#homecmos IRC log for Tuesday, 2011-09-20

« Monday, 2011-09-19 Index Wednesday, 2011-09-21 »
berndjazonenberg, auto-darkening helmets ftw indeed!01:32
berndjazonenberg, some pics of "alligator skin" reminiscent of your fluoride-permeable PR layers http://cool.conservation-us.org/jaic/articles/jaic39-03-005.html02:56
azonenberglol02:58
azonenbergLooks messy02:58
azonenbergwill check it out more in the morning, bedtime for me02:58
azonenbergi have a date with a SEM tomorrow morning :)02:58
berndjbring flowers03:12
azonenbergberndj: flowers? I'm bringing etched wafers :P11:01
azonenbergyou have to know how to treat these things11:02
azonenberg_workberndj: remember, flowers are not high-vac compatible unless dried first :P12:47
B0101hi azonenberg12:47
azonenberg_workhi12:48
B0101a lot research is undergoing now that my lab is finally finished12:51
B0101how your lithography test process?12:53
B0101*hows12:53
lekernelhttp://lekernel.net/blog/2011/09/prywatna-wytwornia-lamp-where-diy-meets-vacuum-electron-devices/13:26
azonenberg_workB0101. i hope you are keeping nnotes for us to see?18:02
azonenberg_workand i'm imaging dies on the sem now18:03
azonenberg_workpics coming soon18:03
bart416azonenberg, couldn't you theoretically do lithography with the SEM? :P18:35
azonenberg_workThey dont have a litho system installed on this one19:10
azonenberg_workuploading pics, h/o19:13
azonenberg_workhttp://colossus.cs.rpi.edu/~azonenberg/images/homecmos/2011-09-20/die_i4_009.jpg19:15
mrdatawhat magnification is that?19:19
azonenberg_workscale bar at bottom19:19
bart416mrdata, 1000x :P19:19
mrdataoh yes19:19
azonenberg_workThat's copper on silicon19:20
azonenberg_workhere's another die with resist still on it http://colossus.cs.rpi.edu/~azonenberg/images/homecmos/2011-09-20/die_i7_005.jpg19:20
mrdatagood job19:20
azonenberg_workwhich means ican definitely use less PR and potentially get better resolution etc19:20
azonenberg_workand use shorter exposures19:21
azonenberg_workI also have to do more work in etching silicon19:22
azonenberg_workmetal patterning and litho at 20um scales is clearly a solved problem at this point19:23
azonenberg_workBut before i can get a working device i need to be able to etch the Si19:23
azonenberg_workOne thing to explore is etching of SiO219:23
mrdatayes19:23
mrdataHF is your friend19:23
mrdatabut oh boy does it burn19:24
azonenberg_workoh, its more complicated than that19:24
azonenberg_workHF penetrates my photoresist :P19:24
azonenberg_workAt least, it does when i etch Ta2O519:24
azonenberg_workSiO2 is easier to etch19:24
mrdatathat would be a problem19:24
azonenberg_worki also have no way of generating SiO219:24
azonenberg_workSo i might buy an oxide-coated wafer to test on19:24
mrdatatry etching glass?19:25
azonenberg_workBorosilicate? Soda-lime?19:25
mrdatahow does the chip fab industry do this?19:25
azonenberg_workNeither has the same characteristics of fused quartz19:25
azonenberg_workThey normally use HF-resistant PR19:26
azonenberg_workor a plasma etch19:26
azonenberg_workAnyway so i'm thinking of picking up one of these http://www.mtixtl.com/thermaloxidewafer300nmsio2layeronsi1002diax050mmtptype1s2wksdelivery-1-1.aspx19:26
azonenberg_workThis is <100> so i can't get vertical etches (V-trenches only)19:26
azonenberg_workWhich means its useless for comb drive work19:26
azonenberg_workBut i could test etching of SiO219:26
azonenberg_workThe other option is to try and bum a few hours of furnace time off a lab on campus19:27
bart416how precise does your temperature have to be?19:27
azonenberg_workbart416: Not very, not at all19:28
azonenberg_workBut you want to be relatively uniform to avoid heat-stress cracking19:28
bart416throw it in a pyrex beaker and put it on the barbecue or your kitchen oven :P19:28
azonenberg_workIts not that simple19:28
azonenberg_workI need to hit 1200C19:28
azonenberg_worklol19:28
bart416ah, 1200C19:28
bart416build a clay furnace19:28
azonenberg_workCould be +/- 20019:28
mrdataSale Price: USD$79.95 ??!? omg that's expensive19:29
bart416If you blow into it with a hairdryer you can reach those temperatures19:29
azonenberg_workmrdata: thats a 4-inch wafer with a whopping 1000nm of oxide19:29
azonenberg_work= THICK19:29
azonenberg_workAnyway so thats one idea19:30
