Home

a methodology for tribological examination of

Contents

1. Z SAFETY IS DISABLED TAKE CARE Use the green Z button to move the tip down to approximately 1 millimeter above the sample Click OK when Finished Coarse Medium Fine Ultra Fine Figure 4 7 Manual height adjustment window From this window the indenter tip is lowered to approximately 1 mm from the sample surface The machine then slowly approaches the sample surface until contact is made and the low load indent is executed The sample is then translated back to the optical CCD microscope and the window in Figure 4 8 pops up Calibrating the Optics E Z SAFETY IS DISABLED TAKE CARE Use the Red X and Blue Y buttons to locate the indent pattern OR Click the Auto Search button to locate the indent pattern Once the indent pattern is located Put the cross hairs over the single dot or if H indent calibration pattern the dot between the 2 legs of the H pattern Click Ok when Finsihed sa Figure 4 8 Search option window At this point no translation is made even though the machine prompts the user to locate the pattern This set is used only to set the z axis offset The x and y offset is set 41 in following steps The ok button is pressed without any x or y translation Now from the Triboscan toolbar the imaging mode is selected The windows in Figure 4 9 will open Initial scan parameters of 50 80 um scan size 2 uN set point and 1 Hz
2. 89 90 As mentioned earlier the Oliver and Pharr model overestimates hardness of materials that exhibit significant plastic flow The Oliver and Pharr model assumes no pile up For indentations on materials that exhibit large plastic flow resulting in pile up the load is supported by a portion of the material that has been exuded Oliver and Pharr s assumption of no pile up results in an underestimated contact area Since hardness is calculated from H P max A where H is the hardness Pmax is the maximum normal load and A is the area underestimating the area will result in overestimating the hardness Indentation tests analyzed with the O P model results show an increase in hardness for each sample below 15 nm in maximum displacement This result could arise from an inaccurate tip area function for these low displacements or from excessive scatter observed at indentation tests performed below 15 nm maximum displacement The exceptionally high hardness shown by the EBE 20 sample 27 nm thick is more representative of hardness values for silicon 100 43 which is due to the maximum displacement for the tests being well beyond 10 of the film thickness Indentation well below 15 nm maximum displacement is required to examine the hardness of the gold film in sample EBE 20 without a substrate contribution Although the manufacturer claims resolution of 0 2 nm displacement on the Hysitron Triboindenter special tips are required to achieve repeat
3. 2 5 at 38 2 20 and 2 5 at 81 7 20 which correspond to the 111 and 79 222 respectively To increase resolution the step size was set at 0 05 The data was entered in the Philips Line Profile software along with scans from standard samples with similar d spacing and the Warren Averbach grain size and strain analysis was performed Figure 6 10 shows results of the film grain size and strain with respect to the film thickness The grain size is increasing with film thickness up to 45 nm grains in films over 300 nm thick All films show 0 1 compressive strain except the sample designated EBE20 with a film thickness of 27 nm and 0 22 compressive strain 50 0 25 s 45 m E E 40 a 0 2 a c be a 30 0 15 25 20 0 1 a L 15 10 0 05 5 0 0 0 100 200 300 400 500 600 0 100 200 300 400 500 600 thickness nm thickness nm Figure 6 10 Plots of grain size and strain versus gold film thickness Scanning Electron Microscopy Samples were imaged to examine wear tracks and determine film failure in an FEI DB235 field emission gun FEG microscope by first locating the test area from the grid lines that were milled prior to testing Once the area was located magnification was increased and the sample was oriented so that the direction of wear was along the horizontal or x axis The sample was then tilted about that axis 52 to provide better perspective Test areas were imaged a
4. crystallographic orientation of the tip is characterized see Figure 6 6 Figure 6 7 shows a schematic representation of the crystallographic orientations of the indenter tip and the direction of reciprocation in the wear tests Once the crystallographic orientation of the indenter tip is characterized elastic properties of the tip can be calculated from the stiffness matrix Single crystals are often elastically anisotropic and therefore the Young s modulus and Poisson ratio are orientation dependent 37 ISG RS SQN ca lt 5 lt BEN 7 Ll Figure 6 6 Schematic representation of the Kikuchi map observed in diffraction mode indexed such that the 100 plane is orthogonal to the tip loading axis The major zone axes are labeled with the sample holder tilt angle readings 010 sliding direction 001 Figure 6 7 Schematic representation of a top down view of the single crystal diamond indenter tip with crystallographic orientations and sliding direction indicated Note the loading axis of the indenter corresponds to the 100 direction as indexed in the image 74 The orientation dependent Young s modulus of a cubic single crystal is given by 1 6 1 1 2 S Sa 5 Sum mn n where Eww is the Young s modulus for the direction of interest S11 S12 and S44 are the coefficients of the compliance matrix and m and n are the direction cosines of the uvw with respect to
5. Ling 2002 New York Springer Verlag 197 Doerner M and W D Nix A method for interpreting the data from depth sensing indentation instruments Journal Of Materials Research 1986 1 4 p 601 Oliver W C and G M Pharr An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments Journal Of Materials Research 1992 7 6 p 1564 1583 Sneddon I N The relation between load and penetration in the axisymmetric boussinesq problem for a punch of arbitrary profile International Journal of Engineering Science 1965 3 1 p 47 Pharr G M W C Oliver and F R Brotzen On the generality of the relationship among contact stiffness contact area and elastic modulus during indentation Journal Of Materials Research 1992 7 3 p 613 617 28 29 30 31 32 33 34 35 36 37 38 39 40 129 Wyrobek T J Triboindenter User s Manual Minneapolis MN Hysitron Inc Stone D S K B Yoder and W D Sproul Hardness and elastic modulus of tin based on continuous indentation technique and new correlation Journal of Vacuum Science and Technology A 1991 9 4 p 2543 2547 Joslin D L and W C Oliver A new method for analyzing data from continuous depth sensing microindentation tests Journal Of Materials Research 1990 5 1 p 123 126 Saha R and W Nix Effects of the substrate on the determination
6. Methods for determining tip geometry include the indentation method the scanning method and electron microscopy imaging Indentation Method This method developed by Oliver and Pharr 25 involves creating a series of indents of various depths on a standard sample of known elastic modulus typically fused quartz By generating a plot of area versus contact depth as described in Chapter 5 a function in the form of Equation 5 14 can be generated to describe the contact area as a function of contact depth This method has proven reliable for samples that exhibit high hardness with little plastic deformation however it greatly underestimates the contact 67 68 area in samples that exhibit pile up which leads to errors in hardness and modulus 32 This method is also lacking in three dimensional information which is necessary in calculating properties derived from lateral translation of the tip Scanning Method The scanning method is an adaptation of scanning tunneling microscopy STM and atomic force microscopy AFM techniques that involve scanning the tip over a sample with well characterized features or features of much smaller radii than the tip 34 36 By scanning a feature with a radius much smaller than the tip the image produced is representative of the tip Figure 6 1 indenter tip a Cena l scan path gt gt sharp asperity scan direction Figure 6 1 A schematic representation of a large radius tip scannin
7. clockwise CW to produce profile images through 120 of rotation A series of images are collected at 40 CCW 0 and 40 CW the images are digitized and points are plotted along the profile The data is fit to a curve to approximate an axisymmetric geometry to describe the tip shape see Figure 6 5 72 40 ccw 0 40 cw y 2 0x10 x 6 0x107 x 7 0x10 x 3 5x10 x 500 400 c 300 lt 200 0 200 400 600 800 1000 a nm Figure 6 5 The upper images are taken at 40 CCW 0 and 40 CW respectively The lower image shows data points plotted along the profile and subsequently curve fit to a 4 order polynomial This method can also provide crystallographic orientation information in addition to tip geometry By entering diffraction mode with the beam focused on the indenter tip a Kikuchi map image is produced The crystallographic planes that are orthogonal to the tilt axis remain parallel to the electron beam and therefore diffract through the range of tilt Since the normal loading axis of the indenter tip is coincident with the tilt axis the planes that remain diffracting through the tilt or the planes indexed to correspond to the Kikuchi band upon which the tilt occurs correspond to the planes normal to the loading axis One degree of freedom remains to describe the crystallographic orientation of the indenter tip By noting the tilt angle at zone axes along the Kikuchi band the 73
8. standard of known elastic modulus In this section the mechanics of generating a tip area function are explained For a detailed mathematical description of the procedure refer to the Data Analysis section A fused quartz standard is loaded on the stage of the Triboindenter A series of approximately 100 indents are made targeting a depth range to cover the series of experiments to be run The series is run and the data is saved A power law curve fit is performed on the unloading portion of the curves A tangent line to the power law curve is used to find the stiffness from which an area is calculated The calculated area versus displacement for all the indents is plotted and a curve fit is performed to generate an area versus displacement function This function is saved for this tip and displacement range A new function must be generated for each different tip and the functions must be verified periodically on a standard to ensure the tip geometry has not changed Machine Compliance Calibration In a depth sensing indentation device the desired measurement is the position of the tip relative to the sample surface This is determined by displacement however measured displacements include compliance of the system along with the tip and the sample 45 There are several factors that can contribute to the system compliance such as Indenter tip the indenter tips are handmade from various materials and adhesives Sample mounting
9. EBE 500 0 1 1 10 100 1000 number of cycles Figure 7 16 Plot showing fraction of coating depth penetrated versus number of cycles for samples run at 500 uN normal load Now it appears the two thinner film samples IBS 100 and EBE 200 behave similarly as do the two thicker film samples IBS 300 and EBE 500 Figure 7 16 shows good agreement with Table 7 4 between the two characterization methods to determine the point of film failure with the exception of Sample IBS 100 appearing to have failed before 3 cycles in Table 7 4 and after 3 cycles in Figure 7 16 This may be explained by the coating first breaking down sporadically along the wear track and slight variation in cycle number at failure from tests run with repeat conditions The fraction of penetration appears to be linear when plotted versus the log of the number of cycles to failure which could lead to a coating lifetime prediction based on an extrapolation The EBE 200 data appears to fall on a line nearly all of the way to failure however the last point plotted for IBS 100 and IBS 300 deviated from a line fit through the rest of the points for these 110 samples The deviation occurs somewhere between 50 and 80 of the film thickness More data is needed to determine if this could be a viable method for life prediction Friction Measurements Metallic friction as described in Chapter 1 is comprised of two components the force of deformation and the force to slide one
10. International Test Conference ITC 2003 2003 Charlotte NC IEEE Computer Society Saunders S R J and H R Vetters Standardisation of test methods for the mechanical properties of thin coatings Thin Solid Films 1997 299 1 2 p 82 Rabinowicz E Friction and Wear of Materials 2nd ed 1995 New York John Wiley amp Sons Inc 315 Bowden F P and D Tabor The Friction and Lubrication of Solids Vol 1 1950 391 Slade P G Electrical Contacts 1999 New York Marcel Dekker 1073 Antler M Processes of metal transfer and wear Wear 1964 7 p 181 204 Antler M Wear friction and electrical noise phenomenon in severe sliding systems ASLE Transactions 1962 5 p 297 307 Antler M Metal transfer and wedge forming mechanism Journal of Applied Physics 1963 34 2 p 438 439 Cocks M Interaction of sliding metal surface Journal of Applied Physics 1962 33 p 2152 2161 Suh N P The delamination theory of wear Wear 1973 25 p 111 124 Tian H N Saka and E Rabinowicz Friction and failure of electroplated sliding contacts Wear 1991 142 1 p 57 Bayer R G A general model for sliding wear in electrical contacts Wear 1993 162 164 Part 2 p 913 Antler M and M Feder Friction and wear of electrodeposited palladium contacts Thin film lubrications with fluids and with gold IEEE Trans 1986 CHMT 9 4 p 485 491 Jang D S and D E Kim Tribological behav
11. University of Florida Alumni F ellowship in 1999 and completed his master s degree in 2001 He worked at the Major Analytica Instrumentation Center throughout his graduate career as a teaching assistant microscope technician and class lecturer 131
12. X ray diffractometer with a representation of Bragg s law ceeeeeeeeeeeeeeeeee 75 6 9 X ray scans of the six different gold films shown with indexed peaks 0 0 77 6 10 Grain size and strain versus gold film thickness ENPA ER TO 6 11 Wear track in center of image to be prepared with cross sectional TEM parallel iew are aa al 6 12 Focused ion beam cross section sample preparation sssesiisesseeireesseerreesseerr seses B2 6 13 Cross section removed from the sample trench with a glass rod cseeeeeeeeeee 83 6 14 Trenched area in preparation of fixed sample method slice and view seria Seb Oni prove UTE sriti 6 15 Slice and view rotating sample method isisiiccsorcnnbwienimnmnamamnr BO 6 16 Gold films with wear tracks run for half paa s at 500 a normal load Film thicknesses are indicated on each IMAGE sssucsccaniscnnisieneionioioriiin O Fel Hardness determined by the Oliver and Pharr method versus maximum tip placement for five samples of varying thickness cc cscseeeeeeereeeeseeeeeeneeeeeee OO 7 2 Stone plot including linear curve fits for sample EBE 20 with a gold film ig ia en asa eee 7 3 Stone plot oo linear curve fits for Aiaia IBS 100 with a aii film thickness of 105 nm PEIEE IENA TANE EE A AAA N EAE A xii 7 4 Stone plots including linear curve fits for samples EBE 200 IBS 300 and EBE 500 with a gold film thickness of 180 nm 315 nm and 537 nm
13. and different tips to determine the compliance for any given testing conditions Figure 5 2 shows a plot of the total compliance measured versus the inverse of the square root of the maximum load for various large loads on a quartz standard an aluminum standard and a 001 silicon wafer 25 gt silicon 001 quartz aluminum 001 x 20 aluminum 001 compliance nm mN ose 0 000 0 002 0 004 0 006 0 008 0 010 0 012 0 014 1 qe uN Figure 5 2 Measured compliance versus square root of maximum load for a load range of 5mN to 10mN on quartz Al 001 and Si 001 A least squares fit line is generated to determine Cm Both the quartz standard and the silicon sample show good agreement in the machine compliance of 2 6 nm mN The 56 aluminum sample shows a slightly higher compliance of 3 0 nm mN which may be attributed to differences in sample mounting A similar yet alternative method to analyze machine compliance and extract hardness and modulus information has been developed by Stone et al 29 Stone et al begin by defining a material compliance parameter H 5 18 Jo PEC T E where Pmax is the maximum load C is the compliance of the sample H is the hardness of the sample and E is the reduced modulus Note that the parameter Jo is independent of the tip geometry This parameter was originally developed by Joslin and Oliver 30 who noted that by substituting for the definition of hardness
14. are by no means new phenomena From the earliest times man has experienced friction and wear To early man friction was beneficial in creating fire by frictional heating of wood and wear was used to sharpen tools for hunting and working From the early 1700s to the early 1900s friction and wear are studied mostly as a scientific curiosity As the industrial revolution progresses friction and wear are seen as engineering problems which lead to energy loss and machine failure Therefore the interest in studying the phenomena greatly increases As the resolution of measuring and characterization devices improve and computer simulation power increases the study of friction and wear progresses from the macro to micro to nano and approaching atomic displacement levels The driving forces for tribological research involve device longevity energy conservation economy and fundamental understanding 3 Early Friction Studies The study of friction and wear dates back to the beginnings of the scientific era Amontons 1699 is credited with the first observation that friction force is proportional to the normal load and independent of the contact area hence Amontons Law F UF 1 1 where Fris the frictional force tangential to the surface F is the force normal to the surface and is the friction coefficient Actually Amontons concluded that the friction force was always equal to one third of the normal load 4 He explained this as
15. from Equation 5 1 and stiffness from Equation 5 11 one could obtain H_4P 5 19 max E x S Stone s parameter is missing the 2 4 term from Joslin and Oliver s analysis This 7 4 term arises from the geometry in Sneddon s 26 analysis of the cylindrical punch which Pharr et al have shown to apply to any axis symmetric indenter 27 Stone incorporates a geometrical factor in his E term which is adjusted depending on tip geometry to account for the 2 4 term The indenter used in this study is modeled as an axisymmetric indenter and no geometrical term is incorporated in the reduced modulus therefore a parameter will be substituted in Stone s analysis to better fit this study This parameter Ko is defined as H 5 20 K PoC 2 2 a 4E 57 By taking Equation 5 15 and multiplying by the square root of Pmax yields C VT a Cms Prax C Fow 5 21 and substituting for the material parameter Cr fP Cn Pan E Now plotting the CtP max versus Pmax yields a line with a slope of the machine compliance and an intercept of Ky provided the material shows no indentation size effect This analysis is useful in that the material parameter can be verified for the calibration standard and material data can be extracted without knowing the indenter tip geometry Just as in the Oliver and Pharr analysis the unloading is assumed to behave as an isotropic elastic half space Modified Winkler Method The Oliver
16. hardest toughest material available A diamond sample of the sharp asperity would be optimum height nm 757 60 1142 42 height nm 38 70 354 86 748 41 1141 97 Figure 6 3 Scanning image produced with a sharp asperity by rastering first in the x direction and then in the y direction 71 Transmission Electron Microscopy Method A direct measure of the tip geometry is the best method for characterization Imaging the tip in a transmission electron microscope provides a high resolution profile of the geometry A fixture to mount the indenter tip in a TEM sample holder was fabricated and is seen in Figure 6 4 The indenter tip is fixed in a polymeric mounting device with a blind threaded hole as received from the equipment manufacturer The polymeric fixture is glued into the TEM mounting fixture using a conductive adhesive The mounting fixture is attached to the TEM sample holder with countersunk set screws mounting fixture Figure 6 4 Schematic drawing of TEM 200 CX sample holder with tip holder fixture and indenter tip mounted The normal loading axis of the indenter tip is oriented such that it is parallel with the sample holder tilt axis The sample holder is inserted into a JEOL 200CX TEM and the tip is imaged at 50 kX and 200 kV The sample is adjusted so the normal loading axis and sample tilt axis are at the eucentric height The indenter tip can be rotated from 60 counter clockwise CCW to 60
