#milkymist IRC log for Saturday, 2011-05-21

wpwrakin the end, you'll have a joystick with which you direct the death ray that picks off those last molecules that resisted everything else you threw at them ;-)00:21
wpwrakazonenberg: you guys lost me at KOH vs. HF. i live in HCl+H2O2 stone age. what do these critters do ? i suppose at least one of them is an etchant, but does this make the other ?00:22
azonenbergwpwrak: HF is hydrofluoric acid, which eats00:25
azonenbergSiO2 (glass) and related materials00:25
azonenbergBut not silicon or polymer materials00:25
wpwraksound like nasty stuff :)00:26
azonenbergHF is quite nasty, thats why i work with 2% solution and am still paranoid with double gloves, a face shield, etc :P00:26
azonenbergOverkill considering the stuff in this concentration is household rust remover00:26
wpwraki usually consider glass and ceramics as my final barrier ;-)00:26
azonenbergBut i've seen what the strong stuff can do00:26
azonenbergMy photoresist is a polymer that dissolves in a weak solution NaOH (household lye) when it's been exposed to light00:26
azonenbergUV in particular00:27
azonenbergSo you shine UV through your mask onto the resist, then develop00:27
azonenbergThen you can use HF to eat away the glassy layer under it00:27
azonenbergbut only where light hit the mask00:27
wpwrak(rust remover) hmm, interesting.00:27
azonenbergIt's rare, most rust remover is phosphoric acid00:27
azonenbergBut Whink brand is HF based00:27
azonenbergKOH is potassium hydroxide (caustic potash) which will eat silicon00:27
azonenbergIt also attacks photoresist so i need the Ta2O5 glass layer as a mask00:28
azonenbergi.e. pattern hardmask with photoresist, then pattern silicon with hardmask00:28
azonenbergusing two etch steps00:28
azonenbergThe nice thing about KOH is that it's sensitive to the bond structure in the silicon crystal00:29
azonenbergIt eats nearly 100x faster along one axis than the other one00:29
azonenbergSo by choosing the angle that you cut the wafer vs the bond planes you can get almost no etching (surface on the <111> vector), 57 degree sloped sidewalls (surface on the <100> vector), or vertical sidewalls (surface on the <110> vector)00:30
wpwrakbrr. messy. i kinda liked whatshername's transistor-making process you once mentioned. obviously, the process control roles are reversed there, but it seems that isn't nothing too horrible to do00:30
azonenbergJeri used HF00:31
azonenbergSame concentration as I am - 2% from Whink rust remover00:31
azonenbergThe main difference is she was doing patterning by hand with electrical-tape masks etc00:31
wpwrakyeah ;-)00:31
azonenbergI'm doing real lithography at several-micron scales00:31
azonenbergAnd i'm not to the point i can do transistors at this scale by far00:31
azonenbergThe first step is MEMS00:32
azonenbergBuild a comb drive00:32
azonenbergAnd before that, basic 2D engraving in Si00:33
azonenbergShallow etching, 10 microns or so00:33
azonenbergjust characterizing the process00:33
wpwraki don't know much about mems or their use. i have some ideas what a transistor can be good for, though :) one problem i'd envision is just connectivity. that's some more layers.00:33
azonenbergConnectivity, well00:33
azonenbergEtching metal is a solved problem, so is alignment00:33
azonenbergThe unsolved problem is how to get the metal there in the first place00:33
azonenbergthere are several deposition processes (sputtering and evaporation are the two big ones) but both need high vacuum00:34
wpwrakwhich requires some alchimist's lab in paris00:34
azonenbergLol, or something i build here in new york00:34
azonenbergThe first step will be etching these patterns http://i.imgur.com/VDW36.png just a few microns deep00:34
wpwrakand people are worred about cern's black holes eating geneva ;-)00:35
azonenbergBlack is informational only (not on the actual mask)00:35
azonenbergThe outer disk around each pattern is the field of view of my exposure system (about 500 microns)00:35
azonenbergBlue areas are masked off, white is to be etched00:36
azonenbergEverything outside the disk is not etched00:36
wpwrakdamn small. i still kinda like the idea of a dual BR direct exposure system.00:36
azonenbergI'm still thinking of direct write but not for exposure on the wafer00:36
azonenbergit'll be used for making metal-on-glass masks00:37
azonenbergthat are then used in my projection system00:37
wpwrakif you have it, why not use it for everything ?00:37
azonenbergBecause it can pattern but has little alignment capability00:37
azonenbergunless you build it in a far more advanced way00:37
