#homecmos IRC log for Thursday, 2011-11-17

horizontallyhomecmos people... do any of you all use spray on photoresist?03:22
azonenberghorizontally: I use spin coating03:31
azonenbergspray on is likely to produce an uneven film03:31
azonenbergits fine for PCB level feature sizes03:31
azonenbergbut for anything below 100um its going to be a bumpy surface03:32
azonenberglike if you've ever looked at spray paint under a microscope03:32
azonenberglooks kind of like sputtered metal but a few million times larger :p03:32
horizontallydamn because that looks easier to get right03:32
azonenbergwhere are you located - USA?03:32
horizontallyi want to do what you did with transparencies03:33
azonenbergthey sell liquid photoresist suitable for spin coating03:33
azonenbergi'd buy the 2 ounce bottle, my 8 is still well over half full and is near the expiration date03:33
horizontallyyou use a jeri ellsworth cpu cooling fan method?03:33
azonenbergi use a drill and a sanding wheel03:33
horizontallyoh okay03:33
azonenbergBut very similar03:33
azonenbergI intend to shrink it, though03:33
horizontallyi have cpu fans and nonworking drills...03:34
azonenbergusing either a cpu fan or, more likely, a model aircraft brushless motor and a custom chuck03:34
azonenbergbut a fan should definitely work03:34
azonenbergyou just wont be able to handle as large of a sample03:34
horizontallybut like how do you know when you're done?03:34
azonenbergmy drill based coater is pretty beefy03:34
horizontallyand how fast do you spin it03:34
azonenberglike, i can easily spin a full 4" wafer03:34
azonenbergThough the largest i realistically work on is a 2"03:34
azonenberg4s are pricey03:34
azonenbergand usually i do single chips03:34
azonenbergthe point is, though, it does have the horsepower for big samples which is nice03:34
horizontallyyou are loaded03:35
azonenbergFinancially? Far from it03:35
azonenbergi'm a grad student03:35
azonenbergi make less than $20k a year03:35
horizontallyoh wow03:35
azonenbergBefore taxes03:36
azonenbergi'm just good at scrounging stuff03:36
azonenbergdumpster diving ftw03:36
horizontallyi'm hungry. gonna go get ramen from the dumpster03:36
azonenbergmy spin coater was made from a drill a friend didn't need03:36
azonenberga 2x403:36
azonenberga sanding wheel from the home depot03:36
azonenberga cardboard box03:36
azonenberga staple gun03:36
azonenbergand one or two random bits of hardware03:36
azonenbergre speed, i kind of go by ear03:37
azonenbergi know what it sounds like at a speed that gives good reuslts :p03:37
azonenbergThe injectorall resist is a little too viscous for my scales of work, it forms a too-thick layer03:37
azonenbergso i dilute it 50% with acetone03:37
azonenbergby volume03:37
horizontallyand like what do you do when you're done with it? like it's stopped spinning03:37
horizontallydoesn't it just settle weirdly?03:37
azonenbergTape the sample down from the back (double sided), form a pool of resist on top03:38
azonenbergspin slowly to let it spread, then fast to thin it03:38
azonenbergkeep on spinning, you'll see diffraction rings appear as the film gets thinner03:38
azonenbergwhen the color stops changing the film thickness has stabilized03:38
azonenbergkeep on spinning another 10sec or so, then slow down gradually03:38
azonenbergthen i pick up the sample by the edge with tweezers03:38
azonenbergstick on a hot plate for 90sec or so to drive off excess solvent and solidify the film03:39
azonenbergand then do litho03:39
azonenbergOne o fthe problems with my current coater is that its a little large03:40
azonenbergso it wont fit into the fume hood i am building easily03:40
horizontallyyou're building a fume hood?03:40
azonenbergThe photoresist i use is dissolved in a somewhat unpleasant organic solvent03:40
horizontallywhy do you do all this?03:40
azonenbergin a well ventilated area its fine03:40
azonenbergBut if you have all the iwndows closed...03:41
azonenbergThe design calls for a 2 foot cube of 1/4" plastic sheet03:41
azonenberghot air welded03:41
azonenbergand venting out a 6 inch duct to the nearest window03:41
horizontallydo you print with inkjet or laser printer btw?03:41
horizontallyoh good03:41
horizontallyi was worried that the laser printers might melt the transparencies03:42
azonenbergMaybe inkjet ones03:42
azonenbergbut mine are rated for laser03:42
horizontallyrated for? aw damn03:42
horizontallyso in general transparencies won't work with laser printers?03:42
azonenbergi went to staples03:44
azonenbergthey sold laser and inkjet transparency03:44
azonenbergi bought laser03:44
horizontallyand so more logistics... how to you get the mask on the photoresist?03:46
horizontallyoh you solidify okay03:46