17. if the sample mounting procedure varies in materials the compliance will vary Transducer mounting the transducer must be mounted the same way each time and should be level and properly seated Therefore it is necessary to determine the system compliance and remove it from the measurements This section only explains the procedure for the compliance calibration The mathematics are explained in detail in the Data Analysis section Once the other procedures have been completed the transducer calibration constant is set so that the machine compliance is 0 00 A set of 15 indents are made with loads varying from 5 mN to 10 mN Curves from the data are ploted and the multiple file analysis is executed A plot is generated of the inverse of the stiffness versus the inverse of the square of the load A linear fit is generated and the machine compliance in nm mN is equal to the y intercept times 1000 This number is entered for machine compliance A new machine compliance is generated for each tip and any time the conditions mentioned above might change Testing Modes The Triboindenter has three basic modes of operation which are illustrated in Figure 4 11 The static indentation mode brings the tip into contact with the surface applies a normal load and records load and displacement The scratch mode brings the tip into contact applies a normal load translates the tip laterally and records normal load normal displacement lateral load and
18. lateral displacement The rastering wear test mode applies a normal load and uses the piezo tube scanner to raster the tip over an area No quantitative data except the prescribed normal load is collected in this test 46 eed indentation mode SF scratch mode SS SY rastering wear mode an A Figure 4 11 Schematic illustration of Triboindenter testing modes Depth sensing indentation Depth sensing indentation tests yield data in the form of loads and displacements which can be analyzed to obtain material properties such as elastic modulus and hardness as described in the data analysis section Once the calibrations have been completed load versus displacement versus time functions are generated in the load function editor The area of interest is located and the test is initiated Indentation tests can be operated in load control displacement control or open loop where a DC bias voltage is set based on the electrostatic force calibration to approximate a maximum load Once the test is started the tip approaches the surface and makes contact at the set point The load is increased and load displacement and time are recorded A general rule of thumb has been established for testing thin films on substrates which suggests maximum indentation 47 depth should not exceed 10 of the film thickness to avoid a contribution from the substrate properties Rastering wear test For data analysis fiducial marks are placed u
19. load versus displacement curve along with a cross section of an indented surface The range of is indicated for various indenter geometries where P is the load h is the depth and hr is the depth of the residual impression is fit to the unloading portion of the data Sneddon derived the following equation to describe the 52 load as a function of displacement for an infinitely rigid cylindrical flat punch into an elastic half space 4Ga Pas l v h 5 3 where P is the load G is the shear modulus a is the radius of the cylinder h is the depth of penetration and v is the Poisson ratio 26 By substituting sa 5 4 2 1 v 6 4 for G Equation 5 3 becomes _ 2Ea T d Rearranging the equation for the area of contact of the indenter to a 5 6 and substituting Equation 5 5 becomes 2 E oa i ot and differentiating with respect to h yields dP 2 E dh Pea where dP dh is by definition the stiffness 5 8 One must now consider the elastic properties of the indenter Defining the modulus of the system as the reduced modulus E yields 1 _d v v E E E r Ss l 5 9 53 where E and v are the Young s modulus and Poisson ratio for the indented surface and E and v are the Young s modulus and Poisson ratio for the indenter Substituting into Equation 5 8 yields E T Tyi 5 10 Solving for E gives E ar 5 11 e OA Oliver Pharr and Brotze
20. loads A 1 cycle B 3 cycles 0 0 0 123 Sample EBE 200 at 500 uN normal loads and 10 cycles cceeeeeeeeeeeereeeeeee 124 Sample IBS 300 at 500 uN normal loads A 1 cycle B 3 cycles 0 0 124 Sample IBS 300 at 500 uN normal loads A 10 cycle B 30 cycles 124 Sample IBS 300 at 500 uN normal loads A 100 cycle B 300 cycle 125 Sample EBE 500 at 500 uN normal loads A 1 cycle B 3 cycle 125 Sample EBE 500 at 500 uN normal loads A 10 cycle B 30 cycle 125 Sample EBE 500 at 500 uN normal loads A 100 cycle B 300 cycle 126 xiv Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy A METHODOLOGY FOR TRIBOLOGICAL EXAMINATION OF THIN FILMS IN THE NANODISPLACEMENT REGIME By Gerald R Bourne August 2006 Chair Luisa A Dempere Cochair W Gregory Sawyer Major Department Materials Science and Engineering This research aims to develop a methodology for studying tribology of thin films on the nanodisplacement level for both fundamental and practical wear applications A Hysitron Triboindenter was chosen to perform the majority of the wear testing of this study Thin film gold on silicon is chosen for its applications in micro electrical mechanical systems MEMS devices electrical switching contacts low surface roughness and its thermo
21. of cycles are plotted on log scale Carbon on gold on silicon A layer of amorphous carbon approximately 60 nm in thickness was deposited on a set of samples as described in Chapter 2 to explore the effects of an electrically conductive friction modifying layer The results from the SEM examination of the wear tests with 100 uN normal load are presented in Table 7 5 In this table the designation of D for coatings that show delamination as described above is included 104 Table 7 5 Results of SEM micrographs from carbon on gold on silicon samples with 100 uN normal load Designations are S for coatings that survived to the number of cycles shown D for coatings that exhibited delamination of the amorphous carbon or F for coatings that have exposed silicon cycles 0 5 The effect of amorphous carbon is significant when comparing Table 7 5 with Table 7 3 The sample EBE 20C appears to have survived for at least 30 cycles with the amorphous carbon where without it failed before one complete cycle With the addition of amorphous carbon the carbon film has delaminated by 30 cycles on the EBE 200C sample and gold film failure is seen at 300 cycles Table 7 6 shows the results from micrographs taken from a sample matrix run on carbon on gold on silicon The results are similar to those seen in Table 7 4 with the exception of the IBS 100C sample With the carbon coating the life of this
22. of thin film mechanical properties by nanoindentation Acta Materialia 2002 50 1 p 23 38 Saha R and W D Nix Soft films on hard substrates nanoindentation of tungsten films on sapphire substrates Materials Science and Engineering A 2001 319 321 p 898 Yoder K D Stone R Hoffman and J Lin Elastic rebound between an indenter and a layered specimen Part ii Using contact stiffness to help ensure reliability of nanoindentation measurements Journal Of Materials Research 1998 13 1 1 p 3214 3220 Reiss G J Vancea H Wittmann J Zweck and H Hoffmann Scanning tunneling microscopy on rough surfaces Tip shape limited resolution Journal of Applied Physics 1990 67 3 p 1156 1159 Villarrubia J S Algorithms for scanned probe microscope image simulation surface reconstruction and tip estimation Journal of Research of the National Insititute of Standards and Technology 1997 102 4 p 425 454 Xu S and M F Arnsdorf Calibration of the scanning atomic force microscope with gold particles Journal of Microscopy 1994 173 p 199 Dieter G E Mechanical Metallurgy 1986 New York McGraw Hill Hertzberg R W Deformation and Fracture Mechanics of Engineering Materials 1996 New York J Wiley amp Sons McSkimin H J and P Andreatch Jr Elastic moduli of diamond as a function of pressure and temperature Journal of Applied Physics 1972 43 7 p 2944 Cullity B D Elem
23. repectively 94 7 5 Hardness determined by four methods plotted versus sample thickness for five POE E eae eee 7 6 Scanning electron micrographs from a rastering wear test matrix with a 40 pu ikai normal load varying tip geometry and number of cycles as indicated 97 7 7 Scanning electron micrograph showing a ra tering wear anomaly The wear scar extends outside of the prescribed area especially in the horizontal direction 98 7 8 Scanning electron micrograph showing a rastering wear anomaly A trench develops on the right and lower edge of the Wear scar o eeeeeeseeeteteeeeeteeeeeeee D8 7 9 Scanning electron micrographs of n IBS 300 ganii and EBE 200 esse run for 30 cycles at 500 uN normal loads eer mao 7 10 Scanning electron micrographs of EBE 200 C run for 3 Dii coe and 30 cycles right at 500 uN normal load sc E alan aia acsod 100 7 11 Scanning electron acai of a IBS 100 after 1 pan of gik with 500 uN normal load AE ETET ANVE E AN T 7 12 Nominal film thickness versus number of cycles prior to film failure 103 7 13 Vertical tip displacement versus lateral displacement for a 30 cycle reciprocating test ocniiiGnmiaiemsitaineeennins TT l 7 14 Transverse cross section of a wear scar run on sample EBE 500 for 10 cycles yyith a 300 aN nomal eG oetnincetneamasatioencet iaieasat le 7 15 Depth of wear scar versus numb
24. scan rate are entered Topography and gradient imaging windows are selected Imaging File Control OO Mem Ge o annl gt 70 000 pm Scan size um o X Offset S 00000 um 1 480 1 110 Y Offset 0 A ym Topograph a ZRange 0 2196 Figure 4 9 Imaging window and control panel with image of modified tip optics offset sample Tip approach is selected and once the Triboindenter has indicated that the tip is in contact with the surface the scanning is started Once one complete raster is finished some of the tip optics offset sample should be visible Now the stage controls are used to translate the sample towards the center of the circles It is critical to use only the stage controls and not the x y offset to translate the sample The x y offset is a bias applied to the piezo tube scanner which will give an inaccurate tip optics offset calibration if used 42 here Determining the direction for translation is easily identified by moving radially inward from the patterned arcs For translation it is easiest to think of the sample being stationary and the reticule moves in the direction indicated on the x y translate arrows Relative direction lengths can be entered and are convenient for homing in on the center As the center is approached the raster size is reduced which effectively zooms in on the image Once the center of the sample is located on the imaging reticule the tip is withdrawn from the sample an
25. the Bragg angle the spacing of the diffracting planes can be determined Samples are scanned from 30 to 130 20 and data is collected to determine crystallographic texture and to choose a family of planes to analyze for grain size approximation 76 Crystallographic Texture Often in processing a certain crystallographic orientation develops due to conditions These conditions may be due to mechanical processing thermal gradients upon solidification or preferred growth orientation For thin films the latter is often the case Since most single crystals exhibit anisotropic mechanical properties it tends to reason that polycrystalline samples with a preferred orientation will also exhibit similar anisotropy Analysis of crystallographic texture by x ray diffraction can range from exceptionally simplistic when only one out of plane orientation exists to rather complex as the orientation becomes more random For the simple case when only one out of plane orientation exists Bragg s law will only be satisfied for planes parallel to the surface and they will be the only intensities that show up in the 20 scan To analyze the texture of a more random sample one must consider the integrated intensities of the various peaks that show up in the scan This is not a simple direct comparison because the intensity is a function of the diffracting planes and the Bragg angle through the structure factor multiplicity factor and the Lorentz polarizati
26. the amount of volume from that incremental displacement over the differential element is transferred to the next element in the outward radial direction The force is summed and the total displacement is compared to the maximum displacement The displacement is incremented until the maximum displacement is reached upon which the unloading cycle begins The loading code is represented in the flow chart in Figure 5 5 62 Loading Calculate incremental pressure Sum dF dF P dA Sum dF df H dA Zero redistributed mass Redistribute plastically deformed mass to next radial dA Schematic of indentation procedure Elastic Region Plastic Region Sum dF Total Force at Depth Max Depth Figure 5 5 Flow chart diagram of loading code Unloading The unloading is assumed to be elastic Each element begins compressed to the point Lmin in Figure 5 4 The material will only recover to the point Ly Figure 5 6 shows a flow chart of the elastic unloading code 63 Unloading Model Complete Figure 5 6 Flow chart of unloading code Rastering Wear Test In the rastering wear test mode no data is collected during the test by the Triboindenter The input normal load is the only quantitative information available without subsequent analysis Often in wear testing a wear rate is reported to quantify the material performance The wear rate K is the volume of material removed per normal lo
27. the test A method for quantitative wear measurements was developed in Chapter 5 This testing mode suffers from difficulties due to machine control and wear debris Figure 7 6 shows SEM micrographs from a test matrix run at a 40 uN normal load while varying tip geometry and number of cycles The 97 images in Figure 7 7 and Figure 7 8 show problems that could not be resolved after numerous attempts discussions and an extended visit to the equipment manufacturer The image in Figure 7 7 shows a problem with wear scar extending beyond the prescribed area This is most likely due to a drift or control problem with the piezo scanner The image in Figure 7 8 shows a trench that sometimes develops during test runs No correlation was made due to sample variation and the phenomenon would occur intermittently on the same sample In addition to the control problems Figure 7 8 shows wear debris that was transferred into the wear scar which could confound a wear volume measurement 150 um tip 50 um tip of passes Figure 7 6 SEM micrographs from a rastering wear test matrix with a 40 uN normal load varying tip geometry and number of cycles as indicated 98 10 um Figure 7 7 SEM micrograph showing a rastering wear anomaly The wear scar extends outside of the prescribed area especially in the horizontal direction 10 um Figure 7 8 SEM micrograph showing a rastering wear anomaly A trench develops on the right and lower edg
28. the unit cell axes 38 Equation 6 1 contains the stiffness matrix for diamond 39 1079 124 124 0 0 0 124 10799 124 0 0 0 C 124 124 10799 0 o0 0 E GPa L 0 0 0 58 0 0 0 0 0 0 578 0 which can be converted to the compliance matix by 0 949 0 0979 0 0979 0 0 0 0 0979 0 949 0 0979 0 0 0 S Co 0 0979 0 0979 0 949 0 0 0 1 n na TPa Y j 0 0 0 1 73 0 0 0 0 0 0 4173 0 0 0 0 0 0 1 73 From figure 6 7 the loading orientation of the indenter tip is characterized and therefore the Young s modulus in the loading direction is calculated at 1053 GPa X ray Diffraction X ray diffraction is used to characterize the grain size and the texture of the gold films A Philips APD 3720 diffractometer with a Cu Ka A 1 54178 A x ray source is used to scan the samples Figure 6 8 shows a schematic drawing of the diffractometer 75 Detector 20m 1352 X ray Source Figure 6 8 Schematic drawing of an x ray diffractometer with a representation of Bragg s law The x ray source is in a fixed position and the sample is rotated through 0 The x ray detector is rotated 2 for every 1 of the sample and is therefore rotated 20 with respect to the x ray source A plot is generated containing the number of counts recorded by the detector versus 20 From Bragg s law n 2d sin 0 6 2 where n is an integer 1 2 3 A is the wavelength of the radiation d is the spacing of the diffracting planes and is
29. tip Using the automated methods feature of the Triboindenter one could set up a similar pattern to that described above and run the pattern with a sharp tip high load and a soft sample Once the sample is produced the sharp tip can be exchange for a blunt tip and the modified tip optics offset procedure could be followed 39 Once the modified tip optics offset sample is produced and mounted on the stage with a safety limits area defined the blunt tip installed on the capacitive transducer and the electrostatic force calibration and zero volt gap calibration completed the modified tip optics offset procedure is initiated The sample s concentric rings are located and centered to the video window reticule using stage translations Figure 4 5 Figure 4 5 Modified tip optics offset sample centered in the video window From the main positioning window tip optics offset is selected Then new single indent is selected from the window seen in Figure 4 6 Calibrating the Optics i Figure 4 6 At this point new single indent is selected In the load function editor a low load of 2 uN or less is programmed By setting a low load the machine will go through the standard procedure of performing an indent but no 40 damage to the sample or the tip will occur The machine will translate the sample so that the tip is over the sample and the user will be prompted by the window in Figure 4 7 Se O
30. to a 4 order polynomial from micrographs see Chap 6 To account for pile up and plastic deformation neglected in the Oliver and Pharr model each discretized element is examined to determine the onset of plastic deformation based on the hardness Upon plastic deformation the volume deformed is transferred to the next differential element Once the maximum displacement is reached elastic unloading is assumed This model is developed to attempt curve fits with load versus displacement data collected from the nanoindenter The first assumption of constraint due to the surrounding material and the indenter tip displacement scale results in no elastic strain in the directions orthogonal to the loading direction such that yy 0 From this assumption and effective modulus can be calculated from te S o 5 23 where s is the strain S is the compliance matrix and o is the stress Expanding yields 59 E EE v 1 v TA E JEEE Os E ee om eee toi ioa E EEE Biz Pal gg E 0 Kz G Ga ANETT al ES P2 G 1 0 0 0 0 0 gt G where E is the Young s modulus Solving for the stiffness and assuming no shear coupling yields E l v Ev Ev 5 25 Cas Ev E l v Ev i v 2v Ev Ev E l v The assumption of no strain in the x and y directions 0 5 26 e 0 E can be substituted into tos C te oe which yields 28 E 1 v 5 28 PORRE se eae v l 2v The coeffici
31. um deep on both sides of the desired cross section area Progressively lower beam currents were then used to remove more material and reduce ion damage Once the sample is thinned to approximately 1 um thick the stage is tilted and the sample is undercut with only small tabs of material keeping the 82 cross section connected to the bulk The sample is then tilted back and the final automated polishing steps are performed at 300 pA current to thin the sample to approximately 300 nm Once the automated script is complete the thinning is controlled manually at 30 pA to thin the sample to less than 100 nm thick The cross section is then cut free with the 30 pA beam current and the sample is removed from the FIB Figure 6 12 illustrates the steps a Figure 6 12 FIB cross section sample preparation A area of interest is located and fiducial marks are milled B platinum is deposited C front trench is milled D back trench is milled E cross section is undercut F cross section is cut free The sample is then transferred to an optical microscope equipped with hydraulic micromanipulators Glass rods are heated and pulled in tension to form sharp radius tips of approximately 1 um radius The rods are attached to the hydraulic micromanipulators for precise movement The cross section area is imaged with the optical microscope and the glass rod is brought into contact with the cross section The cross section is attached to the r