azonenbergThe idea was, stick mask blank into the system00:37
azonenbergzero it *somewhere* on the blank00:37
azonenbergexpose, develop, etch00:37
wpwrakhmm. i'd try to align with perhipheral features. edges, holes made for that purpose, whatever.00:38
azonenbergnow you have your mask, 100nm of (say) nickel plated on glass00:38
azonenbergWith 50 micron (for example) features00:38
azonenbergFeed that into my current exposure system00:38
azonenbergReduce 10x00:39
azonenbergYou now have 5um features and a 2mm FOV, which is enough for a pretty good sized die00:39
azonenbergIt'd be enough to build a 4004 i think00:39
wpwrak2 mm is plenty, yes00:39
azonenberg500um is FOV with 40x objective00:39
azonenberg2mm is FOV with 10x00:39
azonenbergBut that also means my feature size grows 4x so the only way to prevent that is to make the mask 4x smaller00:40
azonenbergi.e. 50um half-pitch vs 200um (the limit of my printer)00:40
azonenbergI could even hit 5mm if i used the 4x obj but that'd be pushing it00:40
wpwrakor just use direct writing ;)00:40
azonenbergBecause i'd need 20um features on the mask00:40
azonenbergAnd those would be so small i wouldn't be able to see them at 40x magnification easily and get good alignment00:41
azonenberg100x is about the minimum to get micron-level alignment00:41
azonenbergAnd like i said, i'd love direct write onto the wafer00:41
azonenbergI just think it's beyond trhe scope of a rev 1 process00:42
wpwrakwhatever rev 1 may be :) for me, your inverted scope is an excellent hack. not only for the idea itself, but also because it breaks down one of those imaginary barriers.00:43
azonenbergThat was the single biggest unsolved problem for me, my group at RPI was puzzling over lithography and mask alignment for months00:43
azonenbergthinknig about building an optical column from scratch00:44
azonenbergThen i realized we already had one, problem solved00:44
wpwrakbut for a more dependable process, it seems to me that BR drives are an excellent opportunity. they have the resolution and, thanks to mass-market volumes, they're relatively cheap.00:44
azonenbergYeah, I fully agree with you there00:44
azonenbergIt's just something that should be saved until the rest of the process is somewhat more mature00:44
azonenbergIf you want, i can set you up with access to the project http://code.google.com/p/homecmos/ so you can hack on the wiki and start throwing thoughts in there00:45
wpwrak(building an optical column) looks like one of those problems that seem ridiculously simple - once you've solved them ;-)00:45
azonenbergExactly, once you have all the numbers and plans anybody can go build it00:45
azonenbergIt's the engineering that's hard00:45
azonenbergAnd, once you've built it, the calibration00:45
azonenberggetting everything precisely in focus00:45
wpwraki have a strong belief in the law of large numbers ;-)00:47
azonenbergAnd this is a solved problem with the microscope - which is a pretty nice piece of glass for the price00:47
azonenbergThe only big problem is chromatic aberration at higher powers00:47
azonenbergBut when using monochromatic light from an LED? Non-issue00:47
wpwrakthe microscope looks more like a mid-way point. it solves the problem at hand, showing that it can be solved, but it suffers many limitations as well (FOV, for one)00:48
azonenbergHence why it's a rev 1 process00:49
wpwrakbesides, it's not really "cheap". while BR is, by and large.00:49
azonenbergThe big problem, like i said, with BR is alignment00:49
azonenbergIt's going to have to be done like they do with steppers00:49
azonenbergyou have purpose-built alignment marks around the edge of the die00:49
azonenbergthat it calibrates off00:50
wpwraki stil think you can just use opportunistic alignment. find any features, and align to them.00:50
azonenbergMaybe for rev 3? But it'd be tricky00:50
azonenbergThis would be quite the hack if we pulled it off though00:50
azonenbergDirect write laser lithography system with 5um or better features00:50
wpwrakyou only care about repeating what you did before. but there's no absolute origin or orientation.00:50
wpwrakindeed. Intel@Home ;-)00:51
azonenbergGood point, only alignment matters00:51
azonenbergBut what about if you have a nearly blank mask?00:51
azonenbergAlso, when you say "align to them"00:51
azonenbergThe detector is easy00:52
azonenbergBut how do you see it?00:52
azonenbergIlluminate with the laser? Bear in mind every laser pulse punches a hole in the photoresist00:52
wpwraki'd align with the chip/fragment. or if that doesn't work, make a hole and align with that.00:52
