted to individual simulators and should be read if you re going to use those simulators 2 3 2 ATHENA Input Output Before proceeding to the ATHENA operation we will discuss how to provide ATHENA with input information and the forms of output information available from ATHENA Input Information The bulk of input information for ATHENA is usually provided in the form of input files An input file is a text file that can be prepared by using DECKBUILD which will be described throughout the rest of the tutorial or any ASCII text editor such as vi on any UNIX system or textedit on a SUN system The individual lines of the text file are called statements Each statement consists of a statement name and a set of parameters that specify a certain step of a process simulation or model coefficients used during subsequent simulation steps See Chapter 6 Statements Section 6 1 Overview for details on statement syntax The remainder of this tutorial will introduce you to the task of creating good input files Since ATHENA uses a great deal of default information much of the default information i
552. ry component of the refractive index DELTA REAL specifies the difference between the real components of the refractive index for completely exposed and unexposed resist This value is used when dose effect on the refractive index is simulated DELTA IMAG specifies the difference between the imaginary components of the refractive index for completely exposed and unexposed resist This value is used when dose effect on the refractive index is simulated Examples The OPTICAL statement is used to load refractive index values into ATHENA for each wavelength The following shows a typical statement OPTICAL SILICON WAVELENGTH 365 REFRAC REAL 4 5 REFRAC IMAG 5 2 You can enter user defined materials in the following format OPTICAL MATERIAL XXX WAVELENGTH 365 REFRAC REAL 1 4 REFRAC IMAG 3 For more examples see EXPOSE and IMAGE 6 72 Silvaco OPTION 6 39 OPTION OPTION specifies the level of run time output Syntax OPTION QUI Description ET NORMAL VERBOSE o EBUG WARNING This statement specifies the level of information sent to the TTY Terminal Window of DECKBUILD QUIET NORMAL VERBOSE DEBUG and WARNING determines the amount of information that is output about errors CPU times and behavior of the algorithms The default is QUIET The V ERBOS E and D EBUG modes are intended mainly for debugging by
553. s SYSTEM rm rf TEST str DIFFUSE TIME 30 EMP 1000 DRYO2 DUMP 2 DUMP PREFIX TEST TONYPLOT st TEST str Advanced Diffusion Model Example The following command performs the boron pre deposition at 950 C for 1 hour The boron concentration in the ambient gas is 107 cem As a result the output files predep1 str predepl0 str predep100 str and predep1000 str will be saved METHOD PLS 1 TSAVE MULT 10 DIFFUSE TIME 1 HOUR TEMP 950 C BORON 1E20 TSAV E DUMP PREFIX predep W IAL METHOD OXIDE TRAP and For more examples see IMPURITY INTERSTITAL MATE VACANCY Silvaco 6 27 DISLOC LOOP ATHENA User s Manual 6 16 DISLOC LOOP DISLOC LOOP defines the scaling parameters and position of dislocation loops Syntax DISLOC LOOP MATERIAL I IMPURITY MIN LOOP CO lt n gt MAX LOOP CO lt n gt Description This command specifies the scaling of dislocation loops during a subsequent IMPLANT step Dislocation loops are used as interstitial sinks whose recombination rate can be determined with the INTERSTITIAL DAMALPHA lt n gt command Note This command will only work if you switch on the dislocation loop model with the METHOD I LOOP SINK command MATERIAL specifies material for which dislocation loops parameters are set see Section 6 2 9 Standard and User D
554. s S OXIDE specifies screen oxide parameter for the SVDP implant model Default is 0 001 microns The screen oxide thickness is not determined from the structure and must be user specified See Chapter 3 SSUPREM4 Models Section 3 5 Ion Implantation Models for more details and the on line examples on how to set this parameter automatically MATCH DOSE RP SCALE synonym is RP EFF and MAX SCALE specify the method for implant calculations in multi material structures see Chapter 3 SSUPREM4 Models Section 3 5 2 Multi Layer Implants Default is MATCH DOSE SCALE MOM specifies that moment scaling to be used with selected multilayer implant model ANY PEARSON specifies no restrictions on the combinations of allowed skewness and kurtosis This is true by default as required for the SVDP models See Chapter 3 SSUPREM4 Models Section 3 5 Ion Implantation Models for details on potential problems with this setting ATHENA versions earlier than 4 0 had this parameter set to false by default Parameters Applicable Only for Monte Carlo BCA Implant Models N ION specifies the number of ion trajectories to be calculated for the Monte Carlo method When the SAMPLING is not specified the default is N ION is 1 000 for 1D structures and 10 000 for 2D structures 6 46 Silvaco IMPLANT MCSEED specifies a seed for the random number generator used for the Monte Carlo calculation TEMPERATURE spe
555. s Ksurp Kran and Kpow are calculated according to the following equations Ksurr KSURF Oexp SSUREE o Kone KRAT Oexp ARATE Kpow KPOW dexp APORE m where the pre exponential factor and exponent terms can be defined in the INTERSTITIAL and VACANCY statements Surface recombination plays an important role in the relaxation of perturbed point defect profiles back to their equilibrium values which cannot happen by bulk recombination alone The surface generation rate G controls the injection of point defects into the silicon during oxidation Two models have been implemented into ATHENA the default model GROWTH INJ and TIME INJ The moving interface can inject point defects into silicon and polysilicon The GROWTH INJ parameter in the VACANCY or INTERSTITIAL statement will activate or deactivate the growth dependent injection model By default this model is always turned on and is described mathematically by the following equation 3 12 Silvaco SSUPREM4 Models 3 40 vi Gpow G 0 vmoLE v L v 1 Max where e 0 is the fraction of silicon atoms consumed during growth that are injected into the bulk as self interstitials e VMOLE is the lattice density of the consumed material e Gpow is a power parameter The values 0 and Gpow are calculated from the following equations 0 THETA Oexp THETAE kT 3 41 Grow GPOW Dexp FOWE 3 42 and the THETA 0 THETA
556. s for the list of materials If no material is specified the STRIP command removes both photoresist and barrier materials Examples The following sequence of statements deposits photoresist patterned with the mask level named CONT etches oxide through the mask and removes the photoresist with the STRIP statement ASK NAME CONT ETCH OXIDE DRY THICK 2 STRIP This example requires the use of MASKVIEWS For more examples see MASK and ETCH 6 106 Silvaco STRUCTURE 6 63 STRUCTURE STRUCTURE writes the mesh and solution information aerial image information or flips or mirrors the structure SAVEF ILE is a synonym for this statement Syntax STRUCTURE OUTFILE lt c gt INFILE lt c gt OPC lt n gt FLIP Y MIRROR LEFT RIGHT TOP BOTTOM INTENSITY MASK REMOVE GAS SIGE CONV TWO DIM Description This statement writes the entire mesh and solution set to a file The saved data is from the current set of solution and impurity values OUTFILE specifies the name of the file to be written Existing files with the same name are overwritten by newly specified files OUT FILE is an alias for this parameter INFILE specifies the name of the section file generated by MASKVIEWS to be imported This file is assumed to contain the unbiased layout structures and will be used as a reference to calculate the percentage area of d
557. s Errors calculated above this value cause points to be added The MIN ERR parameter specifies the minimum error below which points may be deleted from the mesh unitless Error calculated below this value causes points to be removed Both MAX ERR and MIN ERR are calculated using the Bank Weiser Error Estimator which is defined as 2 A C C where h is the average of the edge lengths associated with node i C is the impurity concentration at node i The parameter CONC MIN specifies the minimum impurity concentration below which 2 2 adapting will stop units 1 0 em The ADAPT MESH statement is used to do mesh adaptation for a given device structure without coupling implant diffusion epitaxy to the process Therefore the mesh adaptation module can be used to assist the manual mesh generation process The following parameters are available on the ADAPT MESH statement e ADAPT specifies a stand alone adaptive meshing step should be performed to refine or relax the current mesh based on the material impurity specification given on the ADAPT PAR statement default false e ADAPT COUNT specifies that stand alone annealing be performed during the execution of the ADAPT MESH statement default false e SMTH COUNT specifies the number of smooth loops during the smoothing algorithm 2 88 Silvaco Tutorial Adaptive Meshing Control Adaptive meshing may be used in several different
558. s Also when specifying electrodes in ATHENA it is useful to transfer electrode layer information from layout to electrical tests in a device simulator see the description of the auto electrode capability in the MASKVIEws Chapter of the VWF INTERACTIVE TOOLS USER S MANUAL VOL II ATHENA can attribute an electrode to any metal silicide or polysilicon region A special case is the backside electrode which can be placed at the bottom of the structure without having a metal region there If you deposit 0 1 um aluminum layer on the full structure after reflection Figure 2 22 using DEPOSIT ALUMIN THICK 0 1 and etched the following part of the layer between x 0 8 and x 0 8 using the Any Shape specification in the Athena Etch Menu See Figure 2 15 ETCH ALUMINUM START X 0 8 Y 20 Silvaco Tutorial ETCH CONT X 0 8 Y 20 ETCH CONT X 0 8 Y 20 ETCH DONE X 0 8 Y 20 you will now have the structure shown in Figure 2 23 TonyPlot 2 2 1 File vj i View F C Plot 7 Tools 7 Print 7 Properties 7 Help 7 ATHENA Structure with electrodes GATE 9 in Microns 9 h g in 9 ih S 4 i it
559. s cm at one atmosphere THETA is the number of oxygen atoms incorporated in a cubic centimeter of oxide In the case of dry oxidation it is equal to THETA and in the case of wet oxidation it is equal to 2 THETA Usually the Deal Grove coefficients should be changed instead of HENRY COEFF THETA specifies the concentration of Os atoms incorporated in the material Units are cm ALPHA specifies the volume expansion ratio between MATERIAL1 and MATERIAL2 Only SILICON POLYSILICON and OXIDE make sense here MIN OXIDANT specifies the minimum oxidant concentration for oxidation to occur Units are cm This parameter is active only if METHOD OX THRESH is used Parameters Related to Grid Control INITIAL specifies the thickness of the native initial oxide at the start of oxidation step If any oxidizable surface of the structure is bare an oxide layer of this thickness is deposited before oxidation begins Units are microns Default is 0 002 Note The oxidation algorithm requires selective deposition of a native oxide onto all exposed silicon or polysilicon areas prior to oxidation Grid problems can result in complex structures To resolve these problems adjust INITIAL or use the DEPOSIT statement to create the native oxide SPLIT ANGLE governs the minimum angle at which the oxide will split open one more grid spacing when oxidizing at a triple point i e
560. s in the layers Like in the dose matching method the distribution in the first layer is calculated directly from the moments corresponding to the first layer without any corrections For subsequent layers the implant distribution is calculated by the formulae C x NAR X Xp 3 197 and i l k 1 where N is the normalization factor gt is the total implantation dose and xef is the effective depth calculated as follows In the case of projected range scaling x for the i layer is i l t k k l R 3 199 eff at OP k 1 P where R is the projected range of the specified ion in the material of the k layer For the case of the maximal range scaling x is calculated as i l t PREES eff P Rp 3ARp k 1 where AR is the projected range straggling in the kt layer In this approximation the estimated maximum ion range R 3AR is taken as the measure of the ion penetration into the corresponding material MOM SCALE In all three models described above the range parameters in each layer are considered independent of the presence of other layers But obviously the distribution of ions stopped in the deeper layers may depend on the thickness and stopping characteristics of the upper layers because each ion trajectory passes through these upper layers The Moment Correction Method set by the MOM SCALE parameter of the IMPLANT statement partially accounts for this effect In the SCALE MOM method t
561. s 20 Silvaco E 2 TSUPREM4 and TSUPREM3 Compatibility Features ASSIGN name t n val 1 0 ratio 1 2 diffuse time St temperature 950 dry ASSIGN name tt n val tt t extract name gateox thickness material Si0O 2 mat occno 1 x val 0 1 IF cond gateox gt 100 0 L MODIFY break IF END L END echo Stt E 3 MESH Statement The new MESH statement provides an alternative to standard mesh generation using the LINI statements It also specifies some parameters used in automatic grid generation when layou information is provided by the Mask Data File generated by Taurus Layout and loaded in the MAS statement Gl et aN DX MAX specifies the maximum grid spacing in the horizontal direction It is used when the grid in the x direction is specified using the Mask Data File DX MIN specifies the minimum grid spacing in the horizontal direction It is used when the grid in the x direction is specified using the Mask Data File DX RATIO specifies the maximum interval ratio between adjacent grid points in the horizontal direction You can also specify this parameter in the INITIALIZE statement The default is 1 5 DY ACTIV specifies the grid spacing in y direction at the bottom of the active region DY BOT specifies the grid spacing in y direction at the bottom of the structure DY RATIO specifies maximum interval ratio between adjacent grid points in vertical direction This parameter could be also
562. s Simulator Calibration Examples 27 ATHENA Examples Including Process Topography and or Lithography 28 SSUPREM3 1 D Process Simulation 29 OPTIMIZER General Purpose Optimization 30 VWF_MOS_TESTS VWF MOS Device Tests a1 VWF_BIP_TESTS YWF BIPOLAR Device Tests Figure 2 2 DeckBuild Examples Window Silvaco ATHENA User s Manual Online Help You can find information on ATHENA statements and syntax using the online help facility You can access this facility in interactive mode or through DECKBUILD Typing help at the ATHENA gt prompt either in Interactive Mode or in the Deckbuild Text Subwindow will display a list of valid ATHENA statements The syntax for the help command is shown below HELP lt command name gt This command will give you additional information about parameter names types initial values and a description of the parameters for the specified command To obtain more information on ATHENA default parameters stored in a special file called athenamod select Command Models in the Main Deckbuild Window This opens athenamod in a text editing window making it possible to read the file or copy and paste statements from the file into a DECKBUILD Text Subwindow Select Command Notes to open a special information file that includes the current release notes and a release history For more information about ATHENA syntax statements parameters and their default values see Chapter 6 S
563. s and B E Deal Kinetics of the Thermal Oxidation of Silicon in O2 HCl Mixtures J Electrochem Soc v 124 p 735 1977 R R Razouk L N Lie and B E Deal Kinetics of High Pressure Oxidation of Silicon in Pyrogenic Steam J Electrochem Soc v 128 p 2214 1981 C P Ho and J D Plummer Si SiO2 Interface Oxidation Kinetics A Physical Model for the Influence of High Substrate Doping Levels J Electrochem Soc v 126 p 1516 1979 B E Deal and M Sklar Thermal Oxidation of Heavily Doped Silicon J ELectrochem Soc v 112 p 430 1965 L N Lie R R Razouk and B E Deal High Pressure Oxidation of Silicon and Dry Oxygen J Electrochem Soc v 129 p 2828 1982 N Guillemot A New Analytical Model of the Bird s Beak IEEE Trans on Electron Devices v ED 34 p 1033 1987 C L L S Hung J Gyulai J W Mayer S S Lau and M A Nicolet Kinetics of TiSi2 Formation by Thing Ti Films on Si J Appl Phys v 54 p 5076 1983 BIB 2 Silvaco Bibliography 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 C A Pico and M G Lagally Kinetics of Titanium silicide Formation on Single Crystal Si Experiment and Modeling J Appl Phys v 64 p 4957 1988 S Sen Hou Ko Sh Murarka A R Sitaram Ellipsometric measurements of the CoSi2 formation from very thin co