32. using a carbon arc evaporation system This procedure is similar to the electron beam evaporation except that a bias is applied across two carbon rods to evaporate carbon The resultant condensation yields a thin amorphous carbon film This process is illustrated schematically in Figure 2 5 Bell Jar Graphite Rod Substrates Figure 2 5 Carbon arc evaporation 21 Ion Beam Etched Area of Interest Many of the tests conducted in this research leave residual damage smaller than the resolution limits of optical microscopy Therefore scanning electron microscopy SEM and transmission electron microscopy TEM are used to characterize the damage from testing Even with the resolution obtainable with electron microscopy locating micron and submicron surface features with displacements on the order of tens of nanometers or less is extremely challenging due to the low amount of contrast they produce see Figure 2 6 Scanning at a low magnification to locate such features is not possible because the feature will not be visible Scanning the sample surface at a high magnification without an indication of the approximate area is not practical because the field of view is much smaller than the entire area of the sample A method to accurately position wear tests and to subsequently locate the tests in an SEM is necessary 2 um Figure 2 6 SEM image of 4 um x 4 um rastering wear test with low contact pressure 22 A dual beam focuse
33. 0 24 CJP mm yuN 0 16 0 08 0 00 0 5 10 15 20 25 30 35 40 Pax VUN Figure 7 3 Stone plot including linear curve fits for sample IBS 100 with a gold film thickness of 105 nm Both plots do not fit well to a single line indicating a problem with the model The assumption of an isotropic elastic half space is questionable for multiple reasons It is reasonable to expect some contribution from the substrate on both of these samples considering the ratio of indentation depth to film thickness is 74 330 on sample EBE 20 and 19 84 on sample IBS 100 The assumption of isotropy is problematic in that single crystals and textured surfaces are often elastically anisotropic A measure of the degree of anisotropy is given by 46 Ca C 7 4 2C 44 The compliance matrix for gold is given in Chapter 6 and the value for C11 C12 and C44 for silicon are 165 GPa 64 GPa and 79 2 GPa respectively 41 Calculating the degree of anisotropy for gold and silicon gives 0 46 and 0 64 where 1 0 is the result for an 94 isotropic material In addition to these factors there exists a pressure induced phase transformation in silicon which may occur in the upper bounds of the maximum displacement 47 Figure 7 4 shows Stone plots for the remaining three samples EBE 200 IBS 300 and EBE 500 0 5 0 4 Z 03 E Na y 0 0039x 0 3997 lf 0 2 S i o EBE 200 e IBS 300 o EBE 500 0 5 10 15 20 25 30 35 40 V Faa V
34. 25 30 number of cycles Figure 7 20 Plot of friction coefficient versus number of cycles for gold on silicon samples with carbon coating run at 100 uN normal load CSM microtribometer friction coefficient measurements As mentioned previously an elastic contact cannot be accomplished with the current tip geometry on the Hysitron Triboindenter Although a technique has been developed in this study to target low contact pressures in the Hysitron a CSM Microtribometer was employed as a matter of convenience To target a completely elastic contact the maximum pressure should be below the hardness Sample EBE 200 was chosen for testing From the hardness analysis the lowest calculated hardness for this sample was determined to be 1 4 GPa If the relationship H 3oy holds true for this material the yield strength should be 470 MPa Coarse grained annealed gold is reported to have a yield strength of 100 MPa 120 MPa 41 52 The onset of plasticity can occur at pressures as low as 1 10y 37 With all these considerations a maximum pressure below 100 MPa was targeted to ensure an elastic contact Based on a 2 mm diameter pin of Al O3 with a normal load of 1 5 mN determined by a cantilever flexure in contact with gold the maximum pressure was 115 calculated to be 92 MPa using Herztian elastic contact mechanics Reciprocating tests were run for 1 3 7 10 30 70 and 100 cycles and are plotted in Figure 7 21 0 4 friction coefficie
35. 5 for less than 3 cycles of sliding Both cross sectional microscopy and top down microscopy show the film failure between 1 and 3 cycles for IBS 100 Sample 111 EBE 200 shows failure between 10 and 30 cycles and its friction coefficient drops below 0 25 after 10 cycles Sample IBS 300 has not failed at 300 cycles but the cross sectional image shows the coating is close to penetration see Figure 7 16 The dashed line shown on the plot represents friction values where the gold film has been penetrated and the sliding contact is comprised of a combination of gold and silicon Below the grey region diamond sliding on silicon is a major component of the friction coeffiecent For gold film samples with a 500 uN normal load a friction coefficient below 0 25 indicates film failure 0 6 EBE 20 o IBS 100 0 5 z EBE 200 ar x IBS 300 Z 0 4 o EBE 500 E f 0 0 3 SOSA 5 Ane E Ay film failure E 0 1 1 10 100 1000 number of cycles Figure 7 17 Friction coefficient versus number of cycles in log scale for gold on silicon samples without carbon run at 500 uN normal load Friction results from the gold on silicon samples without carbon coating run at 100 uN normal load are plotted in Figure 7 18 Again the friction coefficient is much lower for the sample EBE 20 that failed as early as the first half cycle see Table 7 3 All other samples survived the entire test at 100 uN normal load and their friction coefficient re
36. 52 from the electron beam A sloped trench is milled to clear material so the sectioning area can be imaged with the electron beam Figure 6 14 Some bulk material from around the area of interest can be removed prior to sectioning to aid in imaging Figure 6 14 SEM image of trenched area in preparation of fixed sample method slice and view serial sectioning procedure A series of ion beam slices each followed by an electron beam image is recorded The fixed sample method can be automated using scripting software which requires initial set up but then runs to completion without user interaction 85 The ion beam is mounted 52 with respect to the electron beam so the image height will need to be corrected if the sample is not rotated when imaging The correction is as follows y Y m ge x 6 4 actual sin 52 Y magt where Yactuai is the actual dimension in the y axis or height of the image area Yimage is the dimension of the projected height in the image The image can be corrected in image processing software by changing the height on the image by 1 27 times while keeping the width the original value In addition to producing a series of images this technique can be used to produce a single cross section image This is much less time consuming and greater emphasis can be place on image quality Rotated sample method An alternative to correcting the image is tilting the sample so that the electron beam is orthogona
37. 7 1 shows Hall Petch calculated hardness using Conrad and Jung s constants and the grain size of the sample measured by XRD from Figure 6 10 The values for hardness range from 1 48 to 1 70 GPa The reduction in grain size does not justify the high hardness values obtained from the Oliver and Pharr analysis method Thus inaccurate approximations of contact area due to material pile up is the likely cause of over estimation of hardness 92 Table 7 1 Hardness values calculated for samples from the Hall Petch equation Sample Hardness GPa EBE 20 2 26 IBS 100 1 70 EBE 200 1 55 IBS 300 1 48 EBE 500 1 49 Stone Analysis Results The same load displacement data that was used in the Oliver and Pharr method above is also analyzed using Stone s approach If the assumptions of no indentation size effect and unloading as an isotropic elastic half space are valid the plots of compliance times the square root of load versus the square root of load should fall on one line as described in Chapter 5 Figure 7 2 and 7 3 show Stone plots for samples EBE 20 and IBS 100 respectively 0 7 0 6 p 8 Jant ee a ee y 0 0001x 0 5911 0 5 9 y 0 0031x 0 4582 0 4 0 3 CJP nm uN 0 2 0 1 0 0 0 10 20 30 40 50 60 70 VP WUN Figure 7 2 Stone plot including linear curve fits for sample EBE 20 with a gold film thickness of 27 nm 93 0 52 0 48 4a 0 40 AB y 0 0058x 0 5087 0 32 y 0 0014x 0 3986
38. A A T 104 Figure page 1 1 ler and arcas from Eguation Pd asssaicintonroeit acd eaten onions 6 1 2 Three slider geometries from Bowden and Tabor s experiment ceeeeeereeerees 6 1 3 Effect of film thickness on friction coefficient of metallic film solid lubricants an harder SSE pnma deena wane 10 2 1 South Bay Technology polishing whegls sssessssnsisnsnisrinaronenanr 16 2 2 Ton beam sputtering prot ge heterodimer EErEE EE EEEa 17 2 3 Gatan Model 681 High Resolution Ion Beam Coater 0 cceccceseceeeeeeeeseeeeseees 17 2 4 Electron beam evaporation process sessssssseseesseesseseessresetssessessrssresseesessresseesee 19 2 5 Carbona Sy acc cated ed Gea a ae ees 20 2 6 Rastering wear test 4 um x 4 um with low contact pressure ceeeeeeeeeteeeeeeeees 21 2 7 Etched grid lines for wear test locationis nnana aia 22 ool Atypical pierce type TT cecum E 23 3 2 Linearrteciprocating TONNE sisane ienee nenn EE EEE ieee 24 3 3 Hysitron Triboindenter with upper thermal acoustic isolation cover removed to ano enor Gl a cele Gane eo eae eens 25 3 4 Capacitive transducer assembly schematic ecstasy accintisesinsriesiinissceeteteeeeseecaidronnets 26 3 5 Enlarged view of Triboindenter components anssen 27 3 6 CSM Microtribometer cc cccsscsssssssessesssessesseessesssessecssesseesseesseesseeseessessssseesseessens 28 4 1 Diamond indenter tip and assembly of a large radius sapphire tip ceeeee 32 4 2 Optical CCD m
39. A METHODOLOGY FOR TRIBOLOGICAL EXAMINATION OF THIN FILMS IN THE NANODISPLACEMENT REGIME By GERALD R BOURNE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006 Copyright 2006 by Gerald R Bourne To Wanda and Gabrielle ACKNOWLEDGMENTS First and foremost I would like to thank my wife Wanda for seeing this long journey to the end with unwavering support Without her encouragement I would not have embarked on this path let alone seen it to completion I would like to thank Greg Sawyer for his advice and support His dedication to his students and research is inspiring and unparalleled Thanks to Carol Sawyer for putting up with Greg Thanks to Amelia Dempere I could not ask for a more understanding and flexible boss Thanks to Mike Kaufman for showing me what fun hovering over a phosphorous screen for several hours could be I appreciate Kevin Jones accepting a committee position at this late stage in the project Thanks to Nam Ho Kim for his contributions in finite element analysis I would like to thank Jeff Bardt for his contribution on many levels from keeping me young with our juvenile banter and back and forth hijinks to the extensive assistance with my computer incompetence Thanks to Ben Boesl Matt Hamilton Pam Dickrell Nicole McCook Kerry Siebein and Junghun Jang for contributio
40. D Introduction The term tribology coined circa 1965 and defined as The branch of science and technology concerned with interacting surfaces in relative motion and with associated matters as friction wear lubrication and the design of bearings by the Oxford English Dictionary is centuries newer than the study of friction and wear For much of history up to and including the present day tribological problems are often handled with an alchemistic approach by choosing a variety of materials and empirically testing One might wonder if some studies were designed with a periodic table of the elements and a handful of darts Much of the problem has been the lack of experimental computational and characterization equipment with the capabilities required to study fundamental tribological interactions The past decade has seen advances in electron microscopy to sub nanometer resolutions computers and software simulation to represent 10 atoms and mechanical testing instrumentation to resolve 10 N loads and nanometer displacement Tribological testing on the nanometer displacement level supported by high resolution characterization is necessary to validate finite element modeling and molecular dynamic MD simulations of surface interactions In addition to fundamental studies of tribology wear on the nanodisplacement level has applications in microelectromechanical system MEMS power transmission and ohmic contact switching Wit
41. F JE 4 1 36 For the relatively soft aluminum sample H 500 MPa With the machine set at the maximum load of 10 mN solving for a yields a 2 5 um Using geometry from Figure 4 3 the following can be written 6 R VR a 4 2 For tips of radius R 5 um the depth of residual impression is 670 nm which is within the limits of resolution in the optical microscope Considering a tip of radius R 100 um the depth of the residual impression is 30 nm which cannot be resolved in an optical microscope Fortunately the Triboindenter has a scanning probe imaging feature which works much like an AFM in that the tip can be held in contact with a surface at a low load and the piezo tube scanner can raster the tip over the surface At each point or pixel a z height is recorded along with the x and y position of the tip This data is used to generate an image by applying contrast values to the z height values This scanning feature has a resolution limit below 1 nm in the z direction because the capacitive transducer is used to record the z height The following method was developed to work around the resolution limits of the optical microscope This method uses a low load which can be beneficial in fields outside of tribology If one were to design a delicate functional tip that could not survive the high load used in the standard optic offset procedure yet still needed accurate tip positioning the modified tip optics offset met
42. UN Figure 7 4 Stone plots including linear curve fits for samples EBE 200 IBS 300 and EBE 500 with a gold film thickness of 180 nm 315 nm and 537 nm repectively The data from three samples from Figure 7 4 fits reasonably well to linear curves suggesting the assumption of no indentation size effect may be valid for these samples The hardness for samples EBE 20 and IBS 100 will be calculated from the curve fits for the lower values of the square root of Pmax because these values correspond to smaller indentation depths Table 7 2 shows the sample versus the hardness calculated by the Stone method 95 Table 7 2 Hardness values calculated for samples from the Stone method Sample Hardness GPa EBE 20 2 57 IBS 100 3 19 EBE 200 1 96 IBS 300 2 83 EBE 500 2 21 The values in Table 7 2 from the Stone analysis are lower than the values determined by Oliver and Pharr but they are still higher than those predicted by the Hall Petch equation Modified Winkler Approach The Modified Winkler method was employed as described in Chapter 5 to account for the material pile up exhibited in this system This model assumes axisymmetric tip geometry based on the TEM characterization Input parameters are film modulus and film thickness The same load versus displacement data is used from the two previous analyses Figure 7 5 shows a plot of the hardness versus the sample thickness for the Oliver and Pharr method the Stone method the Modified
43. Winkler model and the predicted values based on the Hall Petch equation and sample grain size The Modified Winkler predictions show excellent agreement with Hall Petch for sample IBS 100 EBE 200 and IBS 300 Stone s method shows good agreement with Hall Petch for sample EBE 20 however this is probably a coincidence As mentioned previously the assumptions for Stone s model are questionable for large displacement depth to film thickness ratios The more likely agreement shows up on sample EBE 500 where Modified Winkler determines a value somewhat lower than Hall Petch and Stone calculates a higher value The larger thickness of EBE 500 at 537 nm gives an 96 indentation depth to film ratio from 3 7 to 16 6 making the elastic half space assumption much more reasonable 5 m m Oliver Pharr 4 Stone a Modified Winkler A Hall Petch 0 100 200 300 400 500 600 film thickness nm Figure 7 5 Hardness determined by four methods plotted versus sample thickness for five samples Excluding the low film thickness sample due to previously mentioned concerns the Modified Winkler method shows the best agreement with the Hall Petch equation for analyzing highly plastic thin film hardness on elastic substrates Rastering Wear Tests Several sets of test matrices varying tip geometry normal load and number of cycles were run using the rastering wear mode as described in Chapter 4 This mode collects no quantitative data during
44. able results below approximately 10 15 nm Figure 7 1 shows the hardness for the remainder of the samples ranging from over 2 GPa for sample EBE 500 up to 5 GPa for sample IBS 100 Although the general rule of thumb for indentation suggests indenting only 10 of the film thickness to avoid substrate contribution the hardness values are not greatly affected by indentation depth for samples other than EBE 20 even on the IBS 100 sample with a film thickness of 105 91 nm At a maximum displacement of 85 of the film thickness the hardness has only increased by 25 For materials that exhibit significant plastic deformation hardness is frequently approximated by the yield strength with the equation H 30 7 1 where H is the hardness and oy is the yield strength 44 Yield strength values for bulk gold range from 100 MPa to 200 MPa 41 giving a hardness range of 300 MPa to 600 MPa well below the values calculated from Oliver and Pharr s model shown in Figure 7 1 However the published yield strength values from handbooks are for coarse grained microstructures It is well documented that yield strength increases with decreasing grain size by the Hall Petch equation 37 1 O 0 kD 7 2 where oy is the yield strength oo and k are material dependent parameters and D is the grain size Combining Equation 7 1 and 7 2 gives 1 H 3 0 kD 1 3 Conrad and Jung found co 216 MPa and k 0 06 MPa m for gold 45 Table
45. ad per distance of sliding The normal load is the input parameter of the test and the distance of sliding can be calculated by the square root of the area rastered over times 64 256 the number of raster lines times the number of passes over the raster area The volume can be calculated in the following manner e Fiducial marks are placed around the area to be worn prior to testing An area larger than the test area is scanned in the AFM to include the fiducial marks e The desired rastering test area is run for a given load and number of passes The larger area including the fiducial marks is again scanned with the AFM The digitized data from the final scan is aligned to the data from the initial scan using the fiducial marks The initial surface is then subtracted from the final surface e The data below the zero surface plane is integrated to calculate a volume of material removed This volume can now be used in the wear rate calculation The subtracted data can be plotted to reveal wear morphology as shown in Figure 5 7 I i Vi tama v 4 4 Figure 5 7 Rastering wear test data plot showing a trenched area on the right of the test area and a pile up area on the left 65 Triboindenter Reciprocating Wear Test The reciprocating wear test mode or scratch mode is a normal load control and lateral displacement control test The Triboindenter collects normal displacement and lateral force data
46. ad threshold and approaching the sample surface Once the load threshold called the set point is reached the Triboindenter records the z height as the sample height This procedure is performed for each sample safety area Tip Area Function Calibration With depth sensing indentation machines load and displacement are measured however an area of residual impression is necessary for analysis of modulus and hardness as can be seen in Equation 4 1 From the analysis above it is shown that a residual impression cannot be imaged or measured optically It would be convenient to be able to extract area calculations based on displacement data Therefore a tip area function relating the residual area to the depth of indentation is needed For micron and larger scale indentation depths ideal tip geometry can be assumed and an area function 44 can be generated from simple geometric relationships The ideal area function for an indenter with pyramidal Berkovich geometry is A 24 5 h where hs is the contact depth and A is the residual area see Data Analysis section for in depth explanation At the nanometer level the tip of a pyramidal indenter is rounded therefore the assumption that the three faces of a pyramid intersect at a point is invalid and the ideal area function breaks down and greatly underestimates the residual area based on the depth A method has been devised to generate a tip area function by creating a series of indents in a
