#homecmos IRC log for Sunday, 2011-08-21

azonenberglol duct tape01:42
azonenberg_labWoot05:50
azonenberg_labTesting use of 1000nm evaporated Cu over 5nm Cr as a hardmask for KOH wet etch of <110> Si05:50
azonenberg_labearly experiments are promising05:50
azonenberg_labbut i need to leave it in the etch bath longer05:50
azonenberg_labthe Cu isnt attacked visibly but the Si isnt etched by much either :P05:51
azonenberg_labsurface is roughened, which indicates that etching is happening05:51
azonenberg_lab(of the Si)05:51
azonenberg_labIf this works out i may be able to ditch that Ta2O5 for good05:53
azonenberg_labIts been the main thing killing my yields lately :P05:53
azonenberg_labtoo finicky05:53
azonenberg_labaww... it didnt work, there are bubbles in the film06:03
CIA-67homecmos r114 | trunk/lithography-tests/labnotes/azonenberg_labnotes.txt | Today's lab notes06:04
bart416azonenberg, I think I have it ifgured out07:31
bart416Conductive ink07:32
azonenbergFor?07:32
bart416Then I can copper plate whatever I want07:32
azonenbergOoh07:32
azonenbergThats what i tried for via plating07:32
azonenbergbut i tried using it bare07:32
bart416If I find one that can be diluted in water07:32
azonenbergi didnt try electroplating over it07:32
azonenbergDoesnt have to be water - any common solvent07:32
azonenbergethanol, isopropanol, acetone07:32
azonenbergfind one that can be thinned with any of them07:32
bart416I'd like something that I could remove again though07:33
azonenbergSame solvent again?07:33
azonenbergi dilute my PR in acetone before spin coating it07:33
azonenbergand strip with acetone after etching07:33
bart416So I can coat the plexiglass in it07:33
azonenbergas long as its a solvent-evaporation process and not curing you're fine07:33
bart416copper plate07:33
bart416lithography07:34
azonenbergacetone isnt compatible with plexiglass but some of the other solvents may be07:34
bart416remove copper with H2SO4 or another acid07:34
bart416then kill the ink with something that doesn't kill the plexiglass07:34
azonenbergI like what you're thinking but must caution you that you may have adhesion problems07:34
azonenbergcopper is hard to stick to things sometimes07:34
azonenbergif the ink peels off the surface it may not work well07:34
bart416well yeah, it'll be sandwiched between two layers of plexiglass07:34
bart416So it only needs to stick for a few hours07:35
azonenbergHmm07:35
azonenbergExplain your process again?07:35
azonenbergFrom beginning to end07:35
bart416ok, get the circuit on the plexiglass07:35
bart416drill holes where necessary, plate vias, etc...07:35
bart416then stick the plexiglass together07:35
azonenbergSo you want to use a single sheet of PMMA as your board substrate07:35
bart416That's easy enough07:35
azonenbergthis is two layers only?07:36
bart416no07:36
azonenbergtop and bottom of the sheet?07:36
bart416two layers / sheet07:36
bart416You can smash together several sheets though07:36
azonenbergHow thin are these sheets07:36
bart416How thin do you want them?07:36
bart416I was thinking of starting with 1mm07:36
azonenbergHmm, even that is a bit thick07:36
azonenbergyou really want like 250um07:36
bart416I'd rather first get the process to work than start with a fragile material ;)07:37
azonenbergTrue07:37
azonenbergBut that means you have to deal with really high aspect ratios in vias07:37
azonenbergor make the vias gigantically wide07:37
bart416that's why the ink needs to be diluted as well07:38
bart416so it can get in the via through a bath07:38
bart416I want it so I can throw the sheet in the bath for an hour07:38
bart416And it's coated07:38
azonenbergHmm07:39
azonenbergAn hour?07:39
azonenbergYou want something with low surface tension07:39
azonenbergmaybe mix a little surfactant in there07:39
azonenbergsubmerge, pull out, blow dry, done07:39
bart416Just giving a random number :|07:39
bart416yeah ideally you'd have something that chemically binds with the acrylic07:39
bart416I was considering it07:39
bart416But how do I remove it again afterwards eh07:39
azonenbergFirst of all, why do you want to remove it?07:40
bart416You want a conductive substrate on a pcb?07:41