564. s are important parameters to include when setting up SSPUREM4 for an RTA scenario Table 3 6 shows an approximate time for completion of about 95 of the TED at various temperatures Table 3 6 The approximate duration of TED at various annealing temperatures Annealing Temperature C Time for completion of 95 of TED 600 390 hours 700 3 3 hours 750 30 minutes 800 3 7 minutes 850 43 seconds 900 8 3 seconds 950 1 9 seconds 1000 0 48 seconds 1050 0 13 seconds 3 1 6 Electrical Deactivation and Clustering Models When dopants are present at high concentrations the electrically active mobile concentration C may be less than the corresponding chemical concentration Cypjem In order for an impurity to become electrically active in a semiconductor material it must be incorporated into a substitutional lattice site which then will contribute with a carrier to either the valence band an acceptor impurity or the conduction band a donor impurity Above certain dopant concentrations however it is impossible to incorporate more dopants into substitutional lattice sites The excess dopants are said to be non active The threshold where the deactivation occurs is often called the solid solubility limit since impurities can exist in different phases in the crystal But for this section we ll call it deactivation threshold Therefore it isn t well defined which phase transition the solid solu
565. s can now be accessed via the DECKBUILD working environment To run these exam ples run DeckBuild pull down the Main Control Examples menu and select SSUPREM4 from the Section menu Then select an example name from the scrolling list and select the Load button at the bottom of the screen This will copy the example and any associated files to your current working directory and load the example into DECKBUILD You can then run the example The example facility includes a short description of the example that describes how to run it and some description of the results that is similar to the manual description Examples describing interfaces between different simulators are also accessible SSUPREM3 Interface The SSUPREM4 PROFILE statement can read a one dimensional 1D structure file generated by SSUPREM4 The PROFILE statement reads a MASTER file that con tains layer and impurity information from SSUPREMS The interface between this simulators is best accomplished by using DECKBUILD Within DECKBUILD you simply build the SSUPREM3 portion of the input deck Next specify the command GO SSUPREMA Specify the mesh within silicon as you normally would in SSUPREM4 DECKBUILD will automatically insert the appropriate profile state ment following SSUPREM4 initialization User accessible polysilicon oxidation rates In previous releases of SSUPREM4 polysilicon and silicon were assumed to oxidize with similar rates The parameters for polysilicon oxidat
566. s computation is correspondingly more lengthy Parameters related to Grid Control during Oxidation Many grid related problems during oxidation are related to the initial oxide deposition See Section 6 40 OXIDE for more about initial oxides GRID OXIDE specifies the desired thickness in microns of grid layers to be added to the growing oxide It has an effect on time steps refer to OXIDE GDT The default is 0 1 microns GRIDINIT OX specifies the initial oxide grid spacing in microns The default is 0 1 microns GRID SILICI specifies the maximum silicide grid spacing in microns The default is 0 1 microns GRIDINIT SI specifies the initial silicide grid spacing in microns The default is 0 1 microns GLOOP EMIN GLOOP EMAX and GLOOP IMAX controls loop detection during grid manipulation The default value is GLOOP IMAX 170e Loop detection checks for intrusions and extrusions in the boundary The intrusion fixing algorithm is triggered by angles greater than GLOOP IMAX A larger value means that more extreme intrusions can develop and increases the possibility of a tangled grid A smaller value leads to earlier intrusion fixing too small a value will lead to inaccuracy due to premature intervention Similar concerns apply to the other parameters The values are a compromise between safety and accuracy The extrusion fixing algorithm is always triggered by angles greater than GLOOP EMAX It may be triggered by lesser extr
567. s for defining input parameter variations 6 38 Silvaco GO 6 23 GO GO starts the simulator Each ATHENA input file should begin with a GO statement Note The GO command is executed by DECKBUILD and documented in the VWF Interactive Tools User s Manual Vol Examples Two useful features of the GO command are shown here This command initializes ATHENA with a specified version number go athena simflags V 4 3 0 R This command initializes ATHENA with a model file lt install gt lib athena lt version gt common athenamod 97a go athena simflags modfile 97a Note If DECKBUILD encounters a GO statement and there is no change in the version or model file ATHENA will continue running Silvaco 6 39 HELP ATHENA User s Manual 6 24 HELP HELP prints summary of state HELP lt command gt or 2 lt command gt Description ment names and parameters syntax HELP lists the parameters of the specified statement and provides a short description of each If there is no statement name given H Examples ELP will show an introductory help message and will list all statements The following will print a list of valid ATHENA commands to the standard output HELP The following will print a description of the DIFFUSE command and its parameters HELP DIFFUSE 6 40 Silvaco ILLUM FILTER 6 25 ILLUM FILTER ILLUM FILTER specifies the ill
568. s of adding chlorine to the oxidizing ambient is shown in Figure 3 14 where the silicon dioxide thickness increases as more chlorine is added to the ambient TonyPlot V2 6 9 File 7 View Plot Tools Print Properties gt Help file view Plot Tools Print Properties i Help 2 E E a g E i 1 Oxidation Time Minutes Click to place P changes alignment or drag to get leader SILVACO International 1996 Figure 3 14 Silicon Dioxide Thickness Versus Oxidation Time with HCI Percentage and Temperature as Parameters Silvaco SSUPREM4 Models Doping Dependence It is well known that SiO formation on highly doped n type and p type substrates can be enhanced compared to SiO formation on lightly doped substrates 37 The dependence of silicon dioxide growth kinetics on doping concentration is manifested as part of the linear rate constant where the physical significance of the high doping levels has been explained primarily as an electrical effect 37 38 This factor in the linear rate constant is given by O ang 7 4 PAFKO exp BAEKB 15 1 3 159 A doping k T y where V is the equilibrium vacancy concentration in silicon at the Si SiO interface V is the equilibrium vacancy concentration in intrinsic silicon BAF KO and BAF KE are specified on the OXIDE statement The equilibrium vacancy concentration composed of vacancy defects in different charged
569. s stored in several non user specified files These files are as follows e The athenamod file includes default parameters of physical models diffusion and oxidation coefficients default parameters of numerical methods characteristics of predefined deposition and etching machines and optical parameters of materials for lithography simulation e The std_table and several sudp files in the implant_tables directory contain ion implantation look up tables e Several files with suffix mod in the pls and models directories contain parameters for advanced diffusion models Chapter 3 SSUPREM4 Models Section 3 2 Advanced Diffusion Models e The athenares file includes resistivity vs doping concentration data It s important to be aware that information from the athenamod file is loaded into ATHENA each time it starts You can override any of the athenamod default parameters by specifying an alternative parameter in an input file or by specifying the entire models file using the modfile option Silvaco 2 7 ATHENA User s Manual Output Information All run time output generated by ATHENA will appear in the Deckbuild Text Subwindow when running DECKBUILD or in the current window or specified output file when running ATHENA standalone Run time output can be grouped into two categories Standard Output and Standard Error Output Standard Output consists of the output from the PRINT 1D statements or the EXTRAC
570. s the device measurement temperature The base and collector currents are strongly influenced by temperature changes as small as a few degrees centigrade A significant effort should be made to determine the exact temperature of the device during measurements before calibration is attempted This temperature should be input into ATLAS in the MODELS statement using the TEMPERATURE lt n gt parameter An increase in temperature will cause an increase in base and collector currents 2 6 2 Tuning the Base Current All Regions A critical region for poly emitter bipolar devices is the interface between the poly emitter and the mono crystalline silicon This region is difficult to process simulate directly as the interface between the polysilicon emitter and single crystalline silicon usually consists of a thin uneven and possibly non continuous film of oxide This is simulated by calibrating the overall effect of this interface with ATLAS The tuning parameter is the surface recombination velocity at this interface for electrons VSURFN for PNP devices or holes VSURFP for NPN devices This will only be effective for thin emitters where at least a fraction of the holes for NPN devices can reach the emitter before recombination The surface recombination velocity parameter not only affects the base current it also affects the base current in all of the operating regions Therefore it is a powerful parameter to approximately match the base current an
571. s the usual SiGe application 2 7 3 DEPOSIT Statement This statement is used to specify the material silicon thickness gridding and doping parameters Once you specify these parameters specify the germanium dopant concentration either as a constant or as a graded doping profile as shown in the example To calculate the doping required the total number of atoms in silicon is taken to be 5e22 cm3 for simplicity Therefore a doping of 1e22 cm3 of germanium is a 20 concentration Other concentrations are calculated in proportion to this so a final concentration at the end of the deposition of 1e21 represents a 2 concentration of germanium Therefore the example deposits a 200A thick SiGe film with an initial germanium concentration of 20 and a final germanium concentration of 2 2 54 Silvaco Tutorial 2 7 4 DIFFUSE Statement Since this example above used the DEPOSIT statement the thermal budget for deposition is simulated by an inert diffusion for the deposition time A typical deposition temperature is around 650 C Generally a typical SiGe HBT device would have a base profile consisting of boron and non boron doped regions together with a tapered germanium profile at both ends of the base In this case simply specify the DEPOSIT and DIFFUSE statements in several stages You can also use several EPITAXY statements to do the same thing At the end of the process simulation the germanium dopant profil
572. s tntiec hie cites ehh acer ee 3 76 Stress Models Stress Histo eoi be Met at ia 6 69 3 97 D 8 String AlQOFIIM 32 sss2 cade ssi besede steadied saddeatacs doses dcaesches tenet 4 2 3 Structure Exposure Development eresi ies exceiessentaeles 2 83 84 Post Development Bake cccccceeeeseeteeeeeeeeeeeeeeetees 2 84 Post Exposure Bake cccccssseceeeeeeeeeeeeeeeeeeeeeeeeeeees 2 83 Structure Manipulation Tools ATHENA In 1D Mode cceccceeeeeeceeeeeeesseeessseeesseeeess 2 58 OI LLEI O AAEE DE P S E EAT deci ws Bit tdecey maa deds 2 56 Structure FLIP arec neo eee ene eek 2 56 Surface Recombination CNET perene a eE E aE Si nS 3 30 31 SVDP implant models ccscceeseesseeeseeeeseeseeeeeneeees 3 69 70 See Analytic Implant Models T Technical S pport sversi eniris 1 1 TOnyPlotis ete sich a a ati aoe 6 108 ap Equation hicar Mean Asian ihed 3 10 Two Dimensional Model Defect Diffusion i2 c chatting dilate ees 3 16 Dislocation Loop Based Bulk Recombination e10 3 15 Dopaiits sti tei a A eee 3 7 Interstitial Generation eeeeeeceeeseeeeeeeeeeeeeteeeeeeeeeates 3 12 InterstitialS 12 4 nteivads eet toi ee ae 3 9 11 Recombination at Interfaces 3 12 14 VACANCIOS steven a aa atest 3 14 15 Two Dimensional Implant Profiles Convolution Method ccceseeeeeeeeeeeeeeeneeeeeeenretaee 3 72 73 Depth Independent Lateral Distribution eeeeeeeeees 3 73 Gau
573. se only for non zero tilt angles Zero rotation means that the ion beam vector lies in the plane parallel to the 2D simulation plane 90 rotation means that the ion beam vector lies in the plane perpendicular to the simulation plane Selecting Continual rotation causes SSUPREM4 to rotate the wafer i e implantation will be performed at 24 different rotation angles from 0 to 345 in increments of 15 There are several damage models available in SSUPREM4 These models allow you to estimate distributions of various defects generated after ion implantation For more details about the damage models and their effect on subsequent diffusion see Chapter 3 SSUPREM4 Models Section 3 5 5 Ion Implantation Damage When the Monte Carlo model is selected you can specify several additional optional parameters See Figure 2 29 The first three parameters are related to the Damage model Point defects 311 clusters and dislocation loops The three others control Monte Carlo calculation initial random number number of trajectories and smoothing See Table 2 2 for a quick reference of ATHENA implant models Table 2 2 ATHENA Implant Model Reference Process Model Assumption Recommendation Implant SIMS Verified Dual Empirical See Chapter 3 SSUPREM4 Models Pearson SVDP Default Table 3 7 Single Pearson Analytic All other cases Monte Carlo Statistical Multi layer structures angled implants Mon
574. se tables correspond to amorphous approximation and ignore effects of material crystalline structure When accuracy of simulated as implant profiles is important and ion implant channeling is pronounced use the Monte Carlo BCA implant model The Monte Carlo model takes into account both crystalline structure and composition of compound materials The actual composition and density of the default ternary compounds InGaAs and AlGaAs as well as the user defined materials must be specified in the MATERIAL statement 3 94 Silvaco SSUPREM4 Models 3 9 SiGe SiGeC Simulation Several experiments e g 80 81 and 82 revealed that boron diffusion in Si Ge or Sij Ge C alloys may differ from diffusion in pure Si substrates To simulate effects of Ge concentration fraction x and carbon concentration fraction y on boron diffusion several models were introduced in SSUPREM4 To activate these models set the MODEL SIGEC parameter in the METHOD statement 3 9 1 Deposition of SiGe SiGeC Epitaxial Layer In this model Si Ge or Sij Ge C alloys are considered as heavily doped with Ge or Ge and C Usually layers with either constant or graded germanium content are formed by a special epitaxy process You can simulate the formation of the Si Ge C layer with a constant Ge and C content using the DEPOSIT statement with a C GERMANIUM parameter set equal to y Ng where Ng is the atomic 022 density of
575. serted into the ATHENA input file The first line will be located exactly at the edge and the spacing will be 0 05um The second line will be inside the POLY layer 0 3um from the edge and spacing at this line is 0 15um The third line will be outside the POLY layer 0 2um from the edge and its spacing will be set to 0 1um You can choose the current edit layer by selecting the Name button located underneath the Layers Menu for the layer in the key list of MaskViEws Figure 2 48 If you select AAD then only set one line for an edge of the AAD layer because offset distances are equal to 0 0 We recommend that only one line be set for unimportant layers It s also important to validate only those layers that are going to be used in ATHENA MASK statements Silvaco 2 67 ATHENA User s Manual When grid parameters are set for all valid layers a cutline can be chosen Click on the Write File button and the Select First End Of Athena Cross Section Line Popup will appear at the bottom of the MaskViews Window Press the select left mouse button at the desired point in the layout e g within the VIA2 region in the upper left corner of the layout You will be prompted to select another end of the cross section line Then drag the pointer and press the select mouse button on the other end of the selected cutline Figure 2 51 shows the Athena Cutline Popup will then appear This shows the exact location of the cutline You can now preview the mask
