#homecmos IRC log for Sunday, 2011-08-28

azonenbergI think i might try immersion litho later tonight02:28
azonenbergSee if i can get even a small portion of my field to expose nicely at 2um feature size02:28
azonenbergand, if i'm feeling bold, fool with double patterning to hit 1um02:28
azonenbergThough perhaps "bold" isnt the best word to use to describe making lines *thinner* :P02:42
berndjimmersion and double patterning still seem like magic to me02:45
azonenbergDouble patterning isnt that bad02:46
azonenbergin this case the goal would be to create 1/2 sized lines at 1/2 the spacing02:46
azonenbergStart out by exposing a normal line/space mask02:46
azonenbergBut do so for half the usual time02:47
azonenbergDevelop just enough that you can see the marks02:47
azonenbergthen re-expose with the pattern 90 degrees out of phase02:47
azonenbergDevelop again02:48
azonenbergThe only spots that will get removed when you develop are those that were hit by both exposure steps02:48
azonenbergsince each on its own is too weak to fully dissolve the resist02:48
azonenbergNow you have 1/2 sized lines at the usual spacing02:48
azonenbergStrip resist and re-coat02:48
azonenbergRepeat the entire process 180 degrees out of phase02:49
azonenbergnow you have two patterns of half-width lines at normal spacing02:49
azonenbergeach shifted by 180 degrees02:49
azonenbergthe end result is half-width lines at half spacing02:49
azonenbergI might try doing this at the 20um process (which is relatively well established) and immersion litho (for 2um) separately02:50
azonenbergget each working decently well02:50
azonenbergand only then attempt combining the two to hit 1um02:50
azonenbergthen if i can get better mask fab capabilities, 500nm on at least a Poc level might be possible02:50
swkhanazonenberg: still working on copper electroplating of silicon and am reading more before i try anything stupid =p02:50
swkhani can't figure out why chromium is so good at adhesion but i ran into one paper that i really liked that i'd like to share02:51
swkhani also need a bit of advice =)02:51
swkhani really really really like this paper: www.ims-chips.de/content/pdfpublicationorig/471orig.pdf (copper electroplating for integrated rf-devices). there's only one term i don't quite get. the iv. conclusions section on page 6 says "A paddle cell has been successfully used for the electrochemical deposition of smooth, bright Cu blanket films onto 4" Si wafers. The uniformity of the film thickness can be kept within tight limits by activation of a copla02:51
azonenbergand re adhesion, h/o *finds paper*02:51
berndjoh, does double patterning just mean you can more finely control the *phase* of your spacial distribution but you still can't cheat and achieve a higher spatial *frequency*?02:51
azonenbergberndj: No02:51
azonenbergLet me make a graphic, h/o02:51
azonenbergYou can double frequency02:51
berndjsee, that seems like magic to me!02:52
azonenbergIt relies on you having better alignment precision than exposure precision02:52
azonenbergfor example, in my case i am limited by resolution of mask printing02:52
azonenbergbut can do micron or evne submicron alignment02:52
berndjoooh i think i get it - at least in your case (not diffraction limited)02:53
berndjoooh i think i get it even in diffraction limited cases02:53
berndjyou pattern every *2nd* "line" in each step, thereby staying within the diffraction-limited possibility?02:54
azonenbergEffectively, yes02:54
azonenbergRe double patterning03:01
azonenbergLet's say the smallest feature you can do on a mask is 5 um03:01
azonenbergbut you can align down to 1 um03:01
swkhanazonenberg: did you get a chance to see the paper btw?03:01
azonenbergfirst you expose the blue mask halfway03:02
azonenbergthen shift by 2.5um and expose the green03:02
azonenbergthe red area is the only area that's fully exposed03:02
azonenbergThis gives you 2.5um lines on 5um centers03:02
azonenbergetch these, strip resist, re-coat03:02
azonenbergrepeat the process to put another set of 2.5um lines on 5um centers phase-shifted 180 degrees from the existing one03:02
azonenbergthe end result is 2.5um lines on 2.5um centers with four exposures03:02
azonenbergI'm hoping to use a similar technique on my hypothetical 2um process (assuming i get immersion litho working) to shrink from 2 down to 103:03
azonenbergimmersion would let me use the highest objective on my microscope to achieve 100x reduction, taking a 200um sized mask and making 2um features03:04
azonenbergswkhan: looking now03:04
azonenbergThey say copper is resistant to KOH but i had some issues with that03:05
azonenbergi need to try again now that i have a Cr adhesion layer on03:05
azonenbergOk, here we go03:09