azonenberg_workBasically, its all boiling down to one basic problem19:30
azonenberg_workI have no idea how to take a pattern in photoresist and put it into something KOH-resistant19:30
azonenberg_workswkhan: Do you by any chance have a high-temp furnace at your lab?19:31
azonenberg_workAnd if i mailed you a wafer could you grow oxide on it and mail it back? :P19:31
azonenberg_workmrdata: I tried Ta2O5 patterned by liftoff, that failed because it dissolved into my resist19:32
mrdataread the patent literature?19:32
bart416btw azonenberg you were asking about how to cast that plastic19:32
bart416the guys at instructables there used something called smoothcast 300, and I looked it up19:32
azonenberg_workI tried Ta2O5 etched by HF through photoresist, that didnt wok beccause HF penetrated my resist19:32
bart416And it's pretty much what you need to cast those turret parts19:32
azonenberg_workTa2O5 etches slower than SiO2 by far19:32
azonenberg_workso if i use a thick photoresist layer it should survive exposureto the brief HF dose it'll take to eat a micron of thermal oxide19:33
azonenberg_workBut SiO2 is etched by KOH slowly19:33
azonenberg_workIt will work as a mask for shallow etches but not deep (thorugh wafer) ones19:33
azonenberg_workmrdata: Almost nobody does a full-wet etch process for bulk micromachining19:34
mrdatawhat do they do?19:34
azonenberg_workThe de facto standard is to use LPCVD Si3N4 over a thin thermal oxide layer19:34
azonenberg_workNitride can be etched by HF easily, even throuhg a photoresist mask19:35
azonenberg_workBut i have no way of getting a wafer with nitride on it19:35
Action: mrdata let's his imagination wander to nitrides;19:36
azonenberg_workThis whole mess with tantalumfilm was caused by me not being able to generate Si3N419:36
mrdatadont they use harsh things like HCN and NH3 to make nitrides?19:36
azonenberg_workNo19:36
azonenberg_workAmong other things reactive sputtering of a silicon target19:36
azonenberg_workin an Ar + N2 atmosphere19:36
azonenberg_workwill deposit nitride19:37
mrdataoh19:37
azonenberg_workadd a little O2 or sputter SiO2 (i forget which) and you get an oxynitride19:37
azonenberg_workBut like i said, i dont have access to anything like that19:37
azonenberg_workThe last idea i had was to try patterning Ta2O519:37
azonenberg_workby HF etch through a metal hardmask19:37
azonenberg_workI'm going to try that toniht19:38
azonenberg_workphotoresist over Cu+Cr over Ta2O519:38
azonenberg_workexpose, develop19:38
azonenberg_worketch metal in SC219:38
azonenberg_workthen etch Ta2O5 through metal with HF19:38
mrdatabut, the goal is to deposit lines and dots of doped semiconductor onto a substrate19:39
azonenberg_workNo19:39
azonenberg_workThe goal in this case is to etch a vertical pattern into a wafer19:40
azonenberg_workfor MEMS19:40
azonenberg_workDoping is a much easier problem if i have a furnace19:40
mrdatause CO2 laser?19:40
azonenberg_workSpin coat doped SiO2 (Phosphosilicafilm + Borosilicafilm) onto a wafer19:40
azonenberg_workpattern with HF19:40
azonenberg_workbackfill with undoped19:40
azonenberg_workand heat in a furnace to diffuse19:40
azonenberg_workI can evaporate aluminum which i'd use instead of copper since its MOS compatible19:41
azonenberg_workAnd i'd need to use an alternate developer (TMAH based, or possibly ammonia based) since NaOH / KOH kill PN junctions19:42
azonenberg_workBasically, if i had $2K to spare i could probably make transistors with what i have now19:43
azonenberg_work$1.2k for furnace, the rest for dopants19:43
mrdatashould a furnace really cost that much?19:44
azonenberg_workmrdata: a nice one? Yes19:44
azonenberg_workhttp://www.mtixtl.com/compactmufflefurnace4x4x410lupto1200cwith30segmentsprogrammableandgasports.aspx19:44
mrdataif you made some transistors, i would like to have some19:45
azonenberg_workProbably going to buy some of this soon http://www.ebay.com/itm/Tetamethyl-Ammonium-Hydroxide-TMAH-30ML-FERROFLUID-/110702625718?pt=LH_DefaultDomain_0&hash=item19c66403b619:45
azonenberg_workmetal ion free developer19:46
mrdatawould a CO2 cutting laser help?19:46
azonenberg_workmrdata: NOt that i can think of19:47
azonenberg_worktrace metal grade tmah woot19:47
azonenberg_workSo i'd need that, $250 each for P and N type dopants19:48
azonenberg_work$250 for Silicafilm19:48
