#homecmos IRC log for Monday, 2011-08-29

swkhanand thanks for telling me about SC by RCA by the way00:00
swkhani'm so interested in how to clean stuff now00:00
azonenbergberndj: it's 3% HF in water00:00
azonenbergplus a trace amonut of denatonium benzoate00:00
azonenbergthats it00:00
azonenbergWhink brand00:00
azonenbergjeri ellsworth and christopher tarnovsky both use it in their labs00:00
azonenbergwhink rust stain remover, brown bottle00:00
berndjdoes that denatonium benzoate ever cause problems?00:01
azonenbergberndj: Not yet00:01
azonenbergjeri has gotten working CMOS circuits using it as the oxide etch00:01
azonenbergswkhan: you familiar with the details of how it works?00:01
berndjoh, bitterant00:01
azonenbergberndj: of course by that point you're drinking HF and probably have bigger problems :P00:02
smedingyeah i was thinking00:02
swkhanazonenberg: there's general cleaning. there's silica removal. there's organic contaminant removal. i'm not sure what else... oh oxide removal00:02
smedingthat seems like it's not very useful00:02
berndjhehe. no, i thought in terms of organic residue00:02
azonenbergSC1 = remove some organics and oxidize some of the surface00:02
swkhanbut that part is optional if you want to keep the SiO200:02
azonenbergsc2=  remove native oxide including metallic contaminants00:02
azonenbergsc2 = metallic contaminants00:02
swkhanthere's sc3 too right?00:03
azonenbergthe process i'm familiar with is three steps but only the first and last are numbered00:03
swkhani see00:03
azonenbergThe exact parameters i use are00:03
azonenberg1 part household ammonia : 6 parts 3% H2O2, heated (would have to check lab notes for the temp), 10 mins00:03
azonenberg(acetone rinse and water soak before that)00:04
azonenberg3% HF00:04
azonenbergfor around 45 sec00:04
azonenbergthen 1 part conc. HCl : 6 parts 3% H2O2, heated, 10 mins00:04
swkhanwe currently have a process... that i can't justify but i was taught to learn at a time i didn't ask any questions of my mentors00:04
azonenbergdistilled water rinse after each step00:04
azonenbergI usually dont do the full RCA clean though00:04
swkhanit goes DI water, kim-wipe dry, methanol, kim-wipe dry, isopropanol, kim-wipe dry, acetone, kim-wipe dry, DI water, kim-wipe dry00:05
azonenbergsince i am not doing CMOS and thus dont care about trace metals00:05
azonenbergMy stripped down clean is distilled water, blow dry, acetone, blow dry00:05
azonenbergsometimes isopropanol too depending on what i am trying t oremove00:05
azonenbergThis kind of clean is meant to remove organics etc but not trace metals00:06
azonenbergwhich for your application are presumably not problematic either00:06
azonenbergre HF, this is what i use http://www.amazon.com/gp/product/B001229J7K/ref=pd_lpo_k2_dp_sr_2?pf_rd_p=486539851&pf_rd_s=lpo-top-stripe-1&pf_rd_t=201&pf_rd_i=B000LNQNM0&pf_rd_m=ATVPDKIKX0DER&pf_rd_r=1G3VN2CAMVNGV5K9H69G00:06
azonenbergno i am not kidding00:06
swkhanwhy do you need all three? what do they each go after?00:07
azonenbergWater is for salts etc00:07
azonenbergacetone for organics00:07
swkhanwater for salts... makes sense00:08
azonenbergand if you are trying to remove a significant amount of contamination, IPA helps to remove the contaminated acetone00:08
azonenbergthen you blow dry so it doesnt leave streaks00:08
azonenbergyou have to be careful with wipes to avoid scratching it00:08
azonenbergespecially outside a cleanroom00:08
azonenbergsometimes i use microfiber swabs though00:08
swkhanwith a blow dryer wouldn't you need a filter or something?00:09
azonenbergor wipes00:09
azonenbergNot a blow dryer00:09
azonenbergcompressed duster spray, ideally cleanroom filtered00:09
azonenbergthis is basically dust-off except filterered for any >200nm particles00:09
azonenbergprice listed is for a case00:09
azonenbergi use something similar for drying stuff00:10
azonenbergYou want high pressure00:10
azonenberga blow dryer is trying to *evaporate* stuff00:10
azonenbergwith this, the goal is to physically blast the droplets off the surface00:10
azonenbergif it dries on the sample any residual contaminants will be deposited00:10
azonenbergbut if you knock them off while liquid they stay in the solvent00:10
azonenbergmake sense now?00:23
B0101hi azonenberg01:14
B0101I did found a place where they sell wafers, but I think it is mostly asia shipping only, may i add it to the wiki?01:17
azonenbergThe more the merrier01:30
azonenbergFor all you know we'll be getting an asian guy in here soon who wants to know where to buy silicon01:30
azonenbergif you want to add any resources whatsoever, go for it01:32
azonenbergswkhan: how did your plating experiments go?01:34
swkhani am still in the "measure twice" phase of "measure twice cut once" =(01:40
azonenbergswkhan: Well, at least you didnt jump into anything01:41
swkhani came across a good paper that uses copper sulfate (which i can make or buy for cheap), boric acid (apparently i had it in the lab), and hf (don't have... could maybe buy but i want to take the hf safety class first)01:41
azonenbergYeah, especially if you are dealing with concentrated01:41
azonenbergits not something to treat lightly01:41
swkhani don't know what the hf is for and i'm not sure if i can substitute it with sulfuric acid or phosphoric acid01:41
swkhanif i need a weak acid then probably phosphoric acid...01:42
azonenbergMy guess is that its for stripping oxide, in which case you cannot substitute01:42
swkhanhmm, they have two solutions: (a) 0.013 M CuSO4 + 0.50 M Na2SO4 and (b) 0.013 M CuSO4, 0.186 M HF, 0.5 M H3BO301:43
swkhani want to use Na2SO4... i wanted to see if i could make it first but again... everything is so dangerous =\01:43
swkhani was thinking NaCl + H2SO4 -> Na2SO4 + HCl01:43
