Home

John`s AMS Instrument Developoment

1. 41 Vaporization e Optimal temperature e Heater temp vs heater power everyone should generate with the new instruments e Three axes heater current heater voltage and temperature Plot for Heater Temps e IE vs heater temp e IE AB vs heater temp e Sometimes at higher temperature these drop off parabolic shape 42 IPP vs Heater Temp Dioctyl Sulfate DOS also has a peak in the IPP around 550 C Believe in the field it isn t as sharp but it is true that more fragmentation occurs at higher temperature Jose don t know what he asked here Doug That could be part of it but it doesn t explain all of it Heater Experiments 550 C is a special temperature Different results from Frank for ammonium nitrate sodium nitrate and potassium nitrate looking at the width of the particle signal in TOF space See those plateaus by looking at the single particle pulse width lt 150 ms vaporizing all of the particle Also see in the counting efficiency approaching 100 at the same temperature Copper sulfate ammonium sulfate These curves are why we chose 550 C minimize fragmentation of organics and maximize inorganic vaporization Contradiction with some other results Tim sulfuric acid Doug vaporization is even easier with sulfuric acid make ammonia sulfuric acid SO3 and water Should see some sodium nitrate Ann did see sodium nitrate at that temperature but need to check how
2. Correction wave AB measured with MS mode no idea what s going on peaks in Correction Factor Instruments response to the AB should scale linearly with AB change in sensitivity of particle phase should be scaled with AB Every time a EM cal is performed ideally should see AB go down with time and then back up after EM cal in practice it doesn t always work for a bunch of reasons Analysis code just scales the AB Doug wave defined to scale all the masses Jose see sharp jumps like this mass cal of quad is shifting James program is able to deal with that Wave called CorrFac contains all this info 26 MS Averaging Tab Use same batch files to calculate mass loading from Hz Whenever a mass is calculated the cal factors are applied Eiko Ammonium applied from 0 44 How is it done Doug talk about this tomorrow Johannes wrong mass scale how to correct it using James program James can handle changes to slope or off by more than two m z never had that problem before should be relatively easy to stick in Generates legend with mass loadings for each species scaled but average MS shown with nitrate equivalent Can also display signal in logging software choose all for data to average open and background raw data Doug rephrase that anomalies in the data that don t make sense by showing how the sticks were calculated from the raw data John colorize James put in list Jose look at wavefor
3. Jose don t tune the AMS unless really have to Heater setting look at the temperature increase for PSLs forget to decrease Tune ionizer paying attention to m z using 28 or 32 and not 18 28 better if chopper opened any peak background particle etc with chopper closed was old test seemed to be OK Manjula newest version tunes with it open but no one has it yet Doug Bold font for mass used Jose paste whole thing into PowerPoint James old new difference in bits or percent is gone which one Manjula will check that Doug new ionizer control module beginning of the mission after stable tune it 6 times in a row walk better to optimal voltages 63 Performance Checks contd Calibrate multiplier make sure threshold is set OK could have a problem with dark current generating noise Doug real question criteria for setting gain not the knee but how much signal you have look at the signal you have to determine voltage you want Jose change from yesterday kind of big 32 save screen in PowerPoint Doug real issue in MS mode full scale for the data acquisition board the 28 signal is about 2 3 of full scale too high uses up multiplier faster too low run out of dynamic range for big particles change the multiplier voltage to get signal you want then go back into EM cal and by definition the middle voltage is the one you were set at decide based on that if you need to make a change Jose Disagree do
4. s components go to 440 Hz Transformer on board Maybe a UPS will do this 53 Frank Intercomparison of Semi continuous instruments in NY R amp P 8400S Hering ambient particulate sulfate monitor denuder RH controlled impacted on heater flash vaporize in SO2 analyzer 10 minutes HSPH continuous sulfate monitor PM2 5 cyclone denuder tube furnace w SO2 analyzer averaged for 10 min PILS no PM2 5 cut point denuder condense water and impact goes into IC system Also some filter samples 6 hr and 24 hr at Queen s College 24 hr at public school Sampling at Queens College Different sampling inlets for filters and semi cont instruments Overview of all the data Correlation plots all SC compared to PILS all the slopes close to 1 R2 gt 0 9 recovery 1 intercept forced to zero Filter samples slopes close to 1 R2 gt 0 97 recoveries 0 97 to 1 08 6 hr about 6 higher than 24 hr Typically see less sulfate compared to the filters incomplete sampling of M2 5 100 nm cutoff for R amp P AMS inlet transmission PILS losses for large particles and small unactivated particles 2 10 less HSPH conversion loss 5 Filters shorter inlet lines filter artifacts one had denuder but not corrected for blanks Filter samplers had regular cyclones not a very sharp cutoff still measure 10 at 5 microns filter collects everything AMS and PILS had cyclones but intrinsic inlet cut points Other SC had sharp cut cyc
5. 48 best S N not out of the question to just use 48 Jose disagree interferences maybe get rid of 64 Manjula larger mass on the delta 2 scale photochemical events have short chain higher mass more distinct and only from combustion James 57 is more shifted towards the lower mode than 55 but further up get the same sort of ratio but have to check Jose 71 is shifted more toward smaller mode Manjula even go up to 95 Jose more S N at higher masses James background is really bad at 44 Jose watch the fluorine peaks 197 50 69 100 119 150 169 285 is the high one James may have other peaks from other pump oils Phil masses to use if see fire 60 major peak from levoglucosan Number of Masses to Scan 2 5 MHz 30 masses Play with it and see when it crashes do it before you go to the field wait for two saves page lock error Put shortcut onto desktop with automatic start in alternate mode to get program to start after crashing John suggested S M Li Prevent automatic startup when no one is there pumping on Doug get UPS that won t turn back on again Jose UPS turn off computer Eiko RCDs if they loose power they stay off Jose system is robust filaments protected background increases automatic valve that is not hot or little channels that loose particles need compressed air for pneumatic valve S M Li ball valve 112 Next AMS Users Meeting Oct 26 28 2003 at Aerodyne Sun Tues
6. AAAR Oct 20 24 in Anaheim Format Doug keep the structure of this meeting identical with update on hardware and software then scientific issues and results training issues are separate one Training students in April People are interested Doug driven by new orders we would determine the schedule but open it up to anyone not just new users likely necessary John vote on presentations for viewing Hugh no time for posters point to bring people together common areas submit overheads to a single speaker summary of what people have done 113
7. Ambient organic to OC Manjula Houston within 10 Doug extra scaling factor of 1 5 to 2 associated with poor focusing of ambient particles Ann Using delta pattern to refine factor for calculating organic mass from OC measurements with Tim Bates on ship Hugh organic is size dependent need to look at that quite carefully Jose mode dependent Need to be investigated in more detail Hugh nitrate data matches not internally mixed with sulfate Richard Why Hugh acidic sulfate young fresh emissions nitrate forced out Jose nitrate in sodium nitrate mode also calcium nitrate focusing differently Summary Small particles down to 30 40 nm Large particles up to 700 nm Open question for larger particles need converging orifice Aspherical particles need moveable wire and better generation and characterization of larger particles Jose APS is not a black box Ann measure particles optically in front of the lens Alice Darin s working on that James use VOAG to generate particles Jose AMS was the only diagnostic that VOAG had a problem works poorly at 1 2 microns OK above that James worked OK for him at 1 microns Hugh small sizes match to measured distribution non spherical particles with non uniform density effective density of 3 no particles below 90 95 nm Jose depends on aerodynamic diameter Hugh wrapped up in effective density haven t gotten to the bottom of it but that s what we see
8. NYC more likely primary aerosol soot with adsorbed OC Jose going on the surface area mode Hugh morphology Ann constrain Organic with other measurements Doug come a long way from last year 56 Summarize Different Factors nitrate sulfate organics ammonium 1 IE 1 0 87 0 7 4 5 2 Fragments TBD TBD TBD unmeasured 3A Collection 1 if externally NYC 2 34 Small mode Use 1 if shape mixed with sulfate Houston 1 7 externally mixed unless divergence of Or same as Flight 2 5 soot derived same as beam sulfate if internally Vancouver 1 7 1 8 sulfate if internally mixed RAB a Accumulation mode OMBTOwN Es internally mixed same as sulfate not if shifted to smaller sizes 3B Size cut of lens different for each lens f size distribution Discussion Manjula compare with MOUDI Doug mandate size efficiency factor divergence factor explicitly separate them John yes Doug physical reason for losses before and after the lens use size distribution to do closure Thomas constant or controlled conditions check TSI DMA CPC independently APS difference within 5 get sulfate mass overall convert with calculations only thing is to run AMS Dutch guys developing total dissolved carbon OK with stable reference system not field measurements where it s unstable John this is what the user s meeting is about framework for doing those types of studies Ann refractory material
9. balance worked well 30 46 ratio also 3 1 Jon huge pulses of ammonia in California Jose pulses of nitrate in the early morning Atlanta but here it goes into the aerosol because ammonia so high Manjula lots of ammonia in Mexico City Mobile Measurements NYC Buses Try to get emission factors for various vehicles Measure CO2 and aerosol loading Nice correlation between them measuring plume emission index loading of PM per ppmCO2 Nice thing to cover a lot of vehicles and characterize under real world conditions better than dynamometer 92 NYC mode Alternate TOF MS every 2 s MS 6 spectra average TOF 6 masses 1 3 s at each mass Diesel buses unburned fuel oil in the aerosol phase 2 3 particle loading associated with soot Size distributions averaged over event mask out whole event to only look at the data you want Small mode organics buses only large mode sulfate both buses and ambient asphalt particles Discussion James Which masses TOF 57 69 43 44 55 83 48 64 deltaO 2 sulfate and a high negative delta for aromatic 91 or 95 Mass spectra busy all over Johannes was sulfate ever higher than organics Uphill No pm sulfate was less than 0 01 microg m 3 In Mexico City ambient sulfate loading goes down behind the truck Johannes catalysts Some buses with NO NO2 filter to trap particle filter oxidized to clean need to be careful about contamination from other
10. different ambient sampling environments Thomas use the stable data to compare Sept 2002 Comparisons with MOUDI from Pittsburg Looks very close S M Li MOUDI sample need to be careful to invert data issues of particle bouncing Not AMS Stuff we don t measure Crustal metals 5 BC 5 refractory salts NaCl sea salts mineral sulfate etc Water lose 60 measured 40 based on lab studies with sulfuric acid need to do it for ammonium nitrate ammonium sulfate Scaling factor for organics Frank multiply for the 15 lost then the pie is full Doug s point Doug water is not 0 on TEOM Ann SVOC Frank used new differential TEOM agreed well with the other ones so SVOC minor fraction of the total mass 58 Other instrument issues PILS is low Cyclone cut points 5 10 loss of filters compared to PILS Line losses for PILS Frank correction factor determined with filter not PILS Manjula sulfate filter and sulfate PILS compared well in Houston depends on sulfate mass above 2 5 micron Jose Teflon filters not quartz Doug 9 Ipm throwing away routinely put in filters and measure sulfate doesn t cost much to collect John simple colorimetric tests for filters Doug doesn t exist at this level need IC Jose need to be better than a factor of two otherwise useless Filter TEOM MOUDI have artifacts volatile losses or gains reaction of species during sampling Phil AMS vs NIST
11. is that we are adding up all the ions they use calibrated specific ions We know the gain of the multiplier and add up all the fragments to get total signal or absolute mass loading Gas Phase lonization Cross Sections Phil Silva s data Relative ionization cross sections vs Mass Nitrate equivalent cross sections vs mass nitrate organic 0 7 Blue line normalized to 1 nitrate is lower because not counting all the fragments Clearly do more work to extend calibration to more species but this seems to work basis for relative calibration factors Particle collection efficiency taken out but the range of error bars is an indication of that Everything were doing is essentially based on this curve 38 Aerodynamic Lens Focusing Plots from John s paper Streamlines and collection efficiency Since then managed to manufacture transmission efficiency plot using field data Zhang et al lens transmission curves including Best lens that has never been made have impact limit of 40 45 on first lens particle bounce in lens and some fraction is still transmitted no good way of generating particles Small Particle Transmission Curve Transmission vs Da Smaller critical orifice to limit flow rate Show Edinburgh and Manchester data two different lenses NYC data low end cutoff due to transmission high end could be due to transmission Other studies NYC plume chasing not necessarily actual
12. Gizmo 3 D Plotted sulfate as f altitude longitude latitude 15 s MS 15s TOF Image plot July 22 sulfate plume above air mass came from Ohio river valley Systematic testing of Restek Hot first to degas then cool more important to keep temperature constant Doug most important to keep it clean get room up to 35 C over the weekend Jose constrained flow little ventilation John frame had to be enclosed with Plexiglas sides forced air to flow thru system took off toward the end heat the AMS dissipated was supposed to go outside but it didn t work AMS on G 1 not the biggest draw of power 2 6 Amps 600 W 220 V other big pumps on the system Airplane 28 V was very clean on Twin Otter but problem with AC of plane made from DC on plane different from ground changed heater bias when switched over implemented auto tuning every 10 minutes put everything on AC ground was well defined Balzers DC power supply Airplane Discussion Took the AC DC supplies used Balzers box to get 24 V took 20 Amp supply out and it saved space and conserved power worked fine Most of the power load went to the heater chopper and servo took very little Airplane supply of 28 V Could be more efficient Asked to run on 220 V because everyone else using 110 New systems are universal can plug into up to 240 V Earlier version has a switch Doug G 1 everything went through one UPS different power supplies can cause trouble John better