wpwrak(not sure about the cleaning, though)00:52
azonenbergHole meaning drilled?00:52
azonenbergWaaay too imprecise00:52
azonenbergDo you have any idea how small five microns is?00:52
azonenbergHint, typical human hair is 20um across00:53
azonenbergLet's say you drill a hole 1mm across00:53
wpwrakyou only care about finding the same edge several times again00:53
azonenbergYou need to be able to localize to within 1/50 of the hole diameter00:53
azonenbergAnd you'd need >1 hole to get rotation locked down as well as position00:53
azonenbergIt'd just be too imprecise00:54
wpwraksure. make as many holes as you want. the first one is expensive, the rest is free ;-)00:54
azonenbergthe other thing is, in MEMS, puncturing the die can be problematic00:54
azonenbergYou might need a big piece for a heatsink or whatever00:54
azonenbergIt just strikes me as a really bad idea00:54
azonenbergWhereas if the first exposure etched alignment marks into th esurface00:55
azonenbergAt 5um pitch00:55
azonenbergWe could then see those markings and align future stages to them00:55
wpwrakyou don't need the exact position relative to the hole center or such. all you need is an edge you can detect accurately (with significantly less than infinite tries ;). once you have that, you can always reuse that edge.00:56
azonenbergThe other thing is, silicon fractures along cleavage planes00:56
azonenbergA hole is a potential fracture point00:56
wpwrak(mems) yeah, dunno about mems.00:56
azonenbergAnd, on top of that, each edge of the hole is going to be (at the microscale) broken along one of the cleaveage planes00:56
wpwrakcould you try to cut/mill a straight line ?00:57
wpwrakyou'd of course end up with a jagged edge, because you won't match the substrate's orientation00:58
azonenbergWell, the substrate will have at least one edge (or a flat for a wafer) parallel to a cleavage plane00:59
azonenbergyou'd align theta to that00:59
azonenbergSo you're aligned to the crystal structure01:00
wpwrakone degree of freedom solved :)01:00
azonenberg(for the first level) and then etch your main pattern plus alignment markings01:00
azonenbergcrosses, vernier scale, etc01:00
azonenbergAround the edges01:00
azonenbergThen the next level just aligns to them01:00
wpwrakdoesn't sound too bad. you still need at least two edges to align to, though for each step.01:01
azonenbergNo, you need one edge for the first step since you know which plane it's on a priori01:02
azonenbergFrom there on, you have a cross at each side of the die01:02
azonenbergAlign x/y to the center of them01:02
azonenbergand theta so that the lines are parallel to your axes01:02
azonenberggoogle "byu contact aligner" for Bringham Young Universtity's doc on their manual mask aligner01:03
azonenbergWe just need to design an automated system that can read those marks01:03
wpwrakthat's your microscope process. but what about the BR process ? there's not much a priori there. the machine has to discover everything on its own, each time the die is inserted.01:03
azonenbergno, you misunderstand01:03
azonenbergYou stick the die into it, it scans the edge01:04
azonenbergDetermines the crystal orientation (if the first run)01:04
azonenbergi.e. you tell it "this is a <100> wafer, align X axis to a <111> plane01:04
azonenbergThen it'll center your pattern on the fragment approximately01:04
azonenbergFuture exposures will scan the edge, determine the approximate center (to within 100um or so, really coarse) and then locate the alignment crosses01:04
azonenbergwhich it will use for the real fine alignment01:05
azonenbergThe problem is that we need to figure out how to see those marks01:05
azonenbergand get data that a machine vision system can use to determine the actual position of the die and adjust the stage accordingly01:05
wpwrakah, you're trying to identify features on the die01:06
azonenbergIts the only way to get precise alignment01:06
azonenbergbelow a hundred microns or so01:06
wpwraki was thinking of just (re-)identifying the die's shape01:06
azonenbergNot good enough01:06
azonenbergSome of my substrates are precise rectangles 1cm square01:06
azonenbergyou cant tell orientation from that01:06
azonenbergScroll down to the bottom01:07
wpwrakyou don't need a simple euclidian shape. i think it should be sufficient to identify the chaotic edges.01:08
azonenbergTake it from someone who's done a good amount of machine vision01:08
azonenbergSimple euclidean shapes are waaaay easier to work with01:09
azonenbergAnd you can get better precision with them01:09
wpwrakoh, fiducials are great - if you have them ;-)01:09