576. set the correct doping profiles for an analytical implant The syntax for this is shown in Figure C 11 with a comparison of the two different implants in Figure C 12 Photoresist is a special case in ATHENA Although analytical implant tables exist for photoresist they are specific only to one type of photoresist AZ 111 Photoresist materials do vary considerably in density and material abundances Syntax exists in ATHENA to set the required parameters for MC implantation modeling MATERIAL MATERIAL my_resist DENSITY 3 ABUND 1 0 6 AT NUM 1 8 AT MASS 1 16 ABUND 2 0 4 AT NUM 2 6 AT MASS 2 12 ABUND sets the relative abundance of elements in the photoresist AT NUM and AT MASS set the atomic number and weight of the elements respectively DENSITY sets the overall material density From these parameters MC implant can calculate the implanted profile The syntax from Figure C 11 allows the user to fit extract and re use the analytical moments calculated from the MC implant profile A similar technique can be used for implants of non standard species too It is possible for users to build up their own user defined implant moment tables Deckbuild V3 8 4 Release main striker andys dev examples test aniiex07 in e File View Edit gt Find Main Control 7 Commands Tools v H SYNTAX FOR GENERAL PURPOSE MOMENTS EXTRACTION FROM MC IMPLANT raj A R use SET variable to make this approac
577. sh is referred to as a base mesh and options for its formation will be discussed the The Mechanics of the Base Mesh Formation Section on page 2 89 After some initial 1D processing the adaptive meshing function is invoked Subsequently automatically adds mesh that conforms well to the two implanted Boron profiles During the final DIFFUSE statement Boron has been driven down into the substrate and tessellated with the initial simple mesh The mesh adaptation module adapts after each time step This results in meshing conforming to the Boron profile throughout the diffusion process The mesh adaption module is invoked during the simulation by specifying boolean flag ADAPT on the METHOD command preceding IMPLANT DIFFUSE or EPITAXY statements The syntax behind this simple example using the mesh adaption module is shown below Silvaco 2 87 ATHENA User s Manual Three statements are available to access the mesh adaptation module they are briefly described as the following The METHOD statement is used to control numerical algorithms When METHOD ADAPT is specified the mesh adaptation algorithm will be used If you specify METHOD ADAPT false the mesh adaptation algorithm will be turned off ADAPT is off by default ADAPT specifies that the adaptive meshing should be performed on any of the following IMPLANT DIFFUSE or EPITAXY statements Adaptation is performed by following each step on each DIFFUSE
578. shadowed 4 4 R x y Csina i Csina j or R x y Csinw i Csina j 4 5 if point x y is partially shadowed R x y C cos cos i C sina sin j 4 6 if point x y is unshadowed where w and wg are the incident angles and wg are specified on the RATE DEPO command by ANGLE1 and ANGLE2 respectively Silvaco 4 5 ATHENA User s Manual Source Source Li lt lt Figure 4 3 Step Profile with Dual Source 4 3 5 Hemispheric Deposition To use this model specify the HEMISPHERIC parameter in the RATE DEPO statement The flux of vapor is continuously distributed in a range of directions see Figure 4 4 The growth rate can be calculated as R x y C cos cosa i C sin sin j 4 7 where and are the lower and upper bounds respectively of the incident angles of the vapor streams set by parameters ANGLE1 and ANGLE2 respectively To avoid step coverage problems planar sputtering is often used to achieve better film profiles The ideal sputtering source is modeled by means of a hemispheric vapor source with atoms impinging on the substrate from all angles 4 6 Silvaco ELITE Models emispherical Source Figure 4 4 Step Profile with a Hemispherical Vapor Source 4 3 6 Planetary Deposition To use this model specify the PLANETARY parameter in the RATE DEPO statement
579. simulation Second it is always necessary to set a fine grid in the area where ion implantation takes place but the fine grid may be uneccessary after the profile is leveled off during thermal steps So relaxation of an initially fine grid may save simulation time during subsequent steps Parameters for the RELAX statement are set from the ATHENA Relax Menu Figure 2 18 KS Deckbuild ATHENA Relax Area Selected Entire Grid igcation Galert oraties Material 7 Any All Comment Relax everywhere WRITE Figure 2 18 ATHENA Relax Menu 2 22 Silvaco Tutorial To open this menu select Structure Relax in the DECKBUILD Commands menu You can preform grid relaxation over the whole structure if you select Entire Grid or within a selected rectangular area if you choose Selected and specify Xmin Xmax Ymin and Ymax in Location Select Selecting a material from the Material menu specifies which material region will be affected by the grid relax operation The default is all materials within the specified area You can perform the grid line elimination either in one direction or in both directions by selecting X direction Y direction or Both The X direction Relax cannot be performed for individual materials except for the substrate To understand how the relax function changes a grid we will use the structure we have obtained after spacer formation was specified Figure 2 19 If we relax th
580. sity and TRANSMIT By positioning each source element in the source region you can simulate any type of source To simulate a SHRINC source enter the following command lines HHHH If overlapping LUM FILTER CIRCI LUM FILTER CIRCI LUM FILTER CIRCI LUM FILTER CIRCI Li Li Li Ae a Lil SIGMA 0 1 RADIUS 0 25 ANGLE 45 SIGMA 0 1 RADIUS 0 25 ANGLE 1 SIGMA 0 1 RADIUS 0 25 ANGLE 225 SIGMA 0 1 RADIUS 0 25 ANGLE 3 35 15 sources are defined a warning is issued and the most recent source is used If the overlap is partial only the overlap area is overwritten by the most recent source Annular filters can also be superimposed on the source There are two types of annular filters Square and Circle Annular filters have a multiplicative effect on the source Because of this be careful when defining a complex source and a complex filter The following example of an annular source of inner radius 0 4 and outer radius 0 6 is given below IL IL UUM FI UUM FI ITER CIRCLI E a Li iTER CIRCLI SIGMA 0 6 INNER RAD 0 0 OUTER RAD 0 4 TRANSMIT 0 0 In the first statement the SIGMA parameter defines the outer radius In the second statement an opaque spot is defined as an annular filter see Figure 2 63 The source must be described before the filter or the command will be ignored and a warning will be given The
581. specifies concentration at the bottom of the layer and F INTERST specifies concentration at the top of the layer C VACANCY specifies the concentration of vacancies in the epitaxially grown layer in cm 6 30 Silvaco EPITAXY F VACANCY can be specified only together with C VACANCY This parameter generates the linearly graded vacancy concentration in the deposited layer where C VACANCY specifies concentration at the bottom of the layer and F VACANCY specifies concentration at the top of the layer Gridding Parameters DIVISIONS controls the number of vertical grid points in the resulting epitaxial layer This is an optional parameter since it will be generated automatically by default and is related to the surface grid spacing of the original simulation structure before epitaxial process SPACES is an alias for this parameter The default is 10 DY specifies the nominal spacing in microns in the epitaxial layer YDY specifies the depth where the nominal spacing will be applied YDY is calculated relative to the top of the newly grown epitaxial layer MIN DY specifies the minimum spacing allowed between grid lines in the y direction in the new material The default is 0 001 microns 10 Angstroms SI_TO_POLY specifies that the crystalline silicon layer will be grown only over Silicon while Polysilicon will be grown elsewhere Deposition Rate Example The following statement will simulate the
582. ss Parameters Checking 2 47 Tuning Diffusion Parameters eeceeesseeeeseeeeneeteneeees 2 46 Tuning Implantation Parameters sseeeeseeeeeseeeeeneeees 2 45 Tuning Oxidation Parameters ccsceeeeeeeseeseeeeees 2 43 44 Using ATLAS for MOS Process Tuning seeeeeee 2 46 47 Multi Layer Implants DOSE MATCH gi c3 ir at aaa tie i edt tis 3 70 MAX SCALE FEE EAEE EEATT 3 71 MOM SGALE cisien apseiti e teehee 3 71 72 leel E EEA AE AEE E 3 71 N Nuclear Scattering insni aiiai saias 3 77 0 Operation Modes rssi caniiisi erener ahii aeaa 2 5 6 Optical System ET PETEA ET 5 7 8 Ostwald ripening aisi esen aed raa iaaea neiaie 3 11 Oxidation Models escceesesereseteeeseeeeeees 3 44 63 6 67 68 Analytical ivi api ik dedicated chee 3 58 59 EOE E A AE AT 3 47 48 Linear Rate Constant cesceesceseeeeeereseeeeeneeeeeeeees 3 50 51 Mixed Amblents 2 irr E ete 3 58 N mencal cesses aes dash covdea erent oree e Ase eh 3 46 Parabolic Rate Constant cecceeeceseeeeeeeeeteneeeseeeeaees 3 57 VISCOUS eoira a tects 3 48 50 Oxidation Simulation Recommendations Growing Thin OXideS 0 ecceeeeeseeeeeeeeeseeeeeeeeeeeeeeens 3 59 60 Implantation Through Thermally Grown Oxides and Dopant Loss During Subsequent Annealing 3 60 61 Oxidation Enhanced Diffusion OED 3 5 3 61 63 Oxidation Retarded Diffusion ORD seeeeeeeees 3 61 63 P Pair Diffusion ean eins
583. ssian Lateral Distribution Function ccceeseeeeeeeee 3 73 Implant Parameters in the Moments Statement 0 0 3 76 Lateral Standard Deviation cecceeseeeceseeeeeeeeeteeeeees 3 74 Non Gaussian Lateral Distribution Functions 00 3 75 Parabolic Approximation of Depth Dependent Lateral DistribUtion a senetieaediis sei edited even 3 74 75 Using PRINT MOM for Extraction of Spatial Moments 3 76 U USGTIMP tense niesieai eit Gin a il 3 76 Using ATHENA With Other Silvaco Software seeeeeeeeeee 1 3 4 V vacancy flUX EXpress on eeeeeeeeseeeeeeeeeeeeneeetneeeteaeeteaaes 3 14 Silvaco Index 4
584. strate Definition Example The following gives the name we11 to a flat electrode along the bottom edge of the current structure There is no metal required at this location ELECTRODE BACKSIDE NAME WELL For more examples see STRUCTURE Silvaco 6 29 EPITAXY ATHENA User s Manual 6 18 EPITAXY EPITAXY specifies an epitaxial deposition process step Syntax EPITAXY TIME lt n gt HOURS MINUTES SECONDS T EMPERATURE lt n gt T FINAL lt n gt T RATE lt n gt THICKNESS lt n gt GROWTH RATE lt n gt C IMPURITIES lt n gt F IMPURITIES lt n C INTERST lt n gt F INTERST lt n gt C VACANCY lt n gt F VACANCY lt n gt DIVISTONS lt n gt DY lt n gt MIN DY lt n gt YDY lt n gt SI_TO_POLY Description This statement simulates the epitaxial deposition of silicon This model is limited to silicon on silicon applications and should not be used when other materials are present The model is inherently 1D and isn t suitable for selective epitaxial deposition processes Parameters to Define the Epitaxial Step TIME specifies the amount of time for the epitaxial step in specified units HOURS MINUTES and SECONDS specify the units of the TIME parameter Default is MINUTES TEMPERATURE specifies the ambient temperature in C This temperature should fall within the
585. such as conformal deposits oxidation and diffusion run faster in 1D mode The deposition and etch sequences displayed in Figure 2 41 show a sequence of 1D depositions with an automatic conversion to 2D at the first etch Initial silicon Epitaxy or other blanket processing Gate formation or other 2D step Process completion Fast 1D calculation Fast 1D calculation Automatically transformed to 2D calculation A J 2D calculation Figure 2 41 Automatic 1D to 2D Conversion Figure 2 41 shows another aspect of 1D mode In this case the INITIALIZE command is specified with the parameters ONE D and X LOC lt n gt see Chapter 6 Statements Section 6 30 INITIALIZE ONE D specifies that a one dimensional calculation is to be done at the location X LOC In the case of Figure 2 40 1D profiles are generated at different X locations of a complicated BICMOS structure This allows you to quickly check of the overall process using the 1D mode 2 58 Silvaco Tutorial y ram N A que yL Build a complete Wo Cn K A Gull 2D process Nt Nt pt pt description P N i e BiCMOS pt Nt P Substrate y KI Quickly calculate 2 S An S 1D profiles at key g EV 3 c locations using S N Af E e the 1D mode 8 g l S depth depth depth diffuse time 30 temp 9
586. sures are erased from memory You can weigh the exposures by using the DOSE parameter on the EXPOSE command The final four parameters X CROSS Z CROSS CROSS VALUE and X ORIGIN all refer to the aerial image cross section The boolean parameters X CROSS or Z CROSS define the cross section to be parallel to the x axis or the z axis respectively CROSS VALUE specifies the z location of X CROSS or the x location of Z CROSS These parameters are especially useful when several cross sections from the one large two dimensional aerial image will be simulated x CROSS is the default If CROSS VALUE is unspecified the center of the image window defined in the imaging module will be used as CROSS VALUE X ORIGIN allows the aerial image cross section to be shifted laterally in the two dimensional exposure simulation Post Exposure Bake The BAKE command can be invoked by using only one parameter DIFF LENGTH the diffusion length For a post exposure bake of 60 seconds at a temperature of 125 C the recommended diffusion length is between 0 05 and 0 1 micrometers The BAKE command can also be used by specifying TIME and TEMPERATURE in C Development The development module offers a choice of six different development models As mentioned before model parameters are specified in the RATE DEVELOP command After the development model is selected the three primary p
587. surface grid removal has been included in the functionality of the RELAX statement This algorithm allows elimination of surface segments that are smaller than a value specified by parameter DX SURF in microns This is useful for removing excess grid created during high resolution machine etches e A new set of examples is included that illustrate calibration of coefficients for several typical cali bration problems D 14 2 SSUPREM4 Capabilities e Gallium Aluminum and Gold impurities have been added The statement language for DEPOSIT DIFFUSION INITIALIZE and a number of other statements has been modified to include these impurities The IMPURITY statement described above has been added to allow coefficient setting for these impurities e A two stream polysilicon diffusion model has been added This model takes into account the diffu sion of impurities via grain and grain boundary components The relative magnitude of the two components is controlled by the GB VOL RATIO parameter on the MATERIAL statement The grain size of the columnar grains can be set by the parameter GRAIN SIZE on the MATERIAL statement Grain boundary directionality is included in the DEPOSIT calculation Grain size evolu tion is calculated during diffusion and is controlled by the GRAIN SIZE and GB ENERGY parame ters on the MATERIAL statement Impurity segregation into and out of grains is calculated during diffusion The impurities in the grains are treated similar
588. t such as phosphorus or arsenic An Arsenic implantation has been performed with a energy of 2 keV at a dose of 10 at cm and followed by a spike RTA at 950 C with a ramp up estimated at 100 C s In this typical process example for ultra shallow junction dopant is implanted at such a high dose that its concentration reaches the 3 42 Silvaco SSUPREM4 Models solid solubility limit In this case most of the dopant at concentration higher that this limits will precipitate in the early stage of the annealing Moreover all other specific characteristics in arsenic diffusion is taken into account In other words e Arsenic migrate both through interstitial and vacancy mechanisms with roughly the same proportion e Arsenic atoms form with vacancy any clusters AsV or As4V The initial interstitial profile generated by the arsenic implant is modeled though a simple plus n model TonyPlot 2 8 10 R aj File view v Plot Tools gt Print Properties gt Help Arsenic Diffusion after implantation Low energy implant spike anneal As implanted As 2keV 2e15cm 3 As implanted simulation Spike anneal 950C SOC s Spike anneal simulation t 8 2 0 Depth um SILVACO International 2004 Figure 3 9 Simulation of Arsenic diffusion after an implantation at 2 keV with a dose of 1 10 4 cm and a spike anneal at 950 C with a temperature ramp rate estimated at 100