azonenberg_workThats the min order (4 fl oz), if anyone wants to buy some subset of it off me feel free :P19:48
mrdata4 fl oz for $250?19:49
azonenberg_workYes19:49
mrdataseems really expensive19:49
azonenberg_workthats what i was quoted before19:49
azonenberg_workAnd trace metal grade chems are rarely cheap19:49
azonenberg_workBut you dont need nearly that much19:49
azonenberg_workbear in mind the 4 ounces of Tantalumfilm i got, well...19:49
azonenberg_worki've used maybe 2 ml of 250 :p19:49
azonenberg_workactually no19:49
azonenberg_worksub 119:49
azonenberg_workSo lets see - i already have PR19:50
azonenberg_worki can get TMAH and dopants19:50
azonenberg_worki'd need a furnace19:50
azonenberg_worki can evaporate aluminum19:50
azonenberg_worki can etch aluminum with HCl:H2O219:51
azonenberg_workFEOL process needs something to align dopants to, so i'd probably put a sacrificial aluminum layer down and etch alignment crosshairs into it19:52
azonenberg_workThen spin coat N type dopant19:53
azonenberg_workSpin coat PR, HF etch to pattern the N diffusion19:53
azonenberg_workBackfill with Silicafilm, bake to diffuse dopant19:54
azonenberg_workOptionally repeat the process if I'm doing P+ diffusion anywhere (assuming P type wafers)19:54
azonenberg_workHF to strip oxide19:55
azonenberg_workcoat gate oxide19:55
azonenberg_workevaporate metal gates19:55
azonenberg_workspin coat PR, pattern gates19:55
mrdatait's a lot of steps19:55
azonenberg_workspin coat main dielectric layer19:55
azonenberg_workspin coat PR, pattern gate contacts19:56
azonenberg_workevaporate metal 119:56
azonenberg_work*gate and source/drain contacts19:56
azonenberg_workat that point i should have workin gMOS transistors19:56
azonenberg_workgotta go, i'm halfway across campus from my office hours that start in 5 mins19:56
azonenberg_workback in a few19:56
mrdatathat's odd. the logs seem to be missing after 1940 h UTC20:01
mrdatapast 20 minutes is unrecorded?20:01
azonenberg_workback20:07
azonenberg_workAnd the logs are flushed to disk every ~30 mins iirc20:07
azonenberg_workask wolfspraul20:07
azonenberg_workhe set it up20:07
mrdataok, there we go20:12
mrdatalogs agree20:12
azonenberg_workhttp://colossus.cs.rpi.edu/~azonenberg/images/homecmos/2011-09-20/overview_001.jpg20:27
azonenberg_worknice overview of the dies i was working on20:27
azonenberg_workGotta love that insane depth of field20:27
azonenberg_workand here's a long shot of one of my patterns http://colossus.cs.rpi.edu/~azonenberg/images/homecmos/2011-09-20/die_i7_002.jpg20:28
azonenberg_workExtreme closeup http://colossus.cs.rpi.edu/~azonenberg/images/homecmos/2011-09-20/die_i7_010.jpg20:39
berndjmrdata, a CO2 laser is at 10.6µm, so the smallest spot you could make would be of about the same order, if you had a very fast lens.  i think that's getting a bit coarse for what azonenberg_work wants to be able to do20:44
azonenberg_workberndj: Yeah20:45
azonenberg_workIf i was going to do laser direct write litho i'd use UV20:45
azonenberg_work385nm or so20:45
azonenberg_workwith a ~1um spot20:45
azonenberg_workThats actually on the list of methods to research20:45
mrdataok20:46
berndjlol, got a spare excimer laser lying around?20:46
azonenberg_workberndj: if i had one i'd be using it already20:47
Action: mrdata checks his pockets20:47
azonenberg_workI'm thinking bluray actually20:47
mrdatasorry20:47
azonenberg_workdiode laser20:47
berndjgoogle "sam's laser faq" if MTI's prices are too steep20:47
azonenberg_workberndj: i'm aware of the site20:47
mrdatayeah, you can make a N2 laser ithink, from aluminum flashing and air20:47
berndjazonenberg_work, do you really need 1200C to grow oxide, or do you need that only if you want it to grow quickly enough that you don't do it overnight20:47
azonenberg_workbut i doubt they can help me with my problem with MTI's prices20:47
azonenberg_workberndj: it's quadratic with temperature i think, up to a point20:48
azonenberg_worklet me look up the equation20:48
berndjdunno... for someone as disciplined as you, you could probably pull off a DIY Ar/Kr laser20:48
azonenberg_workhttp://fabweb.ece.uiuc.edu/gt/gt/gt8.aspx20:48
berndjexcimer would be a bit harder though :(20:48
azonenberg_workberndj: i dont need a lot of power20:48
azonenberg_work100 mJ/cm^2 or so20:48
azonenberg_workis the complete exposure dose20:48
