#homecmos IRC log for Wednesday, 2011-11-30

B0101azonenberg: you there?06:09
B0101i was thinking about making a mini clean room, in a box06:10
azonenbergYou mean a glove box?06:10
azonenbergYeah, entirely feasible06:10
azonenbergBuilding one is on my to-do list too, for certain critical operations06:10
azonenbergfor example wafer clean followed by spin coating06:11
B0101similar to a glove box, but this time, they can be detached/attached to a process machine...06:11
azonenbergYou mean a sealed container for carrying between tools?06:12
azonenbergYeah, thats doable too06:12
B0101tell me when my ideas get crazy though06:13
azonenbergWhat you're descriibng is SOP for most modern fabs06:13
azonenbergthe cleanrooms are like class 1000 and the tools are all sealed and filtered to below class 106:13
azonenbergnobody ever handles the wafers, they're moved by robot between tools in sealed boxes06:13
azonenbergBasically, around 10 years or so ago06:13
azonenbergSomebody finally realized that humans are dirty06:13
azonenbergand putting them in a clean environment isnt the best idea06:14
azonenbergSo now the outer "container" cleanroom is only used for servicing etc06:14
azonenbergand the tools themselves do all of the interesting stuff06:14
azonenbergSo what are your crazy ideas?06:16
B0101hmm, well I ws thinking about getting to 45nm scale (really crazy idea though)06:16
azonenberglol that *is* crazy :p06:17
azonenbergNot saying impossible, very little is06:17
azonenbergBut it will be far from easy lol06:17
azonenbergFor prototype-scale volumes your best bet would probably be e-beam direct write06:17
azonenbergforget about masks06:17
azonenbergbuild yourself a SEM and then make a direct-write module06:18
azonenbergI suggest you set a more reachable goal first06:18
azonenbergand see how feasible that is06:18
azonenbergsay, 500nm06:18
azonenbergthen 18006:18
azonenbergI mean, certainly doing this kind of stuff on the prototype scale will be easier than mass production06:18
azonenbergbut by how much?06:18
azonenbergI intend to start out by building a nice contact aligner06:18
horizontallybuild yourself a SEM.. azonenberg is one of the few people i know who could probably build it (and somehow acquire a diffusion pump from the lab manager) in his spare time06:19
azonenbergThat will let me hit 12.5 um design rules with professionally made film masks (laserlab.com)06:19
horizontallywhat do you think of photosciences?06:19
azonenbergare they a mask shop?06:19
azonenbergNever heard of them, photronics is the only other one i know of besides laserlab06:19
horizontallyi was told about photosciences from our lab manager06:20
azonenbergAnyway so that'd be a temporary step though06:20
azonenbergThe next step is to build my own laser direct write rig ;p06:20
azonenbergSo i can make my own chrome-on-glass masks06:20
azonenbergThen the most involved tool (on the lithography front at least) would be an H-line stepper06:20
azonenbergprobably 4x reduction of a 12.5um design rule mask06:21
horizontallywhy do you want to make stuff so small?06:21
horizontallywhat is your intended goal from there?06:21
horizontallysure it's nice and all06:21
azonenbergwould put me at a 3.125um design rule06:21
azonenberghorizontally: a) have fun06:21
azonenbergb) advance the state of the art in low-budget fab06:22
azonenbergto the point that a 350nm fab is something you could have in every high school science classroom06:22
horizontallywow, sure, cool06:22
azonenbergThats really the idea, there is not enough people getting into nanotech and having it available might help06:23
azonenbergAlso, just the coolness factor of being able to, say, design a custom ASIC instead of using a CPLD for some glue logic :p06:23
azonenbergimagine having a little project with a PIC and a homemade MEMS accelerometer in a flip-chip BGA package06:23
azonenbergFCBGA is how i plan to package all of my stuff, actually06:24
azonenbergbecause it means i dont need to wirebond06:24
azonenbergjust put the chip on a hot plate, heat to melting point of solder06:24
azonenbergand under a 30x stero microscope use fine tweezers to place a solder ball on each bond pad06:25
azonenbergthen cool06:25
azonenbergflip over, put on PCB, and put board in oven to reflow06:25
horizontallyi don't know much about packaging06:25
horizontallyhow did you learn about it?06:25
azonenbergthis would basically be a chip scale BGA06:25