azonenbergThey used a sputtered Ti + TiN adhesion layer03:09
azonenbergthen a thin Cu layer to start the plating off of03:09
azonenbergand electroplated 6um of Cu on top of that03:09
azonenbergswkhan: Ok, very interesting paper but it doesnt eliminate the need for sputtering03:11
azonenbergit just means you can do thicker layers faster03:12
azonenbergyour question?03:12
swkhansorry i was helping a labmate03:12
swkhani see that they have a sputtered layer too and i'm hoping i don't really need it somehow.. =\03:13
swkhani was more interested in the section on blanket deposition on the wafer for right now03:13
azonenbergThe Ti/TiN is both adhesion and a barrier to prevent the Cu from diffusing into the Si03:13
azonenbergTi is adhesion and TiN is barrier03:13
azonenbergif you dont care about copper contaminating your silicon then you can skip the barrier layer03:14
swkhanthe conclusion and a few other sections say something about the activation of a coplanar auxiliary cathode03:14
swkhanwell i just need a metal contact layer. what happens if silicon is contaminated by copper?03:14
azonenbergSo the wafer is the cathode and the plating source is a flat anode03:14
azonenbergbut it looks like they throw an extra cathode in there somewhere else to modify the shape of the field03:15
swkhani plan to run IV measurements with the bottom contact. later i want to try patterning03:15
azonenbergreference 4 should talk more about it03:15
azonenbergAnd what are you using your silicon for?03:15
azonenberga mechanical substrate or actually as a semiconductor?03:15
azonenbergI dont know the exact effects but afaik transition metals ions in general will reduce charge carrier lifetime03:16
azonenbergWhich is generally bad for any kind of CMOS etc03:16
azonenbergAnd as long as your feature size isnt too small and you have photoresist and some common lab chemicals available, patterning is the easy part03:17
swkhani don't think i'm really using the silicon as anything over than as a mechanical layer to hold everything up03:18
swkhani don't have photoresist, but i'm willing to learn. i think it was this paper that got me really interested in it though. you can remove some photoresist with just acetone? =D03:19
swkhanyou know so much it's incredible03:19
azonenbergI use Shipley SP24 positive photoresist03:19
swkhani don't have a spin coater or anything03:19
azonenberga good source for small volume is http://injectorall.com/photoresist.htm03:19
azonenbergswkhan: Do you have an electric drill, a sanding wheel, and a 2x4?03:20
swkhani was planning on making molds out of aluminum foil03:20
azonenbergYou do now :P03:20
azonenbergThis is another good homebrew spin coater http://pubs.acs.org/doi/abs/10.1021/ed080p80603:20
swkhanwow wow wow03:20
azonenbergFor mask fab and projection apparatus, see http://colossus.cs.rpi.edu/~azonenberg/papers/litho1.pdf03:20
azonenbergi'm now using the second of the two methods described there03:21
swkhanoh right rpi guy =D03:21
azonenbergyeah, i'm at rpi03:21
swkhanwe talked about how cool rpi is03:21
swkhanand i have a friend who just moved there03:21
swkhanyou and me03:21
azonenbergyour friend a grad student?03:21
swkhanhe will be a transfer student (undergrad)03:21
swkhanbut he's 2303:21
swkhanhe's super motivated and interested in physics03:21
Action: azonenberg is 21 and a phd student in comp sci03:22
swkhanoh wow. 22 and to-be PhD student this fall03:22
swkhanin comp sci?03:22
azonenbergYes, i did my undergraduate work in CS (also at RPI)03:22
swkhanEE here, but this is so far removed from what i learned03:22
azonenbergand am now studying cs at the grad level03:22
swkhanwhat do you study / research?03:22
azonenbergi started out doing lithography with the electronics club for PCB fab03:22
azonenberggot into nanotech when i landed a consulting gig summer after freshman year that later led to an ongoing relationship with the company03:23
azonenbergi was just doing numerical modeling of their device at first lol03:23
azonenbergnext thing i knew i was helping with process development, failure analysis, and wafer test03:23
azonenbergand then it became a hobby lol03:23
azonenbergBut its unrelated to my academic work03:23
swkhanthat's so cool03:24
swkhani love people like that =D03:24
azonenbergmy thesis research is on embedded systems security with a focus on smart grid03:24
swkhani'm still trying to work around the adhesion layer issue03:24
azonenbergi suggest you not try to work around having an adhesion layer03:24
azonenbergand instead work around sputtering as the means of getting it there03:24
swkhansee this is quickly getting complicated03:25
azonenbergif you want your film to stick *at all* you need one03:25
swkhani'm doing this primarily because our sputterer is broken03:25