horizontallyi was going to say that spin part would be useless otherwise03:47
horizontallywhat's your light source?03:47
azonenbergIt hardens a bit during the spinning from solvent evaporation03:49
azonenbergthen you bake03:49
azonenbergthe bake is to bring it from tacky to solid, its not liquid by the end of the spinning03:50
azonenbergand my lgiht source is a halogen microscope illuminator wit ha fiber optic head03:50
horizontallyhmm that sounds fancy03:50
horizontallyi was going to just use a little OSRAM halogen bulbs i found and hook it up to a power supply03:51
horizontallyi don't know if i need an enclosure or anything03:51
horizontallydo you just need a light source? does it need to emit UV at all?03:51
horizontallybecause i think this halogen does emit a little UV03:51
azonenbergThat resist is DNQ-novolac based03:51
horizontallywow acroynms03:52
azonenbergand is most sensitive to the mercury vapor H to I lines03:52
horizontallynyms i always mess that up03:52
azonenbergwhcih is 405 to 365nm03:52
azonenbergblue to longwave UV03:52
azonenbergWideband sources like halogen will work but you need longer exposure than a pure UV source of the same intensity03:53
azonenbergthey also suffer chromatic aberration which may blur your exposure, however right now my limit is line edge roughness on the mask so its not too bad03:53
horizontallyhow thin can you get your lines by the laserjet printer btw?03:54
horizontallywhat is 1 px width actually in terms of real world lengths03:55
horizontallylike 1 mm? less?03:55
horizontallywow some links online say 0.1 mm (100 um)03:55
azonenbergat 600DPI a pixel is around 43 um03:56
horizontallyoh wow03:56
azonenbergi found five pixels to be a good design rule for minimum feature size03:56
azonenbergbelow that you risk blurring lines together or having breaks in them03:56
azonenbergymmv, test out the specific printer and transparency combo03:56
azonenbergbut this gives me around 200um features on the mask, reduced 10x by optics to get 20um on the wafer03:57
horizontallyi need to learn more about how current flow is affected by different materials03:57
horizontallylike i know very little about schottky junctions03:57
horizontallyand ohmic contacts03:57
horizontallyand band diagrams03:57
horizontallyi suck03:57
azonenbergi know little about those too03:57
azonenberghence why i've been doing MEMS :p03:57
horizontallyisn't that super hard?03:58
azonenbergi havent made anything yet :p03:59
azonenberggettting closer though03:59
horizontallyi'm a mechanical engineering major trying to get into mat sci03:59
horizontallyand it's tough shit03:59
horizontallyi don't even know how you do this stuff03:59
horizontallyare you a crazy good multitasker?03:59
horizontallydo you have a life outside of this stuff?03:59
azonenbergTypical desktop for me: http://i.imgur.com/QWQHA.jpg03:59
azonenbergdoes that count as crazy multitasking?04:00
azonenbergyes, that is four physical monitors and not virtual desktops04:00
horizontallywhat the hell04:01
azonenbergand a life out of home chip fab? Sure04:01
azonenbergi'm a phd student in comp sci04:01
azonenbergi have homework to grade, office hours to go to, classes to take04:01
azonenbergresearch to do04:01
horizontallyjesus christ04:01
horizontallycan you send me hw4.tex btw?04:01
azonenberglol why :p04:02
horizontallyi think your latex looks good04:02
horizontallyi want to use it as a template04:02
azonenbergits pretty standard04:02
horizontallyyou can strip out the actual homework if you want04:02
horizontallyit is, but i developed a lot of shitty habits04:02
horizontallyusepackage listings... hmm04:02
azonenbergand yeah, makes for nicely formatted source code listings04:03
azonenbergremember my degree is in CS, not EE or something like that04:03
horizontallyi guess i'd just never done \paragraph*{} before04:04
horizontallyit looks really neat04:05
azonenbergyeah, thats pretty much my stnadard homework template04:05
horizontallyamsfonts... never used that04:05
azonenbergthat pulls in some stuff necessary for heavy math04:05
azonenbergsame with amsmat hand amssymb04:05
azonenbergi forget what i needed it for but its now in my standard header04:05
horizontallyi use microtype, no idea why04:05
horizontallysupposedly makes stuff look good04:05
azonenbergi just cut and paste to everything04:05
horizontallyi have little comments reminding me in the template i have04:05
horizontallyit's not really a template04:05
azonenbergi see04:06
horizontallyit's like... grab the old file and remove all the junk04:06
horizontallyput in more junk04:06
azonenbergsame here04:06
azonenbergglad to hear you use latex and not something ugly like, say, ms word04:07