13. N2 CH4 haze formation due to photochemistry Radical initiation ultimately forms polymers Monomer particles lt 0 5 microns Photolysis of CO2 forms O atoms could stop the polymerization of CH4 109 Why use AMS Morphology or composition changes during collection want real time in situ analysis method AMS Lower pressure than atmosphere size calibration Try working at constant pressure Initial results w o CO2 complex mixtures of organics 27 28 peaks separated by 14 amu delta analysis 77 91 105 aromatic species Size distribution is uniform internal mixtures of various organics Increased amount of CO2 C O smaller large peak at 58 44 72 84 86 and 97 More CO2 smallest C O lower total signal higher masses aren t present 58 peak increases and decreases as CO2 increased For similar conditions AMS 60 70 nm SMPS 200 nm shape and density affects High accuracy MS to aid in identification 0 005 amu accuracy could identify species based on isotopes Early Earth Haze Aerosols Manjula 58 cuts the carbon chain formation gets smaller and smaller David Electric discharge produces ions UV produces radicals lightning shock waves cause thermal chemistry different results from different energy sources Doug Upgrade Options Inlet RH P amp T at the inlet Older versions add lens P for inlet flow measure 1S420 ionizer control upgrade compact ionizer High throughput lens Requires
14. Twin Otter exhaust mainly 0 2 relative ratios are different from other aircraft exhaust Ship Plume UMIST 0 2 2 on top of ambient data Jeff Brooks bringing emission info together for Pacific study Eric Williams for NEAQS2002 Winter and summer Manchester pollution look very similar except winter has more 0 than 2 and summer has equal 0 and 2 early morning data not affected by photochemistry rush hour locally dominated pollution qualitatively similar CEH Edinburgh natural gas heating morning peaks own building similar delta 2 0 2 slightly different ratios Prophet organic only event diesel Afternoon delta 2 dominant then delta 0 and 3 comparable 72 Photochemistry Houston 43 is shooting up Aged plumes UMIST Cheju Do 43 is really big delta 3 then 0 and 2 Pacific UMIST went by too fast Crete Mainz sources 1 2 days away again 43 delta 3 is really big also see 0 2 and 2 chain length may be shorter because fall off is sharper Note that all the fragments of sulfate and other organics have been taken out in calculating the deltas Write procedures to calculate delta pattern functions Own little panel that comes up Makes delta fragment waves in batch file copies element over from organic fragment wave in James program calculate time trend as f time and can also get the average delta pattern and colorize mass spec average Way to incorporate into James program What Frank has d
15. and particle counter Jay Harvard Forest BC Whiteface Mtn Summer 2002 Harvard Forest Time trace shows big variations in mass loadings diurnal patterns in the various species see big peak in sulfate in the afternoon Back trajectories for large sulfate looks like long range transport for Aug 20 Other back trajectories also produce single mode at 500 nm with all the species internally mixed Manjula DC offset for sulfate not set right Yes Manjula Issue for the logging software after EM cal Trajectories with a Boston source organic and sulfate strong but offset from the other size distributions James back trajectories all coming from too close to the surface Jose high side on the sulfate tails more than the other species look at the single particle pulses and invert the data Doug large particle mass for sulfate is counted take out the tail by adding the single particle pulses when they appear remember that high particle side means spread out in the distribution Jose calibrate with ammonium sulfate sulfate shifted compared to ammonium similar with something less volatile 103 BC Results amp Discussion BC 24 floor window Sulfate events were much smaller than observed at Harvard Forest systematic error in collection or real Higher loadings of organics than sulfate compared to Harvard Forest Most of the high sulfate organic events with air from the SW Back trajectories on Aug 12 long range transport from e
16. are 0 particles at 0 voltage multiplied by a thousand fix it by changing the operating up end Hugh high concentrations at low voltages need to be very skeptical exactly for that reason Still took measurements with AMS 107 AMS results For the same S content same speed Low power see organic pattern with very low sulfate High power see organic pattern with very high sulfate Catalyst designed to reduce NOx ARI see different results Doug Different technology than anything we ve sampled Conversion efficiency Johannes haven t made quantitative calculations Doug mass 81 is still smaller than 48 and 64 pure sulfuric acid shows mass 81 larger water distribution is from sulfate decomposition H2SO4 to H20 SO3 John Need more organic masses programmed there to show what else with the large mode water Doug pure water at large mode John catalyst temperature studies No Where is the sulfur With more S content don t see a quantitative increase Size distribution very similar to SMPS for high S content high power 100 transmission efficiency Doug Dilute by factor of 100 what does that do to the water nucleation particles only observed after a cold start but too much noise in the size distribution Jay water uptake by Soot Flame source dry DMA condense H2S04 don t see sulfuric acid nucleation then H20 future control future t controlled inlet then AMS AMS shows changes to the pa
17. expensive makes sense if the users change the ion source refurbish with the filaments thing with EM also makes sense not to throw away the ceramics John ionizer cleaning procedure Guenther simple take apart lenses and clean with scotch bright or sand with plastic paper with small grit and supersonic baths Abrasive cleaning followed by chemicals trichloroethylene in old days Ethanol and PA in supersonic bath with 100 C Jose heater can go to 1200 C heat up overnight decreases the background Guenther Discuss with Unexus split possibility to include such devices into the vacuum chamber Jose already have it Guenther might not already be done for ions can get rid of the junk but it has to go somewhere Doug shooting low volatility material directly into the ionizer walls of the ionizer become outgassing source so heating the oven to 1200 C can be effective because it heats the ionizer and walls Guenther but ion source sensitivity could also be changed so need to develop UHV procedures 14 G s Questions contd John Degassing procedure Is 10 6 or 10 7 Torr too high Guenther slightly increase emission current to degas but don t switch degas on do for about 15 20 minutes John emission current or voltage change Guenther both Jose baking with filaments off Guenther baking with filaments on filaments are last to switch off measurements with thermocouples lasts about 1 hour because heat loss is
18. names James Feedback appreciated Can add extra bits read me at the top of the file Manjula thank James and acknowledge in the paper Hugh reference JGR paper please bear in mind that James is starting his final year of his PhD Manjula Tutorials from Wavemetrics Balzers Guenther short discussion of introduction for user of this instrument Licht 1 week course 1 2 day course here Manjula Link for next year s AMS meeting Guenther special questions before we make the course Jose email list 30 Jose Figures of Merit Parameters to keep track of Particle transmission efficiency 100 relative to CPC at 350 nm Sensitivity detection limit 3 sigma of 1 min data when there is a filter on the inlet benchmark for NO3 5 ng m 3 G 1 instrument same as signal to noise ratio Doug compare to prediction S M Li filter Jose any kind of filter Signal Intensity IPP for m z 30 46 when sampling 350 nm mobility diameter pure dry NH4NO3 particles calibrated DMA with PSLs Duke PSLs can be off by 4 in Dp or 15 by mass maybe use NIST PSL size standards instead More Measures of Signal Intensity e lonization Efficiency ions detected per molecule evaporated same as IPP more meaningful units determined by mass calibration oldest 2e 7 ions molecule newest 8e 6 1e 5 related to amps per Torr not used any longer because it is not fully determined Air Beam Signal depends on instrument configura
19. nice peak shape Jose Balzers manuals on the AMS web page Manjula also on the ftp site Results till Oct 2002 Increased emission factor of 4 6 Compact ion source factor of 2 Magnets no improvement Spiraling might be causing peak broadening John electrons the only thing that get perturbed Mass Filter theoretically factor of 2 Lower deflection voltage factor of 2 other deflection unit maybe factor of 2 3 Thoriated filaments 13 G s Questions Eiko Can run 2 filaments at 2 5 mA 1 25 mA each to get linear increase in ionization efficiency without degrading filaments as quickly Guenther two filaments at the same time changes distribution of ions in the source but move up in space charge limits Jose separate controls Guenther discuss with electronics guys about 18420 test if minor change Jose effect of magnets Bigger ionization volume Guenther increases paths of e higher Jose time broadening Guenther distribution of ions greater in time Jose so volume bigger Doug spend time tuning the ionizer better different tuning procedure than now may be able to reduce the tail by playing with the ion optics Eiko done with compact source John Yes but testing with old source G s Questions contd Doug both non modified I1S420 and old ionizer in the field procedures and cost of upgrading Have another Worthwhile to clean Guenther replacement of the ion source need to look into how
20. of the air beam use only that one Ideally use only one calibration but need to determine good way of doing that Won t spend ages doing corrections it s already done Recalculate sticks m z calibration drift and handles open closed MS differently reduces artifacts Error Parameters same as before Doug Units for background James lons counted per square root seconds don t use it Doug put it there James artifacts in logging software calculates baseline from that 25 Using Deluxe for Lab Data contd TOF corrections clean TOF signals removes last few points too high or low Overwrite DC markers checks all the m zs and recalculates TOF traces Ann recalculating Da for whole dataset James velocity cal changes w f pressure ultimately apply to software but not in there yet Ann IE corrections applied to all data James yes Doug this is all James program for correction should do should be up to the individual user to fix James waste of my time to include all kinds of processing everyone should do their own Jose write your own program and execute it on top of James program James open source code appreciate any feedback give acknowledgement if used for paper Paper for JGR accepted that contains the majority of the underlying math in addition to Jose s paper good place to start from Talk to me directly about the nuts and bolts Examples Eiko mark range of runs to apply correction to
21. potassium showing up in MS James multiplier vs filament off Jose can get some surface ionization if filament not off first Example TOF mode look at MS Rel 6 kind of high John any masses in TOF mode Jose yes they should all be the same unless some noise is coming from the quad Doug there will be an easier way to do this with the new EM mode Jose a lot of noise Doug one number to look at threshold number for a mass w o background should be about 0 4 bits figure out why if higher Jose many other parameters that are related and agree how to talk about noise 33 lonizer Tuning on Screen Walk the voltages and look at the signal John do on 32 because 28 is saturated Doug the most important thing during tuning routine look at curve there should be a peak no peak is indication that something is wrong this one is tuning as well as I ve seen size of peak varies once pumped down for fraction of a week this stays stable and shouldn t be checked unless AB changes dramatically John should check but don t accept changes Doug in practice haven t been able to go back to old signal with old values Doug cancel doesn t do anything except close window Jose need to input old values in Balzers menu look at tuning log for old values John was a big change for heater bias Jose biggest influence on signal is heater bias Eiko not seeing a clear peak John you need to talk to me Jose extraction voltag
22. quantitative 43 FP_0801_ 1 amp 2 Study Remnants of propellant when cannon shot off Labeled as sulfate and organic in the spectrum seeing zinc m z 64 66 and 68 in the spectrum Point is that we were able to change the oven temperature in a minute or so Mass spectra at the two temperatures fragmentation pattern changes Organic signal largely disappeared Problem with cycling because background is changing with time So background subtraction doesn t work Very clear that the physics of the vaporization process if the oven temperature is varied we see different things If we don t vary the oven temperature we are never going to see refractory components in the atmosphere Frank Mass Concentration How to calculate Use several correction factors for omitted masses fractionation patterns IE differences IE measurements Sampling Losses 44 Fractionation Patterns Experiment Scheme Results ammonium nitrate and sulfate Ammonium masses 1 2 15 16 and 17 nice pie chart with relative intensity relative error Nitrate masses 14 16 17 18 23 28 etc Sulfate masses even more fragments Biggest unmeasured fragment is 18 for H2O from sulfate Correction factors ammonium 1 05 16 17 Nitrate 1 09 30 46 Sulfate 1 77 lonization Efficiency Nitrate defined as 1 Others Organics 0 7 first correct for all fractions and isotopes for other species 45 Collection Effici
23. recently been able to do this Haven t looked at the sulfate scaling factors based on the impactors at this point Saw three types of aerosol China Korea MBL aged Also saw nitrate in impactor samples that the AMS didn t see associated with the coarse mode Sulfate not too bad Focus on Stage 1 of Berner with AMS 0 2 0 5 microns Haven t folded Frank s new factors in here but sulfate OK nitrate looks very good Water soluble organic James nitrate loading in the accumulation mode might have organic interference Looked at 30 and 46 fragmentation pattern as f Da ratio 30 46 about 3 1 in accumulation mode above 1000 nm the ratio is much higher See significant amounts of this material despite the lens transmission Clearly refractory material has higher 30 46 ratios than ammonium nitrate Doug really vaporizing HNO3 with ammonium nitrate CaNO3 makes NO NO2 leave Ca20 on the oven similar with NaNO3 so there s a really good chemical reason for why this happens Jose signal above 2000 nm PM10 impactor on the inlet Hugh more nitrate than sulfate but it s all in the coarse mode Doug orifice clogging No 81 Jungfraujoch 12000 ft AMS CPC OPC DMPS Carbon HTDMA in ethylene glycol Heated inlet to drive off cloud water from residual aerosol then take in all of the aerosol Other inlet PM1 on the end no cloud droplets Inlets switched every 10 minutes Jungfraujoch Data Summary from total period Very
