#homecmos IRC log for Sunday, 2011-07-31

bart416azonenberg awake or sleeping?07:25
azonenbergawake, for another few minutes07:25
bart416What would you use as substrate to try and make your own pcbs?07:26
azonenbergTrying to finish up some coding for work and make progress on the FPGA board07:26
bart416I meant multilayer eh07:26
bart416What we talked about yesterday07:26
azonenbergAnd hmm07:26
azonenbergWell, one problem is adhesion07:26
bart416you wand it to adhese at the start07:26
azonenbergThe evaporated Cu films i'm using now dont stick well to Si or SiO207:26
bart416And losen afterwards07:26
azonenbergresearch indicates that i need a few nm of Ni or Cr after07:26
azonenbergas an adhesion layer on the substrate07:26
azonenbergbefore doing the Cu07:27
bart416I'm thinking of trying to use polycarbonate07:27
azonenbergIf you're doing evaporation, bad idea07:27
azonenbergi dont think it can handle high vacuum07:27
azonenbergI'd go for glass07:27
bart416Yeah but can't dissolve glass easily without damaging the board07:27
azonenbergno, i mean you build the board on a glass plate07:27
azonenbergrather than FR407:28
azonenbergEvaporate a thin film of copper (500nm - 1um) over it07:28
azonenbergElectroplate that up to the normal 20-30 um07:28
azonenbergpattern by normal lithography07:28
azonenbergThen the question is, what to use as an ILD?07:28
azonenbergHas to be etchable07:28
azonenbergand depositable in relatively thick (tens of microns) layers07:29
azonenbergWhich rules out classic IC processes like CVD and sputtering/evaporation, they're too slow07:29
bart416Can't use polymers cause the developer will attack those as well most likely07:29
azonenbergSo probably a spin coatable polymer07:29
azonenbergActually, no07:29
azonenbergi was thinking a polyimide of some sort might work07:29
azonenbergmost plastics are resistant to dilute NaOH07:29
bart416also, cause it's not micro scale07:29
azonenbergthe harder part is etching it07:29
bart416You could use a focused laser to do the exposure07:29
azonenbergmost plastics can be removed with acetone, but that will kill the resist07:30
azonenbergYou could try a metal hardmask over the polymer but my guess is you'd have prolblems due to the solvent being isotropic07:30
azonenbergAlso, acetone is extremely fast07:30
bart416I know07:30
azonenbergyou'd want to use ethanol and water, or maybe isopropanol07:30
azonenbergboth attack photoresist slowly07:30
azonenbergRead the "bible of etch rates"07:31
bart416How thick of a layer of silicon oxide could you deposit?07:31
azonenbergEtch rates for micromachining processing, part 1 and 207:31
azonenbergand hmm07:31
azonenbergBy spin coating this stuff http://www.emulsitone.com/gl122.html I could do up to a micron per deposition07:31
azonenberggiven baking, a few layers cna be stacked07:31
bart416a micron isn't enough for a pcb07:31
azonenbergWhy do you say that?07:31
bart416You need to be able to solder this07:32
bart416And it has to survive some shocks07:32
azonenbergYou put power planes inside07:32
azonenbergon the back side*07:32
azonenbergthis is just for interconnect layers07:32
azonenbergas an ILD07:32
azonenbergYou could make the top surface thicker07:32
bart416If the heat goes through it might crack the glass though07:32
azonenbergAs long as you heat relatively evenly, nope07:32
bart416Soldering irons don't do even07:33
azonenbergyou'd want to use reflow07:33
bart416even with reflow07:33
bart416You'd need a professional oven to do this07:33
azonenbergre scratching, it can handle nearly 500g on a steel needle before scratching07:33
azonenbergso a soldering iron handled carefully wouldnt damage the film07:33
azonenbergThe bigger issue is the price of the solution07:33
azonenbergidk the exact number but my guess is, "not cheap" :P07:33
azonenbergIn any case i'll be pickign up some of that solution in a month or two07:34
azonenbergi have ideas for stuff using it07:34
bart416Actually I have a idea that might just work07:34
bart416Use a small layer of polymer to protect the metal against HF07:34
azonenbergCopper is resistant to HF07:34
azonenbergIt wont etch07:34
bart416Mhhh, you sure?07:35
bart416Then there is no issue07:35
bart416I can spin coat glass on as well07:35
bart416Just need to melt it07:35
azonenberg200nm of this glass dissolves in 5% HF in 10 sec and 1% in around 100 sec07:35
azonenbergmy 3% should be around 30 sec07:35
azonenbergand you dont want to do that07:35
azonenbergit'll cool unevenly and crack07:35
azonenbergunless you had a heated chuck and everything07:35
bart416mhhh, even if it's spinning in a vacuum?07:35
azonenbergAnd the viscosity would be so high it'd form a very thick layer, i'd guess07:36
azonenbergThe sol-gel mixture i linked you to is a room temperature deposition process07:36
azonenbergthen you heat treat after spinning to densify and harden it07:36
bart416micron density just isn't good enough07:36
bart416The mechanical properties of such a board would be crappy at best07:36