swkhanwhich ultimately happens if provided enough heat i think01:44
azonenbergSemiconductor stuff has quite a few nasties, thats one of the hardest problems to solve in a hobbyist friendly process01:44
azonenbergtrying to eliminate CVD etc01:44
swkhanbut i don't know what i'd do with HCl afterwards01:44
swkhani see01:44
swkhanmetal silicides are still something i'm really interested in01:44
azonenbergI couldnt get around HF but i at least dont need concentrated, just realyl dilute01:44
swkhancould you use HCl?01:44
swkhanHF is a weak acid right?01:44
azonenbergswkhan: yes, it's weak01:45
azonenbergthe H+ isnt what's active in most reactinos you'd use it for01:45
swkhani'll stop teasing you and give you the paper already =p01:45
azonenbergit's the F-01:45
azonenbergSo rather than replacing it with another weak acid you might want to consider an alternative source of fluoride ions01:45
azonenbergpreferably something a little safer than, idk, ClF3 ;)01:45
azonenbergidk if you've heard of it but, well, let me put it this way01:46
azonenbergasbestos and glass ignite on contact with it :P01:46
swkhanignite?! SiO4 + 4ClF3 -> SiCl4 + ... hmm01:46
azonenberg"In an industrial accident, a spill of 900 kg of chlorine trifluoride  burned itself through 30 cm of concrete and 90 cm of gravel beneath"01:46
azonenberg"It is, of course, extremely toxic, but that's the least of the problem. It is hypergolic  with every known fuel, and so rapidly hypergolic that no ignition delay  has ever been measured. It is also hypergolic with such things as  cloth, wood, and test engineers, not to mention asbestos, sand, and  water  with which it reacts explosively"01:47
swkhanwow, that's very useful01:47
azonenberg"hypergolic with test engineers" lol01:47
swkhanwe had lots of problems with contamination and undesirable deposition in our flow lines01:47
swkhani'll see if i can ask my advisor if we can use it at some point01:48
swkhanIn the semiconductor industry, chlorine trifluoride is used to clean chemical vapour deposition chambers.[14] It has the advantage that it can be used to remove semiconductor material from the chamber walls without having to dismantle the chamber.[14] Unlike most of the alternative chemicals used in this role, it does not need to be activated by the use of plasma since the heat of the chamber is enough to make it decompose and react with the semicondu01:48
swkhanlol hypergolic with test engineers. aww poor test engineers01:48
azonenbergLook into alternatives first01:48
azonenbergit's useful but so extremely reactive you dont want to keep it around without a good reason01:48
azonenbergif you thought chlorine or fluorine alone was bad, this stuff is worse01:49
swkhani'm still leaning towards the CuSO4 + Na2SO4 solution01:49
swkhanbut i don't like the electrode they used01:50
azonenbergWhat was it?01:50
azonenbergThats a pretty common plating bath iirc01:50
swkhani am okay with this part... i think01:50
swkhan"technical grade (100) oriented Si wafers n doped for a resistivity of 3 to 7 ohmcm, and electrical contact to each substrate was made through a GaIn or GaAl back contact." << don't like that last part about GaIn and GaAl01:51
swkhan"an adhesive tape was used to mask off all the substrate except for the area on which deposition was desired" << love that part01:51
swkhanthey clean the substrate with 10% HF then put into the electrodeposition cell01:51
swkhanthey had a 3 electrode cell << also something i don't like too much but it seems like a lot of people used it to make for more uniform films01:52
azonenbergI wouldnt use adhesive though01:52
azonenbergif you want to do a lift-off process use photoresist01:52
azonenbergThat will give you features as fine as you want01:52
swkhan"potentials were measured against a saturated calomel electrode (SCE) which was placed as close as possible to the Si surface to minimize the ohmic potential drop in the electrolyte01:52
swkhanhate that part01:53
swkhanphotoresist doesn't like being etched away in strong acids even?01:53
azonenbergyou can do a contact mask on transparency film with a desktop printer01:53
swkhaner.. i guess i don't have strong acids here01:53
azonenbergSure you do - HCl + H2O2?01:53
azonenbergAnd i use my photoresist as a mask for etching in SC2 on a routine basis01:53
azonenbergthats how i do printed circuit boards01:53
azonenberg30 microns copper on fiberglass-epoxy composite, coat in photoresist, expose, develop01:54
swkhanthe paper doesn't use hydrochloric acid or hydrogen peroxide01:54
azonenbergthen etch in HCl:H2O201:54
swkhani see01:54
azonenbergMy point is, the resist is unreactive in acids01:54
swkhani see01:54
azonenbergit will be eaten quickly in strong bases01:54
swkhancheck out saturated calomel electrode btw. it's basically mercury chloride if i remember right01:54
azonenbergand the stuff i use is permeable to fluoride01:54
azonenbergas in HF does not damage the polymer but F- diffuses through it and will attack the underlying surface01:54
swkhanhuh interesting01:55
B0101azonenberg: wait there's a policy. minimum order 1000!01:55
azonenbergB0101: Lol01:55
azonenbergYeah, thats a problem with a lot of the nice places01:55
swkhanwell remember there's 2 electrolytic solutions they used separately01:55
swkhan(a) 0.013 M CuSO4 + 0.50 M Na2SO4 and (b) 0.013 M CuSO4, 0.186 M HF, 0.5 M H3BO301:55
swkhani don't have HF or Na2SO4, but given that.. which one would you go for?01:55
swkhani like the fact that the former doesn't require as much stuff01:55
azonenbergYeah, and specifically doesnt require HF01:56
swkhani like this paper because they are a little bit more upfront about what they did!01:56
azonenbergThat alone is a big point in its favor01:56
azonenberg(no hf)01:56