47. am to the electron beam with a 5 kV accelerating voltage 30 um aperture and 2 spot size This type of image is challenging to produce in conventional SEMs because the technique is sensitive to surface contamination and topography In the FIB a nascent polished surface can be easily produced The microstructure revealed in Figure 7 14 shows columnar grains typical of a deposition process Presumably grains nucleate on the surface with random orientations but will be overtaken by grains oriented in the fast growth direction The results from XRD in Chapter 6 show a strong 111 orientation suggesting this is the preferred growth orientation for gold which is substantiated in the literature 51 A highly deformed plastic zone is visible below the contact area A detailed study of plastic zone evolution could be conducted with this imaging technique Uncoated samples from the wear matrix with a 500 uN normal load were chosen for measurements using the cross sectional FIB SEM technique Figure 7 15 shows a plot of displacement below the original surface versus number of cycles as measured from cross sections Wear tests were run with incremental number of cycles so data could be collected for various cycles The numbers of cycles were as follows 0 5 1 3 10 30 100 and 300 If penetration of the film was observed in a cross section displacement was not plotted because the actual number of cycles at penetration can not be determined Sample EBE 20
48. and Pharr model works well on materials that do not exhibit significant pile up However their model greatly overestimates hardness and modulus in materials that exhibit pile up due to the underestimation of the contact area 31 33 This overestimation of properties is amplified in soft thin films on hard substrates due to excessive pile up The compliance method of Stone et al extracts a material parameter that requires knowledge of either the hardness or the modulus to determine the other The Stone model also assumes no indentation size effect An alternate model is developed based on the Winkler model of a bed of independent springs to account for material pile up Figure 5 3 shows schematic representations of indent cross sections from the Oliver Pharr model the modified Winkler model and the output from an actual indent in a material exhibiting pile up 58 cross section outline Oliver Pharr eeee modified Winkler Figure 5 3 Schematic illustration of indent cross sections from Oliver Pharr and Modified Winkler compared to an actual indent It is assumed that each differential element is constrained orthogonal to the loading direction due to surrounding material The justifications for this assumption are the contact area radius is much larger than the film thickness and the Poisson ratio of gold is 0 42 which is nearly incompressible The indenter tip is assumed to have axisymmetric geometry and is modeled based on a curve fit
49. area of safe travel for the indenter tip 33 Electrostatic Force Constant and Zero Volt Gap Calibration The electrostatic force constant is a conversion used by the Triboindenter to convert an applied voltage to a force This constant is determined by the area of the plates and the distance between them squared Because the center plate is suspended by springs and different mass tips can be used the plate gap spacing can change The calibration procedure is as follows A load of 600 uN is set in the load function editor This load will displace the tip through a large range of motion Perform advanced z axis calibration is selected from the drop down menu The plate is moved through its range of motion and a plot is generated of the electrostatic force calibration based on the spring constant and applied voltage versus displacement The curve is fit to a quadratic and the equilibrium position of the plate is recorded and tared to zero and the calibration is complete Tip Optics Offset Calibration All tests on the Triboindenter are set up on the sample using the optical CCD microscope camera The computer then translates that area to the indenter tip to perform the test Therefore a calibration procedure is needed to record the differences in the x y and z positions of the center of the optical focus and the indenter tip The standard procedure developed by the OEM relies on residual indents large enough to image opticall
50. at the F term from Equation 1 2 or tsA ea term from Equation 1 3 will be negligible for the spade slider and will be a maximum for the cylindrical slider circular spade sphere cylinder Figure 1 2 Schematic drawing of the three slider geometries from Bowden and Tabor s experiment Results from the experiments showed a deformation pressure equal to 1 5 times the material hardness measured by indentation Bowden and Tabor suggest this result may be due to the fast sliding speed relative to the slow indentation speed further suggesting some strain hardening effect The experiment was repeated on four different surfaces indium lead steel and copper and the results of the interfacial shear stress were compared with the shear strength of the material The interfacial shear strength was on the same order of magnitude as the shear strength of the material but 1 5 to 2 times higher than the bulk shear strength Sliding Wear Prow Formation A mechanism for wear of a metallic surface by a linear slider is described by Antler as a severe adhesive process called prow formation 6 A net metal transfer occurs from the surface to the slider This transferred material is described as a lump of severely work hardened metal the prow transferred to the slider which in turn continues to wear the surface by plastic shearing or cutting while the slider remains unworn Prows are removed from the slider by transferring back to the surface or becoming l
51. ation isolation device The working components of the Triboindenter are enclosed in a thermal acoustical isolation chamber with provisions for environmental control 28 CSM Microtribometer The CSM Microtribometer is a commercially available tribological testing device that operates on scales orders of magnitude larger than the Triboindenter but well below those of conventional test equipment Table 3 1 This tribometer is available with a linear reciprocating stage or an interchangeable rotating stage to mimic both reciprocating and pin on disc tribometers Rough stage translations are handled by stepper motors and fine translations and reciprocations are performed by piezos A dual flexure both applies normal load and reacts to frictional forces between the pin and the counterface Mirrors are mounted on this flexure in the horizontal and vertical directions Optical light intensity sensors are positioned at a distance from the mirrors to read the magnitude of deflection of the flexure in the frictional and normal load directions Figure 3 6 test assembly i horizontal hotel e a counterface gt dual flexure mounting stub reciprocating stage oe set screw eae Figure 3 6 Schematic of CSM Microtribometer 22 29 Table 3 1 Comparison of equipment loads and displacements Conventional Reciprocating Hysitron CSM Tribometer Triboindenter Microtribometer maximum normal load 1 6 kN 10 mN 1N maximum lat
52. centimeters F is the normal load in Newtons 5 Rabinowicz and co workers have suggested that materials that show little to no solid solubility and form immiscible liquids would make good candidates for mated materials in adhesive wear sliding applications 3 Antler challenges this theory based on experiments with gold mated with platinum and gold mated with rhodium 5 that show similar performance to gold on gold Antler suggestions for adhesive wear contacts are high hardness to limit initial contact area and low ductility to limit junction growth Delamination Theory Delamination wear occurs in reciprocating sliding when cracks nucleate below the surface and result in liberating sheets of material Crack nucleation can occur in areas of high plastic deformation at dislocation pile ups particularly at hard precipitates in a softer matrix or between material interfaces like a substrate film interface 10 Suh discovered delamination wear typically occurs with harder sliders on soft surfaces Prow formations and delamination wear are different mechanisms and are mutually exclusive 5 Delamination wear has been reported in reciprocating sliding on gold plated contacts 11 Subsurface Wear Gold plated systems have shown subsurface plastic deformation leading to the formation of buckles on the surface layer With continuous sliding exposure of the sub layers can occur with or without thinning of the surface layer Increasing normal load
53. ckness h jj Figure 1 3 Schematic representation of the effect of film thickness on friction coefficient of metallic film solid lubricants on harder substrates reproduced from 17 The critical thickness where u the friction coefficient goes through a minimum is denoted as he Table 1 1 Summary of studies of thin metallic films on substrates film substrate he um p tipradius um load N reference Instel O1 l O5 3000 392 Ao Pb steel 0 5 0 05 3175 29 15 In steel 0 5 0 05 3175 29 15 Sn iron glass 0 1 0 15 3175 2 4 16 Au steel 2 0 3 5 0 1 2375 2 45 17 Pb steel 1 8 2 5 0 1 2375 2 45 17 Co Au Pd Ni 0 05 0 4 N A 1 96 13 Av Si 111 0 01 0 2 1600 0 001 14 Ag Si 111 0 007 0 2 1600 0 001 14 11 Modeling Several attempts have been made to develop models to describe friction and wear phenomena in thin films on substrates Bowers and Zisman modeled the friction coefficient of a thin gold film on a steel substrate based on a pressure dependent shear strength which showed good agreement with experimental data 18 They calculated a friction coefficient of 0 1 assuming no plastic deformation of the steel substrate El Shafei et al investigated thin films of lead on steel with a contact radius larger than the film thickness 19 They found good agreement with a Hertzian elastic contact model once their contact radius exceeded five times the film thickness Finkin sought to explain the ultra thin to
54. d electron beam evaporated gold on silicon wafer These various techniques were chosen to introduce microstructure surface roughness and film thickness variables into the test matrix 15 Bulk Gold Foil Initial experimentation was performed on bulk gold samples A 100 um thick 99 9975 gold foil was cut into 10 mm x 10 mm sections and mounted onto steel gauge blocks using a cyanoacrylate based adhesive Samples were polished using a South Bay Technologies manual polishing wheel Figure 2 1 beginning with grinding discs of 240 grit silicon carbide rotating at a speed of 200 rpm After a short time of material removal the polishing wheel was cleaned and the grinding media was changed to 320 grit and the sample was rotated 90 to the previous polishing direction The sample was polished until no scratches from the previous step were detectable using an optical stereoscope This procedure was repeated using 400 and 600 grit respectively Upon completion of the grinding steps the media was removed and the wheel was cleaned Next a billiard cloth was mounted on the polishing wheel and water based slurry of 15 um alumina Al203 was used as a polishing media The slurry was used to wet the billiard cloth and samples were polished by rotating the sample manually in the opposite direction of rotation of the polishing wheel Samples were polished until scratches from the previous steps were not detectable The wheel was cleaned and the process was repeate
55. d ion beam SEM was used to layout and etch grids on the samples surface Each grid square is 100 um x 100 um and the grid lines are approximately 1 um wide and 1 um deep see Figure 2 7 200 um Figure 2 7 SEM image of etched grid lines for wear test location The grid lines were aligned with lt 110 gt directions on the silicon substrate by setting the intersection of the cleavage planes 111 and the plan view plane 001 to be parallel with the x and y axis in the microscope These markings are easily located in both the SEM and optical microscope and can be used situate the wear test and to identify the area for characterization The markings allow wear tests to be run approximately parallel to a lt 110 gt direction on the substrate which will aid in locating a major zone for high resolution transmission electron microscopy HRTEM analysis in the JEOL 2010F which is limited to 20 tilt CHAPTER 3 TEST EQUIPMENT A wide variety of tribological equipment exists to test various wear conditions and contact geometries By and large this equipment aims to prescribe specific sliding paths while controlling an applied normal load and measuring the frictional forces generated at the sliding interface Most of this equipment does not make in situ measurements of wear rather wear is measured ex situ using gravimetric metrological or microscopy techniques The most common sliding motions are unidirectional with the standard pin on dis
56. d on silicon without carbon E AE E EA T E i Carbon on gold on silicon AIE T Tonita ET ET IE Electron Microscopy Cross Sections of Wear Sears PEENE A 105 Friction Measure M niS sia ctiassieianaasrcnsdsinieeanisinents EATE E mere entre OL Hysitron Triboindenter friction measurements scarp EE PE E eects 110 CSM microtribometer friction coefficient measurements 00 114 e CONCLU ION encon S SA 117 APPENDIX A SCANNING ELECTRON MICROSCOPY IMAGES cccsssccsscsscssscossseroseiacnsvansacese 119 B SCANNING ELECTRON MICROSCOPY CROSS SECTIONS eese 123 LIST OF REFERENCEB ENEE E T PA AE AE E ETAN ssela BIOGRAPHICAL SKETCH iim Eoin OE viii LIST OF TABLES Table page 1 1 Summary of studies of thin metallic films on substrates 0 0 0 0 ces eeseeseereeeeeeeeeenee 10 2 1 PR AT GA RNIN sissano a a dessins acavecnsasieeucanevises 14 7 1 Hardness values calculated for samples from the Hall Petch equation 92 7 2 Hardness values calculated for samples from the Stone method eee 95 7 3 Results of SEM micrographs from gold on silicon samples with 100 uN normal DN hess N AE A E EI E E E E E E E 101 7 4 Results of SEM micrographs from gold on silicon samples with 500 uN normal 7 5 Results of SEM micrographs from carbon on gold on silicon samples with 100 iat E TE EE EE E E E E A E E T 104 7 6 Results of SEM micrographs from carbon on gold on silicon samples with 500 nara pa a
57. d the sample is translated automatically back to the optics window Now the amount of tip optics offset can be visualized in the video window Figure 4 10 Figure 4 10 Actual tip optics offset seen in the video window No stage translations are made at this point Again calibrate optics is selected from the drop down menu and the window in Figure 4 6 appears New single indent is selected again and a low load is entered in the load function editor window The sample 43 is automatically translated beneath the indenter tip and the window in Figure 4 7 appears again The tip is manually lowered to within 1 mm of the surface and the machine performs the low load indent Upon completion the sample is translated back to the optical CCD microscope and the auto search window seen in Figure 4 8 reappears Now stage translations are used to center the tip optics offset sample pattern to the center of the video window reticule and the ok button is pressed in the auto search window The calibration is now complete and can be checked by returning to the scanning imaging mode to verify the pattern is at the center of the imaging reticule This entire procedure was completed with no load greater than 2 uN Quick Approach Now that all calibrations are complete and the samples are mounted with safety zones defined an area of interest is located using grid lines etched by the FIB A quick approach is performed which involves setting a lo
58. d with fresh cloth and 5 um slurry The process was then continued using 1 um 0 3 um and 0 05 um slurries respectively on velvet cloth Upon completion of the final polishing step the samples were examined in a Wyko optical profilometer to verify consistent surface roughness 16 Figure 2 1 South Bay Technology 8 polishing wheel Ion Beam Sputtered Gold on Silicon Ion beam sputter coating is often used to produce thin films for conductivity to aid in imaging in electron microscopy The advantage to ion beam sputtering versus other techniques is that the samples are not exposed to extreme temperatures The low temperature deposition is beneficial in producing and retaining a fine grained or amorphous structure Producing coatings on polished silicon wafers will yield surface roughness values that would be exceptionally challenging to obtain by metallographic polishing techniques The ion beam sputtering process is illustrated in the schematic in Figure 2 2 An argon ion beam is produced by dual Penning guns The guns are aimed to strike the target of desired deposition material Material from the target is sputtered off and deposited on the sample substrate The substrate can be rotated to provide uniform deposition This system is contained in a chamber that can be evacuated to levels of 10 Torr 17 Penning Gun Target beam Substrates eS Vacuum Chamber Figure 2 2 Ion beam sputtering process Commercial
59. ds this difference in the x y and z positions of the center of optical focus and the indenter tip Once this procedure is completed an eighth indent is added to the pattern so not to confuse it with another pattern in a later tip optics offset calibration Figure 4 2 Optical CCD microscope image of residual H pattern indents in Al 001 sample 35 Modified tip optics offset procedure This procedure was developed for large radius tips where the residual indentation left by the maximum load would not be resolvable in the optical microscope By calculating the depth 6 of a residual impression left by a spherical indenter of radius R one can see how indenters of sharp radii less than 5 um leaves an impression that is resolvable by optical microscopy but impressions from indenters of blunt radii greater than 100 um cannot be resolved optically Hardness was defined earlier as the normal load divided by the projected area of the residual impression Figure 4 3 shows a schematic of a spherical indenter on a flat surface and the cross sectional geometry to solve for 5 the depth of impression where F is the normal load 2a is the diameter of the residual impression a is the radius of residual impression and R is the radius of the indenter tip Figure 4 3 Representation of spherical indenter on flat surface and cross sectional geometry to solve for depth of indentation The hardness equation can be rearranged as follows F
60. dulus Hardness is defined as H P A 5 1 where P is the normal load and A is the projected residual area in the direction of the load Microhardness tests are typically performed by applying a known load and calculating the area of the residual indent from measurements taken in an optical microscope Nanoindentation has been defined as tests involving normal displacements on the order of nanometers 23 Residual indents from tests at this scale have dimensions smaller than the resolution limits of optical microscopy and would require electron microscopy for measurements Since electron microscopy is costly and time consuming requiring a subsequent characterization of the sample post indentation methods have been developed to calculate the residual area based on a measured depth of indentation The Oliver Pharr Method The most widely accepted method for depth sensing nanoindentation analysis involves a model first proposed by Doerner and Nix and later refined by Oliver and Pharr 24 25 This analysis is often referred to as the Oliver and Pharr Method Once load and displacement data has been collected and plotted as shown in Figure 5 1 a power law curve fit of the form 50 51 P A h h 5 2 surface profile after load removal indenter G initial urface Fd lt _surface profile oe E a a under load load P possible range for he he ee NS h for e 0 72 displacement h Figure 5 1 Schematic of
61. during the experiment The normal displacement data is subject to errors from sample tilt and thermal drift To compensate for sample tilt a low load pre scan of the wear track is collected and this displacement data is subtracted from the data collected during the test In the indentation mode there is a provision to monitor thermal drift for a selected amount of time and then subtract the recorded drift from the displacement data A typical indentation test is run over a short period of time The drift monitor can be set to record drift over a similar amount of time and wait until the drift reaches some minimum threshold or until the drift is sufficiently linear In the reciprocating wear mode this is not practical due to the length of time that the test is run Therefore the displacement of the test will be verified by cross sections taken from wear tracks and measured in SEM or TEM and adjusted accordingly Lateral force data is also subject to errors due to sample tilt To remove tilt error from the lateral force data the initial low load pre scan friction data is subtracted from the data During the reciprocating tests material piles up at the ends of the tracks and near the endpoints of reciprocation displacement is increased due to smaller initial contact areas The frictional data is filtered to examine sliding conditions excluding these transient areas Figure 5 8 The friction force is the average per sliding pass and can then be exa