azonenbergWhat do you mean?01:09
wpwraki'd add them just for fun. and check them if anything goes wrong ;-)01:09
azonenbergThe first mask level is positioned approximately01:09
azonenbergAdds marks for all subsequent alignment steps01:10
azonenbergAll of the others lock onto those marks01:10
wpwrakwell, our problem is that it's easy to write with high precision and within a large area, but it's hard to see things with high precision within a large area01:10
azonenbergUnless you're built into a microscope01:11
wpwrakwhere "see things" means a generalized kind of vision01:11
azonenbergYou need a camera of some sort, which means an optical column01:11
azonenbergI still think the best option is direct write with little to no alignment onto a mask blank01:11
azonenbergfollowed by manual alignment using the microscope01:12
wpwraknow, there are some things you can see easily, e.g., whether an edge obscures a narrow beam or not01:12
azonenbergBut there's another problem01:12
wpwrakdrift of what ?01:13
azonenbergAs you pan from one side of the die to the other01:13
azonenbergAlignment will gradually be lost01:13
azonenbergdue to imprecisions in the motors etc01:13
azonenbergI suspect you'll need to realign periodically during a long run01:13
azonenbergEspecially if stepping multiple patterns on multiple dies on one wafer01:13
azonenbergIOW, if i move 1000um right and 1000um left01:14
azonenbergI won't be exactly where i started01:14
wpwrakif you lose alignment within the die, you're probably screwed. well, unless you can model the loss accurately :)01:14
wpwrakyeah, big wafers would be an issue. i can see that.01:15
azonenbergWell, i think we can all agree that the first draft of the bluray system will have little to no visio ncapability01:15
wpwrakbut perhaps that's a version 2 problem ? :)01:15
azonenbergand just be direct patterning on mask blanks01:15
azonenbergVision is a v3 problem01:15
azonenbergv2 is where we introduce laser direct write01:15
azonenbergv1 is transparencies on microscope01:15
wpwrakv3 then :)01:15
wpwraki think you still need simple vision for lowering the requirements for v2. 2 x BR player plus a photo diode is still a lot more accessible than a camera-ready microscope01:19
kristianpaulhacking  a camera you mean?01:21
wpwrak(fiducials) looking at the patterns, i wonder if they didn't exaggerate perhaps just a little (towards the end) ;-)01:22
kristianpauli wonder what can you do inverting some lenses on a cheap canon camera, if already on a cheap webcam is not that bad for DIY scopes~01:22
azonenbergExaggerate? What do you mean01:23
azonenbergWhen you're trying to get alignment down to a few nm it's kinda important01:23
wpwrakkristianpaul: ah, i think we covered the "mass market camera" side already :) the somethat disappointing result is that they don't go far enough (sample size = 1, so take this for whatever it means :)01:24
azonenberg+/- 150nm for example01:24
wpwrakazonenberg: (exaggerated) http://www.cnf.cornell.edu/image/stepper alignment global with microscope overlay.jpg01:24
azonenbergwhat about it?01:25
azonenbergYou need to be able to start from any angle and determine the correct position01:25
wpwrakwhat you really need is an edge. two for 2D alignment. i even get 45 deg crosses, but stepped bars, extra-narrow edges of rectangles with rounded corners ? :)01:26
azonenbergThe rounded corners are an artifact of the etch probably, and the long bars are for theta alignment01:27
azonenbergyou need a baseline that's good enough to find the other alignnment mark across the wafer01:27
azonenbergthen the small stepped bars are probably a vernier scale01:27
wpwrakokay, but that's more process optimizations then01:28
azonenbergfor alignment to below the smallest feature scale you can etch01:28
azonenbergi know that the stepper the guys at work are using (not the same model as the one on this page) has a vernier01:28
wpwrakhmm. still looks suspiciously like a process designed for human vision to me.01:32
wpwrakperhaps for some truly advances computer vision as well. but just to find a few edges ?01:32
azonenbergIt's advanced vision01:33
azonenbergThose things cost $40M +01:33
azonenbergFor my process i envision using a much simpler alignment strategy01:33
azonenbergProbably just a couple of crosses at 0/90/180/270 degrees01:33
wpwraki guess that's cheap as far as such things are considered ;-)01:33
wpwraki'd just let the damn machine scan the ragged edge, that's tens of thousands of data points. if you can't mine a decent position from that, what else can you do ? ;-)01:35
