#homecmos IRC log for Tuesday, 2012-01-31

kanzurehi all01:56
kanzurehi azonenberg04:27
gene_hackeranyone home?04:29
gene_hackerwondering how you guys make your masks04:30
gene_hackeror are you using a DLP projector to do maskless lithography?04:31
azonenberghi kanzure06:07
azonenbergwhoops, seems i missed somebody06:07
azonenbergWas having fun with BGA soldering in the living room06:07
azonenbergand in case he comes back, i use a normal office laser printer06:07
azonenbergthen reduce 10x with a microscope objective06:07
kanzureazonenberg: http://www.digikey.com/product-highlights/us/en/texas-instruments-pico/68806:46
kanzurei'm thinking about getting a dmd dev kit06:46
kanzureit would be especially helpful to have find one that does UV06:47
azonenbergkanzure: link doesnt work for me06:47
azonenbergits a 40406:47
kanzureworks for me o.o06:47
kanzuremaybe you block digikey. i wouldn't blame you.06:47
azonenbergi see a dk 404 page06:47
azonenbergi buy stuff from them all the time06:47
kanzureare you doing maskless lithography?06:48
azonenbergPrint on laser printer transparency06:48
azonenbergreduce 10x optically06:48
azonenbergi am looking at doing laser direct write with a bluray diode06:48
azonenbergstraight from the GDS06:48
azonenbergBut thats a ways out06:48
kanzurei'm doing a dna synthesis chip06:49
kanzurealso, any papers you can contribute on #homecmos would be welcomed06:49
azonenbergvery nice06:50
azonenbergand well, my work has actually mostly been in MEMS06:50
azonenbergmy lab notes are in the google code repository in the topic06:50
kanzureetching processes are pretty similar in a few places06:50
kanzuresometimes microfluidics requires gold electrodes and other crap06:51
kanzurebut has the same geometry/layout problems that you need to consider06:51
azonenbergAnd larger06:51
azonenbergWhich is a huge benefit for home / low budget fab06:51
kanzureyes, well, i'm not complaining about that06:51
comeonwhy are nitrides like silicon nitride or titanium nitride such good "barrier" layer materials?07:43
azonenbergcomeon: Mostly because they're denser, i think07:45
azonenbergI'm a little less up on the transistor side of things as i've mostly studied MEMS07:45
azonenbergI dont see Si3N4 used as a barrier as much07:45
azonenbergTiN and TaN i see used07:45
comeonnonstoichiometric silicon nitride is pecvd at low temperatures on gaas after doping gaas so that during an lattice damage removal annealing heat treatment the as doesn't float away07:47
comeonfuckin nazis on ##chemistry and ##physics quieted me07:48
azonenbergI assume by non-stoichiometric they mean Si rich and not N rich07:48
azonenbergbut i'm not sure of the exact characteristics of pecvd nitride's formulation07:48
azonenbergbut makes sense07:48
azonenbergalso just fyi this is the unofficial microfab / semiconductor channel on freenode07:50
azonenbergnot just home fab07:50
comeonthen rename it07:50
azonenbergI might07:50
azonenbergJust saying, discussion of cleanroom fab on "real" tools is not offtopic07:50
comeonwell i only asked because the bigger channels quieted me07:51
comeonbut you're the only person who ever talks here anyway07:51
comeonmight as well call it #azonenberg07:51
azonenbergOther people do talk07:52
azonenbergBut they idle a lot07:52
azonenbergthe community is small07:52
azonenbergNot many people interested in semiconductors07:53
azonenbergand even less know anything about them07:53
azonenbergeven less are on freenode07:53
comeonfreenode has a lot of software cocksuckers07:53
azonenbergHey, dont knock software TOO much07:53
azonenbergmy undergraduate degree is in computer science07:54
azonenbergand i'm going for a PhD in the same07:54
azonenbergMy research is in computer architecture07:54
azonenbergand operating system security07:54
comeonthat sounds super boring07:54
azonenbergNot at all07:54
azonenbergmy advisor is a crypto guy and i'm not too into the intense theory but when it comes time to prototype stuff its fun07:54
comeonwhatever floats your boat dude07:55
azonenbergI just do hardware stuff for fun lol07:55
comeonthere is a dude in ##electronics who literally floats in a boat07:55