berndjhow much coherence do you need?20:48
azonenberg_workbearing in mind that my die is well under that20:48
azonenberg_workZero20:48
azonenberg_workI need a focused spot that i can aim20:49
berndjoh :(20:49
azonenberg_workthat is relatively monochromatic20:49
berndjotherwise i was about to suggest the N2 laser20:49
azonenberg_workThis is basically going to be a raster scan20:49
azonenberg_workwhen i say zero, i mean i dont need coherence at all20:49
berndjif you're willing to sacrifice some power, maybe20:49
azonenberg_workBut it has t obe narrow20:49
azonenberg_workThe plan is basically to raster scan the laser across the die20:49
azonenberg_worka ~1mm field at ~5um resolution is 200x200 pixels20:50
azonenberg_workblankign as necessary20:50
azonenberg_workjust like those cheap laser projectors people mak20:50
azonenberg_workexcept the die is a lot closer20:50
azonenberg_workand the theta tolerance is smaller20:50
berndjwait, are we talking ablation or PR exposure?20:51
azonenberg_workberndj: PR exposure20:51
azonenberg_workablation needs waaaay ore20:51
azonenberg_workmore*20:51
berndjah, hence you need <400nm lines20:51
azonenberg_workenergy20:51
azonenberg_workYes20:51
azonenberg_workBut the spot size can be large as i am making what might be a projection mask rather than a contact one20:51
azonenberg_workI'm not projecting onto the di20:52
azonenberg_workits a mask (metalized microscope slide)20:52
berndjOT: i cut a section through weld beads i ran yesterday, and i'd like to etch them somehow so i can see the various zones.  suggestions?  I have HCl and H2O2, could that work?20:55
berndj(it's a crappy weld, but i want pics for my blog)20:55
azonenberg_workWhat is the material20:55
berndjmild steel20:55
azonenberg_workand whats the goal, show grain boundaries?20:55
azonenberg_workFirst step is to polish it20:55
berndjnot necessarily even that, just to show the differences where crystals are smaller vs where they are larger20:56
azonenberg_workRe oxide growth rate...20:56
berndjpolish - to what degree?  i suppose having filed it flat isn't enough :)20:56
azonenberg_workhttp://www.lelandstanfordjunior.com/thermaloxide.html20:56
azonenberg_workberndj: optical polish is ideal :p20:56
berndjwell, i *do* have SiC and rouge for my telescope making20:57
berndjlol @ your SEM micrograph showing a 5000µm scale (aka 5mm)20:58
berndjthinks look so alien under an SEM!20:58
azonenberg_workberndj: yeah20:58
azonenberg_workgotta love that depth of field20:58
azonenberg_workextreme sharpness20:58
berndjwhat did you put those chip on?  it looks like... salami?21:00
azonenberg_workstandard ted pella carbon tape lol21:00
azonenberg_workits pitch black under visible light21:00
azonenberg_workThese are all <110> orientation as you can see by the edges21:00
azonenberg_workwhich are angled rather than perpendicular (i.e. parallelogram as seen from top down)21:01
azonenberg_workSomebody said http://colossus.cs.rpi.edu/~azonenberg/images/homecmos/2011-09-20/die_i7_002.jpg looked like a sewer grating lol21:02
berndjhttp://www.cleanroom.byu.edu/OxideTimeCalc.phtml another calculator that actually works (for me - the other one barfed when i tried it)21:40
berndjlol, they do look a bit like sewer gratings21:40
berndjouch, a 1000nm oxide will take 9200 hours to grow at 700C21:41
berndjdry.  wet seems faster?  #todayilearned21:42
azonenberg_workberndj: wet is a lot faster21:46
azonenberg_workand yeah lol21:46
azonenberg_workSo at room temp?21:46
azonenberg_workOr hot but not THAT hot?21:46
azonenberg_workIsnt gonna happen21:47
berndjyeah :(  even self-cleaning ovens don't reach 700C21:47
berndji think up to about 500C, but unless you leave your wafer there for a month, not very practical21:48
azonenberg_worklol yeah21:52
azonenberg_workLet me put it this way21:53
azonenberg_workat 1200C the wall of your oven is glowing yellow21:53
bart416You'd have to build your own clay blast furnace azonenberg22:12
azonenberg_workbart416: or buy one22:14
bart416Depends on how big your budget is22:14
bart416A blast furnace isn't that hard to build22:14
CIA-67homecmos r126 | wiki/Vendors.wiki | Add wafer vendor metalink.22:27
--- Wed Sep 21 201100:00
« Monday, 2011-09-19 Index Wednesday, 2011-09-21 »

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!