horizontallyi have a pcb design book i haven't read through yet either06:25
azonenberga) look at premade chips and study them from the outside06:25
horizontallythough it's meant for orcad06:25
azonenbergb) reverse engineering06:25
horizontallyoh jesus, the long road06:26
azonenbergdecapping premade chips and studying how things fit together06:26
azonenbergplus various reading on wiki etc06:26
azonenbergbut once you actually look at a DIP that you've dissolved the plastic from06:26
azonenbergand see how everything fits together06:26
azonenbergit all makes sense06:26
horizontallyhow do you dissolve the plastic but not the chip inside?06:26
horizontallyso there's a bit of wirebonding06:26
azonenbergHot nitric acid is the SOP06:26
horizontallyand some polymer container06:26
horizontallynitric acid really? wow06:27
azonenbergit eats organic mateirals but not silicon, glass, or gold06:27
azonenbergthe chip is usually passivated with SiO2 and has gold bond wires06:27
azonenbergThough if you arent careful it will eat the tin-plated copper legs06:27
azonenbergbut the die itself survives with no problem06:27
azonenbergThe cheap option is to just throw the whole chip into a beaker of sulfuric06:27
azonenbergand cook for a while06:27
azonenbergless selective but if you want a bare die to study its fine06:27
azonenbergYou can do that method with nitric too06:28
azonenbergIn either case, you will want a fume hood or to be working outside06:28
horizontallyi got nitric and sulfuric06:28
horizontallyi'll try nitric soon06:28
horizontallythat sounds cool06:28
azonenbergHot acid vapor is somewhat unpleasant06:28
horizontallyyeah i've smelt it before06:28
azonenbergif you think the fumes are bad at room temp try at 150C06:28
azonenbergThis was the first chip i ever decapped06:30
azonenbergDidnt go too well but it demonstrates the technique06:30
azonenbergyou can see i broke a few bond wires06:30
azonenbergThe blue stuff was a silicone rubber i used to try and keep the acid from falling off the side and damaging the pins06:30
horizontallyhow do they get all of those bond wires on there jesus06:30
horizontallyso tiny and seemingly error prone06:30
azonenbergits some kind of CPLD or FPGA, i dont know details of the part06:30
azonenbergLater i discovered things went better by drilling or milling into the middle of the package06:31
azonenberguntil you're a few hundred um above the die06:31
azonenbergthis makes a well06:31
azonenbergyou then heat the chip on a hot plate, put one drop of acid in the well06:31
azonenberglet it sit, then pull out and swish in a beaker of acetone06:31
azonenbergOr, as I did originally, spray with acetone http://colossus.cs.rpi.edu/pictures/2010/October/10-7-2010%20-%20decapping/S7309756_sm.jpg06:31
azonenbergThe problem with that technique is that it has a habit of dislodging bond wires06:32
azonenbergswishing seems to give better results06:32
horizontallyyou got a clean lab06:34
azonenbergThats my old lab06:34
azonenbergI stopped doing as much wet chem when i moved out of it06:34
azonenbergIt was a storage room in the building next door to my apartment06:34
azonenbergthe landlord told me i could use it until he remodeled it the next year06:35
azonenbergWhich is now happening :p06:35
azonenberghttp://colossus.cs.rpi.edu/pictures/2010/October/10-31-2010%20-%20chiprev/S7300154.JPG is an example of how nice the results can be06:36
azonenbergNo, i didnt eat off the pins, its a 44-QFN06:36
azonenbergseen closer06:36
azonenbergNotice the unbonded pads06:36
azonenbergthe part comes in 44 and 64 pin versions and they use the same silicon for both06:36
azonenbergIt's the Microchip ENC424J600 / 624J60006:36
azonenbergethernet PHY/MAC with SPI+parallel interfaces06:37
azonenbergthe MAC is at top left, that big mess of analog circuitry06:37
azonenbergthe PHY*06:37
azonenbergthe DSP and MAC logic are at the right in that pile of standard cells06:37
azonenbergthen under the CMP filler at center is the RAM06:37
azonenbergi havent fully reversed it though, been busy working on other stuff06:37
azonenberghorizontally: My current lab, as of a few months ago http://imgur.com/a/Bnwst06:38
horizontallyi see your "gold light"06:38
azonenbergYep, i use that for when i do litho06:39
azonenbergthe lamp to the right of the glassware drying rack is yellow too06:39
--- Thu Dec 1 201100:00

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