azonenbergstarting to take EE classes because i've taken most of the ones CS offers adn am not interested in the rest07:55
comeongeckosenator or something07:55
azonenberginteresting, i've heard of the guy but didnt know he lived on a boat07:56
comeonhe cooks his own shit for fuel07:56
comeonsuper greenie weirdo07:56
soul-di din't want to know that07:56
azonenbergi assume he's online via a solar-powered laptop07:56
azonenbergwith a satellite or 3g connection?07:56
comeonand the asshole lives on food stamps07:56
comeoncuz he doesn't want to work a real job07:57
comeonjust float on a boat and leech off society07:57
azonenbergfurrywolf is mostly off-grid too, he/she (i think wolfy is a female) runs almost entirely on solar07:57
soul-di think  most people in here are just to steal azonenberg's  idea's  for world domination07:57
soul-dat least i am07:57
azonenbergsoul-d: they'll be waiting a long time07:57
comeongot to milk them nerds07:57
soul-dno   key is nano bots :P07:57
azonenbergthis isnt ##world-domination07:57
azonenbergmy world domination plans are occurring elsewhere07:57
comeoni'm gonna get shiftaced07:57
comeontired as fuck already07:57
azonenbergcomeon: what real nerds do with alcohol https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-snc7/292037_2309167690240_1280772249_2764502_5609467_n.jpg07:58
azonenberg(me on my 21st birthday)07:58
comeonyou need a life son07:59
soul-dlol he got one07:59
azonenbergi dont drink but i wanted to get a picture of me and a bottle of alcohol on my 21st07:59
soul-dhe does what he likes07:59
comeonyou don't drink?07:59
comeonlike at all?07:59
comeonwhy not?07:59
azonenbergtotally clean07:59
comeonare you a moralfag?07:59
azonenbergand i just dont feel the need to alter my neurotransmitters at all07:59
azonenbergi like 'em just the way they are07:59
azonenbergif it's not broken dont fix it07:59
comeonbut it's fun08:00
azonenberganyway so i showed the pic to my doctoral advisor and he thought it was hilarious08:00
azonenbergcomeon: maybe you think it is, and if so you're welcome to do it08:00
azonenbergwhile you're off puking your guts up i'll be doing FPGA design08:00
comeonhave you tried it at least?08:00
comeonyou don't have to drink to you puke your guts08:00
azonenbergI know i wouldnt like it08:01
azonenbergmy mind cant handle downtime08:01
azonenbergafter i graduated college i went with my family down the shore for a vacation08:01
azonenbergBy the end of the second or third day i couldnt take it anymore08:01
azonenbergpulled out my laptop and started doing mask layout for a MEMS comb drive08:01
comeonyou need to get laid08:01
azonenbergi need to be doing something intellectual every waking hour, basically, or i go nuts08:01
kanzurei don't talk much in here because of the other diybio channel ##hplusroadmap08:02
kanzurebut that one is called ##kanzure not #azonenberg08:02
kanzurealso, that channel is full of neurofags doing nootropics08:02
kanzurebut i think you were talking about alcohol08:03
kanzureso nevermind08:03
Action: kanzure sleeps08:03
comeoni don't know man. you need to lighten up08:03
azonenbergcomeon: lighten up? What is that going to do for me08:04
azonenbergi'm 21, most of my friends havent even graduated yet, and i have companies beating a path to my door basically offering me jobs08:04
comeonthere's more to life than companies and stuff08:05
azonenbergi'm in a PhD program at a well known engineering school doing work i love, with enough spare time to do crazy research and projects on the side08:05
azonenbergcomeon: suffice it to say, ten years from now i think i'll be very happy with where i am08:08
comeonwhere is that?08:08
soul-din a position he can look back08:08
soul-dand be proud on08:09
soul-dinstead of wasted time08:09
soul-dtelling others tehy don't have a lvie becasue you don;t have  one08:09
azonenbergHigh paying job at a R&D lab for some big tech company or national lab, living in a house/lab i built from the ground up with my own two hands08:09
azonenbergloaded with tools and equipment rivaling a small university08:09