24. site Jose changes in the m z calibration of the quad temperature changes in the RF box try to keep that constant only correcting every time it saves so keep that in mind Doug one minor comment make sure you miss a save whenever you do something to create a gap in the dataset to indicate that something happened 66 James Program Mass Spec time series turn off scale to AB time trend of all the species classified Thicken lines and add legend Various events Next scale to AB be aware of how large the corrections are MS Average scale to AB MSSDiff plot new graph add negative log scale One by one go thru species On time trend graph look at ammonium divided by sulfate Lowest loading for ammonium 16 O not corrected from AB correctly So need to go to the batch waves Jose filter on inlet periodically to check O signal in 16 Doug all those traces colors came from batch file AB etc defined Fragment list has how species are determined from m z Scaling factor is at the end of the batch file Going to the fragment waves E g 28 is only for air 14 is only for air 15 from N15 in air is 0 004 times signal at 14 Isotopes are put in here Discussion about 30 contribution of organics to 30 and organic nitrate S M Li will be doing lab studies looking at fragmentation pattern Doug grey in MS is from air colors determined by fragmentation pattern CO2 is from the particles accounting for the air at
25. spectra e How comparable e Systematic biases e Affect on organic analysis Examples Anthracene m z 178 parent also seen with AMS best case Succinic acid m z 100 parent still pulled out of NIST search AMS biased toward lower fragment masses than NIST plus thermal decomposition making 18 and 44 Bigger issue as more oxygenated malic acid almost all 44 and 18 in AMS spectra Correlation Plots Anthracene spectral intensity coded as a function of m z follows close to 1 1 line Oleic acid lower masses not as well correlated Dicarboxylic acids are worst 60 Plot of 44 total signal vs C Roughly 9 44 than compared to NIST Look as f O C ratio water may also be important compounds with more water give less CO2 take CO2 and H2O together as indicator get pretty good correlation with O C Doug Average 44 slope equal amount of the particle CO2 signal is appearing as water get estimate of how much water could be from thermal decomposition of organics small fraction Maybe add into organic mass Summary Aromatics close agreement Heavily fragmented species get more signal at lower m z Oxygenated compounds thermally decompose into H2O and CO2 Frank temperature Hugh peak of sensitivity as f T always around 550 C Frank relative m z 44 as f T Hugh yes Doug can t be simply decomposition not enough CO2 in molecule to give such high signals actually fragmentation after ionization
26. still new some correlation with organics Rodney s new TOC instrument Comparison of Commercial AMSs shown in tutorial Aerosol Climatology Try to have a draft before the next user s meeting Hugh John don t think it s too early themes that run across the data sets James nail technical problems Doug don t believe that that s an issue issue of what we miss compared to 2 5 is not important Hugh remote vs urban if nothing else it provides us with some kind of harmonization across the groups thinks that by next year those things will be addressed Jose start first draft and circulate give him presentations James proposing that Jose would be the first author on it S M Li special issue somewhere and use this paper as an umbrella Jose special session at AAAR was not focused James get other papers out better if with special issues for each campaign Jose include the ones that have already been written up T7 Doug Comparison of predicted NH4 to calculated NH4 Pittsburgh Looks like fully neutralized ammonium sulfate and nitrate high organics maybe some excess ammonium Jose sulfate plumes that were not fully neutralized still see regional aerosol Not allowed to use the command line James to do Time trend mass spectrum image plot of size distribution size distribution all in one macro delta pattern and average concentrations Doug Clarification Vaporizing everything at the same ti
27. time before start up heater turned on as soon as P5 and P6 up to speed vacuum interlock set to nominal power maybe cook it initially at higher temp Progressively step up with ionizer to 2 5 mA Trained someone to start up instrument run it 48 IE AB vs time Multiplier voltage was very stable m z 28 was not in F6 TOF list Constant ratio of IE AB 0 75 pAB Systematic trend of relative gain factor preflight and post flight always lower post flight decays slight interpolation end of day flights enough time to ammonium nitrate calibration and used those numbers for the flight Doug notice the threshold flaw in EM calculation is the threshold John operate in the manual mode Doug note the number because get different values depending on exactly how the threshold was set John this represents an average Jose more stable reason for keeping the pumping multiplier more stable as it s pumped over time John AB complicated unless did single ion we would never know John G 1 Operation End of flight isolated whole system noted Baratron pressure gauge always less than 2 Torr leak rate of less than a few Torr per day Keep it isolated Corollary of background signal m z 55 f time beam closed case ion signals measured after the first few hours saw decay with time in 1 5 hrs dropped about factor of 3 after takeoff background increased due to increase in cabin temperature was this seen when always pump
28. with sulfate sometimes not HYSPLIT model predicts air came from Ohio Valley Blow up of GIZMO plot longitude latitude altitude with sulfate mass loading Nice to develop as we go along Data not corrected for changes in inlet flow 79 Hugh UMIST Studies Long list of 8 campaigns in the past two years starting with Edinburgh Nov 2000 ending with NAMBLEX Ireland Aug 2002 James thesis on urban studies Also looking at various cloud aerosol interactions NAMBLEX chamber studies with seaweed Winter in Manchester Studied two periods with different modes of organic aerosol low winds and high winds Low winds stable air pollution develops over a number of days very common Looked at back trajectories power stations to the south southwest never see sulfate when the back trajectories are not from that area 80 Summer in Manchester High activity and low activity periods Winds from the south high activity and continental outflow low activity Fragmentation analysis looking at the modes of different species One large accumulation mode 600 nm for 44 Delta3 45 Delta4 two modes for 43 higher smaller mode 100 nm for 55 and 57 Weighted size distribution for organics organic distribution is skewed if you don t pick the right peaks Doug note that the 44 and 45 Only need to count three masses to account for most of the variability ACE Asia Comparisons with impactors only
29. 00 signal increase e Small volume ionizer IPP 800 S N increase e Magnets IPP 2000 but impact on size resolution e High emission current ionizer IPP 8000 dynamic range limited Need to evaluation S N increase Performance Upgrades High throughput lens designed needs to be built requires V301 Converging inlet tested to increase transmission of larger particles Conversion dynode tested Balzers SEV218 mass bias eliminated but ion collection efficiency reduced Restek glass coating on detection cylinder implemented not yet quantified may reduce water and organic background Upgrade Plots Summary Converging Nozzle Plot Essentially a cone to channel larger particles Conversion dynode plot SI vs lon mass ETP has mass bias esp above 100 amu but SEM218 doesn t Approaching ng m 3 detection limits but IE 10 5 may be the limit max Operational Upgrades Development Temperature controlled inlet understand scaling factor Inlet temperature pressure and RH measurement fully characterize sampling conditions New rack system to transport easily Beam width probe understand scaling factor need to engineer and software Optical particle detection problem with background scattered light but working prototype get higher size resolution and properties of non volatile particles Plots Design concept for single rack mount right or left handed Wire probe of beam width 75 micron wire ac
30. I on the wing of the twin Otter run as a total filter AMS had lower values for sulfate and nitrate Factor 1 4 for organic mass OC mass Jose at low concentrations may be seeing more mass James R2 negative Negative correlation Don t know 95 Crystal Face July 2002 Results Looked at developing convective systems CPC CCN correlated with AMS spikes in organics and sulfate Image plots smaller mode increasing for sulfate around 100 nm also see in DMA dry volume distribution at 50 60 nm mobility diameter Aerodynamic diameter estimated from DMA volume distribution is reasonable Still looking at the data Dave difficult to explain the sulfate concentration was constant day to day but organics weren t Roya some of this may be noise organics noisier Jose noise level for organics was 5 microg m3 the lower layers were advecting from the surface don t know what to make of those organic layers may not be real also spent some time sampling those layers so we may have better data Discussion of image plots Igor Doug where does white come from at the bottom of the color scale Jose custom color scale to white out James uses a white mask on the color scale James short answer no top parts of the distribution are closer together so higher noise threshold Jose suggestion on how to do it 96 Nobu Ambient Sampling in Tokyo Group also does aircraft balloon measurements of NOx and NOy AMS April 2002 only
31. John s AMS Instrument Development Hardware Upgrades Development Upgrades Adjustable lens system 16 mm quad Differentially pumped ionizer quad region Alcatel hybrid pump Channel apertures Porous conical heater with thermocouple Redesigned electronics boxes reduced cabling Inlet pressure gauge for sample flow Performance Upgrades e List of instruments delivered e Upgrades e High emission current ionizer Balzers 1S420 signals increase linearly up to 8mA tungsten lifetime decreases Reduced volume ionizer 2x increase in NO3 IE selectively improves plume ionization Closed source ionizer tested 50 increase in IE photo lonizer Magnets tested 2x increase in IE but also increases single particle pulse width decreases the size resolution plot Spirals electrons in the ionizer increasing the interaction time Effect of broadening 2X is not as bad with the small volume ionizer Chopper width is still more important in broadening the TOF SP signal High Emission Current Operation e IPP vs emission current plot with AB and IE tungsten signal also increases W IPP AB and IE all increase with emission current Lifetime is very short with higher current Looking into 2 thoriated tungsten filaments but no apparent reduction in tungsten signal with the thoriated alloy vs surface coating Final haven t gotten there yet Summary of lonizer Enhancements e Closed ionizer IPP 4
32. OF to not agree Manjula everyone should be able to upgrade except eddy correlation acquisition Manjula new changes in the next week and then release 4 0 in the all uses directory of ftp site AMS Operating Modes Most users should not be playing around with them Eddy correlation synch TOF with 10 Hz anemometer works for only 1 TOF mass Airplane Caltech twin otter NYC Alternate Mode 2 s MS TOF does it all in one go still buggy only used by us for RT measurements behind busses 0 3 s data for individual mass and 4 s data for entire spectrum Johannes use for aircraft data John TOF masses limited to 6 and mass spec is not toggle Manjula actually had zeros for 28 John 30s Jose down to 20 s make numbers very small Doug latest versions of James program can t deal with data faster than 30 s Light Scattering Mode Doug By George we ve got it John better size resolution smaller peak width Doug see particle from laser earlier than TOF data because different timing 22 Saving Data HDF binary format 5 groups TOF MS Run Mode SP saved as matrices Control via Digital Signals saves on every change of digital state of specified line data averaged after specified delay John saves on low to high high to low transition Manjula yes Hugh Jungfraujoch external trigger for switching inlet one save lasted about 12 hours lost best cloud need an override if longer th
33. V301 and likely Alcatel next three months Moveable wire Inlet T RH control Filter switching need systematic background correction Light scattering module more science for aerosol measurements 110 Future operational issues Heater temperature variations 300 600 1000 C Inlet temperature variations 5 30 C RH 20 or 30 to 100 Moveable wire beam divergence Random automatic in line filter switching Analysis software development Save every minute HDF Accumulate averages over 3 dimensions above ai scattering probe of single particles gt 100 nm TOF Size Distribution Masses Background 11 Air beam 28 32 NH3 H20 16 18 NO3 30 46 S04 48 64 80 81 Org 43 44 55 57 53 delta 2 45 Aromatic 77 91 105 Cl 35 36 Check S N in other peaks 85 suddenly appeared in Alice s data Wish list for adding peaks automatically into data acquisition program Manjula already there 111 Discussion Jose bottom of the MS window biggest peaks Johannes set automatically in F6 window Doug another column in the F6 window that sort of software control is dangerous but maybe practical Manjula 67 79 91 but 55 may have more contribution from photochemical than 67 etc Choose as many inorganic masses as organic masses which have more S N than the inorganics S M Li sulfate fragmentation pattern known well save space for other ions Plausible idea save ratios in the mass spec
34. ab put negative numbers in front of servo values to do manually used to be 110 180 Mass Spec tab Johannes unreasonable values set in mass calibration intercept and slope Rest is pretty normal nothing new unless someone sees something Flow calibration also same Single Particles light scattering operating but should be off not supported for everyone Parameter Menu converts Version 2 to Version 3 less than 3 8 3 cannot read the new menu Log files records date saved menu prm 20 Mass Spec Window Display e Goals help interpret signals e Mass value box rough reference of loadings but doesn t include scaling factors fragmentation isotopes etc Integrate isotopes into mass loadings same as James program by reading fragmentation batch files also have detection limits listed Calibrations Simplify Nitrate IE Calculate IPP from region 2 only can also do ammonium individual fragments as well as species Quad mass resolution calibration automated user friendly procedure don t have to type as fast as Eiko did won t be crashing out stick intensity calibration done automatically around m z 28 works in every case except where signal is high was 18 last summer Eiko 28 Manjula in the mass spec will add choice for air beam mass Way it works maximize broadening but stop when it starts to leak into the previous mass Jose no one has the new software set the width to 1 amu broader will cause T
35. adening James flows around corners Radius for turns at least 4 feet Richard evidence of biogenic Alice maybe but not a lot Jose Web Page Old www its caltech edu jljimene New www cires colorado edu jjose ams html Copies of users meeting files Central repository of papers and conference presentations Research groups and links including other types of particle mass spec groups User s meetings Pictures of the AMS Other resources including manuals and other information 76 Complementary Instruments What are most complementary Will depend on application make sure properly operated and calibrated Fast instruments of things we don t measure Slow instruments that enrich the interpretation Jose s List of Fast measurements Hugh Pandis group SMPS system couple APS system to it because no aerosol number at the top end of the accumulation mode Jose get Don Collins involved Eiko MOUDI huge nitrate issue online IC with different impactor size cuts Jose first approximation use MOUDI for sulfate Frank prefer differential TEOM one filter and ESP to turn aerosol flow on and off correct for evaporation of the volatile stuff Get around PILS if we ramp the heater temperature John small photoelectric probes for PAHs Phil comparisons with the ELPI Manjula higher stages get contaminated by the smaller particles Other groups have problems with this instrument EC OC continuous analysis