62. dynamic stability at ambient temperature and pressure Samples are prepared with a variety of deposition techniques to vary film thickness Various techniques are developed to overcome challenges with testing setup An optics offset procedure solves resolution problems with light optical microscopy Tip holders are designed to alter contact pressures and wear conditions Load functions are designed to operate the equipment in a linear reciprocating motion XV Numerous characterization techniques including scanning electron microscopy SEM focused ion beam FIB transmission electron microscopy TEM and x ray diffraction XRD are employed to characterize testing parameters and results Geometrical and mechanical properties of the tip are characterized using TEM The sample thickness and microstructure are characterized using XRD SEM FIB and TEM From the characterization elastic and plastic properties are approximated using elastic theory and the Hall Petch relationship For quantifying wear SEM and FIB are used for depth measurements An indentation model is proposed to include material pile up exhibited by plastic films on elastic substrates which has been ignored in other models Results show good agreement with the Hall Petch relationship Displacement and friction results show good reproducibility with repeat testing Friction coefficient proves to be a good indication of film failure xvi CHAPTER 1 INTRODUCTION AND BACKGROUN
63. e of the wear scar Debris is also transferred to the wear area The problems with the rastering wear mode make quantification of tests virtually impossible At best this method can be used as a qualitative examination method for a relative study Even in this application the researcher must be cautious about true results 99 versus artifacts created by the problems mentioned above If the rastering problems can be solved this method along with the wear volume measurement described in Chapter 5 could be useful testing tools Reciprocating Wear Tests Film Failure Film failure will be defined as the point where the gold film has been removed and the silicon substrate is exposed The wear track will still contain gold in the contact and gold will be supporting a portion of the load but the indenter tip will be in intimate contact with the silicon substrate at the point of failure Each wear area was examined upon completion of testing ina FEG SEM to determine if the coating survived the test When the film has failed atomic number contrast between the silicon substrate dark and the gold film bright shows in the micrographs of the wear scar see Figure 7 9 Figure 7 9 SEM micrographs of samples IBS 300 left and EBE 200 right run for 30 cycles at 500 uN normal loads Note the dark contrast from the silicon substrate in the center of the wear track on sample EBE 200 due to complete removal of the gold film 100 The re
64. ent 60 E 1 v 4 5 29 l v 0 2v is Lame s constant and can be thought of as an effective modulus Each element of the model film is described in cylindrical coordinates as illustrated in Figure 5 3 For the illustration the tip geometry is shown as spherical however for calculations the 4 order polynomial fit from micrographs of the tip characterization in Chapter 6 is used _ oA dA s d0 ds do dA 2n s ds dO 2m Lass s tan Pmax z D d D R fi os sin z R Figure 5 3 Schematic illustration of a differential element described with cylindrical coordinates and tip displacement based on a spherical tip geometry A MATLAB code is developed to increment tip displacement and calculate differential element elastic plastic response sequentially Figure 5 4 shows the variable designations for the loading and unloading response 61 NE ae le d A I J ml A L nin L y Yy Y aE ee ea pre load post load Figure 5 4 Differential element illustration prior to left and after load right with variable notation indicated At each displacement increment the stress on each element is calculated by 8 5 30 o A L where o is the stress A is the effective modulus 5 is the incremented displacement and Lo is the film thickness The incremental load is calculated by dF odA 5 31 If o H the hardness
65. ents of X ray Diffraction 2nd ed 1959 Reading MA Addison Wesley Publishing Company Inc 41 42 43 44 45 46 47 48 49 50 51 52 130 Brandes E A and G B Brook eds Smithell s Metals Reference Book 7th ed 1998 Butterworth Heinemann Oxford Patterson A L The scherrer formula for x ray particle size determination Physical Review 1939 56 p 978 982 Williams J S Y Chen J Wong Leung A Kerr and M V Swain Ultra micro indentation of silicon and compound semiconductors with spherical indenters Journal Of Materials Research 1999 14 6 p 2338 2343 Kelly A Strong Solids 1966 Oxford Oxford University Press Conrad H and K Jung Effect of grain size from mm to nm on the flow stress and plastic deformation kinetics of au at low homologous temperatures Materials Science and Engineering A 2005 406 1 2 p 78 Nye J F Physical Properties of Crystals 1998 Oxford Claredon Press Hebbache M Nanoindentation Depth dependence of silicon hardness studied within contact theory Physical Review B 2003 68 12 Phaneuf M W Applications of focused ion beam microscopy to materials science specimens 1999 30 3 p 277 Reyntjens S and C Kubel Scanning transmission electron microscopy and dual beam sample preparation for the analysis of crystalline materials Proceedings of the 14th International Conference on Crystal Growth and the 12th In
66. er of cycles for samples run at 500 uN norma Det E E T E a LL 7 16 Fraction of coating depth penetrated versus number of cycles for samples run at 300 HN pormal Ges Oe 17 Friction coefficient versus number of cycles in log scale for gold on silicon samples without carbon run at 500 uN normal load cceeceeeteeteeeeeeeeeeereeel LL 7 18 Friction coefficient versus number of cycles on a log scale for gold on silicon samples without carbon coating run at 100 uN normal load uinieniteeante 7 19 Friction coefficient versus number of cycles for gold on silicon samples with carbon coating run at 500 uN normal load eeeeeseeeeneeeeeeeeseseeeeeeeceeeeneeeeeeereeee LS 7 20 Friction coefficient versus number of cycles for gold on silicon samples with carbon coating run at 100 uN normal load ircsipsinsascmicasstarcideenamndanenmmaisiaions 114 xiii 7 21 A 2 A3 A 4 B 1 B 2 B 4 B 5 B 6 B 8 B 9 Friction y versus number of cycles for a contact pressure of 92 MPa 115 Wear scar of gold on silicon at 100 uN load normal load eel LD Wear scar of gold on silicon at 500 UN load normal load ce eeeeeeeeereeeeee 120 Wear scar of carbon on gold on silicon at 100 uN load normal load 121 Wear scar of carbon on gold on silicon at 500 uN load normal load 122 Sample IBS 100 at 500 uN normal loads A 1 cycle B 3 cycles eee 123 Sample EBE 200 at 500 uN normal
67. eral load 1 6 kN 2 mN IN maximum normal displacement 50 mm 5 um 0 1 mm maximum lateral displacement 0 2m 15 um 0 6 mm CHAPTER 4 EXPERIMENTAL PROCEDURES Hysitron Triboindenter Procedures Tip Preparation For the most part tips in indentation testing are not considered to be a parameter involved in a test Typically tips are chosen so that they are significantly harder than the material being tested so that it can be assumed that they experience no plastic deformation Diamond is the material of choice for indentation testing because it is the hardest material and its elastic properties are well characterized so that any elastic deformation experienced by the tip can be back calculated from the test data Diamond tips are expensive costing approximately 2000 each For indentation testing this cost is not a huge factor because the tips will last for an exceptionally long time In hardness testing a sharp tip is often desirable to reach high contact pressures and induce plastic deformation In wear testing the two materials in intimate contact are both considered Often wear tests consist of self mated materials gold on gold for example or materials from a specific application e g a bearing on a race It is often found that both materials experience some change In the case of a hard pin or tip moving across a softer flat transfer of the softer material to the hard pin is of interest Observing this material transfer in cross section
68. erical geometry 1 flexure 2 vertical displacement sensor 3 mirrors horizontal displacement sensor pin sample reciprocating stage LVDT D EE 1 A e we 5 Figure 4 13 Schematic of the CSM Microtribometer Sample Mounting The sample is mounted on a SEM microscope stub using a cyanoacrylate based adhesive The SEM stub is fixed to the linear reciprocating stage Stepper motors are used for coarse positioning of test area with the aid of a portable microscope 49 Reciprocating Wear Test The Microtribometer has a normal load range of 0 5 mN 1 N This range of applied normal loads is achieved by having three different types of dual flexures with varying stiffness values Low range normal loading 0 5 100 mN is achieved with a glass flexure middle range loading 50 mN 300 mN is achieved with a stainless steel flexure and high range loading 200 mN 1N is achieved with a thicker stainless steel flexure The appropriate flexure for the desired load range is selected and the testing parameters of normal load track length sliding speed and sampling rate are entered in the computer The machine records lateral load and lateral position From this data friction coefficient and sliding speed can be calculated CHAPTER 5 DATA ANALYSIS Indentation Testing Indentation testing is a convenient relatively non destructive method to determine mechanical properties such as hardness and elastic mo
69. es were coated by first sputtering titanium with a 9keV ion beam at 300 uA for 15 minutes followed by gold with a 9 keV ion beam at 300 uA for 60 minutes The lower current was used due to beam stability issues which could not be resolved over several repeated attempts Once the coating process was complete the silicon substrates were glued to 15 mm diameter atomic force microscopy AFM sample disks using a cyanoacrylate based adhesive The AFM disks are ferritic steel which causes them to be attracted to magnets embedded in the testing equipment stage It was discovered early in this project that the magnetic attraction of the stage was not sufficient to rigidly affix the samples to the stage and an adhesive was needed when mounting samples on the stage of the test equipment Late in this study it was determined that the AFM disks significantly hindered scanning electron microscope SEM examination due to the effect of magnetic fields on the electron beam In hindsight the AFM disks should have been eliminated 19 Electron Beam Evaporated Gold on Silicon Electron beam evaporation is another coating technique for producing high quality thin films on substrates In contrast to ion beam coating electron beam evaporation exposes the sample high temperatures due to the energy required to evaporate the metal This leads to grain growth in the film which results in a coarser grained structure The elevated temperature can also result in diffusion be
70. ese reasons displacement is measured using electron microscopy by taking cross sections from interrupted tests Figure 7 14 shows a transverse cross section from a wear scar The image was produced and tilt corrected as described in Chapter 6 4 platinum 4 carbon ki 4 gold 4 titanium silicon Figure 7 14 FIB SEM transverse cross section of a wear scar run on sample EBE 500 for 10 cycles with a 500 uN normal load Measurements of the film thickness and the depth of the wear scar from the original surface are indicated In addition to film thickness and wear scar measurements the cross sectional image shows the morphology of the film damage Note the curling of the material exuded from the contact area on the right side Features such as these and sub surface damage including fracture and delamination could not be investigated with a top down microscopy technique The microstructure of the gold film is revealed in this image through channeling contrast Several researchers have demonstrated and explained FIB secondary electron channeling contrast using the ion beam as the incident source and a secondary electron detector 48 50 Samples from this study could not be characterized using channeling contrast produced with the ion beam due to extreme damage caused 107 even at low beam currents The image in Figure 7 14 is produced by polishing the surface with a low current 30 pA ion beam and switching the incident be
71. f cycles shown or F for coatings that have exposed silicon F F F The designation S F is chosen because the sample shows small exposed areas of silicon sporadically in the track and is the onset of film failure The gold film is still intact through most of the contact and is expected to exhibit the tribological properties of the diamond indenter tip sliding on a gold film see Figure 7 11 E Beam Spot Tilt FWD Mag A um 5 00 kV 2 52 0 5 061 25 0 kx Figure 7 11 SEM micrograph of sample IBS 100 after 1 cycle of sliding with 500 uN normal load Note the small dark spots in the center at each end of the wear track indicating exposed silicon substrate A note of interest from Table 7 4 is that the survival of the gold coating based on number of cycles is not linearly dependent on film thickness Figure 7 12 is a plot of the targeted film thickness versus the maximum number of cycles observed without coating failure Since 300 was the maximum number of cycles in the tests samples IBS 300 and 103 EBE 500 data points are marked with arrows to show the samples may have survived beyond the indicated data point 100 g 10 m O re 1 E 0 1 Fe er E S S E E ia 0 100 200 300 400 500 target film thickness nm Figure 7 12 Nominal film thickness versus number of cycles prior to film failure Note that the 300 nm and 500 nm thick films did not show failure at the maximum 300 test cycles Number
72. g a small radius asperity An FEI DB235 dual beam focused ion beam FIB microscope described later in greater detail was used to produce a sharp pyramidal asperity with a radius of less than 100 nm A silicon wafer was mounted in the FIB and the normal to the sample was tilted to 20 with respect to the ion beam A trench was milled and the sample was rotated 120 about the ion beam Another trench was milled and the sample was again rotated 120 This 69 procedure was repeated at progressively lower beam currents to sharpen the asperity radius Figure 6 2 Figure 6 2 Sharp asperity milled on silicon to scan indenter tip The sharp asperity sample is mounted on the Triboindenter and the desired tip is installed on the transducer The Triboindenter is set to imaging control which places the transducer into load feedback mode An imaging set point load of 1 uN is selected Using the piezo tube scanner the tip rasters over the sharp asperity feature maintaining the 1 uN load and recording x y and z positional data The x y and z data is entered in scanning imaging software to produce the image seen in Figure 6 3 This procedure is attractive because it can be run just prior to and upon completion of testing to verify tip geometry Because it is an in situ method removal of the tip for characterization is not required This method suffers from longevity of the sharp asperity 70 feature The feature should be made from the
73. g imaging function of the Triboindenter is 38 started a rastering pattern of 40 um x 40 um is a typical scan size therefore a radial spacing of 20 um will ensure that at least one line is caught in the initial scan The inner diameter of 40 um was chosen because it could be imaged in the initial scan size once located The two sets of orthogonal crosshairs rotated 45 from each other were chosen because if only one set was present and one of the lines were aligned closely with the rastering direction that line would not be imaged The two sets ensure at least 3 of the 4 lines will be imaged The reason for the 1 um line width is two fold First a line of 1 um in width can be resolved in an optical microscope For the second reason we return to Figure 4 3 and Equation 4 2 Equation 4 2 can be rearranged and solved for a to give a R R 6 4 3 From the geometry of Figure 4 3 2a will be the minimum width of the line spacing required for a given R and 6 The largest radius tip of interest in this study is R 250 um Hysitron publishes the z axis resolution limit to be 0 04 nm with a noise load floor of 0 2nm A value of 0 5 nm should be within the resolution of the equipment Using these values in Equation 4 4 gives a value for a 0 5 um making the line width 2a 1 um Ifa FIB is not available a similar tip optics offset sample could be produced using a soft polished sample such as the Al 001 and a sharp indenter
74. h steady trend towards miniaturization MEMS devices have received much attention and interest Due to the extensive research and development of fabrication technology for integrated circuit IC chips silicon was a convenient material for pioneering work in MEMS devices It was quickly discovered that for mechanical sliding as in power transmission devices silicon components will fail with limited cycles and could not be used outside of laboratory conditions Wear resistant thin film coatings may provide solutions for silicon MEMS and will require tribology testing equipment and methodologies to support research 1 Interest in tribological study of thin films extends beyond MEMS devices Often in successful wear resistant materials combinations a transfer film from one of the materials develops on the other The result is a self mated contact which after an initial run in period exhibits good wear resistance and low friction Typically when these transfer films exhibit desirable wear properties they are thin on the micron to sub micron scale and very adherent A methodology to study and probe properties of transfer films is useful in the design of contacts that will exhibit them A set of standards for the mechanical testing of thin film coatings is necessary for future developments in the field 2 This research aims to develop a methodology for studying tribology on the nanodisplacement level for both fundamental and practical wear applicat
75. hat no correction is needed and the image retains the same resolution in the height and width Since there is no bulk surrounding the area imaging is less challenging because there is no signal loss due to imaging down into a trench The exposed surface can also facilitate electron back scattered diffraction EBSD if the FIB is equipped with such analysis The process has disadvantages due to complications with some geometries and the time required is considerably longer than the fixed sample method Once the images are collected they can be loaded into Macromedia Flash or similar processing software The images can be registered to each other and generate a movie of slices through the material In addition to movie making software three dimensional reconstructure software such as TGS s Amira can be used to produce three dimensional images and slices in different orientations from the slicing orientation These techniques provide information about surface and subsurface damage such as material volume exuded from the contact plastic deformation zone characteristics and sub surface crack formation and delamination that is unobtainable by top down microscopy techniques Transmission Elelctron Microscopy Analysis After cross sections are produce on the FIB they are analyzed ina TEM Film thickness is determined in a JEOL 200CX microscope at 200kV Figure 6 16 shows cross sectional images of the six different film thicknesses with half cycle wea
76. hat when a sapphire sphere is place in the tip holder it will self center Figure 4 1 A small amount of cyanoacrylate based adhesive is used to attach the sapphire sphere to the OEM tip holder The adhesive is allowed to cure overnight The transducer piezo assembly is then removed from the Triboindenter for easier access Once the tip has sufficiently cured it is threaded on to the capacitive transducer with a special torque limiting wrench 32 R 1 um 608 Fi 250m OEM diamond large radius sapphire tip indenter tip Figure 4 1 Illustration of an OEM diamond indenter tip and assembly of a large radius sapphire tip Sample Mounting The Triboindenter has a stage that contains six positions which are magnetized by rare earth magnets Samples mounted on atomic force microscope AFM disks can be placed on the stage and will be sufficiently held in place for indentation testing In preliminary tests it was determined that this mounting system may not be effective for tests involving lateral translational forces Therefore samples are held in position with an alcohol soluble adhesive in addition to the magnetic forces Once the samples are mounted and the stage replaced on the stage motor the machine is powered on The samples are imaged and focused on using the Triboindenter s optical charge coupled device CCD camera and a safety area is generated by plotting the outline of each sample and recording the points This will define an