swkhanthey also did nickel and cobalt plating in this one btw01:56
azonenbergElectroless Cr is likely to be a really good option for you btw01:57
swkhanand multilayers using all of those01:57
azonenbergsince Cr has super good adhesion01:57
azonenbergto glasses etc01:57
B0101azonenberg: here, you cannot even get basic acids like HCL (unless you're like me who posses a license)01:57
azonenbergB0101: what kind of license?01:57
swkhanto spill01:57
B0101Chemical license01:57
azonenbergi mean, whats involved in getting one over there?01:57
azonenberghere the most the supplier would do is mark the package as "adult signature required"01:58
azonenbergexcept for really nasty stuff01:58
azonenbergbut you can buy dilute HF in grocery stores01:58
B0101disposal techniques must be told to authorities01:58
azonenbergand HCl online as an ORM-D consumer commodity (special class midway between normal and hazmat)01:58
azonenbergwhat i meant is, is it difficult to get one?01:58
swkhanif you bake off the water, would you get more concentrated HF or does HF boil away with the H2O?01:58
azonenbergor can any hobbyist who's interested get one?01:58
B0101and you must be working in a lab, or company which uses them and its is very difficult to get01:59
azonenbergswkhan: heating hf = hf vapor01:59
azonenbergAnd you do not want that lol01:59
B0101azonenberg: and yeah, hobbyists cannot buy it01:59
azonenbergWhere is this again?01:59
swkhanHydrogen fluoride is a highly dangerous gas, forming corrosive and penetrating hydrofluoric acid upon contact with tissue. The gas can also cause blindness by rapid destruction of the corneas.02:00
azonenbergswkhan: Yeah, sounds right02:00
azonenbergthe rule in the cleanroom here is that if you see a spill, you assume the worst02:00
azonenbergnamely concentrated HF02:00
swkhantook a lifetime to build up my pretty little eyes and it only takes it a little while to ruin it =(02:00
azonenbergpunch the spill alarm and evacuate02:00
swkhani wish i got training like that02:00
azonenbergthe only exception is if you're the one that spilled it02:00
swkhanmy safety training was stuff everyone mocked02:00
azonenbergand you know its something else02:00
swkhanBoiling point 02:01
swkhan19.5 °C, 293 K, 67 °F02:01
azonenbergYeah, normally hf is used as a solution in water02:01
azonenbergnobody uses it pure02:01
B0101Azonenberg: I put to you this in a simpler way: here in SG. you cannot buy a license unless you're working for a lab or company who uses chemicals. and... hobbyists cannot buy it02:01
azonenbergEven my 3% HF i'm careful with (use no more than 1CC at a time, gloves, lab coat, goggles, and depending on the conditions sometimes a face shield)02:02
azonenbergand if i'm heating it it's in a sealed container to lock in any vapors02:02
swkhanoh singapore02:02
B0101Azonenberg: sorry if i got you confused. earlier02:02
swkhanhere's some gum B0101 =)02:02
swkhanwhat sort of gloves do you use?02:03
swkhanwe have nitrile gloves02:03
azonenbergswkhan: I use thin nitrile most of the time since i'm not handling a significant amount02:03
swkhan4 mil? i think that's what we have02:03
azonenbergand while it would diffuse through, the 3% would take a few seconds at least to penetrate02:03
azonenbergwhich is enough time to peel it off :P02:03
B0101swkhan: its banned, and even if it isn't, i don't eat gum02:04
azonenbergI do have norfoil as well, which i'd use if i felt the risk of a spill was significant02:04
swkhanit's banned?! i knew you weren't supposed to spit it on the ground or you'd get in trouble.02:04
azonenbergcheck out the chemical resistance charts for this stuff, its amazing02:04
B0101Swkhan: they banned it as a idiot stuck the gum to the door of a train02:04
azonenbergswkhan: http://www.labsafety.com/silvershield4h-protective-wear_24530227/02:04
B0101Swkhan: from the inside. causing the sensors to act up02:05
azonenbergthey're a little slippery so i usually double glove with nitrile02:05
azonenbergthe nitrile is just there to make the outside a bit less slippery so i can grab things easily02:05
azonenbergIt also helps reduce the risk of a scratch tearing the norfoil02:06
swkhanoh wow! i could still make this02:06
swkhanNaOH and H2S04 -> Na2SO4 + 2H2O02:07
swkhani love knowing chemistry now. i don't feel crippled02:07
swkhanbut i still do with my lack of physics knowledge02:07
azonenbergand is the bath used for electroplating then?02:07
swkhanthen i have sodium sulfate. i need to make copper sulfate or at least buy it02:08
swkhani'm going to surprise my labmates tomorrow =)02:08
azonenbergIf you are going to synthesize it, dont use copper tape02:08
azonenberguse something cheap02:08
swkhani have copper wire too02:08
swkhanand we have high purity copper as well. i was planning on using it as an electrode02:08
azonenbergUse a bit of that, then02:08
swkhanbut i want to understand the non-solution part of setup02:09
swkhani don't get how they made the electrical contacts all work out02:09
swkhani need to have a procedure so tomorrow i can focus on being safe02:09
azonenbergCant help you there, chem (especially electrochemistry) is one of my big weak points02:09
azonenbergi'm already a comp sci studying EE in his spare time, adding chem to that would take more time than i have :p02:10
B0101HCl + NaOH = H2O + NaCl... quite a good way to "dispose" HCl02:12
azonenbergB0101: Yeah, last time i had extra i needed to get rid of i did just that02:13
azonenbergtitrate to neutral pH and either drain dispose or evaporate the water02:13
azonenbergthing is, most of my acid waste now has metal ions in it02:13
azonenbergso its a lot more complex to get dispose of02:13
azonenberglike i said earlier the easy solutoin is to send it out for professional disposal02:14