36. ake automatically updated now it s not 65 Running James Program Doug the FUN part Loaded the data nothing else haven t looked at the data set Pittsburgh Invoke corrections and auto updates are off Look at diagnostics no MS AB Duty cycle trace should be around 84 Lose some by switching between modes NYC alternate mode brings up duty cycle Jose want to have at least 80 Corrections window turn all off Choose AB reference period calculate AB corrections John why recalculate errors Not apply them Doug that s what I m trying to go thru when it decided to crash do each one by one don t use as a black box After reboot forgot to pick the window for AB reference period James Program AB signal down with time recalibrated EM Ideal case recalibration causes correction factor to return to 1 Next recalculate sticks look at noise in the correction factor to see if any spikes in the noise caused drifting in this case no James this version of the code had a bug in that routine recalculate sticks so it didn t do it Doug do correction for the flow rate relatively small effect Finally smooth it by 4 5 points and recalculate one more time James measured variations in the strength of the AB signal due to variations in the flow rate replacing pinhole with a needle valve AB varies linearly with flow rate with an offset makes it more correct but not a big correction for a ground
37. amp troubleshooting in the lab John manual is evolving Eiko electron multiplier aging Clean John running in the field with m z 28 is killing multiplier shortens the lifetime tested running w o 28 in TOF list and EM didn t drop as rapidly follow up during airplane talk Eiko refurbishing John discrete dynode can be cleaned but better off replacing Jose if cleaned might get some of the gain back Guenther changes the value of the deflection plates helps for the first while until a replacement can be done old ceramic bodies will be discussed Ann operate with the voltage as low as you can Guenther Inficon Balzers Vacuum Current and future improvements on Q pole ionizer Part QMS applications and integration into analytical machines Part Il optimization of the Q pole at Balzers lon source and deflection unit Part Ill future Q pole ionizer improvements Inficon QMS Development 1962 etc 1997 QMG 422 new electrometer preamplifier and auto ranging 1993 Balzers and Leybold were merged 2000 new name Inficon Principle of a QMS Schematic diagram of operation Pressure in the MS lt 10 5 mbar Otherwise degradation of the EM lifetime Generalized diagrams of various ion sources ion optics and energy filters mass filters and detectors and gas inlets Photos of different Mass Filters Show sizes of various mass filters John why 16 mm hollow Guenther weight Jose Hyper
38. an a fixed period valve got stuck Future Updates Continue simplification Computer controlled ramping of oven temperature Jump MS mode scan selected masses Saving in HDF Setting menu parameters via files macro language Integrated mode for switching between the various operating modes to mix and match Doug should become default mode using Jump because increases S N Eiko override other modes 23 Questions Software changes with bigger user base Open source NO Separate programs that interact with AMS e g temperature controlled inlet switching between valves in Prophet moveable wire Specialized modules integrated NYC alternate eddy correlation Comments S M Li online help menu Manjula supposed to be able to get to the various ftp and websites put part 2 of the manual into the code Eiko make it into a html Jose links in PDF Jon not hard code file locations Manjula possible to add that in maybe use two menus 24 James Analysis Software Show current version with trial dataset Lot of improvements with deluxe version 1 2 5 Biggest single improvement by handling data in matrices Ask questions as we go along John 9999 should say testing magnets James first save that it loads Loaded Data from Lab Doug select run interval for air beam at the bottom of the list James easy Split up corrections John type in reference numbers or cursor Do either One point
39. astern seaboard Washington NYC Organic dominated events Aug 4 air from Canada then sitting around Salem area size distribution shows higher loading for organics with relatively more in the smaller size range Doug this instrument has a much larger air beam than the others the ammonium tail at the low sizes is from the air beam and can actually subtract this out be aware of that Jay computer crashes new computer had more noise after July 25 Doug this instrument is serial 0 Jose don t clip zeros indicate noise level for ammonia look at 16 TOF other than the AB Hugh block average some periods Smoothing Jose bring the rest down artificial Hugh all data sigma amount of time spent with 2 sigma attached the best thing is to plot the negatives Johannes show error limits should be above zero when included especially for databases Jose report as they are noisy data S M Li real world there s no noise so need to filter out Johannes averaging but cannot remove it Hugh how to present final form and present in graph Need to put negatives into the database Say what you ve done for a plot Jose don t have the real concentrations some instruments remove zero and bias instruments that s completely wrong Hacene Source Characterization Incinerators aircraft ammunition sources Very high mass loadings hard to sample clog orifice AMS sensitivity too high spectrum different from ambient Boei
40. at Jungfraujoch Hugh outside temp very cold inside temp 20 both cloud and S M Li Brominated species in seaweed aerosols James soup of organics that came off buried in a pile of other things Hugh major peaks bigger than the air beam never seen so much stuff in my life Evidence that the iodine particles are significantly nonspherical we also ran the AMS in line with DMA to size cut them mass fragmentation pattern didn t change but had effective density of 3 4 decreased the size cut of the DMA to 35 nm saw AMS particles around 100 nm below 30 nm w DMA we lost all particles Redo more carefully Roya low cut size Used TSI nano DMA 83 Alice Prophet Data Remote site northern Michigan pollution sources advected in Time trace from the entire period Two basic time periods more polluted with southerly winds and cleaner time periods from North Canada Trends for entire study scatter plots of organics vs sulfate correlated ammonium vs sulfate correlated nitrate vs sulfate not as well correlated nitrate levels are very low Doug the data set that we figured that all out on was Alice s John would expect there to be a correlation Alice no nitrate is iffy because sampling issues Doug color by outside temperature more remarkable was the organic to sulfate ratio was so well correlated southerly Flow Blow up of time trace image plots show 400 nm mode in both sulfate and organics with
41. bolic surfaces Guenther as long as peak form is conserved get better peak shape not designed for UHV glue need to choose other ceramics to make bakeable factory cleaning includes baking Data on Mass Filters Big table Contribution to the neighboring mass He to 3 amu 10 ppb for Mass Filter and applications Table of different applications for the various size filters 6 mm 100 and 200 amu 8 mm 512 1024 and 2048 amu and 16 mm 128 and 340 amu Write the date to know what the name is 10 Optimization of the Q pole Used in the AMS Cross beam ion source Higher ionization efficiency with magnets or increase emission current Field axis technology transmission and precision of the rod system Discrete dynode SEM Current amplification and geometric design Higher Emission Current Plot lon current vs emission current lon sources behave as individuals can t make them 100 equal At 8 mA sensitivity increases by factor of about 4 6 compared to 1 mA need to carefully retune the ionizer as current is increased at every new emission current 11 Cross Beam lon source plots Flight paths of positive ions Flight paths of electrons filament injects e into the formation chamber highest density close to the filament Put together profile of ions generated by e ions extracted more or less efficiently heat sensitive volume in the ion source point is to make the sensitivity b
42. d time Scotland OPC data 88 Scottish Agricultural Grassland e Change height manually using the same inlet not very far because over grassland simultaneous flux measurements of sensible and latent heat CO2 NH3 N2O CH4 height 2 35 0 15 m e Ammonium nitrate fertilization period may also be studied AMS housed in cabin Data amp Discussion Sulfate concentration flux deposition velocities as f time of day for three days Deposition velocities are reasonable numbers 2 8 mm s Size dependence need to average over long periods of time Deposition velocities as f amu 35 37 indicates emission Jose chlorine James not chloride Eiko Emission gradients of HCI not far away could be organic chloride not inconceivable that it s seen Validate techniques with existing technology ordered steam jet aerosol collector filter pack optical particle counters Jon Time 50 minutes at each height next time try 5 minutes with automatic lifting Grassland deposition velocities are much lower than forest moorland data even lower AMS not suitable for measuring fluxes at Edinburgh 89 Manjula Mexico City Dara Salcedo did most of the analysis Fixed sites and drove around the city Sources for most of the pollution is in the north aerosol builds up in the southern part of the city Xalostoc in the north not a good site because lots of local sources horrible smells John so not representative o
43. data from June this year Tokyo usually very hazy sometimes in summer have a clear day Synoptic scale meteorology over Japan Siberian high in winter from China over Korea Spring subtropical high from Indonesia Summer Pacific high Local Met NOx concentrations in city Wind directions from sea during day northerly at night U Tokyo near center of city highway nearby AMS setup on 5t floor of the building SS inlet tubing out the window temporary now have a better design for continuous measurements Simultaneous measurements not available during June but now have them 97 Results Increases in mass loadings probably due to traffic during the early morning before sunrise case1 mid morning case2 Mass spectra very similar need to do the delta analysis One big difference is the 44 43 ratio case2 gt case1 Distribution shifted to smaller diameter for case 1 Separate masses smaller particles primary emissions from traffic fresh plumes larger particles photochemically processed and probably internally mixed Want to do continuous measurements of both gas and aerosols in Tokyo and hopefully aircraft sampling Two more groups are getting AMSs NIES and JCAP together better characterize spatial variation in Tokyo and Sea of Japan Richard Ship and Ambient Sampling Three projects 1 Pacific 2001 Vancouver Filter nitrate and sulfate compared to AMS nitrate reasonably good agreement sulfate AMS sli
44. e John don t ship it any more nobody else needs to know that S M Li model numbers John keep track of what each instrument has by serial number when shipped out the door Jose table at the end John don t want it to be so big Jose website has everyone s info Hugh that is extra information Manjula need to do better job separating AMS Data Acquisition Program Distribution of recent versions via ftp read only site Find newest version according to version numbers current is 3 2 no broadcast when new version is uploaded Incoming site is password protected Alice email comments Eiko correction factors is very brief say what fragment that is based on Doug that will be discussed tomorrow updated 18 Software Changes since last year Using LapLink Harvard Forest during the summer also there are other programs that do the same thing Hugh LapLink does not work thru firewall need a tunnel thru firewall Eiko also used VNC Modifications of Parameter Menu Goal Simplify separate high frequency changes from low v 3 5 8 latest Default contains parameters that shouldn t be changed often factory set but check Example hardware information TOF distance quad in use chopper duty cycle Data Acquisition Saving make sure that the new board is set to the correct device numbers Software no one should play around with this esp w o data acquisition and simulate chopper signal James al
45. e in fragmentation pattern if not ammonium sulfate Should look like Frank s pattern looking at he water signal there s no way we can say anything about the fragmentation pattern Ratio 48 64 etc Eiko distinguishing different composition ratios Doug macro to calculate sulfate fragmentation pattern James Program Batch file Water fragmentation pattern 18 less small correction for O18 and sulfate 18 James Artificially subtracted sulfate from water at 16 and 17 Hugh nightmare Doug correct for organics in sulfate pattern Ann described how to do it by averaging 14 amu from peaks for organic contribution Jose James program is now making it easier to calculate consistently Ann now need to do it for 30 organic contribution Phil only contribution of organics to 30 is from amines James can come up with a few routines to help diagnose issues with batch file Doug share the lists to combine what we are doing check the fragments for each instrument dataset Go to the mass spectrum and force the 16 17 ratio to be correct John Likely the filter zero needs to be redone every time you tune the ionizer Doug next really important concept exception of not adding m z 32 added all the fragments of sulfate from Frank Rename calibration factor column as ionization efficiency relative to nitrate James said technically it s the reciprocal of that Define this appropriately then only the laboratory determined ion
46. e is the only thing that is flat Doug deflection voltages with new IS420 module tunes with finer resolution it s worth upgrading Ann how John Work up a plan Jose pay attention to range of voltages too should be somewhat parabolic EM cal on screen Switch off ionizer and determine threshold above background ions James first point on the curve as the voltage is ramped looks strange Jose some software thing don t pay attention to it John sinusoidal curve on the bottom plot is characteristic of the ETP multiplier and Benchmark Gain between 2 and 4 million Do alt print screen and copy into PowerPoint to determine problems faster 34 Doug AMS Detection Process Key concept of three processes 1 getting particles into ionizer 2 vaporizing once they hit oven 3 ionizing the vapor plume All the ion signals we can detect are quantitative Measure lonization efficiency Need to know fraction vaporized and fraction are getting to oven Need to improve ability to independently look at those three processes Sensitivity lonization efficiency nitrate equivalent mass Vaporization volatility vs heater temperature refractory black carbon crustal oxide Can detect sea salt with high enough temperature Particle transmission and collection lens focusing aspherical particles 35 Jose Coming Back to Background lon Noise e Dominant for m z 30 44 46 48 e Reason why DL is hi