bart416Looks like a bad idea07:41
bart416That's why to use plexiglass azonenberg07:43
bart416It won't die if you try to solder it07:43
azonenbergLet me get this straight07:43
bart416and you can bond layers together with chloroform07:43
azonenbergIsnt the idea, lay down conductive layer07:43
azonenbergElectroplate copper07:43
bart416Etch07:43
azonenbergspin/dip coat photoresist, expose, develop07:43
azonenbergetch all the way through to the plexiglass07:43
bart416yes07:44
bart416I want the ink gone07:44
bart416or whatever I use to make the plexi conductive07:44
azonenbergSo you're talking about removing it from exposed areas07:44
azonenbergIt still seems like the best solution is to have your initial layer be evaporated copper07:44
azonenberga micron or so07:44
azonenbergthen electroplate that up to working thickness of 30um ish07:45
bart416Evaporation is expensive07:46
bart416And hard on large objects07:46
bart416+ you can't get copper to stick to acrylic by evaporation07:46
azonenbergIts not that expensive to run if you have the equipment already07:47
azonenbergBut define "large"07:47
azonenbergare we talking PC mobo size/07:47
bart416Yes07:47
bart416Anything smaller is pointless07:47
azonenbergpointless??07:47
bart416Yes07:47
azonenbergAre you planning mass production or prototypes here?07:48
bart416Cause then you can stretch it out over a larger dual layer board07:48
bart416prototypes07:48
azonenbergIf you're doing prototypes, develop a 6- or 8-layer process07:48
azonenbergand make 'em 3-4 inches across07:48
bart41610 cm is already big07:48
azonenbergThe evaporator i've used has around a 6" wide disk as the usable area for coating07:49
azonenbergif i were to build one of my own it'd be around the same size07:49
bart416I don't have access to an evaporator07:49
bart416And I'm not going to build one07:49
bart416We don't all go to some american university that gets 50 000 / year / student07:50
azonenbergBuilding one doesnt have to be expensive07:50
bart416And most of the funding we do have goes to the stupid management & economy department for their stupid pointless trips07:50
azonenbergCheap vacuum pump as a roughing pump07:50
azonenbergHomebrewed diffusion pump07:50
azonenbergor off ebay07:51
bart416that alone would cost me over 200 euro07:51
bart416Not in the budget07:51
bart416chemicals I can get07:51
azonenberghow?07:51
bart416plexiglas, how much square meter do you want?07:51
azonenbergThe more obscure chems have been my biggest expense07:51
azonenbergor among the biggest alt east07:51
bart416Normal-ish chemicals (not for semiconductor use) I can get through college/university with academic discount07:52
bart416So that's not expensive at all07:52
azonenbergJust out of curiosity07:52
bart416But a vacuum pump alone costs roughly 100 euro here07:52
azonenbergSee what it'd cost you to get 4 fluid ounces of 16% tantalum chloride in ethanol :P07:52
azonenbergI'd be willing to bet a few hundred USD07:52
bart416now in milliliter like the civilised world uses?07:53
bart416the homebrew diffusion pump would set me back another 100 euro in materials at least07:53
bart416Not to mention toolbits to machine the materials07:53
azonenbergYou dont have anything suitable in the engineering department?07:53
azonenbergTo machine it, i mean07:53
bart416I have access to a cnc machine shop, just need to pay for the toolbits myself07:54
azonenberginteresting07:54
azonenbergthey dont charge us for bits here07:54
azonenbergjust stock07:54
azonenbergas in, we're expected to supply our own07:54
bart416Not at college/university07:54
azonenbergoh, i see07:54
bart416Their cnc machines there are constantly in use and time on them is reserved from here till 2040 or something like that :S07:54
bart416Same for the rapid prototyping machines07:55
bart416+ my point still holds07:57
bart416You can't evaporate copper on plexiglas07:57
azonenbergdiff substrate?07:57
bart416Name one that's readily available, cheap, easy to machine and heat resistant?07:57
azonenbergstripping PR ont ork, you need acetone07:58
bart416the chloroform will attack the photoresist as well normally07:58