B0101Wow, people are asking me if they could make HCl from salt and water02:15
azonenbergI think you can do it from acetic acid and salt though02:15
azonenbergturns into HCl and sodium acetate02:16
berndjswkhan, anecdote: in a telescope making list i saw a guy once claim that he basically etched the mirror glass with HF that he derived from HCl + fluorspar (CaF2)02:24
azonenbergberndj: Yeah, that's doable02:24
azonenbergi do not recommend attempting it02:25
azonenberga) purity is poor02:25
azonenbergb) worse yet, it's uncontrolled02:25
azonenbergyou are very likely to be emitting a lot of HF vapor02:25
azonenbergthe one guy i know who attempted the synthesis in a tray outside ended up retreating inside and closing the door when he saw how fast the reaction was proceeding02:25
azonenbergit ended up eating a hole in the patio02:25
Action: B0101 is away: "Lunch"02:26
berndjB0101, don't people have swimming pools where you are? how can you not be able to get HCl??02:26
azonenbergyeah - i get my hcl from a biodiesel supplier and it's a little better purity than pool acid but is the same concentration02:27
azonenbergjust better grade02:27
berndj<azonenberg> the only exception is if you're the one that spilled it // and you know its something else  "don't worry guys it's just ClF3!"02:29
azonenbergStill an exception02:29
azonenbergIn that case, you do the same thing - except you run out a little bit faster than if it was "just" HF02:29
berndjazonenberg, when double-gloving, do the layers glide over each other easily?02:31
berndjdon't want them to chafe!02:31
azonenbergDepends on the materials02:31
azonenbergNorfoil is very slippery02:31
azonenbergso nitrile fits over it smoothly02:31
azonenbergI also use XL for the outside layer over L intner02:32
azonenbergThe idea is that the norfoil is your main protective layer and the nitrile is there both for abrasion protection and improved grip02:32
berndjhow do you recognize nitrile if it isn't labelled?02:36
azonenbergIt's *usually* blue or purple but not always02:36
azonenbergi have white (undyed) nitrile gloves that look a lot like latex02:36
azonenbergtelling the two apart without chemical testing is tricky02:37
azonenbergthe best way is to check the datasheet02:37
azonenbergYou'd want to have that anyway for chemical resistance charts etc02:37
reportingsjrand remember kids, never double bag latex05:01
Action: B0101 is back (gone 02:37:24)05:03
B0101berndj: they don't use HCl05:19
B0101they use Sodium Hypochlorite05:21
azonenbergHmm, interesting05:23
azonenbergi dont remember the details of the reaction05:23
azonenbergi know its fluorite plus something05:23
B0101azonenberg: On the topic of chemicals with electronics, HCl + H2O2 is used as PCB etchants05:32
B0101thats what we all know05:33
azonenbergB0101: It's useful for silicon cleaning (second step in the RCA clean) as well since it removes metal contamination05:34
azonenberggood for wet etching thin films of copper and aluminum05:34
azonenbergHCl (must be without H2O2) works on chromium but you need to touch it with a piece of aluminum first to electrolytically remove the CrO2 passivation layer05:35
swkhangah the most knowledgeable person here can't help me =\05:36
swkhani wish more people were like azonenberg05:36
swkhanwilling to pick up completely different, new fun things =)05:36
azonenbergswkhan: i do too... gets lonely at times lol05:36
azonenberghard to find anyone i can have a technical conversation with that doesn't get lost the second i cross boundaries from one discipline to another05:37
azonenbergheck, half the EE majors here know less about lithography than i do05:37
azonenbergdifficult to find someone i can go to for advice :(05:37
B0101azonenberg: I'm sure there's someone who knows a lot about lithography05:38
azonenbergSure, the guy who runs the stepper in the cleanroom here05:38
azonenbergBut he's too busy to talk to people whose grants aren't paying for the equipment :(05:39
azonenbergi'm talking other students etc05:39
azonenbergits like an asymptotic function05:39
azonenbergas a baby, anyone can teach you what you want to know - how to talk, etc05:40
azonenbergthe more you learn, the less people there are you can learn from in your field of interest05:40
azonenbergThe subset of those people that you know exist is even smaller05:40
azonenbergAnd once you get to the graduate level doing original research, you're now one of them05:40
azonenbergnobody else knows any more than you do about your chosen field05:41
azonenbergin that one narrow area05:41
azonenbergSo they can offer advice but not give you the answer05:41
azonenbergI'm obviously not there in terms of litho, but the point stands - its getting harder to find people who can help lol05:41
azonenbergEspecially considering that as a hobbyist with no budget i'm restricted to those who will help at no cost05:44
azonenbergrather than those who work at a fab and won't look at you if you arent a customer :P05:44
West0nSo for homebrew litho, where do you get you base material?05:44
West0n(The silicon boules)05:44
West0nOr polished cuts05:44
azonenbergWafers, actually - i buy them sliced and polished05:44
azonenbergand MTI (mtixtl.com)05:44
azonenbergthere's a link on the wiki05:44
swkhanoh i didn't realize if i use a copper anode the anode will dissolve and replace the copper ions in the solution proportionally05:45
azonenbergswkhan: Yep, thats how electroplating usually works05:45
azonenbergThats actually how my probe tip sharpening technique works05:45
azonenbergpretty standard, there are a few dozen papers around05:46
swkhani want to try just this: http://en.wikipedia.org/wiki/Electroplating with no platinum electrode or whatever05:46