horizontallywell latex is frankly a waste of time04:12
horizontallyi am using libreoffice more04:12
horizontallyespecially since i don't have so many equations to show as images04:12
horizontallyand typing \includegraphics like a dork sucks04:12
horizontallyespecially for like no benefit either04:12
horizontallyhow thick do you grow your oxide btw?04:13
horizontallyor do you get pregrown oxide wafers?04:14
horizontallyazonenberg: ?04:20
azonenbergwhoops back04:26
azonenbergand i stopped using OO/LO because it was waaay too heavy and slow04:26
azonenbergand my homework/papers are usually math heavy04:26
azonenbergand i dont have oxidation capability atm04:26
azonenberga furnace is on the stuff-to-buy-soon list04:26
azonenbergthe only dielectric i've used so far was high-K04:26
azonenbergsol gel Ta2O504:27
horizontallyheavy and slow with your 4 monitors? :p04:32
horizontallybut yeah math heavy, latex makes a lot of sense04:32
horizontallyit's when figures of any kind are involved it gets kind of iffy04:33
horizontallyit's not that hard, but it's just unnecessarily complicated at that point04:33
horizontallyanyway, you're going to buy a furnace?!04:33
azonenbergsmall one04:51
azonenberg4 inch cube chamber04:51
horizontallyhuh... a cubic furnace?04:52
horizontallyhow much will it cost you?04:53
azonenbergaround $110004:53
azonenberghence why i dont have it yet04:53
azonenbergooh, price went up by $100 since last time i checked04:53
azonenbergactually no, it was always this price04:54
azonenberg1100 is the max *temperature* :P04:54
azonenbergthe furnace costs $128904:54
horizontallyyou know, i've seen people make omega heaters + heavy duty heat tape04:56
horizontallythen it's all wrapped in aluminum foil04:56
horizontallythat'd cost you about $200 maybe less04:56
azonenbergi need a more controlled environment04:56
azonenbergi also need to be able to purge it with N2 or Ar for heating w/o oxidation04:56
horizontallyso another heater + another heat tape would be about 40004:56
azonenbergand i dont think those are going to hit 1200C04:57
azonenbergi mean at that point you are glowing yellow, not even red04:57
horizontallythey hit 1400... i don't remember if it was F or C04:57
horizontallyF never mind04:57
azonenbergaluminum melts at 660C lol04:57
azonenbergbut to get good oxidation of silicon you need to hit >100004:58
horizontallyhmm, so that's just the cost of the furnace right?04:58
azonenbergi hope for a cheaper option, yo ucould probably get one cheaper used04:59
horizontallyyou know what... photolithography sounds really involved and will likely contaminate any devices i might want to make05:40
horizontallyi think i'll try shadow mask lithography05:40
horizontallymy worry is that the shadow masks will get crap deposited on them and then next time i try to deposit something using the shadow mask, i'll be redepositing previously deposited material05:41
horizontallyi'm interested in piezo devices05:42
azonenbergso what contaminants are you worried about05:42
horizontallyi worry that the photoresist step is basically a sort of liquid phase epitaxy05:42
azonenbergno, its a polymer05:42
horizontallyi worry that during the baking, etc. i will get diffusion05:42
azonenbergnot crystalline or even close to it05:42
azonenbergand the bake is done at near room temp05:42
azonenbergyou can even omit the bake and air dry if you want05:42
azonenbergThe biggest contaminant from litho is K+ and Na+ ions if you use KOH or NaOH developer05:43
azonenbergbut there are alternative developers that are metal free05:43
azonenberglike TMAH05:43
azonenbergand i suspect ammonia would work though i havent tested it05:43
horizontallyi guess i don't know enough about polymer chemistry05:48
horizontallyi guess your point is that if it air dries... i don't even really know how diffusion works, but i would think there wouldn't be a large driving force for diffusion...05:49
horizontallyi know diffusion depends on differences in concentration05:49
horizontallybut i don't know how that is related to temperature05:49
azonenbergdiffusion in silicon normally requires high temps05:49
horizontallytemperature would perhaps increase the decrease the mean free path?05:49
horizontallyit's not silicon i'm worried about05:49
horizontallyit's what i deposit on top of it05:49
azonenbergwhoops, wrong listing05:49
azonenberglook at the diffusion coefficients as a function of temperature05:50
azonenbergSo it depends on the density of whatever you are depositing05:51
azonenbergif your piezo material is a ceramic i'd guess it doesnt diffuse much05:51
azonenbergbut the main thing is, photoresist is a huge molecule05:51
azonenbergcompared to an ion like Na+ or F-05:51