77. hod could be employed Since a scanning probe technique would be used to calibrate the tip optics offset a sample was designed and produced to contain features that would be easily imaged and located in both the optical CCD microscope and the scanning probe microscope feature 37 of the Triboindenter A silicon wafer was chosen because of its highly polished surface and its compatibility with the focused ion beam milling technique The wafer was cleaved into a 5 mm x 5 mm section and mounted on a 12 mm AFM disk with a cyanoacrylate based adhesive This specimen will now be referred to as the modified tip optics offset sample and is shown in Figure 4 4 The following features were carefully chosen and patterned on the wafer using an FEI DB 235 focused ion beam FIB SEM series of concentric circles 200 um outer diameter 20 um radial spacing 40 um inner diameter two sets of orthogonal crosshairs rotated 45 from each other 1 um patterned line width lt lt 200 um gt Figure 4 4 SEM image of the patterned modified tip optics offset sample The series of concentric circles were chosen because it is easy to determine from an arc the direction of travel required to head towards the center A pattern size of 200 um outer diameter was chosen because it has been observed that from a tip change the tip optics offset rarely varied by more than 100 um and a 200 um feature is easily located in an optical microscope When the scannin
78. icroscope image of H pattern indents 200 0 eee eeeeseeeeeeeeeeeeneeenee 34 LIST OF FIGURES Spherical indenter on flat surface and cross sectional geometry to solve for depth ME MIAN A E E E E baad uh seen ap abn E o Patterned modified tip optics offset sample y ccciiicrrasiisrnsrieirramnaniiemmnecaiaied T Modified tip optics offset sample centered in the video window scceeee dD e ST a tse adrenal N E e Manual height adjustment windowW sieterinranniendonivnionimasramiannnmiaaercn Search opion WINO W iri eae Imaging window and control panel with image of modified tip optics offset aa eee Actual tip optics offset seen in the video window seseseseeseeceseteeeceeeeeseeeeeees 4 Triboindenter tesknie A nis isirciisBiemesrinoensiencis eae enieeomidaanii einai 2 Rastering wear test showing fiducial indentation marks in circles 0 0000047 3 CSM Microtribometer aise cincssncidicsscartstenceisatoncisiaionisiedayeaielaioncelediseaaab enema Load versus displacement curve along with surface The range of e is indicated for var ious yea tia ia Measured compliance versus square root of maximum load for a load range of 5mN to 10mN on quartz Al 001 and Si 001 siscatnictcsacihcrseieumensoiinissavaairouiete Indent cross sections from Oliver Pharr and Modified Winkler compared to an Ea IE TAE AE E E A PEA A E ase ee ae Differential element described with cylindrical coordinates and tip d
79. interfacial surface 4 Reducing the contact pressure below the yield point of the material will result in an elastic contact and the force of plastic deformation should be eliminated The contact pressure can be reduced by two methods increasing the contact area or reducing the normal load The tests were run using the well characterized diamond indenter previously described and the load resolution is such that a pure elastic contact with this tip geometry is not possible on the Hysitron To examine the friction from an elastic contact tests were run on the CSM Microtribometer in addition to the tests run on the Hysitron Triboindenter Hysitron Triboindenter friction measurements Lateral force is collected during the wear tests on the Hysitron Triboindenter and is normalized with the normal load to determine a friction coefficient Figure 7 17 shows friction coefficient versus the number of cycles for gold on silicon samples without carbon coating run at 500 uN Testing on samples EBE 20 IBS 100 and EBE 200 was terminated at 30 cycles because the film had been compromised by this point on all three Table 7 4 shows that sample EBE 20 had failed by the first half cycle The friction data from this sample is much lower than for the others The friction starts out at 1 6 and decreases to 1 0 This low friction coefficient is probably more representative of diamond sliding on silicon than diamond sliding on gold Sample IBS 100 shows friction above 0 2
80. interlocked asperities on the surfaces which caused a lifting motion to move relative to each other which caused an energy loss and therefore a friction force This theory became known as the roughness hypothesis Subsequent researches found that the friction coefficient did not equal one third for all materials but agreed that the normal load and friction force were proportional The roughness theory remained accepted for hundreds of years Alternate theories involving adhesion were proposed but could not be validated without relating the friction force dependence on contact area 3 A significant contribution to the adhesion theory came from Holm 1938 when he showed that the real area of contact is related to the normal load divided by the yield pressure or hardness of the material 5 It was therefore realized that the real area of contact rather than the apparent area of contact was responsible for the friction force and since the real area of contact was proportional to the normal load the friction force is subsequently proportional to the normal load Mechanism for Metallic Friction The discovery by Holm paved the way for Bowden and Tabor s development of the mechanism for metallic friction Bowden and Tabor showed that the frictional for force to translate a slider along a surface is comprised of two factors the force required to shear the contact between the two intimate surfaces and the force to plough or plastically defor
81. ions A Hysitron Triboindenter was chosen to perform the majority of the wear testing of this study for the following reasons The range of load and displacement The Triboindenter s practical load range is from 1 uN to 10 mN and displacement is from 1 nm to 5 um These ranges cover loads and displacements in MEMS devices thin film applications and modeling simulations The flexibility of tip geometry and materials The tip holder can accept a wide range of materials and geometries There is no need for elaborate manufacturing techniques to produce tips The accurate stage positioning Accurate positioning of test areas is possible to sub micron resolution This allows for precise location of test area prior to testing and post test characterization of the test area The scratch mode The scratch mode of the Triboindenter can be tailored to reproduce motions expected in MEMS devices and other practical applications Reciprocating and unidirectional sliding is a fundamental geometry used in tribological applications and computational simulations The experimental reproducibility Tests on the Triboindenter show good reproducibility This study will present a new approach for tribological testing with a characterized standard which will be available for further studies It will provide a foundation on which to expand research on various materials geometries modes of wear and applications Background Friction and wear
82. ions in MEMS devices and thin films fit in this regime and remain largely unstudied In this study the Hysitron Triboindenter is used to mimic a linear reciprocating tribometer and bridge the gap in scale Hysitron Triboindenter The majority of wear testing done in this study was performed on a Hysitron Triboindenter The Triboindenter is a commercially available nanodisplacement mechanical testing system whose primary function is nanoindentation Figure 3 3 e5 Wig TY al a gt x y stage motor aU lower thermal acoustic isolation cover piezo controller transducer controller stepper motor controller Figure 3 3 Hysitron Triboindenter with upper thermal acoustic isolation cover removed to show interior details 26 Loads displacements and measurements are performed by a capacitive transducer Figure 3 4 The capacitive transducer assembly contains two parallel fixed plates with a parallel center plate supported by springs An indenter tip is mounted to the center plate To apply a load a DC potential is applied between the lower plate and the center plate An electrostatic attraction displaces the center plate towards the bottom plate Based on the spring constant of the support springs the voltage can be calibrated to a force For displacement measurements an AC signal is applied between the center plate and top plate and an AC signal of equal magnitude 180 out of phase is applied to the
83. ior of ultra thin soft metallic deposits on hard substrates Wear 1996 196 1 2 p 171 193 127 15 16 17 18 19 20 pale 22 23 24 25 26 Des 128 Sherbiney M A and J Halling Friction and wear of ion plated soft metallic films Wear 1977 45 2 p 211 220 Shimura Y T Ito Y Taga and K Nakajima Frictional properties of sputtered tin film Wear 1978 49 1 p 179 193 Spalvins T and B Buzek Frictional and morphological characteristics of ion plated soft metallic films Thin Solid Films 1981 84 3 p 267 272 Bowers R C and W A Zisman Pressure effects on the friction coefficient of thin film solid lubricants Journal of Applied Physics 1968 39 12 p 5385 5395 El Shafei T E S R D Arnell and J Halling An experimental study of the hertzian contact of surfaces covered by soft metallic films ASLE Transactions 1983 26 4 p 481 486 Finkin E F A theory for the effects of film thickness and normal load in the friction of thin films Transactions of the ASME Journal of Lubrication Technology 1969 July p 551 556 Schiffmann K I and A Hieke Analysis of microwear experiments on thin dlc coatings Friction wear and plastic deformation Wear 2003 254 5 6 p 565 Dickrell P L Tribological behavior and gas surface interactions of hydrogenated carbon films 2005 Gainesville FL Fischer Cripps A C Nanoindentation ed F F
84. is desirable but would require destruction of the pin Due to the high cost of diamond tips it is not practical to section them It is not uncommon for the pin to experience wear along with the surface For diamond tips wear of the tip should 30 31 be negligible and no transfer film of gold on to the diamond tip has been observed in the scanning electron microscope SEM Self mated materials could be tested by using coating techniques on the tip but again the high tip cost makes this prohibitive An alternative to high cost diamond tips is needed A crucial parameter in wear testing is the contact pressure Contact pressure is used to target desired wear modes and to replicate applications The factors involved in contact pressure are normal load and contact area The normal load can be controlled by the Triboindenter and the contact area varies with the tip geometry In contrast to hardness testing wear tests are often performed with much lower contact pressures requiring a blunt rather than sharp tip By selecting tips of various radii a wide range of contact pressure can be achieved For the reasons mentioned above tips of relatively large radii are produced in addition to the sharp tips available from the manufacturer Sapphire spheres of 100 um to 1 mm are commercially available for a nominal cost of approximately 10 00 each The original equipment manufacturer s OEM tip holder is machined on a lathe with a conical cut such t
85. is not plotted because film penetration was apparent after the first half cycle The figures used to measure the displacement data have been included in Appendix B 108 350 eo 300 250 aa 200 Ei 150 gt A rom o 5 100 F o IBS 100 N 4 EBE 200 50 a9 IBS 300 o EBE 500 0 4 0 1 1 10 100 1000 number of cycles Figure 7 15 Depth of wear scar versus number of cycles for samples run at 500 uN normal load It is interesting to note that both electron beam evaporated EBE samples behave nearly identical up to the failure of EBE 200 The ion beam sputtered IBS samples also behaved similarly with IBS 100 showing slightly more displacement than IBS 300 Considering displacement into the film below the original surface as a measure of wear the EBE samples showed more wear than both IBS samples and IBS 100 showed slightly more wear than IBS 300 A plot of fraction of the coating penetrated versus number of cycles is constructed in Figure 7 16 In this plot the displacement is normalized by the coating thickness to give the fraction of penetration Again if the coating had failed the number of cycles at the observed failure was not plotted and since sample EBE 20 had failed at the first half pass it does not show up in the plot 109 1 0 A y 9 93 o T A o 07 a 2 5 A o 8 5 o 5 S 4 A ie po 6 o IBS 100 E 92 8 4 EBE 200 1 ge IBS 300 j o
86. isplacement based on a spherical tip geometry sates corer anundsnabwuiiakiemGawnate Differential element illustration prior to Lei and after load eT with variable POR Ge ses os cei oo io eee Loading code iow sete ae ee Unloading code Tov WN esorosnsasaia a nie Rastering wear test data plot showing a trenched area on the right of the test area andapileuparca ontis Efi cssnreriineienneisnaieme aA Reciprocating wear test showing the transient areas near the ends of the m and the area over which the friction force data is analyzed 66 6 1 Large radius tip scanning a small radius asperity 00 0 0 eeeeeeseeseeee tenses 8 6 2 Sharp asperity milled on silicon to scan indenter tip s ccsceseecceceseeeceeneeeee OD 6 3 xe produc ed with a sharp ae y first in the x Hon and then in the y direction sinisisi inina omenrmment 6 4 Transmission Electron giaa 200 CX sample holder with a holder fixture and indenter ip MONDO esrrsnsnen in ts 6 5 The upper images are taken at 40 CCW 0 and 40 CW ganita The lower image shows data points plotted along the profile ccceeeeseseeeeeeteteeeee 72 66 Kikuchi map observed in diffraction MoGe cccicssisascsssascnesisarasniiereeesanesiesveecns 79 6 7 Top down view of the single crystal diamond indenter tip with crystallographic orientations and sliding direction indicated cceceseecceteeeetecseeeecneeeecteeseeteeteene TO 6 8
87. k tribometer being a device that prescribes a continuous rotary motion to a disk while holding a loaded pin stationary on the surface Figure 3 1 pin holder A compression 0 35 mm diameter load cell sample 30 000 RPM R spindle S dead weight loads 1 50 N Bia OR Figure 3 1 A typical pin on disc type tribometer 23 24 The reciprocating pin on flat tribometer is another common tribological testing appartus In this configuration the sample reciprocates in a linear direction against a stationary pin through which a normal load is transmitted Figure 3 2 pneumatic cylinder 6 channel load cell Fx Fy Fz Mx My Mz Z stepper motor and ball screw linear table Figure 3 2 Linear reciprocating tribometer Several researchers have begun to use atomic force microscopes AFM to study tribological surface interactions Normal and lateral forces acting upon an AFM tip are calculated from deflections of the tip cantilever From these forces sliding friction is calculated Accurate measurements of sliding friction from AFMs are exceptionally challenging requiring well characterized cantilevers with well know material properties There exists a gap in tribological test equipment from the AFM scale with nanonewton normal loads and nanometer to sub nanometer normal displacements to the 25 microtribometer scale with millinewton normal loads and micrometer displacements Tribological condit
88. l to the imaging surface This requires a free unobstructed view of the surface The following method illustrated in Figure 6 15 was devised to section and image wear tracks The wear test is located with the electron beam and tilted to 52 with respect to the e beam A platinum layer is deposited approximately 1 um thick and twice as long as the wear scar Trenches are milled on each side and in front of the wear scar to about 4 um deep The trench on one side is milled with the sloping cut so an undercut can be performed The sample is tilted to 0 with respect to the electron column and an undercut is made the entire length of the wear scar up to the platinum protection layer 86 The Omniprobe in situ micromanipulator is inserted in the front trench and the material containing the wear scar is lifted up to approximately 45 using the platinum as a plastic hinge Now the sample is tilted 7 and the first slice is taken off with the ion beam The sample is rotated 38 and an image is taken with the electron beam Step 6 and 7 are repeated until the entire wear track has been sliced and imaged Top View Side View pre existing trench from step 3 4 wear scar i l 1 Fi wae Sample undercut cond 5 omniprobe omniprobe 3 Sample 6 Figure 6 15 Schematic illustration of slice and view rotating sample method 87 This process has advantages over the fixed sample process in t
89. le B IBS 100 1 cycle C IBS 100 3 cycle D IBS 100 10 cycle E EBE 200 10 cycle F EBE 200 30 cycle G IBS 300 1 cycle H IBS 300 3 cycle I IBS 300 30 cycle J EBE 500 half cycle K EBE 500 30 cycle APPENDIX B SEM CROSS SECTIONS USED FOR DISPLACEMENT MEASUREMENTS Figure B 1 SEM FIB cross section of sample IBS 100 at 500 uN normal loads A 1 cycle B 3 cycles A Bertie eS ee Figure B 2 SEM FIB cross section of sample EBE 200 at 500 uN normal loads A 1 cycle B 3 cycles 123 124 Figure B 3 SEM FIB cross section of sample EBE 200 at 500 uN normal loads and 10 cycles A enti BSS Figure B 4 SEM FIB cross section of sample IBS 300 at 500 uN normal loads A 1 cycle B 3 cycles AET Sath coo lesol 100x BE aair Figure B 5 SEM FIB cross section of sample IBS 300 at 500 uN normal loads A 10 cycle B 30 cycles 125 AEMET Bha Beer Figure B 6 SEM FIB cross section of sample IBS 300 at 500 uN normal loads A 100 cycle B 300 cycle Figure B 7 SEM FIB cross section of sample EBE 500 at 500 uN normal loads A 1 cycle B 3 cycle AEREE BENER Figure B 8 SEM FIB cross section of sample EBE 500 at 500 uN normal loads A 10 cycle B 30 cycle 126 AXE B Figure B 9 SEM FIB cross section of sample EBE 500 at 500 uN normal loads A 100 cycle B 300 cycle 10 11 12 13 14 LIST OF REFERENCES Walraven J Failure mechanisms in mems in
90. lower plate and the center plate When the center plate is equidistant from the top and bottom plates the net signal is zero When the center plate is displaced a voltage is recorded and calibrated to a distance A similar transducer is mounted to provide lateral displacements lateral transducer Figure 3 4 Capacitive transducer assembly schematic The capacitive transducer assembly is mounted to a three axis piezo tube scanner similar to those found in AFMs The piezo tube scanner allows for precise tip placement and can be used in conjunction with the transducer as a scanning probe microscope 27 Using the tube scanner to raster the indenter tip in contact with a surface the transducer can collect z displacements and produce an AFM like image The tube scanner is mounted to a z axis stepper motor which provides coarse translations for tip approach An optical CCD microscope is also attached to the z axis stepper motor The optical microscope allows for accurate sample location and test placement The sample stage is attached to x y stepper motors for translation Figure 3 5 All systems are computer controlled Z axis stepper motor O T CCD optical G microsco e y EEI S p piezo tube scanner x y stage assembly Figure 3 5 Enlarged view of Triboindenter components The components are mounted on granite to dampen vibrational effects The equipment frame also contains an active vibr
91. ly available 0 6 mm thick single crystal silicon wafer with 001 plan view orientation was scored and cleaved along lt 110 gt directions to produce substrates of approximately 25 mm The substrates were cleaned by sonicating in acetone This was repeated three times using new acetone each time The substrates were then sonicated in methanol three times using new methanol each time Finally the substrates were blown dry using laboratory grade Freon spray Several substrates were mounted for coating in a Gatan Model 681 High Resolution Ion Beam Coater seen in Figure 2 3 Figure 2 3 Gatan Model 681 High Resolution Ion Beam Coater 18 It is well known that gold has poor adhesion when deposited on silicon and a thin titanium bond coat will greatly enhance gold film adhesion A target of 99 99 gold was mounted on one side of the dual target holder using conductive silver epoxy and a target of 99 99 titanium was mounted on the opposite side After sufficient curing time the ion beam coater chamber was pumped down to a level of 2 x 10 Torr and the ion guns were purged with argon following the instruction manual After purging the ion beam was set to 9 keV at 600 uA for 60 minutes with a sample rotation speed of 30 rpm Based on a deposition rate of 3 5 A s the film thickness should be on the order of 1 um The first batch of samples was run without a titanium bonding layer to examine the effect of film substrate adhesion A second set of sampl