589. t deck segment which will be processed several times The number of passes is specified by the STEPS parameter of the LOOP statement The ASSIGN statement allows you to assign a numerical or character value to a name It is similar to the DEFINE statement The main difference is the capability to vary the assigned value within the LOOP cycle using RATIO and DELTA parameters Also only and characters can be used for substitution The L MODIFY statement allows you to alter the processing of the current LOOP cycle For example if you use BREAK parameter the current LOOP cycle gets interrupted and control comes to the first line after the L END statement The following example shows LOOP L END ASSIGN L MODIFY capability It demonstrates a simple way to estimate a diffusion time needed to grow the gate oxide with the thickness of 100 A First the initial value 0 0 is assigned to the name tt Within the loop variable value is assigned to the name t Its inital value is 1 0 and increases by a factor of 1 2 in each subsequent loop iteration The current value of t is used as diffusion time The next ASSIGN statement calculates the total diffusion time The EXTRACT statement finds current oxide thickness gateox If the required condition is that extracted gateox exceeds 100 A the LOOP cycle is interrupted and the total diffusion time is printed ASSIGN name tt n val 0 0 LOOP step
590. t include deposited species reflection sticking coefficient surface diffu sion and density variations A tuning parameter DX MULT lt n gt has been added to the ETCH statement to allow enhanced dis cretization during individual ELITE etch steps Increasing the value of DX MULT from its default of 1 0 will result in larger surface segments and a reduced discretization Decreasing DX MULT will result in better discretization in both space and time during the etch calculation Reducing the value of this parameter allows realistic modeling of wet etches that previously were poorly resolved A new machine type specified by the parameter CUSTOM is now available for ELITE deposits This machine type reads deposit rate vs angle information from a user specified ASCII file This can be used as an interface to deposit rates produced by non Silvaco simulators If a full range of deposit angles is not specified the simulator will interpolate rates between the closest angle and a rate of zero at an angle of 180 degrees The rates at 180 degrees are assumed to be the same The limits on number of regions and number of surface segments for machine etch calculations have been increased The new limits should be adequate for most applications of ELITE D 14 4 OPTOLITH Capabilities A new defocus model that directly couples the imaging module to the exposure module has been introduced The DEFOCUS parameter on the EXPOSE statement is now obsolete and th
591. t model the effect that these defects have on subsequent impurity diffusion ATHENA considers implant damage as point defect generation Point defects are silicon interstitials and lattice vacancies that are created as energetic implanted ions collide with silicon lattice atoms The most practical model for coupling implant damage to subsequent diffusion calculations is the 1 model In its simplest form the 1 model adds exactly one interstitial for each implanted ion This is a reasonable approximation if one assumes that the vacancies and interstitials created by the implant recombine quickly relative to the time scale needed to produce significant diffusion This leaves one extra interstitial for each ion assuming the implanted ion has replaced it on the lattice C 4 Silvaco Hints and Tips This model is applicable to both Monte Carlo and the default analytic implants and can be invoked by including the UNIT DAM parameter on the IMPLANT statement A commonly applied variation to this model is to scale the number of generated interstitials In ATHENA this can be done using the parameter DAM FACT on the IMPLANT statement A corresponding profile of lattice vacancies is introduced in this model with the maximum of zero and 1 DAM FACT times the implanted ion profile The diffusion models that will include the effect of the point defects are either the TWO DIM or FULL CPL models Both models include the local point defect concentra
592. t moments used by the IMPLANT statement is controlled now by the MOMENTS statement Parameter DEF_TABLE specifies that only the default look up implant table athenaimp should be used Parameter USER_TABLE lt filename gt specifies that the user defined table should be used as a first choice for the moment search In other cases parameters specified in the MOMENTS statements if any will be checked first of all D 12 Silvaco ATHENA Version History Template for the User Defined Implant Tables Auxiliary file USERIMP provides template for specifying implant moment sets for all types of analytical implant models from the simplest Gauss to double Pearson with advanced lateral distribution High Energy Implant Tables 1 to 8 MeV implant tables are now available for all major impant species for Silicon Oxide Si Ion Implant into Silicon results in the interstitial distribution which allows estimation of preamorphization effect Trajectories of Primary Ions and Substrate Atoms Knocked on in the Implant Cascade can be now saved in a special TRAJ FILE and subsequently plotted using TonyPlot This frees ATHENA from the last dependency on old graphic library plotlib PRE FACTOR and POW FACTOR Parameters are Eliminated from the IMPLANT Statement Instead PRE FACTOR parameter is added to the IMPURITY statement so electronic stopping can be control for each implant ion substrate material combination separately POW FACTOR does not make a
593. t n gt DPOS 0 lt n gt DPOS E lt n gt KR 0 lt n gt KR E lt n gt IVFACTOR lt n gt IIFACTOR lt n gt KTRAP 0 lt n gt KTRAP E lt n gt DAMALPHA lt n gt MATERIAL TIME INJ GROWTH INJ RECOMB KSURF 0 lt n gt KSURF E lt n gt KRAT 0O lt n gt KRAT E lt n gt KPOW 0 lt n gt KPOW E lt n gt VMOLE lt n gt THETA O lt n gt THETA E lt n gt GPOW 0 lt n gt GPOW E lt n gt WETO2 DRYO2 REC STR lt n gt INJ STR lt n gt A 0 lt n gt A E lt n gt TO 0 lt n gt TO E lt n gt TPOW 0 lt n gt TPOW E lt n gt DCARBON E lt n gt KCARBON 0 lt n gt KCARBON E lt n gt Description These two equivalent commands specify transport and generation recombination coefficients for interstitials and vacancies The statements allow you to specify coefficients for any material though it is only practical for semiconductors ATHENA has measured or calibrated default values only for silicon and some interfaces with silicon MATERIAL specify the material for which the interstitial or vacancy parameters apply as well as MATERIAL1 for the segregation and transport parameters on the boundary between two materials see Section 6 2 9 Standard and User Defined Materials for the list of materials Default is Silicon Defect Diffusion Parameters D 0 and D E specify the interstitial or vacancy diffusion coefficient D O is the pre exponential const
594. t plot of Figure 2 14 The MIN SPACING parameter preserves the horizontal mesh spacing for high aspect ratio grids ATHENA tries to reduce high aspect ratio grids and MIN SPACING stops this To get a finer grid not at the polysilicon surface but in the middle of polysilicon layer change YDY to 0 2 This puts on a finer grid at a distance of 0 2um from the surface of the structure You can do this by positioning the cursor in the input file and backspacing over existing text or entering new text For example DEPOSIT POLY THICK 0 5 C PHOSPHOR 5 0E19 DIVISIONS 10 DY 0 02 YDY 0 2 It is possible to see the effect of changing the YDY parameter within the polysilicon without rerunning the whole input file To do this highlight the previous statement DEPOSIT OXIDE select Main Control Init from History button and press the Cont button The new history file can then be loaded into TonyPLot see the right plot in Figure 2 14 TonyPlot V2 1 beta ATHEMA ATHENA Data trom history 14 Data from history 1 UA a Ea ead SN S N l AAA i Ra j i j NNN i ANNI AA RR a SEESE TAN i l 1 J l aa a AN EEEE PJ PPP m M W aa a a Pal ata al a a Pal AEFIA PERAE LAA Lt ee LI a J Sesesees a ana PLA AAAA Fa a AAAA Le Ge r SO i N NINN
595. tal with the silicon and polysilicon The remaining metal is then etched away The SSUPREM4 syntax used to model silicidation seems very natural to an experienced user For titanium silicide for example the syntax is DEPOSIT TITANIUM THICK 0 1 DIV 8 DIFFUSE TIME 5 TEMP 650 ETCH TITANIUM ALL The results of a salicide simulation are that a titanium layer is formed correctly in the source drain and gate areas with no reaction with the oxide spacer No special model syntax needs to be used with the silicide module in order to achieve the silicidation But a good parameter to be aware of is GRID SIL on the METHOD statement This controls grid spacing within the silicide layer as it grows This is similar to the way the GRID OX parameter controls the grid within thermally grown oxides Question How is implant damage enhanced diffusion modeled by ATHENA Which tuning parameters should be used for matching experimental results Answer The effect of implant damage enhanced diffusion is important in many technologies Typical cases are the source and drain diffusion in MOSFETs and the emitter diffusion in bipolar devices Damage generated by implantation leads to an enhancement to the diffusion of these dopants during subsequent heat cycles Simulation of the enhanced diffusion effects are divided between two processes First ATHENA must simulate the implant damage generated by a given implant and secondly it mus
596. talline materials Two new parameters are introduced e MISCUT TH Target wafer polar angle miscut measured in the XY plane Y being the inward direction e MISCUT PH Target wafer azimuth angle miscut measured in the XZ surface plane Z pointing away from the observer Fixed wrong damage scaling when sampling capability is used Improved algorithm of SSUPREM4 deposition Now it guarantees that non uniform spacing specified by DY and YDY parameters is preserved even when number of divisions is changed due to complex grid The SSUPREM4 deposition is improved for the case when number of DIVISIONS is not specified For thin layers with thickness less than 0 012 microns an uniform grid with spacing of approximately 0 001 microns will be generated A non uniform grid with spacing equal to 0 001 microns at the top and bottom of the deposited layer will be generated for thicker layers The number of divisions is automatically selected dependent on the layer thickness It is 12 for the layers thinner than 0 02 microns and 18 for layers thicker than 2 microns Improve specification of POLY DIFF model The model flag used to be set to false unless the statement METHOD POLY DIFF was immediately before DEPOSIT POLY GR SIZE lt n gt statement Default value for the MIN TEMP in the METHOD statement is returned to original 700 C The manual had always stated that it is 700 C though few previous versions get reduced value of 475C It is more app
597. tatement Added Cc VACANCY and C INTERST to the INITIALIZE statement Added new standard impurity HELIUM The only practical application available in the moment is Monte Carlo ion implantation of Helium Added capability to oxidize materials other than Si and Poly New parameter OXIDIZABLE is added to the MATERIAL statement If OXIDIZABLE is set to TRUE then all oxidation related parameters for the specified material will be set equal to those for Silicon You can specify the different values for oxidation parameters in the OXIDE MATERIAL statement Improved gridding in oxide that results in smaller number of extremely small triangles are generated in areas of slow oxidation For example under the polysilicon gate during reoxidation process D 2 2 ELITE Features 1 Added new parameter OUTF TABLE lt filename gt in the RATE ETCH statement This can be used for detailed analysis of plasma ions and neutrals distributions The old parameter OUTFILE in the RATE ETCH statement has changed to OUTF ANGLE because it specifies a file with ions vs angle distribution output D 2 3 OPTOLITH Features 1 Reimplemented Proximity Printing Model in OPTOLITH which simulates imaging without any reduction lens To use this model specify the GAP parameter in the IMAGE statement The model is implemented for Manhattan Circular Ring and Multi Ring masks Added capabilit
598. tatements 2 4 Silvaco Tutorial 2 2 Operation Modes ATHENA is normally used through the DECKBUILD run time environment which supports both interactive and batch mode operation We recommend that you always use ATHENA through DECKBUILD In this section we present the basic information you need to run ATHENA in the DECKBUILD environment The VWF INTERACTIVE TOOLS USER S MANUAL VOL I provides a more detailed description of the features and capabilities of DECKBUILD 2 2 1 Interactive Mode With DeckBuild To start ATHENA under DECKBUILD type deckbuild an at the UNIX system command prompt The command line option an instructs DECKBUILD to start ATHENA as the default simulator To start with an existing input file type deckbuild an lt input filename gt The run time output shows the execution of each ATHENA command and includes error messages warnings extracted parameters and other important output for evaluating each ATHENA run When you run ATHENA interactively the run time output is sent to the Deckbuild Text Subwindow of the Deckbuild Application Window and you can save it as needed You don t need to save the run time output explicitly The following command line however specifies the file name that will be used for storing the run time output deckbuild an lt input filename gt outfile lt output filename gt In this case the run time output is sent to the output file and to the output section
599. tax BASE PAR MAT ERIAL GRAD SPACE RATIO BOX BASE PAR runs the base mesh for generating the initial grid MATERIAL one of standard materials or user specified material see Section 6 2 9 Standard and User Defined Materials for the list of materials GRAD SPACE specifies the gradient of the adjacent grid spacing in the y direction of this material Default is 1 5 RATIO BOX specifies the approximate aspect ratio of triangle element after base mesh generation in this material Default is 2 0 Examples The following example generates a good quality base mesh for each related material region For more examples see BASE BASE BASE BASE BASE BASE PAR PAR PAR PAR PAR PAR Nn Nn WN DN OXIDE GRAD SILICON GRAD POLYSILICON GRAD OXIDE GRAD SILICON GRAD POLYSILICON GRAD BASE MESH PACE PACE PACE PACE PACE PACE RA RA RA RA RAT RA TIO TIO IO TIO TIO TIO BOX 2 BOX 2 BOX 2 BOX 2 BOX 2 BOX 2 Silvaco BOUNDARY 6 9 BOUNDARY BOUNDARY specifies boundary conditions for the initial material Note For most typical boundary conditions ATHENA has defaults that eliminate the need for BOUNDARY statements The BOUNDARY statement can be used to modify the treatment of the surfaces for special purpose simulations Syntax BOUNDARY REFLECTING EXPOSED B
600. tax PARAM FALSE or PARAM F A mutually exclusive choice among parameters is indicated by parentheses around the parameters and vertical bars between each parameter PAR1 PAR2 Only one parameter in such a group may be specified at a time Specifying more than one parameter in a mutually exclusive group is an invalid operation and will generally prompt a warning or error message Parameters that are optional to a statement are enclosed by brackets Most parameters are assigned default values and so defining them is optional All parameters and parameter values however should be checked in the context of the actual process that will be simulated before relying on the results of any simulation String valued parameters can be specified as a single word e g INFILE FILE1 or as a sequence of words surrounded by double quotes e g MATERIAL Nickel Silicide Real valued parameters can be specified as expressions involving numbers numerical constants the operators and the functions listed in Table 6 2 If an expression contains spaces then enclose it in parentheses 6 2 Silvaco Overview Table 6 2 Functions Function Description abs5 Absolute value active Active portion of the specified dopant erf Error function erfc Complimentary error function exp Exponential gradx Computes the approximate gradient in the x direction grady Computes the approxim