azonenbergand lol, of course its complicated03:25
swkhanit may be fixed now03:25
azonenbergthere is a good reason most hobbyists etc dont even consider doing microfab03:25
azonenbergonly the crazy ones like me :003:25
azonenberg;) *03:25
swkhanthe secondary reason is to learn how to come up with a process of my own03:25
swkhanwe appreciate the crazy ones =)03:25
azonenbergAnd re process development, if you are only doing this to get around the sputter coater being down, unless it's down permanently, thats the wrong reason03:27
azonenbergby the time you get the process working it will almost certainly be fixed03:27
azonenbergHowever, you may want to consider electroplating *anyway* if you need a thick layer03:27
azonenbergsince it's much faster than sputtering03:27
swkhanwell i never get a chance to learn new things like this and just experiment03:28
azonenbergYou have one now03:28
azonenbergI'm just saying03:28
azonenbergif you want to do it for educational value, go ahead03:28
swkhanthe grad student in my lab who supervises me (sort of) always encourages us to use the simplest route possible03:28
azonenbergbut unless your sputtering system is totally borked and will be for >1 month03:29
swkhanit's fine in terms of productivity, but i feel like in 4 years i won't graduate with a meaningful PhD03:29
azonenbergdont do it with the hope it'll be faster03:29
swkhanit's been down for 1 week. i was hoping i could learn enough to get to the point where i could do this on my own by the end of the week03:29
swkhanam i that predictable? =\03:29
azonenbergI got the idea already03:29
azonenbergAnyway so if you do want to put the time into developing the process, go for it03:32
azonenbergJust make sure its for the right reasons03:32
azonenbergIf the sputtering system is going to be repaired soon you may want to try implementing the exact process those guys did, or very similar03:32
azonenbergOr if you have access to an evaporator03:32
swkhanazonenberg: that's what i wanted to do: implement their process as much as possible. i was going to immediately shoot down any part of their process that i couldn't do (sputter) and consider the things i can't do right now but maybe could do with a little effort (photolithography)03:33
swkhani don't have access to an evaporator either03:33
azonenbergLitho is definitely within the reach of a dedicated individual with minimal resources03:34
azonenbergspin coating is the first step which should not be too difficult03:34
swkhanby the way, i'd like to clarify something... i was told that an evaporator can damage silicon03:34
swkhansomething about radiation damage03:34
azonenbergswkhan: um03:34
azonenbergAre you talking e-beam or filament evaporation?03:34
swkhanmaybe i should have read the nuclear chemistry chapter =\03:34
swkhane-beam definitely damages stuff03:34
swkhani don't think filament evaporation should i guess03:34
azonenbergfilament evaporation will not damage your sample for use as a mechanical substrate03:35
azonenbergas long as it doesnt mind a little heating03:35
azonenbergAlso, very interesting03:35
azonenbergi think i figured out why my etch wasnt working03:35
swkhanah i think it was e-beam evaporation that i read about that makes sense03:36
azonenbergswkhan: If you can get access to a filament evaporator that will work nicely03:36
swkhani know i can't get access to it =(03:36
azonenberghttp://www.microtechweb.com/kb/cr_etch.htm has a section on etching of Cr03:36
azonenbergwhich explains one of the problems i had03:36
azonenbergI can etch Cu and Cr but not CrO203:37
azonenbergSo i need to electrolytically remove the CrO2 while in the etch bath03:37
swkhanoh wow03:37
azonenbergWhat this means is that i need to remove a little resist from a corner of the mask03:38
azonenbergetch the copper away03:38
azonenbergthen return to the etch bath and gently tap the corner with a piece of aluminum foil03:38
azonenbergfor a few seconds03:38
azonenbergwhich will electrolytically remove the CrO203:39
azonenbergthen the ~5nm of Cr will almost immediately be eaten by the HCl03:39
swkhani read a book on chemistry and i still don't understand why silicon likes chromium so much03:41
swkhancould you sputter copper instead and skip the chromium part?03:41
swkhanit just seems weird that you always have to have chromium or titanium involved03:41
azonenbergSputtering copper will give you copper on silicon03:41
azonenbergbut it wont stick03:41
swkhanwhy not just use chromium as a contact then?03:41
azonenbergCr and Ti are very popular adhesion layers03:41
azonenbergYou could03:41
azonenbergif it has suitable properties03:41
azonenberggrr... now i want to try this Cr etch technique03:47
azonenbergbut my roommates are in the living room watching TV03:47