azonenbergAgreed, i wouldnt recommend it for anything serious07:37
bart416We need a polymer that can be attacked by something that doesn't eat away photoresist07:37
azonenbergI suggested the idea purely as a curiosity07:37
azonenbergThey have photosensitive SU-8 epoxy polymer07:37
azonenbergit can be spin coated07:38
azonenbergnot sure about the developer etc though07:38
azonenbergAnd its probably harder to get hold of07:38
bart416Yeah but SU-8 is hard to remove07:38
bart416Really really really hard07:38
azonenbergyou dont remove it07:38
bart416Does it have good insolation properties?07:39
azonenbergYou expose, develop, and *then* harden it07:39
azonenbergAnd its a polymer, so i'd assume so07:39
azonenbergvery few are conductive07:39
bart416I mean > 1000V07:39
azonenbergi've never actually worked with it07:39
azonenbergand no idea07:39
azonenbergi never do HV stuff07:39
azonenbergwell, HV by that standard07:40
azonenbergI consider 5 to be "high" since most of my work is 3.3 lol07:40
azonenbergand i'm starting to do 1.207:40
bart416Side advantage of SU-8 is that you could use it to prototype biomedical circuit boards as well07:40
azonenbergbiomedical circuit boards? You mean for implants etc?07:41
azonenbergyeah, iirc su-8 is biocompatible07:41
azonenbergthey use it for microfluidics etc07:41
bart416Well obviously for lab tests with non-human subjects07:41
bart416but it'd be possible07:41
bart416Though you'd have to use titanium instead of copper just to be safe07:41
azonenbergYou can evaporate pretty much any metal that melts at a lower temperature than tungsten07:42
azonenbergin a filament evaporator07:42
azonenbergThe guy i talked to has only failed to do like two metals07:42
bart416And cause this isn't nearly as precise you don't need to go hard vacuum either07:42
azonenbergi think platinum and nickel07:42
azonenbergAnd you need hard vacuum to evaporate07:42
azonenbergotherwise you get oxidation and all kinds of problems07:42
bart416You don't need to go down to the level for semiconductors07:42
azonenbergYou still need 1E-6 torr ish07:43
azonenbergeven if its just for SEM sample prep07:43
azonenbergotherwise the coating isnt even enough07:43
bart416This isn't SEM07:43
bart416It's a PCB07:43
azonenbergor can be contaminated with oxides07:43
bart416the features are easily 1mm07:43
azonenbergOnly way to know for sure is to try it07:43
azonenbergAnd 1mm?07:43
azonenbergThe board i'm designing now has 150um traces07:43
azonenbergon almost all of the signals07:43
bart416Lucky you that you can afford somebody to manufacture it like that07:44
azonenbergAfford? Its the cheapest process i could find07:44
azonenberg$100 for three 4-layer boards 3.8 x 2.5 inch07:44
azonenbergwith silkscreen and soldermask07:44
bart416that's way over my budget for these sort of things07:45
azonenbergOr $55 for three 2-layer, tinned copper with no silkscreen or soldermask07:45
azonenbergi'm putting a lot of time into the design for the board and splurging on the fab because i'll probably be using it for my research07:45
bart416you know, for two layer boards seeedstudio is cheaper :P07:46
azonenbergI'm planning to move to another fab in the near future07:47
azonenbergonce i learn gEDA enough to make serious boards with it07:47
bart416I feel dirty, just read wikipedia page on SU-807:51
bart416"allowing fabrication of relatively thick (hundreds of micrometers) structures with nearly vertical side walls"07:51
bart416That sounds about right :D07:51
azonenbergYeah, its almost transparent iirc07:51
azonenbergWhich means you need high exposure energy (not much absorption)07:52
azonenbergBut also you get very good resolution07:52
bart416Meh, I intended to use a laser :P07:52
azonenbergvery vertical sidewalls07:52
azonenbergI want to try doing vertical sidewalls using KOH in <110> Si07:52
bart416I don't need sub 10┬Ám features so using a laser on a 2D moving platform should be easy enough07:52
azonenbergBut first i need to debug my hardmasking07:52
bart416Assuming I have a way of sealing off the compartment with the laser I could keep the board in the vacuum while working on it eh07:54
azonenbergI see no reason to use vacuum07:54
bart416evaporation of copper07:55
azonenbergYou dont pattern during evaporation07:55
azonenbergYou evaporate a continuous film, spin coat PR, expose, develop, etch07:55
bart416No but that means it'd require human interaction07:55
azonenbergYou are not going to be able to do the entire process continuously07:55
bart416I'd love to make a machine that does all of it07:55
azonenbergYeah, but not in one chamber without breaking vacuum07:55
azonenbergDo you want to use the laser for exposure?07:56
bart416Why not?07:56
azonenbergOr for ablation?07:56
azonenbergYou cant put the photoresist down in a vacuum07:56
bart416It's spincoated on so why not? :S07:56
azonenbergThere's a solvent in the solution07:56
bart416The pump draws it out as well heh07:57
azonenbergthe vapor pressure is high07:57