601. te or BCA into a structure where many ions could be reflected trenches when channeling is not described by SVDP high or very low energy Silicon Amorphous No channeling Most of implant profile is within amor Type effect is phous materials oxide polysilicon pre included amorphized silicon channeling is negli gible or not important Crystal Default Channeling When channeling effects are important effect is light ions boron phosphorus _ zero or included close to 0 tilt implant through thin amorphous layer into crystalline sub strate Silvaco ATHENA User s Manual Impurity Boron Phosphorus Arsenic Bf2 Antimony Silicon Zine Selenium Beryllium Magnesium Aluminum Gallium Carbon Indium Chromium Germanium Dose ions cm2 4 5 1 0 9 9 Exp T 13 Energy Ke 20 O 500 Model Dual Pearson Gauss Full Lateral Monte Carlo Tilt degrees 7 0 jH 90 Rotation degrees 0 ofr 360 Continual rotation _ Material type Crystaline Amorphous Damage Point defects lt 311 gt Clusters Dislocation loops Point defects Scaling factor lt 311 gt Clusters Min cluster thresh 1 1 0 J 99 Exp FT 17 Max cluster thresh 1 5 1 0 99 Exp T 19 Cluster scaling 1 40 0 00 lt a 200 Dislocation loops Min loop conc 1 0 1 0 J 99 Exp T 17 Max loop conc 1 0 10 H 99 Exp T 18 Initial random number 2 2 k 10000 i j s l Number
602. tement syntax description signifies that you can only specify one of impurity names from the list below The generic name IMPURITIES appeared in a statement syntax description signifies that you can specify one or more impurity names from the list below in the statement simultaneously The impurity names below can appear as is or as a part of a parameter name e g I BORON C BORON and F BORON The following is the list of standard impurity names currently available in ATHENA ALUMINUM ANTIMONY ARSENIC BERYLLIUM BORON CARBON CHROMIUM FLUORINE GALLIUM GERMANIUM GOLD HELIUM HYDROGEN INDIUM MAGNESIUM NITROGEN OXYGEN PHOSPHORUS SELENIUM SILICON and ZINC 6 8 Silvaco ABERRATION 6 3 ABERRATION ABERRATION defines aberration parameters of the optical projection system Syntax ABERRATION X FIELD lt n gt Z FIELD lt n gt SPHERICAL lt n gt COMA lt n gt ASTIGMATISM lt n gt CURVATURE lt n gt DISTORTION lt n gt FIFTH SEVENTH NINTH Cl lt n gt C2 lt n gt C3 lt n gt C4 lt n gt C5 lt n gt C6 lt n gt C7 lt n gt C8 lt n gt C9 lt n gt C10 lt n gt C11 lt n gt C12 lt n gt C13 lt n gt C14 lt n gt C15 lt n gt C16 lt n gt C1 7 lt n gt C18 lt n gt C19 lt n gt C20
603. temperature dependence Examples The following command introduces clusters during ion implantation The clusters will have an effective interstitial concentration of 1 4 times the concentration of implanted boron The clusters will lie in the region where Boron is between 10 and 101 cm zZ ETHOD CLUSTER DAM CLUSTER I BORON SILICON MIN CLUS 1e15 MAX CLUST 1e19 CLUST FACT 1 4 IMPLANT DOSE 1e14 ENERGY 50 BORON The example goes on to define the cluster dissolution time and a short thermal cycle Results for each timestep of the diffusion cycle will be stored in files RTA_ CLUSTER I BORON SILICON TAU 311 0 10 TAU 311 E 0 24 DIFFUSE TEMP 1000 TIME 10 60 NITRO DUMP 1 DUMP PREF RTA_ For more examples see METHOD DISLOC LOOP INTERSTITIAL and VACANCY 6 18 Silvaco COMMENT 6 11 COMMENT COMMENT is used to specify character strings for documenting the input deck and ATHENA output Syntax COMMENT or Description The COMMENT statement or are used to document the input file You can insert them in the beginning of any line of the input deck Silvaco 6 19 CPULOG ATHENA User s Manual 6 12 CPULOG CPULOG instructs ATHENA to output CPU statistics Syntax CPULOG LOG CPUFILE lt c gt Description The CPULOG statement logs the CPU time used in various internal operations The CPU time i
604. temperature for the first type of ions which is unitless MC LAT T2 specifies the plasma lateral temperature for the second type of ions which is unitless MC ION CU1 specifies the plasma ion current density for the first type of ions ions second cm MC ION CU2 specifies the plasma ion current density for the second type of ions ions second cm MC PARTSI1 specifies the number of MC simulated particles for the first type of ions MC PARTS2 specifies the number of MC simulated particles for the second type of ions MC ANGLE1 specifies the incident angle for the first type of ions The default is 0 normal incidence MC ANGLE2Z specifies the incident angle for the second type of ions The default is 0 normal incidence Wet Etch Example The following example defines an etch machine that attacks silicon with wet etch characteristics and an etch rate of 0 1 micron minute RATE ETCH MACHINE TEST SILICON WET ETCH ISOTROPIC 1 U M HJ Silvaco 6 93 RATE ETCH ATHENA User s Manual Monte Carlo Plasma Etch Example The following statement defines parameters of Monte Carlo Plasma Etch machine as well as etching characteristics of Silicon associated with this machine RATE ETCH MACHINE MCETCH SILICON MC PLASMA ION TYPES 1 C PARTS1 20000 MC NORM T1 14 0 MC LAT T1 2 0 C ION CU1 15 MC ETCH1 1le 05 MC ALB1 0 2 MC PLM ALB 0 5 C POLYMPT 5000 MC RFLCTDIF 0 5 Silvaco
605. tep on each DIFFUSE statement e Integer IMPLANT MES specifies which adapting algorithm to use on IMPLANT statements cur rently IMPLANT MES 0 corresponds to University of Florida s algorithm This is the default Silvaco D 15 ATHENA User s Manual The parameters available on the ADAPT PAR statement are as follows e Adaptive meshing control variables are available on the ADAPT PAR statement They are MIN ADD IMPL SUB DIFF SMOOTH and IMPL SMOOTH e MIN ADD stops point addition in IMPLANT when the number of points added in the current loop is less than MIN ADD total number of points The default value for MIN ADD 0 05 e IMPL SUB is a boolean flag that stops point removal during IMPLANT adaptive meshing The default value for IMPL SUB false signifies that points are not being removed e Integer DIFF SMOOTH specifies which annealing algorithm to use after each adaption step cur rently DIFF SMOOTH 0 corresponds to no annealing during DIFFUSE DIFF SMOOTH 1 corre sponds to Laplacian smoothing and the dose conservation interpolation algorithm The default is DIFF SMOOTH 0 e Integer IMPL SMOOTH specifies which annealing algorithm to use after each adaption step cur rently IMPL SMOOTH 0 corresponds to no annealing during IMPLANT IMPL SMOOTH 1 cor responds to Laplacian smoothing and the dose conservation interpolation algorithm The default is IMPL SMOOTH 1 e Boolean SILICON OXIDE specify material r
606. ter option to save ion trajectories in a special structure file for subsequent display in TONYPLOT in the IMPLANT statement the multi threading capability will switch off The FULLROTATION parameter of the IMPLANT statement couldn t be specified simultaneously with multi threading Multiple rotation statements should be used instead Implemented BCA implantation models for 110 and 111 silicon Added capability to MC Implant Module that allows to simulate damage or preamorphization induced by arbitrary inert ion bombardment You can specify atomic number Z1 and atomic weight M1 in the IMPLANT statement Only implant damage will be introduced into the structure after the completion of the Z1 ion implant The level of this damage will affect subsequent normal implant profiles Note If M1 is not specified the atomic weight of the main isotope of element Z1 will be used 4 Memory management of all modules is substantially improved As the result the limits on number of grid points nodes and triangles in simulation structure are removed This allows to perform simulation in large multilayer structures without sacrificing accuracy 5 Fixed the capability to specify diffusion through impurity vacancy pairs defined by parameters such as DVX 0 DVX E 6 Added positively charged vacancy impurity pair diffusion parameters to the IMPURITY statement DVP 0 DVP E DVPP 0 and DVPP E 7 Added values for diffusion and a
607. teral Spread of Implanted Ions Theory NASECODE VI Ed J J H Miller Boole Press p 513 1989 D G Ashworth M D J Bowyer and R Oven Representation of Ion Implantation Distributions in Two and Three Dimensions J Phys D v 24 p 1120 1991 G Hobler E Langer and S Selberherr Two Dimensional Modeling of Ion Implantation with Spatial Moments Solid State Electronics v 30 p 445 1987 M Temkin and I Chakarov Computationally Effective Model for 2D Ion Implantation Simulation Semiconductor Process and Device Performance Modeling Eds S T Dunham J S Nelson MRS p 27 1998 R Oven D G Ashworth and M D J Bowyer Formulas for the Distribution of Ions Under an Ideal Mask J Phys D v 25 p 1235 1992 J E Gibbons W S Johnson and S W Mylroie Projected Range Statistics 2nd edition Stroudsburg Pennsylvania Dowden Hutchinson amp Ross Inc 1975 A F Burenkov F F Komarov M A Kumakhov and M M Temkin Tables of Ion Implantation Spatial Distributions Gordon amp Breach Science Publishers 1986 O B Firsov Soviet Physics JETP v 33 p 696 1957 J F Ziegler J P Biersack U Littmark The stopping and range of ions in solids v 1 Pergamon Press 1985 UT Marlowe Version 5 0 User manual University of Texas Austin USA 1999 O B Firsov Soviet Physics JETP v 36 p 1517 1959 W Brandt and M Kitagawa Effective Stopping Power Charges of Swift Ions i
608. tes of mask specification which define the left and right of simulation space e The number of mask in the file e The first line of each mask description includes the mask name e g Poly and number of opaque segments Each subsequent line gives the minimum and maximum coordinates of each of these segments DECKBUILD will generate the LINE X statements which are used by ATHENA according the following rules 1 The lines will be generated at each mask edge 2 The grid spacing at these lines will be equal to DX MIN specified in the MESH statement 3 If none of the mask edges coincides with left or right boundary of the simulation space the LINE statement corresponding to such boundaries will be without spacing 4 Additional one or two LINE statements will be generated between the lines corresponding to mask edges The SPAC parameters at these additional lines will be minimum of DX MAX and DX MIN DX RATIO n where DX MIN DX MAX and DX RATIO are parameters specified in the MESH statements This will guarantee that grid spacings in the horizontal grid will be increased far from mask edges These rules are illustrated by the following example of a structure with two POLY gates If the Mask Data File has the following fragment 1 000000E 03 0 3000 J POLY 2 800 1200 1800 2200 and the MESH statement MESH dx min 0 01 dx max 0 1 dx ratio 2 follows then DECKBUILD will generate the next sequence of the
609. that you let the device simulator calculate the work function of the gate electrode from the simulated doping profile rather than assigning a value to it This means making sure that the polysilicon gate is not itself defined as an electrode but rather a layer of metal usually aluminum is deposited on top of the polysilicon gate Therefore this metal layer is the film defined as the electrode Do not assign a work function to this deposited metal electrode to ensure that it behaves as an ohmic contact rather than a Schottky contact The effective work function of the poly gate will then be correctly calculated from the doping profile in the polysilicon An important area for accuracy in MOSFETs is modeling the inversion region under the gate As it is this charge that is responsible for current conduction in the device The inversion region charge under the gate only extends approximately 30 Angstroms into the silicon The inversion region charge density under the gate falls off rapidly with depth into the silicon It is imperative that there are several mesh points in the Y direction in this inversion region to model the drain current correctly Accordingly we recommend that the mesh spacing under the gate be no more than 10 Angstroms 1 nm You would think that a 10 Angstrom mesh under the gate would result in a huge number of mesh points But there only needs to be approximately three mesh points within the inversion region in the Y direction T
610. the moments are not found in the specified file ATHENA will proceed to the standard tables Finally the set of MOMENTS statement can be used to specify all spatial moments for any ion material energy dose combination ATHENA will use parameters from this set before proceeding to a standard search sequence If the moments for certain implant conditions are unavailable then you can use the Monte Carlo simulation for these conditions Using PRINT MOM for Extraction of Spatial Moments The PRINT MOM parameter in the IMPLANT statement prints the calculated or extracted from the tables moments into output and also saves the moments in the standard structure file The last capability allows you to use the extract EXTRACT statement and substitution functions of DECKBUILD for automatic generation of the MOMENTS statement If spatial lateral and mixed moments need to be found from Monte Carlo calculation use the IMPCT POINT parameter because it forces all trajectories to be started in one point This not only allows the spatial moments to be found but also the building of a Monte Carlo calculated source point 2D distribution function which can be useful for comparison purposes 3 5 4 Monte Carlo Implants The analytical models described in the previous section give very good results when applied to ion implantation into simple planar structures bare silicon or silicon covered with thin layer of other mat
611. the Athena Cutline Popup see Figure 2 51 is decreased from 0 05 to 0 025 a finer grid will be obtained at both POLY gate edges see Figure 2 52 When the location of a cutline and the corresponding grid are satisfactory the cutline information can be stored or used either as a Cut file or as a cutline object You can save the Cut file by pressing the Write button in the ATHENA Cutline Popup You can then load this file into DECKBUILD for use in ATHENA by selecting MaskViews gt Cutfiles from the Tools menu in the DECKBUILD window The MaskViews Cut Files popup Figure 2 54 will then appear Select the desired sec file and press Load Silvaco 2 69 ATHENA User s Manual MaskViews Cut Files Category F Disk Files Directory export main mishat athman Filter sec default sect defaultseco Co E x Filename defaultsect Figure 2 54 MaskViews Cutline Files Window You can now select any preview as shown in Figure 2 52 Press the Select mouse button anywhere within the Display Masks window and the cutline icon will appear Without releasing the Select mouse button drag the icon into the MASKVIEWS Cut Files Window and drop it by releasing the Select mouse button Several cutlines with different locations and grids can be dragged and dropped in this fashion You can then load them into DECKBUILD for use by ATHENA When ATHENA is loaded with a cutline DECKBUILD will comm
612. the METHOD statement is used for two purposes The first purpose is to specify models for how damage is induced during processes such as implantation or oxidation The second purpose is to specify how that damage anneals and diffuses in subsequent or concurrent thermal processes It s important to realize that the METHOD statement must be placed above the line specifying the process step or steps to which it refers in the input file Any number of method statements can be used in an input file allowing you to change the models at will during the process flow to optimize the speed and accuracy of the simulation The models specified in the METHOD statement will hold true for all processes that follow it until it s updated by a subsequent method statement 2 30 Silvaco Tutorial Table 2 1 below indicates a recommended method statement for typical processes It should be realized that these statements are hierarchical so there is no accuracy lost if a more complicated model is used where a simpler one would suffice The only downside here is a longer simulation time The table below starts off with the simplest of models and progresses to the more complicated ones Table 2 1 Recommended Method Statements for Typical Processes Method Statement Syntax Suitability of using this method syntax method fermi Use only before undamaged silicon diffusions where doping concentrations are less than 1e20 cm3 and no oxidizing