azonenbergall the toys i could ever want, from a SEM to a class 1000 cleanroom to a machine shop08:09
azonenbergby that point i will likely have large submicron (750nm or so) CMOS fab working08:10
azonenbergoh, and out in the country so i can enjoy the outdoors and not be bothered by neighbors living too close08:11
soul-dazonenberg,   whenever your company put's out stocks   give me pm id like to take some08:11
azonenbergthough still within a commuting distance of wherever i work08:11
azonenbergsoul-d: lol, the last company i owned closed its doors when my partner took a full time offer and i left for grad school08:11
azonenbergso thats a ways out08:11
comeondo you have friends in real life?08:11
azonenbergcomeon: sure - my roommates, most of the electronics club08:12
soul-dgiven your personality comeon  i doubt you have any real friends08:12
azonenbergi dont go out drinking with them, more likely i'll wander into the living room at 2 AM and find one of them working in the lab08:12
soul-djust friendly to your face till you are gone08:12
azonenbergor etching a PCB in the fume hood08:12
azonenbergAnd you know what? I wouldnt have it any other way08:13
comeonsoul-d: i don't get teetotalers08:15
soul-dwho said  he was  and  why are you obseesed by it08:16
soul-dmaybe he understands the chemical formula08:16
soul-dand knows it's basicaly poision08:16
azonenbergAlso, for the record, i made this FPGA dev board http://i.imgur.com/6pYLF.jpg today to relax08:17
comeonin small quantities it's healthier than completely abstaining08:17
azonenbergand get away from studying for my upcoming PhD qualifying exam08:17
soul-ddid he say he did  and why are you intrested in  why  are you obsessed with an online persons life ?08:18
soul-dand why do you want to change it08:18
soul-dsomthing to do with just respecting people's choices08:18
comeonwhat's with the whiteknight?08:19
azonenbergi dont need anyone to defend me, he's just having fun poking the troll08:20
comeonyou don't have to respect people's choices especially if they are super different than everyone else08:20
Action: azonenberg pops lid of "troll food" can and throws it toward ##conspiracy-theory08:21
soul-di don't take people that use the word "super" very seriously08:21
comeoni don't take people that use the word "very" super seriously08:21
azonenbergAnd for the record i do not care what you think about me whatsoever, i am perfectly happy with where i am in life and am not "missing out" on anything08:22
azonenbergas a result of spending my evenings designing microprocessors instead of going out drinking08:22
soul-done establisches brain networks and makes you measurably smarter  other08:23
soul-dyou can guess08:23
azonenbergIf you enjoy going out and ingesting near-toxic levels of ethanol, thats your decision08:24
azonenbergOne less person to compete for my job08:24
azonenbergBut i find joy in intellectual stimulation and there's absolutely nothing wrong with that08:25
soul-dand i can tell from experience08:27
soul-dthat it's better tehn waking up in a ditch08:27
soul-dafhter   to much alchol08:27
soul-dstill wonder how i got home08:28
soul-di hate missing parts of memory :P08:28
Synckanzure: I probably can get you a DLP devkit11:25
kanzureSync: what are the details18:22
kanzurealso, can you possibly get me one that can also do UV18:23
Syncthey are sillicon imaging dev boards iirc18:23
kanzureazonenberg: i'm a little disappointed that you're in school :)18:24
azonenbergkanzure: why? Lol18:24
azonenbergthought i was some 14-year-old kid?18:24
kanzureSync: link?18:24
kanzureazonenberg: that would be the ideal case yes18:25
kanzureazonenberg: but more to the point that you feel university is necessary at all18:25
azonenbergI needed a piece of paper that says "bachelor of science" to get some employers to even look at me18:25
azonenbergcant say i've learned much from class, i dont find lectures a good way to learn18:25
kanzureare those the types of people that you really want to work for18:26
kanzureyour employment requirements should just be money, imho18:26
azonenbergNo, but18:26