azonenbergThe thing is that not all of the dies have ragged edges01:35
azonenberg1cm square01:36
wpwrakhey, if it's perfect, even better ;-) find the edge on X, find it one Y, and you're done for the day ;-)01:36
azonenbergThats the thing, the edges are accurate to maybe 10um or 25um01:37
wpwrak(well, twice each - unless you eliminated rotation as well :)01:37
azonenbergi need alignment to 5um01:37
azonenbergand you still need optical to determine which way is up01:37
wpwrakah, so there's the great return of the ragged edge then :)01:37
azonenbergAlso, http://www.laserlab.com/plotprices.php01:37
azonenbergThis is the price to beat, you can get commercial photoplots for $34 for a 12x18 inch film sheet01:38
azonenbergat 4000 DPI01:38
azonenbergFeature sizes on the masks are maybe 25 or 50 um01:38
azonenbergif not better01:38
azonenbergAnd 12x18 inch film is enough for a complete mask set01:38
azonenbergseveral times over01:39
azonenbergconsidering that my exposure system can handle up to a 2cm square mask01:39
wpwrakcomparing price is tricky. since we're talking about extremely low duty cycles, almost all of the cost is in up-front investment01:39
azonenbergI'm just saying, unless you are making a lot it may be the case that masks are not viable to homebrew01:40
azonenbergOn the other hand, the cost of wafer fab is insanely high and homebrew is the only affordable way to experiment01:40
wpwraki think direct write is a lot more exciting. also because it reduces the number of steps. each step increases the probability of failure.01:41
azonenbergI think I am going to place a single order from these guys over the summer01:42
azonenbergInclude a bunch of test patterns etc01:42
azonenbergAnd see how small i can actually reach01:43
azonenbergIOW my current lambda is limited by printer resolution01:43
azonenbergSo if i have a super high res mask i can push the rest of the process to its limit01:43
wpwraksounds like a good plan. if there's something you can already get industrially made, even better01:43
azonenbergI'll do the CAD myself since i want to know all of the details of the mask layout01:44
azonenberg$35 plus shipping for 12x18 inches is like 100 test patterns01:44
azonenbergheck, for scale01:44
azonenberg$35 is the cost of a 4-inch wafer01:44
wpwrakthe price is certainly affordable enough. well, if you're in the US. seems they don't even do international.01:46
azonenbergThey do canada and mexico01:46
azonenbergnot sure about overseas01:46
wpwrakdoes the panama channel make Argentina overseas ? ;)01:47
azonenbergNo idea, they just mentioned canada and mexico as being extra fees for shipping etc01:47
azonenbergThe very-high-res is $5201:49
azonenbergLine widths of 12.7um for axially aligned and 25um for diagonal01:49
azonenbergThe normal is limited ot 38um for axially aligned and 75 for diagonal01:49
wpwrakline width = resolution * 2 ?01:56
wpwrakno, wait01:56
wpwrakunits 1mil/4 micron -> 6.3501:56
wpwrak(for 4000 dpi)01:57
wpwrakyour factor would be dpi*601:58
azonenbergYeah, looks like it01:59
azonenbergso 6pix is the smallest line they can reliably resolve01:59
azonenbergAnd my printer is 1200 DPI = 21um per pixel02:00
azonenbergtimes 6 pix = 127 microns02:00
azonenberg150 is the smallest i've actually tried doing, but 200 is the smallest that gives goodr results last time i tried02:01
azonenbergActually no, that explains it02:01
azonenbergThe printer is 600dpi, not 120002:01
wpwrak5 pixels/line then. still a lot. but okay, considering rounding errors, blur, and all that05:41
lekernelwhat's a BR drive?08:52
lekernelbtw this paper http://www.physics.rutgers.edu/ugrad/387/388s06/film_deposition/Musgraves05.pdf uses direct write with a DLP, which solves the alignment problem nicely08:56
lekernel"The patterning of this actuator was done by synchrotron x-ray lithography" lol why?09:05
lekernelit's 30 micron09:05
[florian]lekernel: blu-ray ?10:00
lekernelah, yeah probably10:01
[florian]morning btw :)10:02
wpwrak(BR drive) yes. the idea is to make a direct high-resolution writer from the head assembly of two Blu-Ray readers: one that provides laser and movement in the X direction, and other to move the target in the Y direction.11:05
lekernelroh: how are the rc3 cases going?13:48
kristianpaullekernel: had you made some tests with OpenCV and mm1?16:00
kristianpaulor and idea where to start with, i think i can econrage a univesiry student to take a look..16:01
kristianpaulor you have own plans for "Open Vision" on mm1 may be?16:01
kristianpaul:D ok16:06
--- Sun May 22 201100:00

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