613. the SSUPREM4 model statements such as METHOD OXIDE MATERIAL and IMPURITY For more information about SSUPREM4 see Chapter 3 SSUPREM4 Models For more information about the MODEL statements see Chapter 6 Statements When simulating any process involving dopant or its diffusion or both it is absolutely critical for simulation accuracy to use the appropriate model Process steps where correct choice of models are vital include implantation diffusion rapid thermal annealing oxidation and epitaxy This section provides specific advice on which models should be used for each process step 2 4 2 The Reason for Multiple Models for Each Process The key to simulating any dopant related process is to accurately account for damage in the semiconductor For example in silicon processing typical implantation doses can cause sufficient damage to the substrate to enhance dopant diffusion rates by three orders of magnitude or more so choosing the wrong model in this instance will result in inaccurate results Well known device anomalies such as the Reverse Short Channel Effect in MOS processing or the emitter push effect in bipolar processing are wholly the result of such damage enhanced diffusion Other processes that consume the semiconductor such as oxidation and silicidation also inject damage into the substrate This must be accounted for if accurate dopant profiles are a requirement This section
614. the laboratory coordinate system Silvaco 3 83 ATHENA User s Manual Specifying different ROT SUB will have an effect on channeling But remember that ion propagation is three dimensional and there are some channeling patterns that remain the same or become stronger weaker because of favorable unfavorable initial impact conditions TILT ROTATION ROT SUB combination If other substrate material is used say 4H SiC the simulation plane X Y plane in Figure 3 23 in ATHENA coincides with 4H SiC 1100 crystal plane in Figure 3 24 This is specified by ROT SUB 0 major flat simulation plane 7 1 gt pans e AWOOThS Pr t i o i Figure 3 24 4H SiC 1100 Crystal Plane If you want to specify the 1120 crystallographic plane as being the simulation X Y plane in ATHENA then set it to ROT SUB 90 MC Implant in ATHENA requires that ROT SUB should be always less than 90 Therefore you need to use other equivalent crystallographic planes for example 2110 which could be specified by ROT SUB 30 Amorphous Material Monte Carlo In the doping of semiconductors the rest distribution of the implantations is of principal importance The penetration of ions into amorphous targets is most simply described by using a Statistical Transport Model which is the solution of Transport Equations or Monte Carlo Simulation Among the two approaches Monte Carlo is more convenient for multiple components and two
615. the silicon substrate by the advancing interface The first parameter to tune is the fraction of consumed silicon atoms that are re injected back into the substrate as interstitials In ATHENA the related tuning parameter is called THETA 0 and is defined in the INTERSTITIAL statement THETA 0 has been found to be slightly different for wet and dry oxides The default value is reasonably accurate for dry oxides but some tuning may be required for wet oxidation The major effect of interstitial injection during gate oxidation is to create enhanced diffusion of the threshold adjust implant The measured threshold voltage of the final device is very sensitive to the dopant concentration near the silicon gate oxide interface Consequently threshold voltage measurements are a sensitive indicator of interstitial behavior Oxidation however is not the only source of interstitial injection The source drain and LDD implants also induce a large concentration of interstitials In order to isolate oxidation enhanced diffusion the threshold voltage of a long gate Silvaco 2 43 ATHENA User s Manual length device is used preferably where L 20 um or more so that the threshold voltage will be little influenced by damage near the source drain regions Interstitials injected by source drain implant damage can travel up to 10 um along the surface before recombination takes place A gate length of 20 um is recommended as the minimum gate length for
616. the timesteps during oxidation and uses additional equations is not present in the geometrical mode Therefore the DIFFUSION statements usually execute much faster when only the oxidation is being simulated This mode can be used to check the geometry generated by etching and deposition processes as well as the validity of mask steps Since oxidation still occurs oxide thicknesses as well as bird s beak shapes can be estimated But you should be aware that dopant enhanced oxidation effects are not taken into account in this mode 3 Coarse Grid Mode In this mode you can alter the number of grid points without changing the LINE statements It can be done by changing the Parameter Spacing factor in the ATHENA Mesh Initialize menu This will change the parameter SPACE MULT in the INITIALIZE statement The value of SPACE MULT is the amount by which the grid spacing specified in the ATHENA Mesh Define menu is multiplied A value for SPACE MULT that is greater than 1 0 will reduce the total number of grid points A SPACE MULT value that is less than 1 0 will create a finer mesh throughout the initial structure Reducing the number of grid nodes greatly increases speed You can still observe dopant diffusion in 2D and get valuable information about the accuracy of the input file before committing to the full simulation Each of these three fast modes of operation have the advantage They only require minor modification during mesh i
617. they react Therefore the probability of their sticking is considered You can define the sticking coefficient using STICK parameter in the RATE DEPO statement Random X coordinate Relaxation Figure 4 7 Deposition and Relaxation Model used in Ballistic Deposition model MONTE2 4 10 Silvaco ELITE Models The model uses an analytical approach to calculate a surface diffusion through a normalized gaussian distribution nd 2 nd exp _ 4 12 SIGMA DEP where x is the point of contact with the surface as shown in Figure 4 7 MONTE2 invokes a ballistic deposition model which simulates film growth by the random irreversible deposition of hard two dimensional discs launched with linear trajectories from a random point at the top of the simulation area towards the structure surface At the point of contact with the growing film the incident discs are relaxed to the nearest cradle point with the highest coordination number contacting the largest number of neighbor discs within a radius equal to SIGMA DEP which is four disc diameters by default The profile was initialized using a series of discs In order to inhibit unrealistic epitaxial growth from a closest packed surface 91 and 92 the initial series of discs was spaced with centers approximately 1 3 diameters apart This relaxation process simulates limited surface diffusion that occurs
618. ting wafer mount User definable materials added The capability to define new materials in SSUPREM4 has been included in this release This allows separate treatment of materials deposited using different processes Ramped DIFFUSION syntax change The RAMP parameter has been removed from the diffu sion statement If the parameter is present it will be ignored Temperature ramps for thermal diffu sions can now be specified by adding either the T FINAL or T RATE parameter to any DIFFUSION statement The initial temperature must be specified using the TEMPERATURE parameter Line continuation syntax change Line continuation is now supported in a manner consistent with use within DeckBuild The line continuation character for SSUPREM4 as well as other simula tors running under DeckBuild is backslash The character should be the last character on a line that is to be continued on the following line ETCH statement default change The TOP LAYER parameter on the ETCH statement defaults to true This parameter can be set to false to etch underlying material layers simultaneously with exposed layers of a particular material SSUPREM4 will now give information warning messages for etches that create voids within a structure In addition unexposed materials will not be etched unless TOP LAYER is set to false TonyPlot and go syntax supported The command TonyPlot can be included in a SSUPREM4 input deck and will initiate a ToNyPLoT of the str
619. tion displaying the results from the TWO DIM and FERMI models 3 62 Silvaco SSUPREM4 Models TonyPlot V2 6 9 Ir ll Files View Plot Tools Print Properties Help 7 si 5 E k 6 A Ey i L gt 0 OOF 008 O12 O16 02 O24 028 4 8 Depth Microns Depth Microns SILVACO International 1996 Figure 3 20 a Antimony Concentration Versus Depth b Corresponding Vacancy Concentration Versus Depth Silvaco 3 63 ATHENA User s Manual 3 4 Silicidation Model Silicide modeling capability is implemented in SSUPREM4 Silicides are formed when a metal reacts with silicon or polysilicon to create an intermediate phase The conductivity of silicides is typically orders of magnitude greater than that of highly doped n and p regions Modern CMOS technologies use silicides to reduce contact and interconnect resistances Also the use of SALICIDE technology self aligned silicides is a practical way to reduce poly gate resistance and source and drain sheet resistance Silicidation process is invoked by depositing refractory metal layers on the exposed silicon poly surface and then specifying a thermal cycle in the DIFFUSE statement There are four standard refractory metals in ATHENA Titanium Tungsten Platinum and Cobalt Corresponding silicides are called TISIX WSIX PTSIX and COSIX User defined metal and corresponding silicide can be also specified using parameters MTTYPE and
620. tion in the diffusion rate of the dopants Both interstitials and vacancies diffuse quickly compared with dopant ions The point defects also recombine as the implant damage is annealed out When it comes to tuning to match measured doping profiles two approaches are possible Either the damage during implant or the diffusion effect of the point defects could be used The amount of point defects generated during an implant is extremely difficult to measure Similarly the model parameters for both diffusion and recombination rates for point defects are uncertain All are areas of current academic research Typically the most effective tuning parameter in this type of simulation is the DAM FACT value itself Figure C 1 shows how fairly small changes in this parameter affect the doping profile A value of 0 01 is typical An Athena implant statement for an MOS source drain might be IMPLANT ARSENIC DOSE 3 0E15 ENERGY 60 UNIT DAMAGE DAM FACT 0 01 ATHENA Effect st Implant Damage Factor on N Source Drain Profile 2 ee pte Messer Be Bis Bis teat tatiana eee x Arsenic cm3 Ge J DAMAGE FACTOR oe 0 0 x 0 005 0 01 0 02 0 05 o 1 Figure C 1 Variations in diffusion due to tuning of DAM FACT parameter Figure C 2 illustrates how the damage produced by source drain implants affects the center of a MOS transistor with varying gate length For shorter gate length devices the damage at the source drain
621. tity oes Activation Energy actor Kr KTRAP 0 KTRAP E e FRAC O0 FRAC E Cr TOTAL The trap equation is either derived from the simple reaction I TS IT 3 30 or posed as a rate equation Cer ot K7C Cpr t K Cr Cer 3 31 where K is the trap emission rate In equilibrium the left hand side of Equation 3 31 must vanish giving 3 10 Silvaco SSUPREM4 Models A C oC xe KyCy Cony KCC SRS a 3 32 ioe G Substituting this value for K into Equation 3 31 leads to the expression in Equation 3 32 The recombination rate of 311 clusters R g11 in Equation 3 24 accounts for the release rate of 311 interstitial clusters which are small rod like defects that have been observed in Transmission Electron Microscopy TEM studies after medium to high dose implantation of impurities into silicon Since a large fraction if not all of the excess interstitials after implantation are believed to exist in this form the time scale for dissolution of 311 clusters plays an important role for the duration of any Transient Enhanced Diffusion TED Think of these volume defects as small pockets of interstitials distributed throughout certain parts of the doped regions which are released during annealing thus acting as bulk sources of point defects Note Actually 311 defects are believed to be created from excess free interstitials during the earliest part of the annealing cycle throu
622. to diffusion in silicon Impurities in the grain boundary diffuse more quickly as set by the GB DIX 0 and GB DIX E parameters on the IMPURITY statement The advanced polysilicon diffusion model is invoked by specifying the POLY DIFF parameter on the METHOD statement The METHOD POLY DIFF statement should precede the deposition of the polysilicon The CRYSTAL parameter on the IMPLANT statement is now true by default This parameter determines whether silicon materials will be treated with a full crystal representation during Monte Carlo ion implant calculations The previous default can be obtained by including CRYS TAL f on the IMPLANT statement Monte Carlo implant calculations will now take longer to per form due to the use of the more complete crystalline model The AMORPH parameter can now be used instead of CRYSTAL f to determine which model for Monte Carlo ion implant will be used Either AMORPH or CRYSTAL f can now be used to specify that statistics for amorphous silicon be used for analytic ion implant calculations The UNIT DAMAGE model now has a default value for a DAM FACTOR of 0 01 e Dynamic amorphization is now included in the Monte Carlo ion implant capability This models the amorphization that takes place during implantation e The MATERIAL statement includes the boolean parameter DAM THRESH that specifies the implant damage threshold in eV This can be used to control the extent of amorphization that occurs during implant The par
623. to give overall agreement with the available experimental data The figure also demonstrates that there can be a possible disagreement with individual set of measurements Similar stopping powers validations were performed for other important materials The accuracy of the calculated ranges in ATHENA is within 10 for majority of ion material combinations which is close to the best possible achievements of today s theory of stopping and ranges 3 90 Silvaco SSUPREM4 Models 3 6 Deposition Models A deposition step is simulated by the DEPOSIT statement where the material and the thickness THICKNESS parameter of the layer to be deposited must be specified The deposited layer is constructed by a simple algorithm that describes conformal deposition In this algorithm the whole layer is divided into a number of sublayers with thicknesses equal to grid spacings calculated according to the grid control algorithm see Section 3 6 2 Grid Control During Deposit Each sublayer is deposited and triangulated separately More complete physically based models for deposition are available in the ELITE module as described in Chapter 4 ELITE Models 3 6 1 Deposition of Doped Layers You can add the uniform or graded concentration of impurities or defects or both to each node of the deposited material by using the C BORON F BORON C INTERST parameters in the DEPOSIT statement 3 6 2 Grid Control During Deposit You
624. to non uniform rectangular elements defined by the LINE X and LINE Y statements By default each box element is then divided into two rectangular triangles by a diagonal going from the bottom left to upper right corner of the box In some applications it is preferable to have a symmetrical grid triangle orientation locally One of the examples is etching of a non vertical trench If TRI RIGHT default is specified all boxes between this line and the line with TRI LEFT or the last line will be divided by a bottom left to upper right diagonal If TRI LEFT is specified the boxes will be divided by a diagonal going from upper left to bottom right corner of the box 4 H Example In the following specifications there are 3 user specified x lines and 2 user specified y lines Spacing of the x lines is finer in the center than at the edges After processing ATHENA produces a mesh with x lines at 0 0 0 42 0 69 0 88 1 0 1 12 1 31 1 58 2 0 Around the center the spacing is 0 12 approximately what was requested At the edge the spacing is 0 42 because that was as coarse as the line spacing could get without having an interval ratio greater than 1 5 If the interval ratio is set to 9 then we would have one interval of 0 9 and one interval of 0 1 on each side In this example specifying a spacing of 1 would produce an x line at 0 0 and 1 0