azonenbergthey'd pay the bills while i looked for the *right* people18:26
SyncI don't know which ones gotta find them first18:27
kanzurei saw some pretty cheap dlp dev kits for $300ish18:27
kanzurebut they didn't do UV18:27
kanzureUV would be nice so that i can do DNA quantification by raman absorption spectroscopy or something18:28
azonenbergThey specifically said no uv, or they didnt say they did?18:28
azonenbergAnd depends on the wavelength18:28
kanzuretheir support varies, i've looked at some spec sheets.. whatever18:28
azonenbergfor lithography using DNQ-novolac photoresist you can get stuff that responds at 405 or even 450nm (my resist is 405 peak)18:28
kanzurethere's a few problems with dna synthesis, maybe you have some ideas18:28
azonenbergwhich is actually on the edge of the visible spectrum18:29
kanzureif i explain what the process is.18:29
azonenbergWell, i understand at a very basic level how dna replication works18:29
azonenbergbut when you say synthesis18:29
kanzurenah this is different18:29
azonenbergyou want to take a bunch of raw nucleotides and make a piece of dna?18:29
azonenbergwith a specific sequence?18:29
kanzureyou can use dna polymerase to make new sequences but you can't control what nucleotides it chooses18:29
kanzurethere's a few different chemistries for adding nucleotides to a growing single strand18:29
kanzurethis is generally called oligonucleotide synthesis18:29
kanzurethe most popular solid phase oligonucleotide synthesis method is phosphoramidite chemistry18:30
kanzurethere's also a version of this phosphoramidite chemistry that uses photolabile protecting groups on the ddNTPs18:30
azonenbergBear in mind my ochem knowledge is relatively limited18:30
azonenbergi have more experience with inorganic and even that is somewhat limitd18:30
kanzureso you shine some light to decap, and hope a reaction occurs to add that nucleotide to the end of the growing strand18:31
kanzurewash step. then some other capping/decapping steps.18:31
kanzurethe wash step is the main pain in the ass for me18:31
kanzureit's not a complicated step, but i mean from a practical standpoint18:31
azonenbergOk, let me see if i have this straight18:32
azonenbergStart with a single nucleotide, attach chemical A to one end and B to the other18:32
azonenbergB is stable and A degrades under UV18:32
azonenbergUse UV to remove A, add a small amount of your second nucleotide and attach to that end of the strand18:33
azonenbergremove unreacted nucleotide somehow, add more A to cap it?18:33
kanzuremore or less, yes18:33
kanzure"remove unreacted nucleotide" is a giant wash step18:34
azonenbergYeah, i can imagine that'd be fun to do without taking out your target strands18:34
azonenbergno clue where to start myself lol18:34
kanzuremost microfluidic devices use continuous flow, so they just flow the wash chemistry by all of the 'array units'18:34
azonenbergAnyway so why do you need lithography-style patterning for this?18:34
azonenbergare you talking for building a microfluidic unit?18:34
kanzuremost of the time this chemistry is done on solid supports (beads)18:34
kanzureso you can imagine an array of beads and DLP/DMD using different wavelengths to activate different nucleotides at different locations (beads in physically separate chambers)18:35
kanzurehrm, i'm not explaining this well. there's a few more points to mention18:35
kanzure1) you can do this in continuous flow with an array of these reactions, but they have to all be growing the same sequence so that you don't contaminate different reactions18:35
kanzure2) you theoretically want to do dna synthesis of different sequences in parallel; but you'd have to keep each reaction physically isolated18:36
kanzureif you are doing this in continous flow with a shared reagent/wash bath for all beads (on which this reaction is occurring),18:36
kanzurehmm no. the point i am trying to make is that because of the error rate of this chemistry you want to monitor each bead and make sure the sequence is correct18:38
azonenbergOoh, interesting18:38