47. ed MS Diff is still useable signal Outgassing Yes heat entire chamber tried heating entire chamber with heating tapes but that didn t work unless heat evenly Restek coated chamber should reduce organic and water absorption still saw outgassing With AMS program running ran James on the same computer macro to push load button another button to add tags computer system ultimately was stable first computer problematic and was replaced autoloader was useful 49 Summary Hot National Instruments Board may be the weak link 45 C max General cabin temp an issue forced air flow coaxial along sample line helped but did not solve Try to maintain constant temperature could be harder in fixed cabin pressure James program on top of AMS exe worked Startup pumping procedure Background governed by cabin temperature not vacuum condition Leak free system M z 28 not scanned increases multiplier life Measured inlet pressure still need to account for changes in flow rate Usually got hot inside while cold outside Discussion Jose changes seen when heater turned on in the aircraft draw as much flow as possible sheath flow John TC on inlet line and sample stream downstream always warmer than the delivery line despite cooling huge thermal mass of instrument Jose missing volatiles Hugh Swiss see same thing in size distributions ran ethylene glycol to maintain constant temperature 50 Igor
48. ency Compare to PILS nitrate in PMTACS NT 2001 2 34 Nitrate internally mixed with sulfate therefore 2 34 Ammonium also internally mixed therefore 2 34 Organics no useful comparison done yet correction factor is unknown measuring beam divergence or lens with better transmission to heater James Paul Ziemann Doug Paul has never done calibration Ann Jim Smith CIMS with ion trap done for organics and sulfate maybe compare Average Mass Concentration The rest in the pie chart is about 36 Large particles low inlet transmission Maybe water Sulfate nitrate ammonium already accounted for large particles by comparison with PILS Thomas cutoff with PILS Frank most of it is accounted for Eiko was TEOM corrected for nitrate loss Frank should have been corrected S M Li sulfate may not have been totally accounted maybe sulfate losses on the high side Frank correction factor of 2 34 check as f size Refractory Only 5 was EC and 5 metals crustals Water Water lost in vacuum but TEOM also dries particles 46 To Do List To get mass closure Good relative IE measurements Collection efficiency Lens that transmits 100 Need to know transmission function to correct for particles not transmitted Jose step function at 2 5 Doug sulfate filters were 20 higher Thomas water 24 Doug can tell from fragmentation pattern in the sulfate getting water hard to make pure sulfuric acid Jay Slo
49. etter by matching the volumes Doug oven is just behind this volume so that the vapor plume is coming out right into the ion source volume want to maximize the ion density there John Alice walked the particle beam across the ion source region have mapped out that density crudely Guenther extraction efficiency gets less as you move the vapor towards the filament Designed compact ion source passed around room Deflection Unit Pair of two concentric plates non spherical shape in the direction of the SEM from the mass filter Deflection inner V6 and outer V7 voltages Maybe use a 90 degree deflection unit with spherical device causes mechanical problems Inline version increases signal and noise maybe a slight angle instead 12 Field Axis Voltage and Electrically Isolated Analyzers Electrons can impact on chamber walls and cause desorption but with positive bias e can only impact onto the anode well defined surface get more reliable emission current and background is reduced High energy ions 170 eV can be accelerated with high voltages to cross the fringing fields ions have 10 eV in the mass filter after being decelerated by filter voltage bias peak shape is also better with the field axis technology Stop doing it before the peak shape looks like a bird s wing Doug useable field axis voltage 10 14 V been running too low should be closer to 13 14 V Guenther increase the field axis voltage to maintain
50. f Mexico City Pedregal in the south posh area highest pollution there Fragments Checked consistency of the data Nitrate 30 vs 46 linear relationship except for a few points with high organic loading Sulfate colored by site looks OK except deviation only seen increases in 64 at Xalostoc Everyone needs to look at it make sure that fragmentation pattern is OK Chloride fits 36 38 non sea salt chloride 90 Nitrate Sulfate Ammonium Predicted vs calculated In general good agreement suggesting that ammonium was mostly neutralized lots of XS ammonia in Mexico City Time trends See a lot of nitrate in Pedregal in the afternoon corresponds to when there is a lot of organic AAAR Paul Ziemann talked about large alkenes can form alkyl nitrates easily have to check this out Pedregal aerosol mass loading correlate with photochemical markers ARI TDL instruments Mobile lab drive around where we wanted 91 Speciated Size distributions During the day see sulfate in the morning is in the large mode in the afternoon see new mode growing in that goes away also contains oxidized organics 44 marker consistent display of large mode with sulfate and oxidized organic processed mode during the day and small mode of organic Jose similar to Houston But Mexico City aromatic precursors are not that high Jon aerosol organic nitrates huge Doug signature of ammonium nitrate ammonia to sulfate nitrate
51. f in AMS also seen with MOUDI S M Li need to plot in the same way Hugh ultrafine nucleation Yes volume convolved Doug number density plot looks the same Jose volume concentration cut off around 10 nm Eiko AMS measures stuff growing on small particles not nucleation Hugh volume collapsing around 9 am breakdown of nocturnal boundary layer Jose cartoon of event plume of SO2 above depletion of ammonium and SO2 at the same time growth of particles but also nucleate in parallel Size distribution before and during nucleation event on Sept 12 Growth of sulfate in fine mode Particles before were organic within detection limit organics are not growing in the smaller mode Doug organics are the remnant of the traffic mode Hugh prediction based on the breakdown of the nocturnal boundary layer Doug this is predicted the night before based on the weather every 3 days or so Jose lots of SO2 everywhere James sulfate in the low diameter mode instead of his measurements Hugh sulfate levels are much higher in Pittsburgh Hugh accumulation mode is regionally dominated Doug standardize total on every plot Johannes HAZE Hohenpeissenberg mountain in Germany 1000 m surrounding mountains 300 m below May 2002 Nucleation events sometimes observed there but not during the time they were there 2 3 per year not like Pittsburgh List of measurements project not funded hasn t seen any other data See Alps during c
52. filters Frank not all because mass filter loses nitrate poug if we can account for roughly half of 36 10 is BC and dust 2 5 inlet losses in the 0 Frank filter sulfate is 85 Doug disagree with error because AMS scaled with the PILS within 10 or so do we think we have mass closure Frank organics Doug next point organics nitrate and ammonium looks pretty good organic size distribution is different shift in size distribution relative to sulfate organics undetermined with pie chart some of the organic should be scaled by PM2 5 Discussion contd Hugh assume internally mixed get upper bound Fit distribution Frank not necessarily assumes 2 34 accounts for large particles but nitrate in Houston is 1 Manjula pure nitrate particles no sulfate Frank not focused as good Hugh basis of agreement with PILS Doug issue is 2 34 contains two processes large particles lost at critical orifice nonspherical particles not focused no way of quantifying which is which beam profiling bounce detected by multiple bounces in TOF trace second order effect in new oven design large particle transmission will depend on particle type in NYC data is 2 34 in the Houston data is 1 9 or 1 8 reason is Frank saw larger particles shifted to the right compared to Houston Jose spike in nitrate has different size distribution than sulfate try to look for different populations John small mode organics in Houston are secondary
53. gh for these species instead of organics 7 e Look at check AB graph Relative background graph much cleaner at 30 than 184 Lib Peaks background up when heater on but should go down over time Useful to have for your instrument to monitor changes Jose AMS calibration Math e Look in James paper or Atlanta paper for details and equations e 1 lon flux correction for quad transmission and multiplier response haven t been doing systematically e 2 Molecular flux vs ion signal e 3 species mass concentration 36 AMS calibration Math II 4 Empirical relationship between relative IE of different species Nitrate equivalent mass IE for species is linear as f MW Show graph Electron impact ionization cross section vs number of electrons in molecule slope of line is IE vs MW Phil has done quite a bit of work determine this 5 final expression to calculate species mass concentration AMS post calibration Evaporated flux Changes to correction factor due to particles not focused due to shape size cut of lens critical orifice slow vaporization of low volatility species Cs scaled Remp Cs Report results as non refractory species e g NR sulfate Doug get relative response which is a scaling factor and want to go beyond empirical value 37 Doug El lonization cross sections Ultimately separate that calibration factor into two components Compared to other MS systems particularly GS MS
54. ghtly higher 2 C SOLAS Alcuma Mexican over Pacific Ocean went northwest of Seattle in NE Pacific Ocean dump iron into water and look at biological activity afterwards Fe limited sulfur quantities DMS GC MS SO2 AMS plus DMA er ees PSAP MOUDI quartz filters 98 SOLAS Results Time trend concentrations all less than 2 microg m3 sulfate typically 1 much less for organics ship plume bow into wind sometimes had exhaust in sampling lines DMA volume multiplied by 1 78 remarkable agreement with AMS sampling right next to AMS same inlet Mass distribution for sulfate constant through out project Organics SO4 fairly constant except during exhaust sampling Jose cutoff in the lens Didn t look at transmission Johannes compare with DMA data but only up to 300 nm geometric diameter T 550C Jose evaporation losses in inlet Outside copper room temperature differential not always the same direction Jose try scanning DMA up to higher masses Ann steeper cutoff with new inlet Doug better transmission of larger particles when the smaller cut point is moved up from 40 60 nm might be seeing a measurable effect Jose try switching 100 micron orifice with 120 micron orifice 3 Egbert Ambient Sampling in Canada Gas phase and aerosol instrumentation Time trend mass loadings dominated by events with diurnal cycles of nitrate early on later sulfate events Comparison with TEOM amp nephelometer everyth
55. gs Jon write for expert and put in front of it a section for the new user Jose troubleshooting section John recommends table format started that Jose should circulate that Manual Discussion contd Frank addresses and phone numbers of various support including suppliers Alice technical information has list of companies part outdated price and phone number Johannes depends on where the campaign takes place Manjula Anything else missing John chopper zero later on in manual Manjula quick start gory details Jose summary Eiko different version of instruments John older instruments shouldn t be included Keith for teaching John clunky Alice have old instrument tried to put in numbers but don t have them John archive manual for older instrument Jon upgrades make several Alice reference different sections or versions of manual Jon beginning of manual say which version that it covers 17 Manual Discussion contd Hugh tables of figures of merit depending on each version of the instrument put in method and referred number for particular instrument Eiko calibration sheet for reference Hugh helps to determine qualitative changes Alice John ships a sheet for the instrument when it is shipped Manjula pump powers from Jose s sheet as a function of time use it in the field plus power point screen captures Eiko problem for new people trying to determine baseline default on the ion gaug
56. ing co varies but TEOM mass is substantially more than AMS look at how the data are processed with James program Correction factors as in the program 1 2 5 0 7 Close up of one high TEOM event after fog at night with high levels of NOx oxidation of NO2 with O3 at night producing particulate nitrate NO3 or N205 lots of ammonium in the region ammonium nitrate in the aerosol 99 Qi Pittsburgh Supersite North edge of CMU PAQS Characterize Pittsburgh PM quantify sources evaluate new PM techniques Sept 7 22 2002 with G1 AMS Time trends use NR for Non refractory Clean air from Canada dirty air also seen Sulfate is the dominant component then organics nitrate is low Results Size distributions all on the same y scale Put all the plots for each day for each species Increases in peak size bimodal to monomodal chose 4 typical days 24 hr averaged that clearly show the two modes Clean days dominated by small mode polluted days with the larger mode distribution shapes mostly similar for the different components Comparison with SMPS indication of nucleation in SMPS trimodal distribution in the morning of Sept 12 similar size range for each mode Peak in surface area seen at the same time 100 Results amp Discussion MoupI comparison July 2001 with Sept 2002 looks very good for the distribution of sulfate ammonium and otal James MOUDI normalized to dlogDp space Yes Eiko cutof
57. it Keith need to design a support to take care of it Hugh shipping quad separate from instrument baffle holding it into chamber need shipping collar Jose never take quad off John ship units all assembled just plug it in Keith ship John in it as well 15 Manual Tool for teaching new users Repository for combined knowledge of users group Hopefully looked at the manual Structure Operating procedures hardware Software General comments Ann add printouts of all menus Frank details vs basic summary John too much detail overwhelming Frank quick start guide Manjula format to separate it Manjula missing parts Jon shipping the instrument Ann standby mode section Hacene right after delivery Manjula John has one James emergency shut down procedure Jose close inlet John switch off all valves 16 Manual Discussion Erik check sheet John Jose s got that Manjula MacDonald s sheet very detailed version Alice disagreement about timing of procedures hard to generalize maybe one for each application Jose working on that nice to have structure great tool for teaching people Manjula integrate some version of that Ann initial procedures jumped around venting last also names of procedures need to be the same as in cal section Erik non quantified statements e g not pumping down fully signal low John didn t give quantifiable user should make note of the settin
58. ium appeared too high immediately sees that the fragmentation pattern from ammonia NH NH2 NH3 the ratio from Frank is 1 1 16 Problem is the correction for O at 16 is not enough completely empirical but can be measured with a filter or go to the mass loading Forced the ammonium sulfate ratio to be 0 3 make this correction Ann disagree with forcing because don t know that ammonium sulfate is present in the ambient air Eiko go to time of flight mode Doug Yes John multiply 16 air signal by 1 3 Doug not the best or the only way to do this completely empirical Manjula check the right fragmentation pattern in the ammonium nitrate pattern Jose pure water filter hard because using nitrate like crazy and keep it clean in principle it can be done Doug procedure is justification for doing a filter argue that it doesn t work is when there is a problem in 16 17 18 Self consistently getting ammonium and sulfate Thomas no nitrate in the aerosol then the ammonium sulfate is not necessarily neutralized How would it look if ammonium bisulfate is measured Ann also organic acids neutralized by ammonia Doug this procedure is not the definitive one not by coincidence getting the right fragmentation pattern of ammonia Do filter measurement for at least an hour Jose get fragmentation pattern John add procedure that verifies everything adds up to 100 James will also add warning about circular logic Doug see chang