bart416I know cause I tried07:57
bart416I did that when I built that stm07:57
azonenbergIt'd evpaporate insanely fast, i mean07:57
bart416Though I'm not sure if it'll work with SU-807:57
azonenbergYou wouldnt be able to coat it07:57
azonenbergThe only way it'd work is if you had photoresist that was evaporatable07:58
bart416Does SU-8 really have that bad properties?07:58
azonenbergNot that i know of07:58
azonenbergIt seems like a good idea07:58
azonenbergAnyway so you'd need evaporatable PR07:59
bart416Lets see if I can get my hands on another vacuum pump for starters07:59
azonenbergBut most photoactivators are heat sensitive07:59
azonenbergSo it wouldnt survive the process even if coated08:00
azonenbergit'd decompose08:00
azonenbergAnd you'd need an aqueous or solvent based developer08:00
azonenbergI'm not saying you cant automate the process - modern IC fabs are almost lights-out level automation08:00
azonenbergBut i dont think you can do it without breaking vacuum08:00
bart416well, did you build a fillament evaporator already?08:01
azonenbergI've been borrowing time on one in the electron microscopy lab at my school's mat sci department08:01
azonenbergwhile saving up and seeking out parts to build one of my own08:01
azonenbergIts on the todo list but is a few months out08:02
azonenbergmy next major purchases will be an oscilloscope/logic analyzer (for general purpose use, not specifically fab) and parts for a small fume hood08:02
azonenbergIn case you havent guessed my plans are bigger than *just* basic home fab lol08:06
azonenbergThe wishlist for my dream home includes a CNC machine shop and a class 1,000 clean room with a class 100 glove box (assuming that full class 100 is prohibitively expensive/difficult)08:07
bart416the clean room you could build yourself actually08:07
azonenbergI intend to08:07
azonenbergBut it will require extensive enough modifications i cant do it in a rented apartment08:07
bart416Alluminium frames08:07
bart416Perspex windows08:08
azonenbergSo it has to wait until i graduate and buy a place of my own08:08
bart416then a good filter and HVAC system to create positive pressure08:08
azonenbergI also wont have the budget for it until then :P08:08
azonenbergBut yeah, thats basically the plan08:08
azonenbergexcept i was thinking of glass for the windows08:09
azonenbergwall materials are TBD, most likely plastic or masonry08:09
azonenbergthe only thing i know for sure is drywall is out of the question :P08:09
bart416I'd use a plastic08:09
bart416Something scratch resistant08:09
bart416+ easier to seal08:09
azonenbergBut its not solvent resistant08:10
bart416Some are08:10
azonenbergNot many clear ones08:10
azonenbergpolycarbonate and acrylic are both degraded by acetone08:10
azonenbergWhich is what i'd use to clean up spatter from, say, a spin coating gone wrong08:10
bart416Ideally you'd use the same polymers they use to make glasses these days08:12
azonenbergI doubt those like acetone though08:12
azonenbergAlso, arent many glasses these days still actual glass?08:12
azonenbergjust with scratch resistant coatings?08:12
bart416Heh, I've gotten aceton splashes on mine, nothing wrong with that08:12
bart416Hell no08:12
bart416They're moving away from glass08:12
bart416too fragile08:12
Action: azonenberg doesnt wear glasses, as you can probably tell08:12
bart416+ polymers are thinner for the same effect08:12
azonenbergYou mean higher refractive index?08:13
azonenbergin that case, use something insane like tantalum oxide :P08:13
bart416My glasses would be 5 times as thick without using a polymer08:13
azonenbergPolycarbonate - 1.5841.58608:13
azonenbergFused quarts - 1.45808:14
azonenbergTa2O5 is around 2.14 lol08:15
azonenbergThats almost as high as diamond08:15
azonenbergand about on par with cubic zirconia08:16
azonenbergIn any case, off to bed with me - i've accomplished what little i could to tonight given i was also busy with work08:17
azonenberg1 part HCl / 6 parts H2O2, dilute the resulting mixture 4:1 with distilled water08:17
azonenbergetches my copper film in a more reasonable 8-10 seconds08:17
bart416Read up on CR-39 quickly :P08:18
azonenbergi.e. controllable enough to probably not overetch08:18
bart416That's what they use for glasses according to google heh08:18
bart416Most descriptions mention it's resistant to solvents08:18
azonenbergYeah, polycarbonate and PMMA are the only two clear plastics i'm particularly familiar with08:19
bart416"Taking into account that CR-39, one of the most sensitive ion track detectors, can not be removed by any solvent without altering the replica materials, the technique had to be adapted for using with this material."08:21
bart416I have a feeling it'll get the job done :')08:21
Action: azonenberg cats "standby" to /home/azonenberg/bedroom/azonenberg/sys/power/state08:21
azonenbergbe back on tomorrow :P08:21
bart416azonenberg, http://www.homalite.com/cr39.htm09:45
--- Mon Aug 1 201100:00

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