kanzureif you are in a continous flow situation, how do you physically move an individual bead to lead it down the path to, say, sequencing18:38
azonenbergBy raman spectroscopy?18:38
kanzurewell, that's one way, but i haven't seen a paper doing that yet18:38
kanzurethe most basic version of all of this is don't use DLP/DMD, do only one oligonucleotide synthesis at a time on a single solid support bead, don't do anything parallel and hope for the best..18:39
kanzurebut this loses the advantages of microscopic features18:39
kanzurei mean why not just pick up a syringe and do it, at that point ;)18:39
kanzuresyringe method: http://diyhpl.us/~bryan/papers2/DNA/Syringe%20method%20for%20stepwise%20chemical%20synthesis%20of%20oligonucleotides.pdf18:39
azonenbergHmm, i'm not familiar enough with the techniques to suggest much at this point lol18:40
kanzureyeh i just need to rant to someone heh18:40
kanzurei mean, i guess it's not too disastrous if i have an unmonitored collection of a million beads simultaneously (but if it's unmonitored, and under continous flow, i don't need the photochemistry and all the sequences would (hopefully) be the same)18:41
kanzureand then hope that at least one of the beads has at least one strand of DNA that ends up being correct18:43
azonenbergHow do you find it'18:44
azonenbergand separate it18:44
kanzurewhether or not "finding it and separating it" is a problem depends on which method you're using18:46
kanzurefor instance, if you are under continous flow where all beads get the same reagants/wash simultaneously, you hope for the best that all beads are synthesizing/growing the same sequence18:47
kanzurewhen you force the beads out of the system, you wouldn't care which one is which18:47
kanzuresince they all theoretically are the same18:47
azonenbergBut in practice?18:48
kanzurein practice- in that method- there are definitely going to be certain strands that are wrong18:48
kanzureoligonucleotide chemistry has error rates as bad as 1 in 250 bp wrong18:48
kanzure(enzymes like dna polymerase make 1 error in 10,000...)18:48
kanzureso what if you're synthesizing a 800 bp strand.. and the 750th bp is wrong. you're going to be pissed off.18:49
azonenbergLol, yeah18:49
azonenbergCan you make say four 200bp strands separately18:49
azonenbergverify each18:49
azonenbergthen concatenate?18:50
azonenbergkinda like Xilinx does with their virtex 7 chips18:50
kanzureyes you can concatenate18:50
kanzurethis is called ligation18:50
kanzureverification is its own can of worms18:51
azonenbergIs it worth doing to boost yieldss?18:51
azonenbergAnd the way i'd verify would be18:51
kanzureyou can verify by dna sequencing on a microfluidic chip ;)18:51
azonenbergto PCR each of the test strands18:51
azonenbergthen sequence some of it18:51
kanzurepcr on a chip is really amazing.. my favorite method is LED-assisted heating of a droplet of water. it's like 20x faster than lab equipment since the volume is so small18:52
azonenbergvery fast thermal cycling?18:52
kanzureso, dna sequencing needs as much optimizing as dna synthesis, so it basically doubles the complexity of the project18:54
kanzurebut dna sequencing is important to me so i'm willing to overlook that18:54
azonenbergi see18:55
azonenbergYeah, i've never done full sequencing18:55
azonenbergonly simple fingerprinting by gel electrophoresis18:55
kanzureyou have to select what you are going to sequence. in a continous flow situation with DLP, how would you move beads out to individually test (or rather- how would you send the "detach from bead" chemistry to only that bead)18:56
azonenberglooking for presence or absence of one or two SNPs18:56
kanzureif i can't come up with a "access a specific bead under continuous flow" solution, then the DLP is more or less useless :P18:56
azonenbergHmm, i see18:56
kanzure"just put 1000s of valves in your circuit and it'll be fine"18:57
kanzureone possibility is to use droplets instead (water-in-oil).. a set of beads per droplet. store each droplet at an xy location. the wash step would involve.. uh.18:58
kanzurei guess a droplet wash step would involve (1) a clean drop of water + (2) wash chemistry + (3) a way to dilute the bead/droplet's water enough times over?18:59