625. tride viscosity depends on the oxidation temperature as well You may use the parameter VISC E when the temperature dependence of the oxide shape is considered 7 Model parameters VC VR and VD see Equations 3 147 3 149 can also be used for tuning Default parameter values are reasonable for temperatures of 1000 C and higher For several test structures the alternative set of these parameter values VC 300 VD 60 and VR 12 5 are more appropriate for lower temperatures 950 C Silvaco C 3 ATHENA User s Manual Question How can a self aligned silicide process be modeled in SSUPREM4 Are there any special model parameters required Answer The formation of metal silicides can be simulated using the optional silicide module in SSUPREM4 In a typical self aligned silicide salicide process the goal is to form a silicide layer on the polysilicon gate and MOS source drain regions The silicide layer in the source and drain regions permits device designs with shallow junctions that still have low n or p sheet resistances On the gate the silicide layer forms a low resistance interconnect The process is self aligned since the oxide spacer on the gate sidewall is used to prevent the silicide shorting gate to drain The usual sequence for salicide is to deposit a refractory metal layer Commonly used metals are titanium tungsten and platinum Then a short fairly low temperature heat cycle is applied to react the me
626. ture Manipulation Tools yikes ea sce ed Shee heated eek Det ee bccn 2 56 2 8 2 Deposition and Wet Dry Etching using the Physical Models in ATHENAVELITE 0 05 2 59 2 8 3 MaskViews Interface 0 02 enindt pees yee toed ebeeede adit ved bewet ines eed eee eee bee xs 2 65 2 9 Using ATHENA OPTOLITH s22siiessiee scence stare ee hee dee eee Eanes Se eee eae ss 2 74 29 VWVOVEWIOW fois niki cele es tele eee eu leet pete ste ber tai hil eee beets cts 2 74 2 92 Creating A Maska ede tur letigem nee eatin cutee el yee elare emit ead REEDE 2 74 29 3 Illumination System sess seen e wows ee ehive dhe alee EEEE EE 2 77 2 9 4 The Projection SISIOM cence carte adei cnr EE een Meath Mune ee h r eae Rw AL O dee ty 2 79 2 9 5 IMAGING CONO lma tars penn k etal NS ou tania Aa e AE AE Uh eR ete eta Ree nk Seat At Sed 2 80 2 9 6 Defining Material Properties 2 oo Sotto teat eh ee So ee ih Re Seg 2 82 27S OUMIOTUNE EXPOSURE citi tae oN Ged tact Shee A Maat ciate cial ATi ola Hie E te a Ua atl 2 82 2 9 8 CD Extraction Smile Plots And Looping Procedures 0c e cece eee eens 2 84 2 10 Adaptive MESHING oir ena rnai na e a wle aetna Was w apa es aie aad A toate wan 2 86 2 10 1 Introduction to Mesh Adaption ite inte tela eden t ote cotta ea aed teh eae Rens 2 86 210 2 Interface Mesh Control mie ke as cin ae cee need Pets OO E ace EERE EEN che area ae ef 2 89 Chapter 3 SSUPREM4 Models oac 2 21 50 guava EEE ees etter oe prea t er
627. ty conforming to an estimated profile Graphical tools such as DEVEDIT can make this easier But it can t totally eliminate the process With the Adaptive Meshing module you can overcome these difficulties to a larger extent The program uses an iterative algorithm to determine the required mesh density distribution to accurately conform to the implanted profile and will automatically generate the additional required mesh The algorithm is illustrated with the flow chart depicted in Figure 2 67 Previous Mesh Dopant Implant Mesh Adaption Clear Dopant Increase Adaptive Meshing critera Accurate Implanted Profile Distribution Continue Figure 2 67 Flow Chart of Mesh Adaption Algorithm 2 86 Silvaco Tutorial Adaption During A Heat Cycle During the diffusion oxidation epitaxy processes impurity profiles are usually changing continually with each elapsed time step An initially generated optimal mesh will not conform to the time varying dopant profile If the impurity profiles change substantially during the process the mesh density distribution will be different from the dopant contour distribution causing both accuracy and speed problems During simulation the total time of a diffusion oxidation epitaxy process is usually divided into many small time steps with profiles changing gradually between time steps Using the Adaptive Meshing module you can perform a mesh adaption after each time st
628. ual Pearson moments selection from the implant tables It s up to you to select its value accordingly The effect of this parameter is that it represents ion implantation through a thin 0 50nm surface oxide layer The present algorithm in ATHENA when encountering a multi layered structure see Section 3 5 2 Multi Layer Implants For example oxide silicon switches to multi material scaling technique for evaluating the depth profile This technique will combine two profiles single Pearson for the oxide and dual Pearson for silicon with S OXIDE preferably set to the thickness of the oxide There are two reasons why this separation between the surface oxide is present in the structure before the IMPLANT statement and the S OXIDE parameter The first reason is because the flexibility of using this parameter for different thin surface layers other than oxide with appropriate scaling of their thickness for stopping The second reason the currently restricted availability of moments with screen oxide in the tables 0 50nm 15 80keV and for boron only If you need a more precise dependence of the implantation profiles on the surface screen oxide use a single layer of silicon with S OXIDE set to an appropriate value 3 5 2 Multi Layer Implants To apply any of the described analytical distribution functions for structures that are comprised from several different material layers use a scaling method that s mentioned in this section
629. ucture if run under DeckBuild For SSUPREM4 standalone operation the TonyPlot statement is ignored Also under DeckBuild the command GO SSUPREM4 will initiate SSUPREM4 execution This statement is ignored in standalone operation Manual improvements and additional examples The manual for SSUPREM4 has been refor matted and thoroughly revised to be more readable and provide the user with more important guidelines for effective use of SSUPREM4 The Tutorial section and Getting Started sections have been added to provide an introduction to the use of SSUPREM4 Additional examples detail the use of the user defined material capability bipolar device fabrication and EEPROM device fabrication D 19 SSUPREM4 Version 5 0 Version 5 0 of SSUPREM4 represents a new standard for 2D process simulation SSUPREM4 Version 5 0 incorporates a number of new models and convenience features briefly described in this chapter One dimensional mode Version 5 0 offers a significant enhancement for speed and ease of use by incorporating a one dimensional 1D mode This may be specified within a conventional two dimensional 2D input deck This allows fast analyses of particular points in a 2D structure prior to complete 2D analysis with the same input deck The use of this feature is described in the INITIAL IZE statement description Analytic angled implant The implant capabilities of SSUPREM4 have been enhanced by the inclusion of analytic angled implant mode
630. ular feature Z HIGH specifies the maximum z coordinate of the rectangular feature X TRI specifies the x coordinate of the right angle corner of the triangular feature Z TRI specifies the z coordinate of the right angle corner of the triangular feature HEIGHT specifies the height of the right angle triangle feature WIDTH specifies the base width of the right angle triangle feature ROT ANGLE specifies the angle of rotation of the feature 180 lt ROT ANGLE lt 180 with respect to the x axis The default value is 0 The center of rotation is at the center of the rectangle and at the right angle corner of the triangle respectively Note You can only use the X TRI Z TRI HEIGHT WIDTH and ROT ANGLE parameters for projection printing model X CIRCLE specifies the x coordinate of the center of the circular or ring feature Z CIRCLE specifies the z coordinate of the center of the circular or ring feature RADIUS specifies the radius of the circular feature or the outermost radius of the ring or multiring feature RINGWIDTH specifies the width of ring s of the mask feature shaped as a single ring or a series of concentrated rings MULTIRING specifies that mask feature consists from series of concentered rings with ring widths and distances between rings specified by the RINGWIDTH parameter The number N of the rings in the feature should satisfy the following relation 2N RINGWIDTH lt RADIUS
631. ulation including 2D aerial imag ing non planar photoresist exposure and post exposure bake and development 1 1 1 Using This Manual This chapter is an overview of ATHENA For new users read Chapter 2 Tutorial especially the sections that describe the simulator or modules that you have licensed This chapter will give you a basic understanding of what ATHENA can do and how it s used The remaining chapters will give you a detailed understanding of ATHENA s capabilities and how to access them Appendix D ATHENA Version History gives information about the current version of ATHENA ATHENA is supplied with a number of example problem descriptions You can access them through DECKBUILD as described in the VWF INTERACTIVE TOOLS USER S MANUAL VOL 1 These examples demonstrate the capabilities of ATHENA The input files provided as part of these examples can provide an excellent starting point for developing your own ATHENA input files 1 1 2 Technical Support If you have difficulties or questions relating using ATHENA e mail SiLvAco Support at support silvaco com When you send us an e mail message please 1 Explain the problem or question in detail 2 Include any input files that you have created 3 Provide us with the version number of ATHENA and the version numbers of the VWF INTERACTIVE TooLs that you are using 4 Include your business telephone number and fax number SILVACO support will contact you promptl
632. umination source shape and illumination source filtering in OPTOLITH Syntax ILLUM FILTER CIRCLE SQUARE GAUSSIAN ANTIGAUSS SHRINC GAMMA lt n gt RADIUS lt n gt ANGLE lt n gt SIGMA lt n gt IN RADIUS lt n gt OUT RADIUS lt n gt PHASE lt n gt TRANSMIT lt n gt CLEAR FIL Description This statement specifies the illumination source options as well as illumination source filtering CIRCLE SQUARE GAUSSIAN ANTIGAUSS and SHRINC define or change the shape of the exit pupil of the illumination system SHRINC can be used to define the illumination system only not annular filters GAMMA defines or changes the GAMMA value for GAUSSIAN or ANTIGAUSS source transmittance GAMMA is a parameter that defines the truncation of the GAUSSIAN by the pupil In the limit of GAMMA gt 0 the source will be uniform RADIUS specifies the radius of a single source if you define the SHRINC illuminator concept This parameter must be entered in fractions of unity ANGLE specifies the angular location for the SHRINC illuminator SIGMA defines or changes the filling factor for the combination of the illumination and projection systems The value of SIGMA is expected to vary but it will not be reset Also specifies the radius of a single source if you specify the SHRINC illuminator concept This parameter must be entered in fractions of unity assuming a unit pupil radius IN RADIUS and OUT RADIUS def
633. upled from impurity diffusion for the TWO DIM model e A new parameter has been added to the OXIDE statement It is called SPLIT ANGLE It governs the minimum angle at which the oxide will split open one more grid spacing when oxidizing at a tri ple point i e where silicon oxide and nitride coincide together at a point The default for the split angle is 22 5 degrees The SPLIT ANGLE parameter for triple point oxidation is material depen dent Specify the oxidizing material without a and the second material with a using the follow ing format OXIDE SPLIT ANGLE 35 SILICON NITRIDE There are only three possible combinations SILICON NITRIDE SILICON POLY and POLY NITRIDE e Anew parameter for scaling analytic implants has been added to the MATERIAL statement A mul tiplicative factor IMPL SCALE is specified on the MATERIAL statement along with the material name in which the implant is to be scaled An example format would be MATERIAL IMP L SCALE 0 5 PHOTORESIST This scales the implant RANGE STD DEV SRANGE and SSTD DEV parameters with this factor when they are take from the implant moments file athenaimp This is intended to be a convenient way to modify these tables with a constant multiplicative factor Monte Carlo Implant Capabilities e Secondary recoil in Monte Carlo implantation model has been implemented The model is invoked by specifying REC FRAC lt number gt together with the
634. ure and then stretch the active or non active or both areas to the actual widths This will also save a tremendous amount of simulation time TonyPlot 2 2 1 File vj i View F C Plot 7 Tools 7 Print 7 Properties 7 Help 7 ATHENA Structure before stretch EEEF E a a Fa Ea a E P Fa a PP ry T F J J J J EI EJ EJ EJ EI FJ lJ Pz a Ei Fa Fa a Fa Fa a Fa Fa Fa lef i A a LI l E l 1 Ly cy cy ct Fa a A AANT AE RARER oe LS A A aS NN A aa AVIAN AAA 1 VV WAA AAA OAL WANA PRE REREE ARERR BRRERBERER BER EE PF Pl Pf PP PFJ fff rs ra rT e J PF PFJ PJ FF ESS VA AAA aa TN BR RERBEEES REESE lJ FJ FJ FJ J Pf r A T E f r F WACO International Figure 2 40 Using Stretch Function for a MOSFET Structure Note The stretch function can save a great deal of CPU time Silvaco 2 57 ATHENA User s Manual Using ATHENA In 1D Mode You can increase the simulation speed by running ATHENA in 1D mode ATHENA automatically runs in 1D mode by default initially The simulation will automatically converts to 2D mode when you perform a two dimensional simulation process such as ETCH or EXPOSE Simple operations
635. uring Silane Epitaxy J Electrochem Soc v 124 p 591 1977 124 G L Vick and M Whittle Solid Solubility and Diffusion Coefficients of Boron in Silicon J Electrochem Soc v 116 p 1142 1969 125 F A Trumbore Solid Solubilities of Impurity Elements in Germanium and Silicon Bell System Tech J v 39 p 205 1960 126 H Park and M Law Point Defect Based Modeling of Low Dose Silicon Implant Damage and Oxidation Effects on Phosphorus and Boron Diffusion in Silicon J of Appl Phys v 72 p 3481 1992 127 F J Morin and J P Maita Electrical Properties of Silicon Containing Arsenic and Boron Phys Rev v 96 p 28 1954 128 M Giles Transient Phosphorus Diffusion Below the Amorphization Threshold J Electrochem Soc v 138 p 1160 1991 129 P Gas et al Diffusion of Sb Ga Ge and As in TiSi2 J Appl Phys v 63 p 5335 1988 130 P Gas et al Boron Phosporus and Arsenic Diffusion in TiSi2 J Appl Phys v 60 p 1634 1986 131 C M Osburn et al The Effect of Titanium Silicide Formation on dopant Redistribution J Electrochem Soc v 135 p 1490 1988 132 Chi On Chui et al Activation and Diffusion Studies of Ion implanted p and n Dopants in Germanium Appl Phys Lett v 83 p 3275 2003 133 8 Uppal Diffusion of Ion implanted Boron in germanium J Appl Phys v 90 p 4293 2001 BIB 6 Silvaco Index A Beam Propagation Method
636. using with double negative vacancies DVMM 0 is the pre exponential constant DVMM E is the activation energy DVP 0 and DVP E specify the impurity diffusing with single positive vacancies DVP 0O is the pre exponential constant DVP E is the activation energy DVPP 0 and DVPP E specify the impurity diffusing with double positive vacancies DVPP 0 is the pre exponential constant DVPP E is the activation energy FI O and FI E are the fractional interstitialey parameters that determine whether the impurity diffuses through interaction with interstitials or vacancies Once the expression for total FI is evaluated from these coefficients the value of total FI can vary between 0 and 1 FI equal to 1 corresponds to a pure interstitial based diffusion while value of 0 corresponds to a pure vacancy mechanism Activation Model Parameters SOL SOLUB specifies that solid solubility model and solid solubility tables will be used for calculation of active concentration of the specified impurity in the specified material This is default for all cases except Arsenic in Si and Poly CLUSTER ACT specifies that cluster activation will be used It is default only for AS in Si and Poly SS CLEAR SS TEMP and SS CONC are the parameters for solid solubility data SS CLEAR clears the currently stored solid solubility data for the specified impurity in the specified material SS TEMP and SS CONC add a single temperature and an associated