59. ization efficiencies are put in here Difference between what we are doing and what Frank is doing is by including the organic contributions 68 James Program Batch file Scaling factor for sulfate is 1 47 1 9 didn t follow Doug here Eiko different calibration factors for TOF data Doug no reason for normalizing TOF data to MS data TOF all species average for whole mission put in run numbers don t normalize to MS scale to MS AB and TOF AB Manjula Analysis of Organics No separation of individual species prior to vaporization and ionization Overlapping patterns of multiple molecules Examine time trends and size distributions Example 48 and 64 organic class1 44 and class2 55 and 57 Correlations of fragments with other species in the gas phase Example 44 and 43 signatures of oxidized species in the aerosol with PAN PILS ammonium etc 69 Analysis of Organics Marker for oxygenated organics is 44 Hydrocarbons have long envelope with a bunch of peaks separated with a regular pattern characteristic of specific species like fluorocarbon oil contamination Delta Analysis lon Series in Mass Spectra See McLafferty and Turecek Interpretation of Mass Spectra R CH2 CH2 CH3 CH2 groups form backbone of molecule generates a series of fragments separated by 14 amu delta peak mass 14n 1 Different deltas correspond to different types of organic compounds unsaturated aroma
60. ks better with single CPU but want dual to support light scattering system working in house and trying to come up with a stable design User manual troubleshooting and debugging section Chopper Zero Issue Plot of ion signal vs TOF time for various frequencies is t 0 measured properly Systematic offset means zero isn t correct Peak TOF s vs 1 chopper frequency plot Need to carefully evaluate effect on size calibration when chopper freq is changed Velocity cal accurate for only the chopper freq that was used Maintenance Pin hole assembly check o rings and replace with 2 006 After field mission clean the chamber and filters on all systems Last thing don t know what he said here Questions Johannes Problem with turbo controller on aircraft Theirs failed back in the lab Didn t restart pump talked to Varian for an exchange and thought it might have been due to temperature differentials John temperature spec rating says it should be OK Keith also had problems overheating in Switzerland Jose mounted another fan and that solved the problem flashing light is a warning James Alcatel was cutting out even with fans working TSI 3800 might have accentuated the problem John Only happened when two controllers stacked Alice Alcatel controller did fail during one flight restarted the pump to solve Feedback for John s Presentation Jon issue of nominal pin outs problem of pre
61. l sorts of problems because simulate chopper signal is sometimes yes if EM cal is stopped before finished Manjula if that happens it should appear in the status log in the main menu need to figure out the most important errors and reduce the log size try to focus on just warnings AMS operating mode discussed later 19 Regular Parameter Menu NOT Crashing program is getting better Analog input channels is now being saved with each itx file use them Should be able to read volts as well as conversion using calibration in the panel Desired voltage range used to say desired gain plus minus implied and she will put it in new Jon saving Manjula auto save interval of slow boards for faster saving Averaging and Saving needless ones removed efficient data saving mode not yet implemented soon moving toward HDF Data Acquisition Boards varying rate noise with high frequency can be reduced using fast acquisition rate but less masses can be studied because of buffer size E g 1 MHz 40 masses with 4 MHz only 17 John put max masses in Johannes Windows 2000 instead of 98 Ann stay with 2000 Manjula yes John if the computer is stable don t change it Manjula page lock error Doug extremely important to make those numbers sampling rate as large as possible S N and ion packground reduced works at 4 5 MHz but can crash intermittently if too fast on slower computers Parameter Menu Multiplier and chopper t
62. lear days Aerosol inlet outside 101 Results amp Discussion Time trends Sharp cutoffs after rainy days During the day nitrate and organics have similar slope except where nitrate shoots up After rain nitrate didn t change the same way as the rest of the species Need to look in more detail More HAZE Results Image plots size distributions messed up during rain any ideas Doug mass 18 No Ann look at the complete picture when it rains WD and other gases because could be change in the air mass not always changes in aerosol due to rain itself James distribution also drops out after rainy period might be particles creeping into the second DC region John several micron size hard to get in that lens Doug hunch these are water particles Doug as soon as you see a region where it doesn t make sense plot the size distribution look at the raw data in TOF mainly to make sure its working properly Jose red spikes spots with green around it symptom of noisy data inlet T and RH compare to outside and see if you might have condensation in the inlet by calculating dew point or water vapor concentration Pie charts for days with without rain Higher nitrate concentration on rainy days 102 Falcon deployment In March 2003 up to 12000 m Configuration with CO2 tandem DMA and MS Arnold Removed 80 20 and built own rack with the quad upside down inlet sticks out into the next rack Maybe add UPS
63. little from the local pollution Shift in the organic size distribution to the lowest size Predominantly 43 44 57 in the accumulation mode Inlet differences plotted with cloud droplet number for nitrate sulfate nice cancellation when the droplet number falls Reduction of 43 and 44 need to look at more thoroughly Some evidence of activation differences for different species in different or similar modes nitrate organics sulfate a lot of explanations Complicated because advected synoptic system over the site with vigorous updraft velocity both processes could account for the differences in activation of different species organics and nitrate from upslope sulfate from synoptic need to look at much more data 82 NAMBLEX most too early Add on to Jose and Roya s lab studies of new particle formation with diiodomethane in the presence of ozone and saw huge concentrations of particles Linked in an associated way to organic iodine emissions in the coastal zone Bell jar study of seaweed particle production Yes over 1e6 particles cc Mass spectrum contains many iodine and iodine oxide peaks similar to Jose s 12 and I peaks are the largest James didn t label anything he didn t know what it was 56 especially but saw that in Ireland too Could use help identifying 56 58 and 100 one of them in the coarse mode Discussion of UMIST data Eiko question about differences in sulfate size distributions
64. lly fragmentation ratio is within certain peaks put in math for that but not tested yet 29 Discussion Jose link to website to get newest version make list of people s versions James stop using old versions few buggy versions floating around use unified version if can t get this one to work Eiko correct data as often as calibrated James problem is that it s complicated not apparent correct method to do that need to look at each data set individually use another program to generate correct calibration wave John supply wave for user to put in CorrFac wave default 1 separate from program s CorrFac wave Eiko HDF data James not usually need to put that in Jose Don t touch James program need to keep track of changes do all programming in default procedure email Jose if you have general program that others can use James Write in C and plug in Jon spike of zero in TOF data by taking lowest quartile as zero James apply to the end of the TOF signal haven t done anything to the beginning Jon spikes in the beginning of the signal Jose noise in the chopper speed is causing some of this problem James might be worth sticking in Additional Discussion Ann A d channels input read into program James now in the load data tab Doug read and generate waves from A D channels in the itx files James that s what it does Doug names for A D channels in data acquisition use to create wave
65. lones Also possible oxidation on the filters Doug what does the TEOM sample It had a PM2 5 sharp cut cyclone Not made by URG R amp P 54 Nitrate Intercomparisons R amp P 8400 N AMS PILS 24 hr filters Time series messy but typically PILS has the highest concentrations AMS smaller R amp P significantly smaller Nitrate Comparisons Slopes vs PILS after scaling with 2 34 AMS 0 87 RP 0 57 Jose Houston events externally mixed Doug most of Houston data need the same scaling factor Slopes vs Filters about 90 Losses caused by non complete sampling of PM2 5 RP decrease in efficiency of Mo converter Jose nitrate events in the morning CF 1 James Pacific nitrate correction factor of 1 comparison with filters was consistently good 55 Discussion Jose correction factor depends on shape particle focusing stayed constant better than it is Manjula sulfate constant internally mixed Doug put up the size distribution with statistics to look at variability John table of scaling factors Frank putting in size distribution Doug Vancouver data make that same plot Jose summarize 2 3 factor shape dependent effect same for all populations modes of particles internally or externally Doug close the mass pie chart how do we interpret pie chart 36 rest sulfate nitrate and ammonium normalized to PILS size determined by the TEOM agrees with filters 15 discrepancy between PILS and
66. low anything below is interpreted as noise straight line as f TOF and then overlays white there in image plot negative values below SD value offset is big black spot John classic picture of lab air airplane use struggled Da max was too large automatically determined James put in default values depending on velocity cal it can get confused will put in safety net Eiko don t know what he asked here James Improve that integration between minimum and maximum diameter stick something onto the panel about that Misc Tab Remap Time Series Use to compare with other time series on different time basis Custom Colors table needs names to color traces to apply specific colors in order of priority only match part of the name Doug list for James 1 view list make that view color list 2 Make default thick and then thin the control button 3 add zero line to TOF plots Options time and smoothing are obvious Save Batch button for fragmentation batch and color waves one more to put on would contain comments to make library for each campaign Johannes load batch Doug Jose load by clicking on itx file Doug have to become familiar with Igor New bits selectively average runs in the MS average and TOF average by some other criteria wave of 1s and Os mechanism is there to write external macro to create wave not been tested yet Batch list and fragmentation list with time dependency in waves occasiona
67. me Separation of organics and inorganics and what we can say about them Colors on MS attempt to separate organics and inorganics fairly straightforward All inorganic fragmentation patterns are understandable solved problem make sure it s implemented properly need to write a tutorial We need to somehow build an editor to check that table Next order issue what do we do with all this green organics Manjula talking about delta analysis all the organics at the same time Delta analysis is really the first cut want to get specific compounds Old version of James program is in units of Hz want ng m 3 Any GC MS analysis of individual compounds ng m 3 or 10s of ng m 3 about X Hz Parent ion is small fraction of the total signal ideal world 1 of the signal is in the parent ion 1 ng m 3 small signal mixed with everything else at once Impossible to speciate individual organic molecules To compare to literature how many ng m 3 do you really expect to see will you have any chance of seeing it in terms of percentages of the total complex problem to say the least Review MS and delta analysis from yesterday first cut show Manchester winder data Basically the ratio of all the deltas greater than 2 to the other reflects the degree of heteroatoms to the rest of the molecule The ratio isn t precise as Jose mentioned see alkyl chain in the fragmentation pattern Combustion aerosol and processed aerosol Example from CEH gro
68. mospheric CO2 could become very important go thru and check the fragmentation patterns James Program Sulfate example only assigned major peaks 48 SO 50 has S34 isotope 64 is O2 assuming the sulfate at 48 is leftover after subtracting organic contribution to 48 calculate 50 from S34 isotope of 48 etc 80 is SO3 81 is HSO3 red is the contribution of sulfate to the signal at that mass Delta pattern is underneath the sulfate pattern 48 from organics is low because it s in the trough of the delta pattern not set up to completely analyze this properly agrees well with everything Frank talked about for the fragmentation pattern of sulfate Based on Frank s measurements we can estimate 65 Frank is seeing HSO2 fragment which is 5 7 of 64 Recalculate with new 65 for sulfate Need to subtract 65 sulfate from organic frag_organic 65 frag_sulphate 65 Idea is to go thru this systematically and convince yourself that you are calculating the right things Other major change with fragmentation list is 25 of sulfate is coming from water Referenced to 64 18 78 16 of 64 Hugh nothing to stop the user from creating circular arguments Jose check peaks up to 100 James OK put that in James program checks batch list and evaluates everything until it can identify everything by the appropriate m zs 67 James Program Calculated sulfate contribution to water at 16 17 and 18 Look at the ammonia pattern ammon
69. ms if TOF doesn t look like size distribution it s wrong especially in the limit of low signals MS Time Series Tab Simple time trends can also do simple math e g 48 64 etc Mass calculations works the same as it always has Code shows axis labels as constants in the code change mass concentration to Japanese German etc in one place for all the graphs under Axis Labels in code 27 Load Data tab Eiko load discard update List James load new all or specific runs Discard deletes specified runs from memory Update checks each one and deletes or reloads as appropriate List just makes a table of the run times and numbers TOF Traces Tab Problem loading TOF traces because 28 32 need to be in TOF masses unclick scale to TOF air beam Hasn t changed much over the past year or so Data to plot 4 choices Smoothing normalize to MS aes normalize vs scale to MS air beam Scaling to both s James TOF AB as f TOF AB vs MS AB departure causes TOF data to be different Doug Normalize to MS always applied neither one of the scaling buttons mean anything to us so remove and make it clearer how these are connected Johannes scaling to AB Jose recovered changes by looking at AB in TOF mode 28 TOF Image Plots Johannes default settings Not really varies campaign to campaign SDs 1 or 2 Spike in the data causes strange automatic scaling do manually SDs to ignore be