92. m the material in front of the slider Their equation for friction force takes the form F F E 1 2 where F is the frictional force F is the interfacial shearing force and Fy is the force of plastic deformation or ploughing They expand this equation by breaking down the interfacial shearing force to an interfacial shear stress times the area it is acting upon and breaking down the force of plastic deformation to the pressure required to yield the material times the projected area of the slider in contact with the surface in the direction of sliding and Equation 1 2 takes the form F HA 7 A 1 3 real where H is the hardness of the material which is taken as the pressure required to deform the material A is the projected area of the slider in the direction of sliding and in contact with the surface ts is the interfacial shear stress and Aj ej is the area of the slider in contact with the surface Figure 1 1 lateral frontal area A eal normal area Ap Figure 1 1 A schematic diagram representing the slider and areas from Equation 1 3 Bowden and Tabor tested this equation with a set of three sliders one with the geometry of a circular spade one spherical and one cylindrical all three with the same radius Figure 1 2 The tests were made with steel sliders on an indium surface with a slow sliding speed of 0 1 mm s to avoid strain rate sensitivity From the geometry of the sliders it can be seen th
93. mained above 0 25 112 0 6 EBE 20 IBS 100 0 5 4 EBE 200 w T so l x IBS 300 U 0 4 Q9 4 o EBE 500 a A z Saa w ios re SX 9 ue ane 03 2 o DRA an i oan iS rk T E ene 02 film failure o _ 0 1 1 10 100 1000 number of cycles Figure 7 18 Plot of friction coefficient versus number of cycles on a log scale for gold on silicon samples without carbon coating run at 100 uN normal load Figure 7 19 shows friction coefficient results for gold on silicon samples with carbon coating run at 500 uN normal load The dashed line indicates a friction coefficient of 0 25 Again samples exhibit friction coefficients below 0 25 at or near the number of cycles when the gold film is penetrated Comparing Table 7 6 to Figure 7 19 EBE 20C shows failure immediately and the friction coefficient begins below 0 25 Sample IBS 100C fails between 3 and 10 cycles from the table and the friction coefficient drops below 0 25 at 5 cycles Table 7 6 shows sample EBE 200C failed between 10 and 30 cycles and the friction coefficient drops below 0 25 at 14 cycles The remaining two samples survived beyond 30 cycles and their friction coefficients remained above 0 25 Friction data was lost for these samples beyond 30 cycles due to an error in the wear cycle load function code Friction data does not seem to be sensitive to the carbon coating at the 500 uN normal load and the data does not show a correlation to the carbon delamination eve
94. mined versus sliding cycle 66 60 transient steady state sliding range transient lateral force uN 0 0 2 0 40 sliding position um Figure 5 8 Data from a 10 cycle reciprocating wear test showing the transient areas near the ends of the reciprocation and the area over which the friction force data is analyzed CSM Microtribometer Frictional force data is collected and analyzed in a similar manner to the reciprocating test in the Hysitron Triboindenter An extensive development of data collection and analysis was presented by Dickrell 22 CHAPTER 6 CHARACTERIZATION Tip Characterization The indenter tip used for the majority of testing in this study is a single crystal diamond Diamond is a convenient material due to its high hardness and resistance to wear In addition it is well known that gold does not wet carbon which will be beneficial to avoid material transfer from the film to the indenter tip Due to high costs and limited available geometries indenter tips of various materials may be sought out The following methods can be applied to most tip materials Tip geometry is critical in analysis of indentation and wear testing Properties such as elastic modulus hardness shear stress and friction coefficient are all calculated from contact areas based on tip geometry Inaccuracies in tip geometry will cause significant errors in contact area calculations which will propagate to the above mentioned properties
95. n have shown that this equation holds true for any E axisymmetric indenter and will extend to pyramidal indenters with a correction factor for different geometries 27 A function to describe contact area from displacement along with Equation 5 11 gives a convenient method to measure the reduced modulus from data shown in Figure 5 1 The Poisson ratio of the material of interest is often unknown therefore reduced modulus rather than Young s modulus is often reported For a Berkovich indenter of ideal geometry the relationship between contact area and contact depth can be calculated from simple geometry A 24 5 h 5 12 However due to the inability to produce a perfectly sharp indenter some radius is expected at the tip apex For large indents this radius is negligible but for nano displacements it cannot be neglected An area function to relate contact area as a function of contact depth can be generated by indenting on a fused quartz sample of know elastic modulus E 72 GPa A series of indentation tests of increasing displacement over the desired range are run on the standard sample Load versus 54 displacement plots are generated to find the stiffness as described above By solving for h from the relationship in Figure 5 1 P h F Ries E re 2 and Equation 5 10 solved for area and plotting A versus hg a curve fit is generated in the form A h C h Ch Ch C hZ Ch Ch 5 14 Indentation
96. near testing was demonstrated in the Hysitron Triboindenter and proved to be the best method for data collection and reproducibility A method to mount a wide range of tip materials and geometries was developed to test various contacts and pressures A systematic approach was designed to locate wear regions throughout testing and characterization An optics offset calibration method was developed to overcome the limitations of light optical microscopy Wear test placement was repeatable within 0 5 uum A variety of characterization techniques including nanoindentation x ray diffraction ion beam milling and electron microscopy were employed to analyze sample properties and quantify results Established indentation models overestimate hardness in materials that exhibit pile up An alternative model was formulated and applied to load displacement data Results of this alternative model show good agreement with the Hall 117 118 Petch relationship for the grain size of the films tested Wear test lateral load was measured with the Hysitron Triboindenter transducer and showed good repeatability Displacement data was characterized by direct measurement from SEM micrographs due to drift issues during extended testing times Friction coefficient showed good agreement with an alternate testing method and proved to be a good indication of film failure The methodology developed in this study gives well characterized reproducible quantitative data f
97. ns to and useful discussions about this project Thanks to the students of the Tribology Lab for support and comic relief Thanks to the MAIC facility and its staff for use of the instruments financial support and discussions about techniques and analysis Finally I would like to thank the faculty at the Materials Science and Engineering Department of the University of Florida As an undergraduate I was inspired to pursue research in this field from the knowledge enthusiasm and support from Richard Connell Robert DeHoff Mike Kaufman and Fereshteh Ebrahimi TABLE OF CONTENTS page ACKNOWLEDGMENTS rriena ES iv CET OF TABLE ain eS ix LETOFTFIGURES sapii r T x ee as Ta ss aso ane ae a estes i a es XV CHAPTER 1 INTRODUCTION AND BACKGROUND iiss cxiushsalrosnisoccunninsnienbaiah aiaabysavoeieniocianes 1 Tne eaaa 1 e T A AS ve EDR RED L E NNN eg HET UOT Slur A E UG AAT EEN eT reef 3 Early Fachon SWIC a eae ae oO tee NS ele A E Mechanist for Metalle 0 essorer EASE 5 SE WON a a EA EEE 7 Pron TOD ii ee ee 7 SG i ieee icin cedar ect degree aie 8 OI 5s acerinessniwipalarcieaniers toadsasatecinbsinortieussie inion Gyseta el aeeremasndasiinee 9 Film Thickness Wear and Friction Dependence 0i ccnccecissadeoacisasconnrrccnerssnceassins 9 M ALT ERA E E ATE E E a T E E E AE EAEE TE E 11 2 EAPERIMENTAL Meee epee eer etre ieai 13 Rift i telat crs elt See eee eee eee eee eE IT fate One RTT Tne ate ee mT pEtren nnn Ie ere en 14 PE eect a
98. nt 113 0 6 0 5 m o z A a oa o o E 0 4 j w a ee O x R A So ia x i ees 593 Eriak PS Ora ONG Gig Ono Ole or akg eD a wes ea We aide ee ay er E x AOE N EO E RON E ER R ee oe T film failure w 0 2 u A AAAAAA a a ee 0 1 e ee E 2 4 0 5 10 15 20 25 30 number of cycles Figure 7 19 Plot of friction coefficient versus number of cycles for gold on silicon samples with carbon coating run at 500 uN normal load Dashed line indicates a friction coefficient of 0 25 The friction data for gold on silicon samples coated with carbon run at 100 uN load shows significant differences from the previous results Table 7 5 shows the carbon coating survived on all samples up to 30 cycles with the exception of a delamination of EBE 200C at the 30 cycle micrograph The friction data for these tests is plotted in Figure 7 20 The friction coefficient for all samples ranged from 0 12 to 0 17 and the point of delamination on sample EBE 200 was not detected in the friction data Friction data beyond 30 cycles was lost for these samples due to the same error in the wear cycle load function mentioned previously The lower normal load of 100 uN allows friction measurements of the diamond indenter tip sliding against the amorphous carbon coating to be measured whereas the 500 uN normal load friction is dominated by properties of the gold 114 0 3 friction coefficient S o D a N a 9 0 05 0 5 10 15 20
99. nt 0 1 0 0 1 10 100 number of cycles Figure 7 21 Friction coefficient versus number of cycles plotted on log scale for a contact pressure of 92 MPa run on sample EBE 200 The friction coefficient from Figure 7 21 should contain no plastic deformation component and should be representative of the interfacial sliding term from Bowden and Tabor s model 4 Virtually no deformation was detected in SEM examination of the wear area on this test sample This result suggests for the samples that failed on the Hysitron the contact remained plastic removing material until failure A combination of work hardening and increased contact area might eventually develop an elastic contact on 116 thicker samples resulting in a more wear resistant contact but testing for higher numbers of cycles is required to investigate this possibility CHAPTER 8 CONCLUSIONS Considerations for testing thin films on the nanodisplacement level were examined and addressed A test matrix was designed to investigate thin film wear testing over a range of film thickness normal load and number of repeat cycles Gold was chosen as a test film for its well characterized properties and applications in MEMS A variety of sample preparation techniques were employed with best results obtained from commercially processed electron beam evaporated and in house ion beam sputtered coatings Various testing modes and parameters were explored Reciprocating li
100. od by static forces and it is lifted out of the sample trench The sample is removed 83 and replaced with a carbon film coated copper TEM grid The static attraction between the carbon film and the cross section is greater than the attraction between the cross section and the glass rod so the cross section can be deposited on the carbon film The process is illustrated in Figure 6 13 The cross section is now ready for TEM examination Figure 6 13 Cross section is removed from the sample trench with a glass rod attached to a hydraulic micromanipulator and then deposited on a carbon film coated copper TEM grid Slice and View Serial Sectioning The FEI DB235 FIB microscope is capable of milling with the ion beam followed by imaging with the electron beam Using these techniques a series of images can be produced similar to tomography From these images movies can be assembled to produce a view of moving through the bulk of a sample and with the aid of computer software three dimensional reconstructions of the sectioned sample can be generated There are alternate methods for producing a series of slice and view images The first 84 described will be referred to as the fixed sample method The second technique will be referred to as the rotated sample method Fixed sample method In this method the sample is not rotated once the sectioning process has begun The sample is oriented orthogonal to the ion beam or
101. on factor 40 A simple method to approximate the sample texture is obtained in the JCPDS files Each file contains space group lattice parameter planar spacings and relative peak intensities for perfect powder samples A perfect powder sample is equiaxed or randomly oriented therefore by comparing relative peak intensities collected from a sample of interest with those of the JCPDS file one can draw conclusions about the texture Samples were scanned from 30 to 135 20 with a step size of 0 02 Figure 6 9 shows 20 scans for the six different gold coatings used in the test matrix All six show a strong 111 texture with no other significant orientations 3 lt x 20 40 60 80 100 120 140 20 EBE 200 500 000 400 000 silicon 100 5 _ 300 000 200 000 100 000 0 20 40 60 80 100 120 140 20 EBE 400 1 400 000 gold 111 5 1 000 000 lt x 11 IBS 100 10 000 silicon 100 5 lt 0 20 40 60 98 100 120 140 20 IBS 300 100 000 silicon 100 80 000 gt 60 000 40 000 20 000 silicon gold 100 222 Figure 6 9 Scans of the six different gold films shown with indexed peaks Equiaxed polycrystalline gold has a Young s modulus of 78 5 GPa 41 whereas single crystal gold is elastically anisotropic with the lt 111 gt orientation being the stiffest It is reasonable to assume that the textured film will exhibit the elastic properties of the single crystal rather than those of an equiaxed
102. oose debris If the slider continues to ride over virgin material prow formation continues throughout the wear test With dissimilar materials friction and contact resistance characteristics of the system are determined by the prow metal therefore a diamond slider on a gold surface should behave as a self mated gold contact 7 Antler found that the size of prows is inversely related to the hardness and metals that do not work harden fail to form prows He attributes this to the prow becoming harder than the surface from the severe deformation and then being able to cut the surface 8 Cocks proposed the following mechanism 9 for prow formation adhesion occurs between the slider and the surface plastic deformation occurs in a volume of metal from the surface adhesive forces are sufficient to remove the metal from the surface the metal transfers to the slider as a chip the process repeats on the chip the chip becomes too large and detaches from the slider Antler found that with a linear reciprocating slider prows would be deposited at the ends of the tracks and prow formation would eventually stop when the surface material work hardened to a hardness higher than that of the slider in which case would cause the slider to begin wearing From empirical observations the following equation was developed _ 4 X cm 1 4 my O RN where Nprow is the number of cycles for which prow formation will continue x is the track length in
103. polycrystalline sample therefore the Young s modulus is calculated to be Equciti gt 117 GPa using Equation 6 1 and the stiffness matrix for gold 78 190 42 3 423 0 42 3 190 423 0 C TEF 42 3 42 3 190 0 L 0 0 0 161 0 0 0 GPa 0 gt O o 0 0 0 0 161 0 0 0 0 0 161 Grain Size The grain size of a metal affects the yield strength by the Hall Petch relationship 37 2l o 0 kD 6 3 where oy is the yield strength D is the grain size and oo and k are material parameters To elucidate the plastic response of a material the grain size should be well characterized Scherrer first showed that decreasing particle size resulted in slight deviations from Bragg s law in diffraction due to incomplete destructive interference which results in peak broadening 42 Peak broadening can be due to particle size or grain size and residual strain and is usually a combination of the two Warren and Averbach devised a method to represent diffraction peaks as Fourier series They showed that by comparing families of peaks 111 and 222 for example coefficients due to strain and grain size could be separated because the strain coefficient is dependent upon the index of the planes whereas the grain size coefficient is not These calculations have been incorporated into software packages for stain and grain size analysis To collect data for the grain size and strain measurements the diffractometer was set to scan
104. r scars run at 500 uN normal load Samples are named based on the deposition techniques electron beam evaporation EBE and ion beam sputtered IBS and the targeted coating thickness Acutal coating thickness is indicated on the images 88 IBS 100 EBE 200 IBS 300 EBE 500 Figure 6 16 Cross sectional TEM images of gold films with wear tracks run for half cycles at 500 uN normal load Film thicknesses are indicated on each image CHAPTER 7 RESULTS AND DISCUSSION Static Indentation Testing A series of fifty displacement control indents were performed from 10 nm to 100 nm on each sample varying the displacement incrementally from test to test The tests were analyzed using all three methods mentioned in Chapter 5 namely Oliver and Pharr O P Stone and Modified Winkler MW For each sample hardness is plotted versus maximum tip displacement Oliver and Pharr Analysis Results A plot of the hardness results based on the Oliver and Pharr model is shown in Figure 1 16 s e00 14 eee e poet e e ee eo eo Sa e i j e EBE 20 o IBS 100 i e EBE 200 e IBS 300 g 8q o EBE 500 a g A x 6 eg o T e er a a008 8S0 4 o P gV P 00 90 o 208 000r Se K e a ad one 2Ps 898000000 oh P9 A 2 a e po oP oe 0000o goee C0 M O 0 maximum displacement nm Figure 7 1 Plot of hardness determined by the Oliver and Pharr method versus maximum tip displacement for five samples of varying thickness
105. results in less number of cycles to produce buckling and exposure of the sub layer It was found that increasing layer thickness can reduce or eliminate the mechanism 12 Film Thickness Wear and Friction Dependence Several research groups have investigated thin metallic films as a solid lubricant in sliding contact applications 4 13 17 All found that friction and wear were initially high for ultra thin coatings but as coating thickness was increased both friction and wear went through minima values at some critical coating thickness Figure 1 3 Table 1 1 summarizes the results from the various research groups In some studies tribometer geometries were not well defined and experimental details were not mentioned Two of the research groups studied sputtered coatings versus evaporated coatings and both found the sputtered coatings exhibited lower friction coefficients and longer life than the evaporated coatings 14 17 Both groups attributed this behavior to better film adhesion and finer grain structure resulting in higher coating hardness exhibited by sputtered coatings Antler improved his coating by alloying gold with cobalt 13 The cobalt acts as a solid solution strengthener to harden the gold film 10 u for bare metals 3 c Ro 4 U jes Qe O C hs critical or dikea is minimum p ick coating O U Umin ultra thin thin region region critical thickness increasing film thi
106. rom reciprocating wear tests on the nanodisplacement level The data set from this study will be extended to predict wear rates and friction coefficients using finite element analysis APPENDIX A SEM IMAGES OF EXPERIMENTAL WEAR MATRIX Figure A 1 Images showing the wear scar of gold on silicon at 100 uN load normal load Conditions showing transitions from Table 7 3 are included A EBE 20 half cycle B IBS 100 300 cycle C EBE 200 300 cycle D IBS 300 300 cycle E EBE 500 300 cycle B C D E 119 120 Figure A 2 Images showing the wear scar of gold on silicon at 500 uN load normal load Conditions showing transitions from Table 7 4 are included A EBE 20 half cycle B IBS 100 half cycle C IBS 100 1 cycle D EBE 200 10 cycle E EBE 200 30 cycle F IBS 300 300 cycle G EBE 500 300 cycle 121 J Figure A 3 Images showing the wear scar of carbon on gold on silicon at 100 uN load normal load Conditions showing transitions from Table 7 5 are included A EBE 20 30 cycle B EBE 20 100 cycle C IBS 100 300 cycle D EBE 200 10 cycle E EBE 200 30 cycle F EBE 200 100 cycle G EBE 200 300 cycle H IBS 300 300 cycle I EBE 500 100 cycle J EBE 500 300 cycle 122 B E H 7 J K Figure A 4 Images showing the wear scar of carbonon gold on silicon at 500 UN load normal load Conditions showing transitions from Table 7 6 are included A EBE 20 half cyc