azonenbergwhich penetrates much more rapidly05:51
azonenbergits this giant organic compound05:51
azonenbergif anything you should be worried about contaminants from the developer05:52
azonenbergNot the resist05:52
azonenbergand if you use a dev like TMAH or ammonium hydroxide there should not be any problems05:52
bart416horizontally, to diffuse a material into a crystal lattice you need considerable amounts of energy05:55
bart416Most normal chemical reactions are not capable to muster the entropy change required to do such things without some assistance from outside05:55
horizontallywhat about my worry about sputtering a shadow mask onto a sample?05:56
horizontallyis that feasible?05:56
horizontallyi guess with ion beam sputtering in particular05:57
azonenberghorizontally: that implies making a mask that's small enough05:57
azonenbergAnd in that case you should be evaporating, not sputtering05:57
azonenbergbetter shadowing properties05:58
azonenbergbut that basically degenerates to a lift-off process05:58
azonenbergat the max resolution (mask in contact with wafer)05:58
azonenbergAt which point your mask is probably photoresist05:58
horizontallymm, the evaporator is busted. some 3 phase thing issue05:59
horizontallyi guess i have to use photolithography then05:59
horizontallyor maybe not i should revisit a few things05:59
horizontally1. quantify the issue of contamination due to photoresist06:00
azonenbergIt's virtually nil06:00
azonenbergthe issue is developer06:00
azonenbergand a suitable choice of developer can minimize that06:00
horizontally2. quantify the issue of contamination due to a shadow mask with an ion beam sputtering system06:00
horizontallyknowing that it's argon @ 8 kV and 6 mA06:00
azonenbergi dont know enough about the particulars of sputtering to give you any kind of estimates06:01
azonenbergbut sputtering is generally conformal06:01
azonenbergi dont think it'd work well for masking06:01
azonenbergand how would you make a mask at that size without photoresist and optical reduction?06:01
horizontallywell i've managed like 500 um - 1 mm line widths06:01
horizontallynothing great06:01
azonenbergyeah, when you hit 20um without photolithography come talk to me lol06:02
horizontallyjust using tape, then just using paper, then aluminum foil06:02
azonenbergyou are going to need a reduction step in there somewhere06:02
horizontallymaybe, maybe not06:02
azonenbergDepends on your desired feature size i guess06:03
azonenbergabout to hit the sack, class tomorrow and i want to be at least somewhat awake for it :p06:03
horizontallyall right thanks a lot06:03
horizontallyi got gobs to think about06:03
horizontallybart416: sorry did not mean to ignore you. i read what you said and i'll just have to learn more about diffusion and polymers06:03
bart416A crystal lattice has what we call grid energy06:06
bart416to diffuse particles into the crystal you need at least the grid energy in most cases06:06
bart416the grid energy being the entropy difference between the bound and unbound state of the atoms or molecules in the crystal06:07
horizontallygoing back to the sputtering just out of curiosity, if you knew the energy with which the ion beam struck a target and you knew the shortest distance between the target and the sample, could you figure out whether the sputtered atoms will undesirably etch the shadow mask?06:58
horizontallyyou're just giving me stronger resolve to read through the fundamentals of materials science and engineering06:58
horizontallyaw man i forgot to ask azonenberg how he aligns the mask07:05
horizontallyor lock the mask for the matter07:05
azonenberghorizontally: i mostly do projection litho14:06
azonenbergusing my microscope14:06
azonenbergso i turn the lamp down low, align the mask to the wafer using the microscope stage to move the wafer around14:07
azonenbergthen turn it up high to expose14:07
horizontallyazonenberg: how do you "hold the mask down"21:06
azonenberghorizontally: hold it down? What do you mean22:33
azonenbergon the projection plate?22:33
azonenbergIt's small enough that i just let it sit22:34
azonenbergfor contact i'd need something like vacuum22:34
horizontallylike how do you get the mask stay on top of the thing you're covering and not fidget?22:34
azonenbergThe photomask just sits there22:34
azonenbergsometimes i tape the corners to the slide22:34
horizontallyokay, that's what i was thinking of doing too22:34
azonenbergbut i'm projecting so i can handle pretty coarse errors22:34
azonenbergas the error gets 10x smaller on the wafr22:34
horizontallyi was planning on doing it without a microscope22:35
--- Fri Nov 18 201100:00

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