70. n t need to keep 28 on scale can use 32 Measurement of the gain is wrong want to be on the knee or higher otherwise measurements can be biased Doug changing the criteria this instrument has the lowest water vapor I ve ever seen want to know how close to saturation you are James example threshold too high because no knee try setting voltage with the preamplifier turned off then the shape of the curve changes with knee at lower voltage Jose see the same curve at higher voltage if higher threshold sometimes the dark current from multiplier could affect it John Balzers multipliers have the dark count ETP doesn t depends on the Balzers multiplier better get a spare ETP Doug Balzers can t keep on the shelf indefinitely John ETP can be there a year or more made from different technology Balzers copper beryllium Doug failure of multipliers is increase in the dark current Jose multiplier on off with background mass look at the threshold Performance Checks contd Jose make sure AB is not saturated Doug disagree measure AB and calibration with ammonium nitrate particles under conditions where the gain is defined if the gain drifts IE should scale with AB trend in AB over time won t know if something is changed or if EM cal is changed procedure is not fool proof Jose agree in general but AB takes care of it funny problems different failure modes Ann higher gain EM shorter lifetime Doug absolu
71. near in log space Ann small effect when subtracting averages Doug no log in batch file of James program Jose other comment spectra from McLafferty 71 Diesel Discussion Hacene 2x unsaturated Manjula no because cycloalkanes can have delta 0 and some alkanes have some delta 0 Doug rather than unsaturation this could be branching cycloalkane is the extreme Manjula more unsaturations lower in the scale More heteroatoms higher in the scale Jose this way it represents loss of H Manjula it s as though you add hydrogens Doug alkanes can t give any signal here delta bigger than 2 or 3 For delta 3 subtract C13 Dave is right that it s C13 maybe Qi How to get the delta 2 Add all the peaks that generate a delta 2 value together and divide by the total organic signal to get relative delta 2 signal Keith why assume CH2 and not functional group Jose that s why 43 and 57 show up Doug not assuming anything Phil knocking off CH2 makes sense for long chain molecules but got to be careful for organics with lots of functional groups James Rami s work functional group is bimodal in the delta pattern Manjula what is the most common part of the delta pattern Do a bunch of lab experiments to determine that Or go to sources Other Combustion Sources Aircraft exhaust delta 2 0 and 2 Eiko what was combusted James how did you make it Manjula Nebulize sometimes in hexane
72. ng 757 results Total organic carbon as f distance from engine engine power spikes when switching from idle to on and on to idle takes 8 minutes to get back to background conditions real spikes up to 40 mg m3 85 113 127 155 peaks in MS different populations transient period has both fuel and oil steady conditions just oil Emission index as f distance for sulfate and organics they increase Size distributions shift to larger masses at longer distance but organics shift up larger than sulfate look internally mixed at short distances definitely not at long distances How is Da affected by AMS flow rate Errors in flow rate are magnified in aerodynamic diameter 104 Other experiments Ammunitions propellant List of compounds present Mass loadings 3 4 orders of magnitude higher than ambient air Oven temperature ramping saw sulfate increase at higher temperature additional peaks due to zinc and other metals Zn has isotopes at 64 66 68 correct spectrum for zinc contribution Organic decreased while 44 increased with time Doug green is not really organic Other Experiments contd SS incinerator Manchester Still need to correct for all the variation in oven temperature flow rate Vary many parameters just hoping to understand what is going on there Every time you increase the temperature of the oven the total organics increase then go down need to let it stabilized and then get the measurement D
73. nt 1 minute transition 5 minutes at new temperature then transition again Organic signals decrease but subtraction problem which can be fixed we have to do it and find out how practical it is Water content was very high but for the first one it was very low wet scrubber after that used a compressor dew point 150 C heated inlet line to 150 C water condensation went away Johannes Diesel Exhaust Particle filters put on exhaust Primary vs secondary secondary may not be removed Low power settings soot particles Fuel sulfur emitted as SO2 High power converts SO2 to SO3 sulfate which condenses on soot or forms new particles New 30 nm particles composed of H2S04 H20 106 Diesel Experiments All experiments at Achen are for diesels gasoline studies in the US Showed setup Mechanical rotating diluter 2 3 sampling lines James size range measured 10 nm 1000 nm The problem mass calibration was off Balzers control menu not their problem Beers Instruments fast board output voltage factor of Couldn t correct mass scale by AMS software Manjula high mass Johannes tried everything Results Varying fuel content power loadings looked at size distributions with SMPS Took two 20 min samples to make sure conditions were stable Higher speed higher particle production More S content more sulfate SMPS comparisons they might have had a problem Thomas Need to check that there
74. one Delta 0 and Delta 2 patterns as f TOD in NYC 0 high in morning 2 high in afternoon Delta Pattern Discussion David All these 70 eV spectra Doug Can do whatever you want but lose some intensity S M Li how well can you separate Doug add delta 3 and 4 are oxygenated glib assumption that amines are negligible Houston photochemistry data hard to know how to quantify these fraction above delta 0 is rather large long chain will give lots of HC peaks qualitatively Mainz delta 3 is the signature of oxygenated organics Aromatics show up down here 7 5 3 77 91 and 105 unmistakably aromatic S M Li sulfur containing compounds Manjula heteroatoms will pull it up Jose many heteroatoms will loop it around the delta pattern consistent with it being hydrocarbon but not it is HC Manjula plan is to do this for a bunch of lab species and NIST spectra Hugh Rami s made a start of this HCs and oxidized organics can t do organic nitrates S M Li halocarbons too Jose Important to publish 73 Jon Brief Discussion of HDF HDF new format doesn t save the redundant data space savings isn t all that big can put into Mat Lab or other database program James big advantage is the random access Catalog of itx files Manjula will send out email again Aerosol Sampling in Prophet From sampling and transport of aerosols John Brockman Aerosol parameters Relaxation time settling velocit
75. oug size distributions can show prediction of when it will plateau sharpens up because vaporizing all of it E g 64 gets sharper and sharper at higher temperatures Better to decrease the emission current and keep the EM voltage the same to decrease the sensitivity when the mass spectrum gets saturated 105 Discussion James degassing from the oven should get subtracted out when you take the difference spectrum Doug background has a ramp on it so that it s changing with time systematic change on the wrong time scale changing quickly 0 2 Hz average against things that are faster than that Jose can only measure times scales that are faster Doug account for that ramp if you just assume whole other level of analysis Calculate the organic Doug turn the oven back down have to keep ramping oven up and down and reach steady state One way to solve all this the biggest problem is clogging the orifice need to dilute the sample can we dilute the sample quantitatively even still using a cyclone Doug showed the blow up of the 757 more efficient less organic aerosol mass at higher engine powers after dividing by CO2 sulfate showed no difference More refractory compounds vaporize at higher temperatures all mineral sulfate not ammonium ambient sampling no idea how important refractory materials are colder temperatures we might change the fragmentation pattern for organics but have to do it systematically in the ambie
76. plumes 93 Roya Twin Otter Studies Not much room Velocity changes with ambient pressure need to calibrate Sensitivity Corrections Lens pressure varies with sampling pressure changes in the flow rate James flow pressure in the lens of Differential Pressure gauge Results from ACE Asia AMS mass and DMA volume very well correlated except when dust levels were high density about 1 During dust periods density ratio averages about 0 72 Looked at back trajectories of air masses some from Mongolia China during dust events 94 Sulfate Layers As altitude increased saw sulfate concentration go down after correcting to standard volume After converting to appropriate size at that altitude the distributions are very uniform When reporting mass concentrations standardize the volume for aircraft measurements but not for health effects Report both Should know define the number 1 microg m3 0 5 ppbv with MW of 100 See p 28 of Finlayson Pitts book ACE Asia Results Vertical profiles of organics and sulfate Mixing state indicates some neutralization of sulfate ia appeared in the same mode missed whatever was below nm Pie chart for contribution of different fragments to total organic mass mostly oxygenated but some aliphatic organics AMS vs PILS on C130 factor 2 5 for everything correlation about 0 9 for sulfate maybe due to impactor cut off nitrate and ammonium also shown MOUD
77. put lots of oxygen into C analyzers so not like them S M Li analysis for short molecules except anthracene still see a lot of CO2 with longer molecules Phil 44 lesser contributor to total mass for C9 molecule S M Li obviously see a lot of that Doug mass 44 CO2 most important organic component in the field and also 43 acetate Jose paper by Paul Ziemann decomposition into smaller fragments 61 Jose s Worksheets Initial checks Provide a general one and one for new users Webpage utility for network Atomic clock James save times get screwed up two clocks Windows and computer Windows update John separate PCI card for clock Doug drift is temp dependent James clock slows down temp effect Save electronic file of computer screen printouts Space left on hard drive Antivirus check run before campaign plus defrag Airplane weather affects flow rate Weather Service has website to calibration Baratron Flow meter not very sensitive pressure inside the lens is important and needs to be calibrated drifts due to calibration drift or ambient P drift Check inlet flow critical orifice clogged Gillibrator works dry cal has problems due to significant pressure drop calibrate DMA too 2500 Thomas correction for water vapor on the order of the vapor pressure Vacuum checks Health of the vacuum system run thru at the beginning of the campaign condensed daily version Pump speed and current curren
78. ross heater to measure transmission try to use routinely to determine how much is actually sampled Temperature controlled inlet liquid coolant used during ITCT UMIST peltier cooled being tested at BC Light scattering module 532 nm with 25 mW ellipsoid mirror Most important improvements Beam width probe and converging aperture Hardware Failures Premature bearing failure of turbo pumps on the front end inlet and back pressure too high solution bigger pump and Alcatel hybrid pump Alcatel hybrid bearing problems new release planned for Oct 2002 National Instruments fast board data acquisition failure solution by replacement Balzers components RF QMH 410 5 18420 recessed pins in the ionizer connector John We ll do what we gotta do Hardware Failures contd V301 premature failure infant mortality case Vacuum Interlock TP6 shut down of multiplier heater and heater bias V70 controller reverts to low speed operation Pin hole assembly tightening over tightening deformation of disc low pressure side o ring disrupts particle transmission 5V not sure what he said here Difficulties Cabling need to explain better some left unconnected AMS acquisition program corrupt AMSmenu prm don t close with X hit the enter key Balzers 18420 loss of V6 and V7 power glitch that resets SEM to FC and disables V6 and V7 discussed in manual Computer instabilities dual vs single CPU wor
79. rticle size distribution m z 202 shown No uptake w o H2S04 Measure H2SO4 condensed with AMS physically increase particles in size and also in density Density factor when adding water hard to quantify addition of water Use MS to do it See carbon backbone of soot and poly aromatic functional groups doubly charged ions with H2SO4 see 871 is the largest this sulfate is not entirely pure sulfuric acid and 81 is a lot bigger when it is pure Effect of H2SO4 loading on water uptake of activated soot with more sulfuric acid get more water Thomas why isn t there an increase for the magenta data slight decrease intermediate sulfate Water evaporation in the AMS inlet And didn t go to high enough RH In between 50 75 water loss compared to predicted from vapor pressure of water Doug fragmentation pattern pure sulfuric acid has less water than ammonium sulfate need to look at quantitatively Propane soot effective density of 1 2 dirty propane effective density more like 1 distributions like Johannes showed 108 Melissa Tholins Couple billion years ago BYA early earth Geologic record indicates that molecular oxygen was present 2 4 2 2 BYA Solar intensity much lower than today Larger greenhouse effect otherwise temperature too low for liquid water Moderate levels of methane and ammonia would increase temperature but they would be destroyed rapidly Titan as a model P 1 5x earth
80. size but transmission cutoff 39 Pinhole vs Converging Orifices Streamlines diverging more for pinhole than for converging orifice especially for 2 micron particles Demonstrated in a crude way by sampling an asthma inhaler Drug particles larger than carrier gas particles transmitted better thru glass converging orifice than thru pinhole Still work in progress Plotting Particle collection efficiency Walk beam across the heater by tilting beam AMS collection efficiency as f beam position Nitrate detection efficiency is also flat On one side of the oven the efficiency increases because it s on the filament side Improved by using the small ionizer Overall collection efficiency is actually now very sharp define edge of oven really sharply therefore beam is very well focused Ammonium nitrate particles well focused Ammonium sulfate is not as well focused Beam is wider than the oven and ions per particle is the same even when hitting the edge of the particle beam Still need a scaling factor on the order of 1 5 to 2 Eiko why is it 2 Transmission is 15 for sulfate particles but close to 70 for nitrate S M Li Doug Particles in the field are more spherical See the same thing in the field for various studies scaling factor is fairly constant Need to implement beam width on ambient particles in the field 40 Particle Collection Discussion Phil organics showed similarities to sulfate
81. slightly higher sizes at the peak of the pollution episodes Sulfate is scaled down organic peaks separated by m z Average MS full scale shows sulfate and 44 higher than 43 mix of stuff delta 0 is dominant 84 Northerly Flow Time trend image plot for sulfate 300 400 nm mode and some times with a smaller mode Average mass spec shows the delta 0 and 2 one spike in organic due to diesel truck Small particle events Especially large event midday on 7 26 01 Saw higher sulfate but not organics in the smaller particle mode DMA number distribution also shows event with growth of smaller particles Detail of size distributions for 3 events distributions broad or with 2 modes Also had gas phase data 85 Elementary School in Denver Showed time trends measuring off the roof of the school 2 story not much in the nearby area Every other day from midnight 2 am increase in the organics Also day where there was a larger event Still going on starting to look at this Jon Eddy Correlation with AMS End of Prophet Aug 2 9 eddy flux from the top of the tower Aug 4 relatively low sulfate concentrations Needed to synchronize AMS and anemometer triggered at 200 Hz average for least 0 1 s Data load would have needed 10ks of computer equipment for data figured out what needed to be saved and new format for saving it 86 Prophet Sampled above the canopy from the top of the to