107. sample has been extended beyond 3 cycles Table 7 6 Results of SEM micrographs from carbon on gold on silicon samples with 500 uN normal load cycles 105 Electron Microscopy Cross Sections of Wear Scars A dual beam FIB SEM was used to make cross sections of the wear tracks as described in Chapter 6 While the Hysitron Triboindenter is capable of measuring displacement the instrument is sensitive to thermal drift during the extended amount of time required to run a reciprocating wear test Figure 7 13 shows a plot of vertical tip displacement versus lateral tip displacement for a 30 cycle test 200 8 oO vertical tip displacement nm 0 2 4 lateral tip displacement um Figure 7 13 Plot of vertical tip displacement versus lateral displacement for a 30 cycle reciprocating test Points are plotted in red at time 0 and progress in color through the spectrum ROYGB as time increases The data points are shown in color beginning in red at the start of the test and as time goes on the color of the points transition through the spectrum to end in blue at the end of the test The initial vertical displacement of the tip is in the 100 nm range and 106 increases in displacement initially Thermal drift eventually overtakes the vertical displacement and the tip appears to be at a negative displacement at the end of the test This contradicts force measurements and electron microscopy of cross sections For th
108. shows results of the micrographs from the entire test matrix Selected micrographs showing transitions from survived coatings to delamination and or to failure are included in Appendix A Gold on silicon without carbon Five different target thicknesses of gold on silicon were produced by either electron beam evaporation or ion beam sputtering as described in Chapter 2 Table 7 3 summarizes the micrograph results for tests with a 100 uN normal load using the designations described above Table 7 3 Results of SEM micrographs from gold on silicon samples with 100 uN normal load Designations are S for coatings that survived to the number of cycles shown or F for coatings that have exposed silicon F F As seen in Table 7 3 all coatings survived up to 300 cycles with the exception of the EBE 20 coating with a target thickness of 20 nm which had failed by the first 4 cycle Table 7 4 summarizes the microscopy results for tests with a higher normal load of 500 uN Here again the sample EBE 20 did not survive any number of reciprocations This is an expected result because as in most applications higher normal loads result in more material removal Sample IBS 100 shows that the coating survived the first 2 cycle but the table entry for the one cycle test is S F 102 Table 7 4 Results of SEM micrographs from gold on silicon samples with 500 uN normal load Designations are S for coatings that survived to the number o
109. sing the indentation mode prior to running the test An automated pattern of eight indents surrounds the test area Figure 4 12 An atomic force microscope AFM scan of this area with the fiducial marks is necessary prior to testing In this test a normal load is prescribed in the wear function editor along with an area to raster over and a scan rate The number of raster lines is set at 256 The area of interest is selected in the optical CCD microscope Once the test is started the tip approaches and contacts the surface applies the desired load and the piezo rasters the tip over the prescribed area for a given number of cycles Figure 4 12 SEM image of a rastering wear test showing fiducial indentation marks in circles 48 Reciprocating Wear Test The scratch test mode is modified to replicate a reciprocating linear tribometer This mode was originally designed to place a normal load on the tip and using the 2D transducer translate the tip to produce a scratch The test mode is often employed for testing relative coating adhesion By using this method with a low normal load and programming multiple passes in the function editor a nanodisplacement scale reciprocating tribometer is emulated CSM Microtribometer Tip Preparation Tips on the CSM Microtribometer were mounted on the cantilever flexure using a cyanoacrylate based adhesive 22 Figure 4 13 The tips are commercially available BK7 optical glass lenses with a semi sph
110. structures film thicknesses number of reciprocations and normal load effects on wear resistance of gold films on silicon substrates Gold was chosen for its applications in MEMS devices electrical switching contacts and its thermodynamic stability at ambient temperature and pressure Standard metallographic polishing techniques were initially used to prepare sample surfaces Wear testing and profilometry revealed surface roughness effects which excluded the use of bulk gold Polishing media particles were embedded into the surface changing the properties of the material and the roughness produced was on the scale of the measurements taken Therefore commercially available polished silicon wafer was chosen as a substrate material for its low surface roughness In addition silicon has direct applications in MEMS research A variety of microstructures and sample thicknesses were produced using deposition techniques including ion beam sputtering IBS and electron beam evaporation EBE An amorphous carbon layer of approximately 60 nm in thickness was deposited on half of the samples to explore an electrically conductive friction modifying substance The samples with and without carbon will be referred to as coated and uncoated respectively Samples were named based on the deposition technique and the targeted approximate film thickness in nanometers Table 2 1 Normal loads were chosen of 100 uN and 500 uN because the instrument shows good repeatabilit
111. sults in Figure 7 9 will be denoted with an S for coatings that survive the number of cycles of the given test and denoted with an F for coatings that show removal of the gold to expose the silicon substrate One half of the test matrix was run with an evaporated carbon coating on top of the gold coating Some of these samples exhibited a delamination of the carbon coating which could occur with or without failure of the gold film Figure 7 10 illustrates each case Figure 7 10 SEM micrographs of sample EBE 200 C run for 3 cycles left and 30 cycles right at 500 uN normal load On the left the carbon film has delaminated and the gold is worn but not compromised On the right the gold film has been worn through and the silicon substrate is visible Results from Figure 7 10 will be denoted with a D for samples that show delamination of the carbon without exposure of silicon below the gold or denoted with an F if the silicon is exposed Each test condition was repeated four times to produce a test matrix with twelve different load and cycle combinations on five different samples both with and without carbon coating on top of the gold coating yielding a total of 480 tests Every wear scar was imaged to determine whether the coating survived delaminated or failed Micrographs from the repeat tests were very consistent with similar results in at 101 least three of every set of four Table 7 3 through Table 7 6
112. t 20 25 kX magnification using a secondary electron detector 30 um aperture 5 keV accelerating voltage and number 2 spot size Some samples displayed beam sensitivity so image quality was sacrificed to protect the 80 sample from repeated rastering with the beam Each wear test from the matrix was imaged and examined for coating damage and breakdown Focused Ion Beam Milling A Dual Beam FEI focused ion beam FIB DB235 microscope was used for grid line etching as mentioned earlier This microscope is equipped with a field emission gun FEG high resolution SEM column situated vertical Mounted 52 with respect to the electron column is a liquid metal ion source LMIS FIB column using gallium for the ion source The microscope has a gas injection system GIS which allows platinum deposition by heating the organic metallic compound methylcyclopentadienyl trimethyl platinum to 313 K and then opening the injection system in the microscope chamber The low pressure of the microscope chamber 10 Torr causes the compound to out gas and adsorb on the sample surface Scanning with the ion beam reduces the compound leaving a platinum layer on the surface The area of platinum deposition can be controlled by the rastering area The microscope is also equipped with an Omniprobe Autoprobe 200 which provides in situ mechanical micro manipulation Transmission Electron Microscopy Sample Thinning Samples for TEM examination must be at least less
113. ta eatcoiear ie eae eee ate 15 lon cai Sputtered Gold On SiO wien ac ne ee 16 Electron Beam Evaporated Gold on Silicon sisiascosinctarrsscnicisianacerphedicraaiebes Gadorsieds 19 Carbon Are FO isan aier snare naar EAEE 20 Ton B ara Etched Area of Interest eissecar ea 21 a TESTEOUMMEN cate doeccosea ia eee 23 Hysiton Tibondem ei aseensa 25 TE ci isin teaches cite bs ae earache ab napaeceeapllins 28 d4 EXPERIMENTAL PROCEDURES ereaicssaaevirscoeetcicen eeinersanleunnaeseenns 30 vi Nn Hysitron T Triboindenter Procedures Tip ee Offset pamena Standard tip optics offset pro Modified tip optics offset proced cae E Ta Mod CS x Depth sensing eee Rastering wear test ao wear test CSM Microtribometer Tip Preparation PE war OK E E aA Kies aissa it Indentation Testing iver Phar Meth d CSM ae ea CHA Scanning Electron Microscopy EE E ta via ed wanes voetacaauees e pir Bem aan Fixed ate method Rotated sample thee nission Electron Microscopy A vii Giane Weare ree VR rosin a aedaeaed eae emma 89 Oliver and Pharr Analysis Restis nurunin naene 89 Stone Analysis Results ceeee eshte panne sents cates ule reuse T E Modified Winkler Approach E E ena eteneTe 95 Rastering Wear Tests mien assis E TEETE AEE TE EETA P s Reciprocating Wear TosiSusnssnasn E E E ieee Pilm Fa a encore apres ieLbieimanmaemide 99 Gol
114. ternational Conference on Vapor Growth and Epitaxy 2005 275 1 2 p e1849 Uchic M D M A Groeber D M Dimiduk and J P Simmons 3d microstructural characterization of nickel superalloys via serial sectioning using a dual beam fib sem In Press Corrected Proof Barrett C S and T B Massalski Structure of Metals 3rd ed 1966 New York McGraw Hill Davis J R P Allen S R Lampman and T B Zorc eds Asm Handbook Volume 2 Properties and Selection Nonferrous Alloys and Special Purpose Materials 1998 ASM International Materials Park OH BIOGRAPHICAL SKETCH Gerald Bourne graduated high school in 1982 and worked in the automotive parts industry until 1986 He then began an entry level position in a serigraphic textile company He learned the trade and opened his own company in 1988 By 1990 competition with foreign manufacturers lead to declining profits and caused him to seek other avenues He joined a small service company and helped them to expand to a nationwide parts and service supplier In 1992 he met Wanda Hix and fell in love They married in the spring of 1993 and their daughter Gabrielle was born in November Wanda had the opportunity for a new position in 1995 so they moved to Gainesville in the Fall Gerald went to Santa Fe Community College and completed his associate s degree in 1996 In 1997 he transferred to the University of Florida He completed his bachelor s degree in 1999 He accepted the
115. testing data can now be analyzed from eqn 5 1 and 5 11 to obtain hardness and reduced modulus The Oliver and Pharr method includes the following assumptions Material does not pile up outside the indent The unloading behaves as an isotropic elastic half space The tip area function is well approximated over the range of displacements Compliance Method Stone s Method When designing mechanical testing equipment engineers choose stiff materials to reduce the effect of machine compliance upon sample data however machine compliance cannot be ignored Compliance is defined as the inverse of stiffness Compliance of the entire system Cr can be written as C C C 3513 where Cmn is the compliance of the load frame machine and all other compliances except the sample and C is the compliance of the sample The load frame compliance includes sample mounting tip frame and stage The following method is suggested by Hysitron to determine the compliance of the machine 28 The inverse of Equation 5 10 can be substituted in Equation 5 15 for C to give 55 5 16 C C or 2E JA solving Equation 5 1 for area and substituting yields i 5 17 2E P max Cr C Assuming hardness is constant at large indentations a series of large indents can be run and plotting the inverse of the total measured compliance versus 1 Pmax will give y intercept equal to Cm This method can be run on various mounting techniques
116. than 200 nm thick for most materials For HRTEM samples must be typically less than 100 nm thick Producing cross sectional TEM samples of a uniform thin film using conventional techniques is challenging Attempting to produce site specific cross sectional TEM samples with conventional techniques of micron sized features would require a degree of luck that would be better suited for playing the Florida Lottery Fortunately the FIB microscope has the capability of locating micron sized sites and milling thin cross sections from them 81 Once the wear tests were finished samples were selected for TEM examination On those samples the wear tests were located using the grids lines The wear test was oriented such that the cross section would be orthogonal to the sample surface and either orthogonal or parallel to the wear direction For cross sections that were parallel to the wear direction indicator marks were etched for final thinning process Figure 6 11 Fiducial x marks were milled for automated milling Figure 6 11 SEM micrograph of wear track in center of image to be prepared with cross sectional TEM parallel to the wear direction Both indicator and fiducial marks milled by ion beam are indicated with arrows Platinum was deposited 2 um x 15 um x um high to protect the area of interest while milling Trenches approximately 15 um x 5 um were milled at 5 nA beam current to 6 um deep at the base and sloped out to about 1
117. thin film transition phenomenon seen in Figure 1 3 20 He developed two theories to explain the two regimes The ultra thin regime is explained as stiffening due to coupled stresses with a quantitative expression verified by data from literature The coefficient of friction in thin film regime is shown to be proportional to the square root of the film thickness divided by the normal load based on a thin film with modulus much less than the substrate Jang and Kim and Schiffmann and Hieke incorporate a ploughing term in their wear models which takes plastic deformation into account 14 21 For these models tip geometry contact geometry and plastic properties of the film and substrate are necessary parameters Difficulty in measuring these parameters is attributed to discrepancies with the models and experimental results The combinations of substrate properties film properties and thicknesses sliders and geometry lead to different modes of wear Due to the complexity of the problem no single model or solution is expected to explain all possible wear combinations and geometries A model that describes the friction coefficient of a thin soft film on an 12 elastic substrate of significantly higher modulus is not expected to describe a hard elastic film on a soft compliant substrate A variety of models need to be developed to explain specific cases CHAPTER 2 EXPERIMENTAL MATRIX An experimental matrix was designed to test various micro
118. tween the film and substrate Again a bonding layer is used to enhance film adhesion Titanium and chromium layers are often used for gold deposition on silicon Since titantum has a lower diffusivity in gold than chromium titanium was selected The electron beam evaporation process is illustrated schematically in Figure 2 4 An electron beam is generated by producing a large electrical bias between the electrode and the crucible The crucible contained the desired film material The electron beam heats the material in the crucible to a liquid and then to a gas which fills the chamber The gas condenses on the substrate thereby producing a thin film on the substrate Vacuum Chamber Substrate _ so Electrode E beam Figure 2 4 Electron beam evaporation process 20 Attempts were made to use an on site electron beam evaporation system but coatings were of a poor quality with many spheres of gold on the sample surface Fortunately high quality electron beam evaporated coatings on silicon 001 substrate are available commercially at economical prices Wafers of 100 mm diameter with gold films of 20 nm and 200 nm were purchased and sectioned into samples as described previously The sectioned samples were then glued to AFM disks using a cyanoacrylate based adhesive Carbon Arc Evaporation To investigate an electrically conductive friction modifying coating all samples were duplicated with a thin amorphous carbon layer
119. y and low signal to noise ratio at these levels With the 100 uN normal load setting number of reciprocations was 13 14 varied from one forward pass referred to as a half cycle to 300 cycles in the following increments half cycle one cycle three cycles 10 cycles 30 cycles 100 cycles and 300 cycles Although the reciprocating tip is diamond and is therefore much harder and more wear resistant than the gold foils some precautions were taken to avoid possible tip damage For the 500 uN normal load setting the 100 cycle and 300 cycle tests were omitted if film failure was detected on samples at 30 cycles to avoid excessive number of reciprocations on the harder silicon substrate The combination of loads and number of cycles results in 12 different test parameters Each set of tests was repeated four times on all 12 samples for a total of 576 wear tests Table 2 1 Sample naming designations Approximate sample Without amorphous With amorphous thickness carbon carbon 20 nm EBE 20 EBE 20C 100 nm IBS 100 IBS 100C 200 nm EBE 200 EBE 200C 300 nm IBS 300 IBS 300C 400 nm EBE 400 EBE 400C 500 nm EBE 500 EBE 500C Sample Preparation Gold was chosen as the test material for its applications in the electronics industry and its mechanical and chemical properties described previously A variety of samples were produced and can be categorized into three basic types including bulk gold foils ion beam sputtered gold on silicon wafer an
120. y and works well in indentation applications with sharp small radius tips For tribological applications a blunt radius tip is often desired so an alternate method had to be developed Standard tip optics offset procedure As mentioned above this optics offset procedure involves optically imaging residual indents A relatively soft polished sample is required for this procedure A single crystal aluminum sample with 001 out of plane orientation is included with the 34 Triboindenter This sample is mounted on the stage The open loop load function editor is set with a maximum load of 10 mN This load with a sharp tip of less than 5 um radius will leave a residual indent that can be easily imaged with the optical CCD microscope A safety zone is defined around the aluminum sample and an area free of indents is chosen to perform the calibration The optics offset procedure is chosen from the software menu and the machine automatically translates the tip to the approximate position above the center of the optical focus The user is prompted to use the z height control to lower the indenter tip to within 1 mm of the sample surface Once the user does so an automated pattern of seven indents in the shape of an H is performed on the sample at the prescribed load Figure 4 2 The machine then translates the sample back to the optical microscope and the user is prompted to set the microscope reticule over the center indent The computer then recor

Download Pdf Manuals

Related Search

Related Contents

FUJITSU Public Sector Solution 農業用水管理システム  Samsung PL90 manual de utilizador  XBA-30 - ソニー製品情報  HD 10/25 (440 V)  Checkolite P1006-71  Multimedia Interface MERCEDES Bedienungsanleitung  Schwa User Manual Version 1.0  LM-X1 user manual  ATTENTION - AquaGlide  Maximatic User Manual  

Copyright © All rights reserved. Failed to retrieve file