#homecmos IRC log for Monday, 2012-07-30

kristianpaulCNC? you could start tring some FDM/Electrolitic building block system too ;-)00:54
azonenbergone thing at a time01:50
azonenbergI want a joystick that lets me move it around and snap pictures on command01:50
azonenbergthen the ability to script it for scanning of large areas01:51
azonenbergso i can image an entire chip (or even wafer) without human interaction01:51
kanzureazonenberg: are you practiced in the art of chip decapitation?04:59
azonenbergkanzure: I wrote this article http://siliconpr0n.org/wiki/doku.php?id=tutorial:tutorial_on_epoxy_decapsulation05:04
azonenbergyou tell me05:04
kanzuretough call, hmmmm05:05
azonenbergI don't currently have any HNO3 in my lab, I usually let my friend john (the other guy behind siliconpr0n) do the wet lab work05:06
azonenbergHe also has a computer-controlled microscope which is able to do automated mass imaging of a large die while i have to take all my pictures by hand05:06
azonenbergi'm catching up, ordered the first batch of parts to CNC my microscope today05:06
azonenbergand i now have a homebrewed fume hood so i'm less concerned about fumes from decapping05:06
azonenbergWhy do you ask?05:07
hideoazonenberg: are you walter white?05:11
azonenberghideo: lol no05:11
azonenbergI do have blue crystals in a cabinet05:12
azonenbergBut they're blue raspberry rock candy05:12
azonenbergOne of my roommates was watching the show and i freaked him out tossing the packet at him :P05:12
azonenbergyou should have seen the look on his face05:12
hideoyou don't watch bb?05:13
azonenbergNot regularly05:13
azonenbergi've seen a bit here and there05:13
hideoit's really good05:13
azonenbergMost of what I "watch" these days is man pages05:13
azonenbergand journal articles05:13
hideoit's the only tv i watch05:13
azonenbergI probably have another month or so of work to do on general lab tooling05:14
azonenbergthen build the spin coater and mask aligner05:15
azonenbergat which point i can start pushing hard05:15
azonenbergsee if i can get good patterning in, say, spin-on glass across a full 2" wafer05:15
azonenbergI also need to get a good anisotropic silicon etchant and photoresist developer that are metal-free05:15
azonenbergI've been using NaOH and KOH in the past since MEMS aren't ion sensitive05:16
azonenbergbut if i want to do CMOS that'll have to stop05:16
azonenbergTMAH is a bit nasty but i have an interesting idea05:16
azonenbergSince in, i think, both cases it's the OH- that does the dirty work05:16
azonenbergwhat's a nice nonmetallic organic base that's easy to find? Ammonium hydroxide05:16
azonenbergSomewhat more volatile than TMAH but much easier to get hold of05:16
hideotry it and see05:19
azonenbergFor developer, you dont need very aggressive05:19
azonenbergAnd for etch, well05:19
azonenbergi dont expect to do through-wafer etches with it05:19
azonenbergfor MEMS i can use KOH05:20
hideotmah is usually diluted as developer05:20
azonenbergWhat i mean is, i want to put alignment marks in the silicon05:20
azonenbergto line implants up against05:20
azonenbergThat will also function as a test etch for finding orientation if i'm doing KOH etc later on05:20
azonenbergSo i only need to go down maybe half a micron, enough to leave a visible edge05:20
azonenbergThats a more difficult problem i can deal with later05:21
azonenbergbut if straight ammonia, perhaps diluted, is usable as developer i'm fine05:21
azonenbergJust have to get something with the same OH- concentration as ~1% TMAH05:21
azonenbergiirc thats whats usually used05:21
azonenbergbut for aggressive Si etching you'd use 30% KOH and probably TMAH at a similar concentration05:22
hideodo you have DI water?05:22
azonenbergAs of now, just distilled05:22
azonenbergI've focused on lithography and it's pure enough for that05:22
azonenbergi haven't attempted to make transistors05:22
azonenbergit's sat in the plastic jugs for enough months it probably isnt DI grade05:22
azonenbergThey're HDPE though so it probably hasnt picked up many metal ions05:23
azonenbergCMOS is a whole other ball game in terms of the necessary purity etc05:25
azonenbergthats why i'm trying to do a comb drive as a first step05:25
azonenbergit demonstrates patterning at the necessary scales but is much more tolerant of ionic contamination05:25