82. slow not so much radiation suggest filament current on first then bake heater off then filament off Jose cleaning summary Remove ion source to see if additional cleaning will help Doug My experience first time with Balzers most reliable and robust simplest believe key to this besides quad ionizer focusing ions fraction of ions formed is larger and this is key to the high sensitivity system is incredibly stable over days better than he expected Jose support RF failed in field replaced with already existing replacement If not available what could be done Tried to return unit to Balzers but not end user given the run around need to set up some system so that a replacement could be sent quickly Guenther job for me and Steve to talk about we ll find a way Open Discussion S M Li cleaning John under vacuum with heater and filaments on see increase in background during sampling but goes down self cleaning Explore degassing operation to help speed up process S M Li benchmark for background value John keep it pumping monitor background in program with check status convenient way to check background levels Jose compile series of benchmarks talk about later Manjula other failures Keith transportation Rods loose the top end tighten OK or does the position need to be fixed critically John rods aren t changing because on ceramic holders deflection plates should take care of
83. t increases as failing before pump actually fails inlet closed and open Lens pressure is diagnostic if the lens is clogged more precise than flow meter Ann depends on met conditions Jose look at it and determine if its meaningful James best hit light on the RF box Balzers manual wasn t flashing on the fault line Jose uses a little bit more power when flashing but OK 62 Performance Checks Software version kept in the files Always write in systematic way then other people can see it e g Houston data didn t keep track can t remember Slow board analog voltages make sure working Switches for the electronics Measure electronic noise everything on but filament and multiplier off James routinely run scanning m z 11 in TOF mode shouldn t detect ions there so all electronic noise useful in ITCT by measuring it take out chopper interferences Jose numbers not meaningful when running so good idea James depends on fast board sampling rate Performance Checks contd Balzers error messages see manual Cross check resolution and mass scale calibration default corrects this in use between saves still check it Can recover from MS but not from TOF Doug F6 window look at 28 44 peaks Jose sometimes the serial box gets confused restart program to fix Offset parameter in the F6 window should be the same for all the masses Check that there is a MS James only really update it if significantly shifted
84. te signal look at full scale John talk about bits not Hz which includes gain 64 Performance Checks contd Check air beam with summary window toggle mode Molecular velocity dependence on MW in air beam John and Eiko measured Doug current versions of software crashes due to operators exit program after calibrations and restart program Jose check repeatability of EM cal If AB is strange check servo motor not common anymore but still fails want open position at peak center John failures Just Jose Vancouver didn t actually fail but was replaced David CEH Group members in TOF window needed 28 as B 32 is another letter is for EM cal can only have one m z as B Jose TOF AB and MS AB should be similar Displaying too often reduces duty cycle David CEH chopper freq is red warning Jose always pay attention to red warnings in this case chopper speed is too fast for maximum diameter Check that the chopper is not being simulated set chopper speed control to automatic otherwise it may oscillate Keep track of heater bias on the multiplier on the electronics box Jose Calibration with NH4NO3 Go thru standard steps for calibration John threshold routine too low with the magnets manually make them higher Jose increase the width of the single particles James how do you remove noise Manjula hit N in stick mode to remove calculated noise questions about how that is done m
85. tics have deltas of 4 5 Clustering of different types of compounds deltas do not correspond to specific groups but may be used to distinguish different degrees of saturation oxidation 70 Example of Delta Patterns Delta 2 alkane e g 29 43 etc Delta 0 increasingly unsaturated 27 41 etc ARI lab spectrum of pentacosane long chain HC strongest contribution is to the delta 2 group Aromatic dicarboxylic acid phthalic acid Add up all the fragments that correspond to a delta value and determining their relative contribution to total organic signal Maleic acid C4 unsaturated acid biggest delta contribution is 1 Malonic acid C4 dicarboxylic acid largest component is from delta 3 indicative of oxidation 44 ions from dicarboxylic acids all give large delta 3 Black shows the percent of 43 that is in delta 2 and percent of 44 that is in delta 3 Diesel Fuel Example Diesel fuel show entire spectrum with different colors for different delta groups Intensities as a function of increasing carbon numbers try to make sense of envelope Straight line slopes didn t fully characterize fall off with increasing c number do a fit to describe it fall off plot Composition is pink for other 15 16 17 18 etc green for organic 44 and 43 are included in fall off plot Doug diesel fuel looks like delta 2 but also contains delta 0 in HC 43 is a small part of delta 2 small delta 2 Jose decays li
86. tion newest 1 4 e7 Hz for m z 28 John thinks it s saturated oldest 1 0 e6 Hz for m z 28 Subtle things can go wrong and screw up calibration Crystal Face 30 change in AB due to orifice problem Use 32 if 28 is saturated 31 More measures of Signal Intensity lonization efficiency to Air Beam ratio should be constant and change with ionizer tuning newest 8 0 e 6 1 4e7 0 57 e 12 0 57 pAB pico Air Beams Doug roughly within a factor of 2 of that AB is a pretty good indicator of the ionizer sensitivity Jose spatial distribution of AB molecules is similar Show Worsnop Jayne s Law detection limit of NO3 vs time 5 x sensitivity per year W J Law Part II e X3 signal every year looking at IPP nitrate vs time e Other plot AMS single particle detection Increasing particle TOF Da sample TOF trace as f time Background ion noise at some frequency sensitivity m z dependent makes it harder to distinguish ions from particles electronic noise 100 kHz Power line switching instrumentation Electronic Noise Dominant for high masses Affects sensitivity if noise large need higher threshold Example in MS mode switch off filament and multiplier look at third RMS number 11 5 with filament off 3 with multiplier off relative to noise with Crystal Face instrument changed from 0 5 to 2 every day if 10 or constant there is a problem with electronic noise John toggle mode Jose little peak of
87. to separate them solution to have a system keeping lens at optimum pressure Johannes DLR Falcon Up to 12 km altitude AMS MS Arnold DMA CO2 Two racks Quad down inlet outside go thru other rack 3k aluminum withstand 9 G forces John make sure SS of quad is directly bolted to the Al frame Johannes need to prove every single part of it will withstand John wire shock mounts Johannes whole rack will be shock mounted Jose vibration wasn t bad truck worse In the air not bad worst during takeoff and landing 52 Next Flights Johannes Flying March 2003 John G 1 Mexico City April 2003 Jose No plans for the Twin Otter Hugh UMIST test flights in March 2003 optimistic actual date is moving target Ann P3 Summer 2004 John rack mounted with buried inside somewhere Keith Rack System Electronics on the bottom Hugh met office used to fly with perk racks this is 9 G forward 6 G back 2 G side AMS swappable with that in mind operating for on port or aft on starboard Quad suspended from horizontal beams in the middle of the rack Barry mounted aircraft approved mounts 4 places on top 4 on bottom plate for CPC and RF box on top Can t protrude on walkway side pretty flush against cabin wall Power 110 V 400 Hz switchible power key straight into power system John Is 400 Hz OK Computer is questionable Main bits OK Computer is 47 to 63 Hz Balzers is also 47 to 63 Hz John
88. to keep on one circuit single AC plug Jose APC works well John model 1400 220 V Doug bought 240 V European rated between 150 260 V John s UPS weight 50 60 Ibs Jose s 70 Ibs Entire package John s 500 Ibs Jose s 410 John only person taking data during takeoff and landing UPS registered switch from ground to aircraft power Watch DC generators failed 5 minutes before landing UPS covered it until on the ground definitely want UPS Jose switching power was very clean on Twin Otter instrument never detected anything airplane dependent John dragged power cords to the plane other places maybe can t do that consider bigger UPS fueled up with the plane plugged in Hugh UMIST guys struggling with these details pilot watched them put it in quality of the work no formal procedure to approve instrument s08e allowed me to keep power on the plane daisy chained up to 1 5 hrs with the valve closed pain in the neck 0 do it 51 Jose Inlet Pressure Effects As P changes flow changes size cal changes AB changes transmission vs size shape changes P pressure constant run at 160 Torr instead of 760 orr Chuck Brock has some design Ann pressure controlled on PALMS will talk to them to borrow design Sensitivity corrections gain of the multiplier changes with time need to take into account sensitivity changes John measured before and after flight Heater bias suddenly changes by several volts need
89. up building exhaust ratio of oxygenated is somewhat different but picket fence goes out to higher masses Ann may not have enough signal for the Manchester data Real point is the delta pattern ignores that so really only started to look at that information Degree of structure is important but difficult to interpret We as a group need to deal with this We are going to figure it out by making these measurements and combine them with others Clearly seeing systematic similarities and differences Framework for comparing them Hopefully next year we will have a clearer and more defined version of what we did yesterday 78 John G 1 data DOE SBIR to upgrade aerosol measurement capability on the G1 aircraft Background of the study New England Air Quality Study Show the package for the AMS specified rack system everything else up to them Aerosol inlet designed by Fred Brechtel double diffuser cone with actively flow controlled MFC S M Li inlet sample with pump and T John no pump but Venturi pumped adjusted velocity to match plane speed with MFC S M Li design from Fred Talk to Fred Doug polenta practical and useful design one flight speed 400 knots 100 mph G1 Twin turbo prop room inside to stand up but tight and hot Sample data from July 22 Showed all the data Blow up of AMS date amp time trace of sulfate up to 60 microg m 3 and size distribution SO2 sometimes nicely correlated
90. wer no SMPS no mixing Eddy correlation Learning micrometeorology Calculate fluxes as a covariance with vertical wind speed Deposition velocity for each species normalized for the number concentration E g 48 for sulfate Sulfate Deposition with time Deposition velocities positive toward ground Looking at one day s worth data Pick best delay time between anemometer and AMS sampling times Noisy data in the afternoon with about 0 6 cm s 0 cm s until about 10 am As f size don t believe this data because error bars are large Chopper 100 Hz every 10 cycles Also had size distributions but huge amount of noise in that data S N 1 for 1s Goal was to look at nitrate deposition but levels were really low so weren t successful with that have to look at other days Eiko Forest deciduous topped trees as the forest grew Alice Wasn t a lot of variation as a f height need to look at much faster time scales 87 Eiko Current projects with AMS Micrometeorological flux measurements F height and combine with met measurements Eddy covariance Virtual disjunct eddy covariance Fast sampling or sensor counting statistics important More masses spend less time on each of them New sampling mode Implement new mode to select certain masses and use MS mode to get better S N Combine with light scattering probe later Deposition velocity removal rate of particles by the ground changes with size an
91. wick 81 jumps up water signal just happens to give you mass balance Ann 18 as f RH Doug seen sulfuric acid losing somewhat half of the water compare to the phase diagram that we see still working on that Ann suggest looking at time periods when discrepancy is high vs low Relative ionization efficiency for ammonia is 4x larger for the small quad Basing ammonia concentrations on mass 16 Jose Airplane Monday morning Issues Pumping Inlet pressure temperature Other issues rack system power supply 47 Background Testing Plot of signal as f time Two time decays 3hrs several days At one point shut off turbos and kept pumping with backing pump Waited 10 minutes before starting MS Same pressure as venting with He He for 3 hours still high background Baking made the background worse Maybe better with new systems need to be tested In clean areas like Crystal Face can t afford to lose signal due to high background Doug change to sensitivity calculation advantage need to know IE and AB good EM cal ng m 3 in 1 min keep the system clean pumping gate valve on the detector pump don t have UHV vacuum no gate valves everywhere but one would help us John Aircraft Instrument Inlet system double diffuser cone with anodized piece Brechtel actively flow controlled with MFC Cabin had room to stand AMS in the forward cabin Vacuum startup procedure day and evening flights 3 4 hrs pumping
92. y stokes number Plot of relaxation time vs Da Sampling from ambient into still air List of avoidable negligible sampling losses 74 Aerosol Transformation Not covered well by Brockman interested in the chemistry of particle List of issues Gas wall loss includes diffusional and turbulent loss of gases can also have out gassing contaminated tubing not important for bulk measurements Diffusional losses lower for laminar flow with Re as close to 2000 as we are comfortable with Prophet Tower Three inlets Used 2 tubing cyclone gentle copper tube bend and down the tower 12 lpm flow rate total with isokinetic take off most of the flow goes out critical orifice radial heterogeneity introduced by long lag time so passively mixed flow then another isokinetic take off for SMPS and AMS Remove large particles laminar flow large excess flow passive flow control mixing is questionable temperature control lines residence of the centerline was 10 s Tested by generating particles measure w CPC and then put into inlet losses probably due to mixing 75 Discussion Richard losses Low flow thru something with sharp corners could cause dead spots Johannes problems with copper tubing At least a day s trajectory from acidic particles for logistic reasons couldn t use SS James mixing down pipe temporal broadening Parabola on the centerline for laminar flow so very little temporal bro

John`s AMS Instrument Developoment

Contents

Download Pdf Manuals

Related Search

Related Contents