bart416so while bonding layers it will destroy the remains of the photoresist if there are any07:58
bart416maybe mylar08:03
bart416But that again brings the issue of how to coat it08:03
bart416Then again, mylar I could probably buy coated08:05
bart416buy coated mylar, electroplate with copper08:05
bart416etch08:05
bart416attack with whatever I want really08:05
bart416Actually that's the best thing I can come up with08:07
bart416buy mylar with a metal coat on it already08:07
bart416That's cheap, readily available and strong08:07
bart416Actually I'll settle on that08:10
bart416Mainly cause I can electroplate it08:10
bart416side bonus is that I could potentially laser cut thin sheets of mylar08:12
bart416never ever try to laser cut plexiglas ;)08:12
azonenberglol08:13
bart416what do you think about using mylar that already comes with a metalic layer?08:13
Action: azonenberg liek08:15
bart416azonenberg, side bonus is that the industry actually uses mylar for this :P08:24
bart416My only issue with this is how to do vias08:24
bart416But maybe the fact that there are already metals available might act as a seed for the plating process08:25
bart416also tantalum chloride azonenberg08:27
bart416high purity powder?08:27
azonenbergyes08:27
bart416So >99.8%?08:27
azonenbergyep08:28
bart416Bit more than 100 euro / 10g08:28
bart416and with bit more I mean like a few cents, lol08:28
azonenbergso 4 fl oz @ 16% solids =?08:28
bart416meh, that's up to you :|08:28
bart416catalog doesn't list liquid solution08:29
azonenbergondr if i got ripped off lol08:29
bart416academic pricing though08:29
bart416I doubt you can easily get this08:30
bart416A bit less pure it's 74 euro / 10g08:32
bart416(still 99.9%)08:32
bart416also, if I use regular electroplating I can finally get rid of that damned copper tubing I've been wanting to get rid off for ages08:57
azonenbergll09:00
bart416azonenberg, I think I found something suitable to power a portable railgun cheaply http://be.farnell.com/epcos/b41231a9568m000/capacitor-snap-in-100v-5600uf/dp/183929611:21
bart416I just did something scary11:52
bart416Mathz!11:52
bart416with 50 of those capacitors I should be able to speed up a bullet sized projectile to speeds in excess of a hunting rifle11:53
bart416total weight would be about 4kg11:53
bart416(this is assuming a gross lack of efficiency)11:53
berndjbart416, i always thought you want to drive railguns with *thousands* of volts just to get over the inductance13:23
bart416berndj, depends on the design really13:26
berndjand your level of ambition!13:28
bart416That too13:30
bart416lol13:30
bart416Nah, but the induction of a railgun isn't nearly as bad as you think13:30
berndjoh? ok13:30
berndjmaybe it's more applicable when you're building one you hear about on the news13:30
bart416at least, haven't really had any trouble with that before13:30
berndjas in, a very big one13:30
bart416Yeah, that's probably going to be bad13:30
berndji was doing some calcs for a spot welder the other day, that's also a bit depressing13:31
berndjyou need either a huge transformer or loads of copper13:31
bart416Yeah, and cooling is pretty nasty as well13:32
berndjis it?  i assumed you could manage that with a low duty cycle13:32
berndjsince the xformer windings are your first heat sink!13:32
bart416well yeah, I meant of the capacitor bank :P13:33
bart416We've tried to build some rapid fire coilguns at college13:33
bart416and the main point of failure is always the capacitor self destructing due to heat13:34
berndji was just gonna cheat and do mains -> bridge -> igbt crowbar as the "switch"13:34
berndjself destructing due to one shot or a duty cycle issue?13:34
bart416well yeah, you're pretty much short circuiting a lot of them close to each other through a rail of copper13:39
bart416what do you expect to happen? :P13:39
berndjto warm up, of course, but i've always assumed that a single discharge isn't enough to kill it?13:41
bart416with cheap chinese capacitors it is13:44
bart416Or it won't be able to build up enough charge until its cooled down again13:44
--- Mon Aug 22 201100:00

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