azonenbergAnd i think i can pull it off in <110> Si with KOH if i angle things right05:39
hideogah unfortunately i can't say much05:40
azonenbergMy plan is to start out by taking a wafer with a single metal layer over oxide, pattern that single metal layer05:41
azonenbergthen put overglass over it05:42
azonenbergdo flip-chip BGA bonding05:42
azonenbergJust metal, no active components05:42
azonenbergThat will let me test metal patterning as well as packaging05:42
azonenbergThe next step will be two or three metal layers with CMP in between05:42
azonenbergI'm tentatively thinking copper damascene in spin-on glass05:43
hideocmp? chem-mechanical polishing?05:43
azonenbergif i use thick SOG as the ILD, i might get enough planarity from that05:43
azonenbergAnyway, once i demonstrate a full BEOL process including packaging and flip-chip bonding05:45
azonenbergthen i'm going to try doing a MEMS FEOL process for doing the comb drive05:45
azonenbergAt the end of that i should be able to make a comb drive in FCBGA05:45
azonenbergfour balls, two on each electrode (for stability)05:46
azonenberg500um wafer thickness on the rim for stability, then back-thinned to ~50um in the active area05:46
azonenbergthen through wafer etch with the finger pattern05:46
azonenbergit wont be nearly as nice as if i had DRIE but i dont know of any homemade MEMS ever05:46
azonenbergThe next step will be to duplicate Jeri Ellsworth's CMOS FEOL process and shrink to the 12.5¼m node using my litho process05:47
azonenbergand stick my BEOL on top05:47
azonenbergAt which point i should be able to make a 2" wafer full of CD4000 chips and hopefully at least a few of them will work05:48
azonenbergcomplete with overglass around the bond pads and solder bumps05:48
azonenbergThat will be the biggest milestone05:48
azonenbergfrom then on it's just improving yields until i can make something as big as a microprocessor05:49
azonenbergDebating between making an i4004 (from the original released mask art)05:49
azonenbergand trying to make a custom CPU05:49
azonenbergDo you know if anyone's ever made a BGA-packaged i4004? ;)05:50
azonenbergThe other possibility, like i said, is a custom CPU05:53
azonenbergDepends in part on where my yields are at that point and what process node i'm at05:53
azonenbergi can get cheap plastic film masks down to 12.5¼m design rules (3.125¼m pixels, 4 pixel min-feature design rules)05:54
azonenbergif i want to go finer i'm probably going to have to either kick out major $$$ for chrome on glass05:54
azonenbergOr ditch contact lithography05:54
azonenbergmove to projection and build myself a stepper05:54
azonenbergwith 4x reduction or so05:54
azonenbergRight now i'm using projection but there's no stepping/scanning05:54
azonenbergso i'm limited to one objective FOV, which is tiny05:55
azonenbergi'm moving to full-wafer contact in the short term but i want to make a stepper/scanner in the long term05:55
azonenbergIt also depends on where i am in my education/career at that point05:56
azonenbergonce i get out of school and save up for a few years i'm planning to build myself a house (if you can call it that) with a class 1000 cleanroom and a few other nice lab facilities05:56
azonenbergi could quite plausibly be able to do submicron features reliably, though likely not with great yields05:56
azonenbergRIE isn't even out of the question in the long term05:57
azonenbergif i could get the proper plumbing installed and find some cheap stuff used from an old 500nm fab someone is shutting down etc05:57
azonenbergSF6 only would probably be a good initial setup05:58
azonenbergSince if i could do nice vertical trenches in oxide, i could do copper damascene05:59
azonenbergThe first step would be sputtering or evaporation, i NEED in-house metal deposition capability more than really sharp etches05:59
azonenbergi'm pretty sure i can do at least 12¼m features reliably with wet etching05:59
Syncazonenberg: for that feature density you don't really need a cleanroom13:05
Syncwe do most stuff on laminar flow benches and yields are good13:05
azonenbergSync: good to know17:45
azonenbergi want to go smaller thoguh17:45
azonenberg500nm or even 350 in the long run17:45
--- Tue Jul 31 201200:00

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