azonenbergyou're electroplating the tungsten wire onto the electrode05:46
azonenbergand as a result the material dissolves and it forms a point05:46
swkhanoh cool05:46
azonenberg"reproducible electrochemical etching of tungsten probe tips", nano letters 200205:46
azonenbergis the one i designed my process off of05:46
azonenbergwith some modifications for the equipment i had on hand05:46
B0101ooooooh, "purchase wafers for just $0.99, no minimum order amount" this seems good05:48
azonenberg$0.99? Doesnt sound possible05:48
azonenberggotta be unpolished05:48
swkhanhave you ever tried to make your own wafers?05:48
azonenbergi've gotten unpolished 4-inch for $2 each so half that is reasonable for 2-inch05:48
azonenbergswkhan: No, i dont have the apparatus for that05:48
azonenbergno high-temperature stuff05:48
azonenbergi cant even do thermal oxidation in fact05:48
azonenbergmy hottest oven goes up to 225C05:49
azonenbergi need to hit around 800 to get an appreciable oxidation rate05:49
azonenbergand until i have $1.2K to spare i wont be getting a furnace :P05:49
swkhancould you make a joule heater cheaply?05:49
swkhanand safely i guess05:49
West0nHow much do polished wafers usualy cost?05:49
azonenbergIn theory, but it gets more difficult05:49
azonenbergfor example i'd need to be working under inert gas to avoid oxidation05:50
azonenbergWest0n: Depends on size05:50
azonenbergwe talking a 2" or an 18"?05:50
azonenbergone or two sides polished?05:50
azonenbergcoinroll, test grade, prime grade?05:50
B0101azonenberg: yeah, unpolished wafers05:50
West0n2" one side polished05:50
azonenbergAs a rough example, http://www.mtixtl.com/siwafer1004diax05mm1spptypebdopedresistivities7-13ohm-cm-1-3-1.aspx05:50
azonenberg2" i have found to not be a great deal price wise05:51
azonenbergthis one is <110>, other was <100>05:51
azonenberg3/4 the price for around 1/4 the area05:51
West0nThat's not as bad as I was expecting05:52
azonenbergRight now my inventory is one untouched 4-inch <100>05:52
azonenbergtwo <110>, one single side and one double side polished05:52
azonenbergthat had some films deposited on them and are gradually being broken up into dies and being experimented with05:53
azonenbergand two <100>05:53
azonenbergall but the first are 2-inch05:53
B0101azonenberg: have you heard of homemade developer solutions?05:53
azonenberg1% w/v technical grade NaOH crystals in distilled water05:54
azonenbergworks fine for most positive photoresists05:54
azonenbergthats what i use05:54
West0nSo what are you all making?05:54
azonenbergi bought one 2-pound jar of this, used maybe a few dozen grams05:55
azonenbergcomes with certificate of analysis etc05:55
azonenbergnot SUPER pure but you know what you're getting05:55
azonenbergWest0n: Working toward a comb drive05:55
azonenberggradually improving my lithography process05:56
azonenbergtrying to improve yields (and reduce overetch) on my 5um process http://colossus.cs.rpi.edu/~azonenberg/images/homecmos/2011-08-26/5h10_004_annotated.jpg05:56
West0nWhat's your genral setup now?05:56
B0101WestOn: making etchings of art on wafers (after Azonenberg's Nyan cat experiment)05:56
azonenbergnote the broken lins and heavy pitting05:56
West0nWhat's the setup you are using?05:56
azonenbergthe second method described in the paper (camera port)05:57
West0nI am guess its a bit more advanced then normal pcb photo exposure setup?05:57
azonenbergvery similar05:57
B0101sodium carbonate as developer?05:57
azonenbergbut its projection rather than contact05:57
azonenbergB0101: I've heard it works too05:57
azonenbergNever used anything but NaOH myself05:57
West0nContact for pcb etch?05:58
azonenbergYou may have to fool around a bit to find the right concentration05:58
azonenbergWest0n: Most people doing PCB stuff use contact litho05:58
azonenbergwhich by definition is 1;1 scale05:58
azonenbergI use projection litho05:58
azonenbergmask is placed behind a lens and projected onto the wafer/die05:58
azonenbergdepending on the objective you can get different levels of reduction05:58
azonenbergthe image i just pasted was designed and printed at 600DPI on a laser printer, lines are 5 pixels wide05:58
azonenberg(thats the smallest that resolves clearly)05:59
azonenbergThat forms ~200um wide lines on the transparency05:59
B0101azonenberg: if that is true, then i will curse myself for buying positive developer solution for $605:59
azonenbergit's then reduced optically by 10x (which gives pretty good results) or 40x (still trying to improve yields, havent gotten anythign to work)05:59
azonenbergor whatever else your objective is05:59
azonenbergI have a 100x immersion objective but i dont want to try immersion lithography until i have good yields on the larger processes05:59
azonenbergthe immersion objective could in theory give me 2um design rules06:00
azonenbergWhich i could then shrink to 1 by use of double patterning06:00
azonenbergAnd if i switched my 600DPI printer out for a 1200, i could theoretically hit 500nm06:01
azonenbergi know my optics are capable of it, i can see 500nm features using the same objective06:01
azonenbergThat's at least a year out and i expect it will start being impacted by dust etc at that point (if not before)06:02
azonenbergWest0n: Did you see the pic i pasted?06:02
azonenbergOh, and the bigger problem is that the aspect ratio i can hit with KOH is such that i can't make deep features of that size06:03
azonenbergAnd i dont have DRIE capabilities for obvious reasons06:03
azonenbergRealistically i think most of my MEMS work will be done at significantly larger sizes - 5-20 um design rules06:04
azonenbergThe main reason i'm continuing to shrink rather than focusing on improving yields is that i enjoy the challenge :P06:04