637. usion Model Set in the most recent METHOD statement If you need another METHOD statement include it before the EPITAXY statement Deckbuild ATHENA Epitaxy Display Time Temp Thickness Grid Impurities Time temperature Time minutes 30 0 500 Temperature C 900 End temperature C 1000 Temperature rate C min 3 333 500 e Rate Variable Thickness rate Thickness um 2 00 0 00 10 00 Deposit rate um min 0 0667 Grid specification Total number of grid layers settings vv Grid spacing location um 0 00 Minimum grid spacing umh Dei Amtimony iC Arse nig Boron fC Phosphorus Silicon 1 Zine ie Selealumn 1 6 Beryllium iC Magnes iii Aluminum 1 6 Gallias LC Carbom Chromium 1 Germanium fE indium 1 Temp Constant OO F 1300 Ramped 20 vv Nominal grid spacing um 0 10 0 00 lt jH 1 00 1 00 L t LG Comment Epi Layer WRITE Figure 2 32 ATHENA Epitaxy Menu Silvaco 2 41 ATHENA User s Manual 2 5 Calibrating ATHENA for a Typical MOSFET Flow This section of the manual provides information on which parameters should be tuned in the input file to provide predictive simulations using a typical MOSFET process flow We assume you are now familiar with the mechanics of making an input file and using the correct methods and models see Section 2 4 Choosing Models In SSUPREM4 For example
638. usions anything greater than GLOOP EMIN if the extrusion is a single triangle error in the boundary The default value is GLOOP EMIN 130 Neither of these parameters should be less than 90 because the rectangular edges of the simulation space would be smoothed OXIDE EARLY OXIDE LATE and OXIDE REL should not normally be modified They relate to internal numerical mechanisms and are described here only for the sake of completeness A node whose spacing decreases proportionally by more than OXIDE LATE is marked for removal Also if any nodes are removed then all nodes greater than OXIDE EARLY are also removed For earlier node removal fewer obtuse triangles try OXIDE LATE 0 3 and OXIDE EARLY 0 1 Though not logical it is harmless for OXIDE EARLY to be greater than OXIDE LATE The OXIDE REL parameter is the percentage error in velocities for the non linear viscous model The default is 1 0x107 that is a 1 0 percent error OXIDE REL can be increased for a faster solution OX OBFIX specifies the cosine squared of the worst angle allowed during oxidation FILL and PERIMETER specify which action to apply to voids that may form during oxidation FILL specifies that you must fill the voids with oxide materials Default is false PERIMETER specifies the maximum perimeter of the voids to fill Default is 0 2 microns
639. usually part of the same molecule In chemically amplified resists the reaction kinetics are more complicated The inhibitor concentration still however is considered to be the key quantity for the development process In positive tone Novolac resists to determine the inhibitor concentration from exposure simulations use Dill s model as previously described This model applies when the resist material undergoes a transition between two chemical states during the exposure step The actual development process is treated as a surface limited etching process which is dependent on the particular resist developer chemistry and on the local concentration of the dissolution inhibitor at the surface of the resist that has been decomposed to a degree during the exposure step If the resist developer chemistry is held constant the dissolution rate is assumed to be a function of the inhibitor concentration only The rate function r x y is determined experimentally and usually fitted by an empirical function to experimental development rate data as a function of the remaining PAC concentration M x y You can use one of the following models to simulate the development process for the specific resist developer combination e Dill e Kim e Mack e Trefonas e Hirai Each model assumes a specific rate function type to describe the rate inhibitor concentration relation These models are described in the following sections 5 6 1 Dill s Development
640. ven an HCl percentage a look up table is used to determine an enhancement factor for the parabolic rate constant Figure 3 14 shows the SiO thickness dependency on HCl percentage Silvaco 3 57 ATHENA User s Manual 3 3 6 Mixed Ambient Oxidation In practice an oxidizing ambient may be a gas mixture consisting of more than one oxidant and other impurities The total oxidation rate will be the combined effect of all these species To simulate oxidation under a multi gas ambient SSUPREM4 simultaneously calculates the diffusion and oxidation of several ambient gases The capability is invoked by specifying the gas flow parameters F 02 F H2 F H20 F N2 and F HCL on the DIFFUSE statement From the gas flow the partial pressure of each gas is calculated as F 2n where P and F are partial pressure and gas flow rate for the jt gas respectively and P otal 18 the total pressure of the gas mixture specified by the PRESSURE parameter on the DIFFUSE statement P J Prota Z priog If only one oxidant gas is specified in the gas flow i e only Os or HO with other gases oxidation is then modeled as previously described Equation 3 169 determines the pressure of the oxidant gas If both F H2 and F O2 are specified the reaction of Hy and Os to form H3O is assumed to occur The partial pressure of H O is then calculated before solving the oxidation equations For ambients containing more than
641. wing documentation is provided for each statement e The statement name e A list of all of the parameters of the statement and their type e A description of each parameter or group of similar parameters e An example of the correct usage of each statement The ATHENA command language encompassed by this document describes each of the modules of ATHENA namely ELITE OPTOLITH SSUPREM4 and their submodules Depending on which of the ATHENA modules have been purchased some of the capabilities described may not be available as part of the ATHENA installation Note You can print a summary of statement names and parameters by using the HELP statement The following list classify ATHENA statements and provide their brief description and use 6 2 1 Structure and Grid Initialization Statements These statements define the dimensions boundary conditions grid density and material type of the initial structure Typically only LINE and INITIALIZE statements are required e BASE MESH specifies parameters of the base mesh used for initial grid generation e BOUNDARY specifies which lines in a rectangular grid are exposed to gas e INITIALIZE sets up the initial grid and specifies background doping concentrations and mate rial type e LINE specifies the positioning of x and y grid lines for a rectangular mesh e REGION specifies corresponding sections of the rectangular mesh and material 6 2 2 Structure and Mesh Manipulation Statements Th
642. y USLE EEL L cL fc DE R EN fsck LOL aE DES EN N RAL ER Ea eck E a oa fo 1 2 3 4 5 6 7 8 9 10 Gate Length um Click to place P changes alignment or drag to get leader SILVACO International 1996 Figure 2 35 How Changing the clust fact parameter affects the threshold voltage The second implantation parameter that can now be tuned is the lateral spread of the implant near the surface In ATHENA this parameter is called LAT RATIO1 and is defined in the IMPLANT statement The lateral spread of the source drain and LDD dopant is responsible for the classical short channel effect where the threshold voltage reduces for very short channel lengths Simply tune the LAT RATIO1 parameter until the onset of classical short channel effects of simulated and measured data correspond If the LAT RATIO1 is increased the onset of the classical short channel effect will occur for longer gate lengths Silvaco 2 45 ATHENA User s Manual 2 5 4 Tuning Diffusion Parameters The final part of the threshold voltage versus gate length curve can now be used to tune the surface recombination rate of interstitials In ATHENA this parameter is called KSURF 0 and is specified in the INTERSTITIAL statement The surface recombination of interstitials will dictate the roll off rate of threshold voltage from its peak value reverse short channel effect to the long gate length value Once again simply tune KSURF 0 until the long channel threshold v
643. y and resolve your problem as quickly as possible User feedback helps further develop ATHENA Please send your comments on the programs suggestions for improvements and additional feature requests to support silvaco com Silvaco 1 1 ATHENA User s Manual 1 2 Athena Features and Capabilities Table 1 1 shows the features and capabilities of Athena Table 1 1 Athena Features and Capabilities Features Capabilities Bake e Time and temperature bake specification e Models photoresist material flow e Models photo active compound diffusion C Intepreter e Allows user defined models for implant damage Monte Carlo plasma etching and diffusion in SiGeC CMP e Models Chemical Mechanical Polishing e Hard and soft models or a combination of both e Includes isotropical etch component Deposition e Conformal deposition model e Hemispherical planetary and conical metallization models e Unidirectional or dual directional deposition models e CVD model e Surface diffusion migration effects e Ballistic deposition models including atomistic positioning effects e User definable models e Default deposition machine definitions Development e Five different photoresist development models Diffusion e Impurity diffusion in general 2D structures including diffusion in all material layers e Fully coupled point defect diffusion model e Oxidation enhanced retarded diffusion effects e Rap
644. y of default photoresists has been extended with the inclusion of more resists and param eters describing the new models D 14 5 FLASH Module e The FLASH module has been introduced with this release of ATHENA The FLASH module provides the ability to model gallium arsenide materials This involves a number of changes A partial list of the FLASH capabilities is provided here for reference e GaAs material is now included on the INITIALIZE and DEPOSIT statements as well as a number of model coefficient statements e Impurities appropriate for GaAs processing namely beryllium chrome germanium magnesium selenium silicon and zinc have been added to a number of statements e Ion implantation moments tables have been added that describe implant of these species into mate rials typical of GaAs processing e Monte Carlo ion implant capabilities have been extended to accommodate the new impurities and GaAs material including crystal effects e A diffusion model for impurities in GaAs has been included This model can be accessed by specify ing the DIFFUSE statement D 15 ATHENA Version 1 0 e Version 1 0 incorporates a number of new models as well as convenience features The maximum number of grid points has been increased to 20000 Dynamic allocation of critical arrays makes this limit practical A slight slowdown while dynamic allocation is being performed may be observed during execution of INITIAL statements If the grid definiti
645. y of oxide elements and u is the oxide viscosity The oxide viscosity is calculated from the following equation YOUNG M e 3 141 2 2 POISS R where YOUNG M is Young s modulus which is specified in the MATERIAL statement and POISS Ris Poisson s ratio which is specified in the MATERIAL statement The oxide flow is treated as an incompressible fluid By doing this it implies the density of the oxide is constant with respect to time Applying this fact to the mass continuity equation the incompressibility condition is given as V V 0 3 142 The incompressibility condition in Equation 3 142 is implemented by allowing a slight compressibility of the flowing oxide Thus Equation 3 142 is modified to give the following equation u The solution of Equation 3 143 at each time step gives the velocity field of the flowing oxide elements The Compress Model is recommended for simulations of planar and non planar structures where stress effects play a minor role in determining the oxide shape When stress effects are important you can use the Viscous oxidation model Figure 3 10 shows a two dimensional cross section of the structure resulting from a LOCOS oxidation using the Compress Model Silvaco 3 47 ATHENA User s Manual TonyPlot 2 8 18 A Files View Plot Tools Print Properties Help ATHENA Compress Model Materials SiO2 Silicon Si3N4 1 0 8 06 04 02 Midons 02 04 0 6 Figure 3
646. y segregation between grain interior and grain boundaries g C b a i le 3 69 i Pseg where Pseg is segregation coefficient and q is the rate of segregation specified as PD TAU in the IMPURITY statement Initial conditions for Equations 3 66 and 3 67 are determined by setting the PD CRATIO parameter in the IMPURITY statement This parameter specifies the initial ratio between impurity concentration in the grain boundary cz and total concentration Cf biy CE The grain boundary segregation is calculated according the model suggested in 17 Q seg LANs kT Silvaco 3 21 ATHENA User s Manual where Q is the density of segregation sites at the grain boundary specified by the PD SEGSITES 022 parameter in the IMPURITY statement Ng is the atomic density of crystalline silicon 2 5 10 atoms cm A is the entropy factor specified by the PD EFACT parameter in the IMPURITY statement Qo is the segregation activation energy specified by the PD SEG E parameter in the IMPURITY statement and L t is the time dependent grain size calculated according to the grain growth model suggested in 16 iz PDGROWTH O _ _PD GROWTH E Et fe i kT A E i where Lo is the initial polysilicon grain size which should be specified by GR SIZE parameter in the DEPOSIT POLYSILICON statement GR SIZE F parameter allows to have linearly graded grain size within deposited polysilicon layer
647. y to load mask information directly from MAsKViEws layout file for image calculations in OPTOLITH This capability has several advantages when comparing with old interface through a special section file The section file approximates an arbitrary shaped mask features with only rectangulars The new interface doesn t do any approximations and internally divides mask polygons into triangles and rectangulars for exact image calculations Considerable speeded up and improved accuracy of image calculations for big area mask layouts when complex geometry light sources are used Added additional standard wavelengths to the ILLUMINATION statement KRF LASER alias is DUV LINE ARF LASER and F2 LASER D 2 Silvaco ATHENA Version History 5 Added capability to save intensity and mask information separately into the structure file after OPTOLITH image calculations Now the MASK parameter in the STRUCTURE statement will save only mask layout information The INTENSITY parameter now saves only intensity distribution When you specify both parameters both mask layout and intensity will be saved D 3 ATHENA Version 5 10 7 R Release Notes 1 Additional TSUPREM4 compatibility feature is implemented If you set the E FIELD parameter to FALSE in the MATERIAL statement the electrical term will be ignored during diffusion simulation in the specified material Added the boolean parameter CENTER to
648. ymp Proc Si Front end processingphysics and technology of dopant defect interactions lll edited by E C Jones K S Jones M D Giles P Stolk J Matsuo vol 669 no pp J5 6 2001 N Cowern G Mannino P A Stolk F Roozeboom H G A Huizing J G M van Berkum F Cristiano A Claverie and M Jaraiz Energetics of self interstitial clusters in si Phys Rev Lett vol 82 no 22 p 4460 1999 Silvaco BIB 1 ATHENA User s Manual 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 T J Lenosky B Sadigh S K Theiss M J Caturla and T D de la Rubia Ab initio energetics of boron interstitial clusters in crystalline si Appl Phys Lett vol 77 no 12 p 1834 2000 P H Keys Phosphorus defect interactions during thermal annealing of ion implanted silicon Ph d thesis University of Florida 2001 A H Gencer Modelling and simulation of transient enhanced diusion based on interaction of point and extended defects Ph d thesis Boston University 1999 L Pelaz M Jaraiz G Gilmer H Gossmann C Raerty D Eaglesham and J M Poate B diusion and clustering in ion implanted si The role of b cluster precursors Appl Phys Lett vol 70 no 17 p 2285 1997 W Orr Arienzo R Glang R F Lever R K Lewis and F F Morehead Boron diusion in silicon at high concentrations J Appl Phys

ATHENA

Contents

Download Pdf Manuals

Related Search

Related Contents