kanzuremy connection sucks today.19:28
kanzurei think i've found a good solution19:28
kanzurei'll store 100~ beads per droplet. water-in-oil microfluidics.19:29
kanzurestore each droplet in a register specific to the next reaction that it needs to have. i.e., store all the "give this a G" together.19:29
kanzuremove the droplets out of the register into a continuous flow channel. space them out by distance, so that 100 beads from droplet 1 are physically separated from 100 beads from droplet 219:30
kanzuremagnetize the beads, do continuous flow microfluidics to do the synthesis cycle/steps, waste water output19:30
kanzureturn off waste; redirect flow to a feature that re-forms a droplet based on timing, so that the beads from the original droplets end up in separate droplets again19:31
kanzuredepending on the length of the reaction channel maybe you'll react 100s of different sequences simultaneously19:31
kanzurethen you cycle between doing A, C, T and G reactions in the continuous flow "reaction channel" to build up your final sequence, based on what you have loaded into the reaction channel at the moment19:32
azonenbergSo each of those 100 beads are considered equivalent?19:34
kanzureyeah; there can be any number (but hopefully at least one) growing oligo per bead19:35
kanzuremaybe there's 10,000 strands of dna growing on a single bead19:36
kanzurebut adding beads is nice for redundancy's sake19:36
azonenbergBut you cant grab one bead separately from the others19:37
azonenbergonly drop from drop19:37
kanzureyes, because of the spacing/timing when the other beads were released into the reaction channel19:39
kanzurei.e. release first droplet, let the droplet break up and go downstream for 10-20 ms (or whatever the bead formation time constant is...)19:39
kanzure*droplet formation19:39
azonenbergYeah, makes sense19:40
kanzurein this scenario, you hope that each bead within the same droplet is growing the same oligo19:40
kanzureso if you want to sequence a droplet, you take just one bead out and go pcr it and do other things19:40
azonenbergAnd then keep the other beads going19:41
azonenberghow big are these beads?19:41
azonenbergtens of microns?19:41
azonenbergsingle microns?19:41
kanzuretons of companies sell tons of different beads19:42
kanzurelook up polystyrene beads on google scholar19:42
kanzurebut yes i've seen 10 micron beads, definitely19:42
kanzureinterestingly enough there are microfluidic devices capable of manufacturing polystyrene beads as well...19:43
azonenbergwhat i meant was, more, what is a typical size19:44
azonenbergor size range19:44
azonenbergand are they porous or pretty solid19:44
kanzuredunno. 1 micron beads exist too.19:44
azonenbergAnd the DNA just sticks on the surface?19:44
kanzurethere's some chemistry for that19:46
kanzurei'm really bad at magnetism, i don't know if a magnetic microbead could be kept in place with an electromagnet directly under it, or if the flow forces would sweep it away19:48
azonenbergDepends on the conditions, there's no definite yes or no19:49
azonenbergdepends on cross section, flow velocity, amoutn of magnetic mateiral in the bead, strength of the electromagnet's field19:49
kanzureflow velocity can be varied by me, so that's not a relevant variable19:50
kanzureactually, if the overall flow velocity is low enough, i bet a bead couldn't be moved19:50
azonenbergWell, in that case i'd say it can definitely be done19:50
azonenbergthe question is how strong a magnet will you need19:50
azonenbergand how magnetic you can make the bead19:50
kanzurethey wouldn't have to be magnetic if i'm ok with very low velocities19:51
kanzureand then increase the flow rate when i want to move them19:51
kanzureah but that would make the beads closer to the "top" move first, sadly19:51
kanzurehi gene_hacker19:54
kanzuregene_hacker: azonenberg is around now, you had some questions..19:54
gene_hackeris azonenberg here right now?20:03
azonenberglol, missed him again21:42
--- Wed Feb 1 201200:00

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