azonenbergRealistically i have no idea what i would do with a 500nm fab process lol06:09
azonenbergAt least not until i get the budget for dopants and all of the trimmings and trappings i'd use for CMOS06:10
azonenbergAnd yields will likely be poor as that's really pushing the limits of what i could do with the setup06:10
azonenberg2um, without double patterning or the high DPI printer, is still going to be tough because i'd need to find an immersion medium that is compatible with the process and can be cleaned off06:10
azonenbergfor example most oils are out06:11
azonenbergwater might work but i'd need a water compatible objective06:11
azonenbergAnd in any case the first step is to work on improving the 5um process to the point that i get workable yields06:12
azonenbergAnybody here good with cross section microscopy sample prep? :P15:51
azonenbergI started out having 1500 grit sandpaper and 60nm colloidal silica15:52
azonenbergjust ordered a selection of diamond polishing compounds from ted pella that are in the 6um to 250nm range so hopefully that will be enough15:52
swkhansandpaper works?15:52
swkhanwouldn't that give you a lot of silica dust?15:53
swkhani don't have any experience... i'm just wondering aloud15:53
azonenbergi actually started out with a really coarse abrasive, like 300 grit sandpaper15:53
azonenbergYes, it makes dust15:53
azonenbergbut you can clean it off15:53
azonenbergThat was to grind halfway throuhg the sample15:53
azonenbergso i'd get to the rough area i wanted15:53
azonenbergsince i didnt have a suitable wire saw on hand15:53
azonenbergthen 1500 grit to get a smooth but not polished surface15:53
swkhanso what do you want to see? thickness?15:53
swkhanof each layer*?15:53
azonenbergthen i realized i needed something finer than that but coarser than the 60nm15:54
azonenbergso i got the diamond stuff which hasnt arrived yet15:54
azonenbergThickness and slopes15:54
azonenbergi want to see how steep the vertical edges of an etched layer are15:54
azonenbergmeasure undercut15:54
swkhani wonder if your abrasive would affect the slopes15:54
bart416Mhhh azonenberg is it something round?15:54
azonenbergThe abrasive? I'm not sure15:54
bart416No I meant what you're trying to cut15:54
azonenbergthe colloidal silica almost certainly is15:54
bart416If it's a round shape, use a lathe15:54
azonenbergno, its a silicon die lol15:55
azonenbergrectangular and being cross-sectioned along the thin axis15:55
azonenbergAnd i need it polished down to submicron15:55
bart416yeah, you'd need an electron beam cutter :P15:55
azonenbergideally, i want to be able to see the 5nm thick Cr adhesion layer :P15:55
azonenbergbut thats not likely15:55
azonenbergwhat i do want is to be able to see the 200-1000nm of copper, the ~50nm of tantalum oxide, and the few microns of photoresist15:56
azonenbergalong with edge profiles for each (or at least overetch estimates)15:56
swkhanwhen would you use an e-beam cutter versus an ion-beam cutter?15:56
swkhanlike e-beam versus fib15:56
azonenbergswkhan: Not familiar with e-beam for cutting15:56
azonenbergonly for lithography exposure15:56
azonenbergor evaporation15:56
azonenbergso i dont know details of what they're good for15:56
azonenbergThe only nanoscale cutting tool i'm even somewhat familiar with is a FIB and while i know what they are in theory i've never used one myself15:57
azonenbergi got to sit next to one though :P15:57
azonenbergwhile using the SEM in the cleanroom15:57
azonenbergfib is right behind it15:57
swkhanoh cool =)15:57
azonenbergAgain, this is all part of the process characterization step15:57
swkhanwe're getting a fib soon. rie too =)15:57
azonenbergonce it have this figured out, you wont need it for actual processing15:57
azonenbergonce i*15:57
azonenbergand cool15:57
swkhani see15:57
azonenbergDRIE or just regular?15:57
swkhani think both eventually15:58
swkhanjust regular for now15:58
azonenbergand if you get to use the fib have fun... those things are amazing15:58
azonenbergre RIE, i want that kind of capability in my fab :P15:58
azonenbergbut dont expect it any time soon15:58
azonenberganisotropic silicon etch: poor man's DRIE ;)15:58
azonenbergyay KOH!15:58
swkhanlol. but you don't have to pay for your lab once it is setup16:00
swkhanwe have to essentially "rent" services16:00
swkhanwhich... we perform, so at least we save on labor16:00
swkhanbecause i'm worthless =(16:01
azonenbergYou've been around long enough you should know by know16:03
azonenbergcheap foreign labor, academic style16:03
azonenberg= grad student16:03
bart416azonenberg, doesn't even have to be a grad student16:11
bart416cheap labor, academic style = students16:11
azonenbergTrue, sometimes undergads are abused worse lol16:12
azonenbergBut they also know less and are less valuable to the faculty16:12
azonenbergso they wont necessarily make you spend as much time on the project16:12
bart416The issue I'm currently finding is that we're actually not allowed to participate in any research projects as undergraduates >_>16:12
bart416So it's hard to score your name on publications16:13
azonenbergnot allowed to? o_O16:14
bart416Maybe if you ask really really really nice16:17
bart416But not really no16:17
bart416They just employ full time researchers16:17
bart416But not even I with my near perfect grades get offered a chance to participate...16:18
azonenbergAround here you usually have to ask, but they're pretty willing16:18
azonenbergAlso, wow... looks like i was pretty lucky to have not lost power16:19
bart416Doesn't your college/university have some generators?16:20
azonenbergpower company reports 277 people w/o power just a few blocks south of me16:20
azonenbergAnd i live off campus in an apartment downtown16:20
bart416Wow, somebody in the US that lives off campus!16:20
azonenbergThe school has both battery and generator backups on some critical infrastructure16:20
azonenberglike the core servers, chillers in the biotech, fume hoods that were not scheduled to be shut down, etc16:20
azonenbergbut it doesnt cover everything16:21
azonenbergMost of their utilities are underground so their impact was minimal16:21
azonenbergas i understand there were a couple of brief glitches but no long-term outages16:21
azonenbergMy immediate area seems to have been surprisingly lightly impacted16:22
bart416You guys should really put your electric cables underground16:23
azonenbergOn campus, they are16:23
azonenbergAs are many of the distribution lines16:23
azonenbergbut it doesnt help if the tunnel gets flooded :P16:23
azonenbergthats what took out a lot of power in NYC afaik16:23
bart416I'm not talking about putting them in the underground transportation tunnels16:24
bart416I'm talking about putting them 1-3m under the street16:24
azonenbergI dont know exact details16:24
azonenbergbut also, transformer sites etc are potential weak points16:24
bart416In water tight shielding16:24
bart416And if you bring it up to surface level it first has to go up 2 meters at the very least and then as soon as possible enter a fuse box16:25
bart416So no water can get to it16:25
azonenbergyeah, makes sense16:25
azonenbergi think the problem is that there's a lot of older equipment around16:25
bart416Here their comment was "get new equipment" :P16:26
azonenberg... http://www.northjersey.com/news/crime_courts/80-year-old_without_power_shoots_transformer_with_WW_II.html16:26
azonenbergi'm sure THAT helped get his power back :P16:29
bart416But that's just a pole transformer, they have those in stock luckily16:32
azonenberghitting a substation would have been worse probably16:35
azonenbergBut still its a rather ridiculous situation16:35
bart416But since when do you expect 80 year old americans to use common sense?16:40
azonenberglol, most of the ones i know do16:40
bart416I haven't met a single american in real life that I wouldn't like to punch down to the ground after 5 hours16:43
bart416That's about the time it takes you to hear their entire life story on general, at least with tourists16:43
bart416After that they start repeating themselves or complain about how socialist Europe is16:43
azonenbergThe same, or something similar, could be said of a lot of people in general16:46
azonenbergregardless of origin16:46
Action: azonenberg cant really stand "normal people" from anywhere :P16:46
azonenbergnow that is just stupid16:51
Action: azonenberg continues hacking on FPGA16:57
bart416Honnestly, jogging pants, clearly artificial fabric tshirt considering it melted into his skin o_O16:59
bart416And then trying to saw a 20kV line, lol16:59
bart416Bah, it sucks having to wait for a 800 euro paycheck >_>17:27
bart416+ another 60 euro from another place >_>17:27
azonenbergLol i know the feeling17:29
azonenbergMy school is great at taking your money17:29
azonenbergless good at giving it out17:29
azonenberghopefully my teaching assistantship will start paying up soon because i have rent due in ~3 weeks17:29
lekernelI thought you were in US17:31
azonenberglekernel: i am but i dont think bart is17:31
swkhanazonenberg: very personal question... how much do you make in a year?17:33
swkhani will be happy if i get about 12k a year starting next year17:34
swkhani think the highest i can go is 22k17:34
azonenbergIn theory? min stipend for grad students here is 18k before taxes17:34
swkhanthat's after taxes btw17:34
azonenbergwhich is what i was offered17:34
swkhani hate how we're so low on the totem pole =(17:34
azonenbergin practice? we'll see :P17:34
azonenbergMy guess is "not much" lol17:35
bart416Here TAs get about 1800 euro / month17:35
azonenbergDepends a lot on cost of living though17:36
bart416At the start TA pays about as much as the industry here actually :S17:37
bart416But after a few years working in industry your paycheck is way way waaaaaaaaaaaaaaaaaaaaaaaaaaay higher17:38
azonenbergUm, yeah17:38
bart416So how much hours / week do you have to teach azonenberg ? :P18:11
azonenbergbart416: nominally? 2018:13
azonenbergrealistically? no idea18:13
bart416What subjects?18:13
azonenbergi will be havnig office hours twice a week for two horus18:13
azonenbergeach session will have one hour for one class and one for the other18:14
azonenbergcomputer organization / crypto 118:14
bart416Sounds boring :P18:15
bart416Did you have any choice or was it simply assigned?18:17
bart416A friend of mine is Ph.D. student/TA for chemistry18:18
bart416And the professor he was assigned to got sick (cancer)18:18
bart416He had to do his own classes + the professor's18:19
bart416+ the administration + the conventions + everything else you can think off :')18:19
bart416And his wife was pregnant at the same time >_>18:19
bart416(Mind you, he was happy with all of those things)18:22
azonenbergBoth professors requested me as their TAs18:30
azonenbergand i was ok with it18:30
azonenbergboth were among my favorite classes as undergrads18:31
azonenbergand i know the material inside out and backwards18:31
azonenbergcomporg covers x86 and mips plus low level cpu arch18:31
azonenbergi'm implementing mips in an FPGA :P18:31
bart416The name is weird :|18:31
azonenbergand know x86 decently well as well18:31
bart416Here thats called computer achitecture, lol18:31
kristianpaullm32? ;-)18:31
azonenbergcrypto, well, i'm kind of a phd student in security18:31
azonenbergso if i dont like crypto i'm in the wrong field lol18:32
bart416Thought it was boring network topology :P18:32
azonenbergkristianpaul: I'm implementing several softcores18:32
azonenbergthe simplest is a custom 8-bit 3-stage pipelined architecture18:32
azonenbergthe complex one is MIPS 118:32
azonenbergwith 5-stage pipeline18:32
swkhanazonenberg: why is this stuff your hobby and CS your bread and butter?18:32
swkhaninstead of the other way around?18:32
swkhanCS is cheaper to do on your own certainly18:33
azonenbergswkhan: I like all of it and couldnt decide exactly what i wanted to study18:33
swkhani see. same here =(18:33
azonenbergi'm not a big fan of math which meant i didnt want to take the intro EE classes that covered analog stuff18:33
azonenbergSo CS it was18:33
swkhanlol, isn't CS full of math too?18:33
azonenbergit's a different kind of math18:34
kristianpaulanalog will hunt you fure sure18:34
kristianpaultime time time :)18:34
swkhanso set theory type stuff over vector calculus / differential equations?18:34
azonenbergkristianpaul: I like to say that if a circuit i built has anything but 0 and 3.3V on it (except during clock edges), i probably messed up in the design somewhere18:34
azonenbergi've since added 5, 1.8, and 1.2 to the acceptable list18:34
azonenbergswkhan: and yes18:35
swkhanlol acceptable! you're so funny =p18:35
berndjyikes azonenberg i hope you didn't do low level org of x86!18:36
azonenbergberndj: Lol, no we didnt do too much of that18:36
berndji'd be half interested, but that would be cruelty against students18:36
azonenbergmost of the really low level stuff was mips18:36
azonenbergand having implemented a good chunk of mips 1 in an FPGA i know it at the pipeline internals level lol18:37
berndjyeah, we did that too.  patterson & henessy?18:37
berndjtimeless book18:37
azonenbergI used the 3rd edition18:37
azonenbergthe class is now on the 4th18:37
azonenbergand my mips implementation is currently on hold as it wont fit in the tiny XC3S50A i have on hand lol18:37
azonenbergSo i'm doing an 8-bit design until i finish the dev board for a larger processor lol18:38
lekernelyou could get a M1 instead :-)18:41
azonenberglekernel: XC6SLX75 looks yummy all right18:42
azonenbergbut i want to design the board and pick out the components18:42
berndjbuild a subtract-and-branch-if-negative processor. that should fit18:42
azonenbergberndj: lol, the design i'm doing is a little more complex18:42
azonenbergits heavily inspired by MIPS with just a little bit of x86 and 8-bit PIC in there18:43
azonenbergshrunk down to 16 bit instructions and an 8-bit datapath18:43
lekernelsounds challenging in a 3s50a18:43
azonenberglekernel: actually, i have everything implemented except the shifter and memory controller and it's using just over half18:43
azonenbergi was hoping i could fit two of them in a 50a but that might not happen18:44
azonenbergthis is the version with 8x registers and 4-level hardware stack18:44
azonenbergbut its parameterizable to 4-8-16 regs and 2-4-8 level stck18:44
azonenbergsame binary instruction format18:44
azonenbergright now its using 204/1408 FFs, 779/1408 LUTs, one block ram for instructions18:45
bart416<azonenberg> i'm not a big fan of math which meant i didnt want to take the intro EE classes that covered analog stuff <-- The analog part of electronics is the most fun :P18:46
swkhani agree! =D18:46
azonenbergbart416: and thats why i stick to CS and digital whenever possible :p18:46
bart416You can do horribly complicated math with some opamps, resistors and diodes, heh18:47
lekernelah, 1400 LUTs? I thought it had more like 500...18:47
azonenberglekernel: the 50a has 1408 FFS/LUTs, 704 slices, 68 IOBs, 24 BUFGMUX, 3 block RAM, 3 MULT18x1818:47
azonenbergStill hoping i can fit two copies of the area optimized (4 registers, 2 level stack) version into the 50a18:48
azonenbergthis is the general purpose design18:48
azonenbergthen the 16-reg 8-level is even bigger18:48
azonenbergoh, and this design passes static timing for 90 MHz right now18:48
azonenbergwith a 3-stage pipeline18:48
bart416swkhan, have you ever attempted to build an analog computer?18:49
swkhani made some components of one like a differentiator, a summer, an integrator, etc.18:49
azonenberglekernel: in the all-out configuration this thing is going to rival most of the 8-bit PICs i think18:49
azonenbergwaay better cpu performance18:50
swkhanwhat constitutes an analog computer? i'm missing memory i guess...18:50
swkhanpower efficiency?18:50
azonenbergnot even close18:50
swkhani see18:50
azonenbergbut in price it would actually be comparable to some of the bigger 18Fs18:50
azonenbergYou'd lose the analog peripherals of course, i dont have an ADC on the fpga :P18:51
azonenbergIts not cost effective if you are JUST running the softcore18:51
azonenbergwhat its useful for is if your design fills 3/4 of the fpga and you cram it in the other 1/418:51
azonenbergrather than putting another chip on the board18:51
bart416My swkhan most of it is fairly easy to build actually18:55
bart416Putting it together is the hard part18:56
bart416At least in a way that can be somewhat "programmed"18:56
bart416Another guy here on freenode also named Bart is working on an analog-ish computer19:00
bart416Fully analog pong and tetris, lol19:00
berndjpong i can still get, but tetris? wow19:01
bart416He doesn't have any scopes so he has to wait until college starts again19:02
bart416He's going to the same-ish place I'm going to, heh19:02
bart416Well yeah, I go to a university technically, he goes to a college19:03
Action: kristianpaul needs a PIC18 clone..19:08
Action: kristianpaul looks at azonenberg 19:08
kristianpauloh, http://opencores.org/project,ae18, same developer as aemb it seems19:09
kristianpaulazonenberg: indeed, putting chips and chips never ends..19:11
--- Tue Aug 30 201100:00

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