azonenbergi need them to get out so i can turn the lights out for litho :P03:47
azonenbergone of the many hazards associated with home fab lol03:48
azonenbergon a different note i plan to try a new exposure technique03:58
azonenbergusing a light source that is nice and even but, unfortunately, only about 1cm across03:58
azonenbergbut is bright and uniform03:58
azonenbergyou might have one on hand - halogen fiber-optic microscope illuminator03:59
azonenbergStick it on top of a few books etc until it's high enough to shine down the camera port of a trinocular scope03:59
azonenbergthen put a glass slide on the camera tube, stick the mask on top04:00
azonenbergand shine the fiber lamp down it04:00
azonenbergrough guess says 5-8 minute exposures04:00
azonenberg*wants to try it now*04:00
azonenbergi did a dummy test - projecting onto a blank die with no resist on it - a minute or two ago04:01
azonenbergbut i need them to get out of the living room so i can shut the lights off and do a live test :P04:01
azonenbergOh, and i dont know how thick the Cr layer is04:09
azonenbergWhich will definitely "help" my process :P04:09
azonenbergAnother chrome etch i see is 3 parts HCl to 1 part water04:13
azonenbergSo it looks like even though Cr and Cu seem similar at first04:13
azonenbergCu is *not* etched by normal acids - and needs an oxidizing acid like HCl : H2O204:13
azonenbergCr will passivate given the slightest access to an oxidizer04:13
azonenbergand needs a straight-up acid04:13
azonenbergthis could be interesting...04:14
azonenbergAnd ok, this is the electrochemical method04:15
azonenbergthe original long-expired patent http://www.freepatentsonline.com/3616349.html04:15
swkhanwow, i feel like you're doing all this for me =D04:16
azonenbergswkhan: i'm doing a similar process04:16
azonenbergthats what half of science is about - finding someone who's done something similar before04:17
azonenbergthen making some tweaks04:17
azonenbergthen you share your results so others can do the same with your work04:18
azonenbergswkhan: one of the skills necessary to survive as a grad student is getting good at locating prior work in the field04:18
azonenberggoogle scholar is your friend04:18
swkhani'm reading a lot of papers to hopefully get good at it04:19
swkhanwhat i'd like to learn is what causes good adhesion between materials04:19
azonenbergReading is all well and good but you need to not be spoon fed them04:19
azonenbergyou have to be able to locate them too04:19
azonenbergand that's a very interesting topic that i havent studied much about04:19
azonenbergif you find anything good, paste links04:20
azonenbergre etching, here are some starting points http://en.wikibooks.org/wiki/Microtechnology/Etching_Processes#cite_note-704:20
swkhani think my problem is that papers are not a good way for me to learn yet04:20
azonenbergThe etch bible is another good one http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=125735404:20
swkhanit used to be i couldn't learn from anything but really good lectures. but video lectures are time consuming and i have to really scrub through to find something good. textbooks are my current obsession04:20
swkhani learn a lot from them and i'm getting better at reading through them04:20
swkhani'm simultaneously reading a few papers04:20
swkhanbut papers really vary in quality04:20
swkhanand going from one paper to the reference can mean going from a paper that totally resonates with me to a paper that doesn't at all04:21
swkhanmy basics are not so good either04:21
swkhanso i'm strangely strongest in chemistry right now04:21
swkhani'm reading about solid state physics (ashcroft and mermin)04:21
azonenbergchem is one of my big weak points lol04:21
swkhanbut i'm reading a paper right now that claims it does not need a seed layer and they get some level (haven't read it yet) of improved adhesion between copper and silicon04:22
azonenbergmy theoretical EE knowledge is mostly in digital stuff (given my comp sci background) and much of what i know about microfab etc is applied04:22
swkhanmaybe i'm lying. maybe i'm really better with EE stuff right now digital, analog stuff04:22
swkhanbut i feel like the only thing i've totally taught myself is chemistry04:22
azonenbergand analog stuff is naother hole in my skills :P04:22
swkhananalog guides my intuition about what a device "looks like"04:22
azonenbergcome on, show me the paper :P04:22
swkhansorry sorry =p04:23
swkhanphew good thing i run screen. accidentally killed xterm04:24
azonenbergcopper compound, vitamin c, ammonium fluoride, and something else? And no adhesion layer required?04:24
azonenbergvvery interesting04:24
swkhanammonium fluoride... i'll have to check how expensive that is04:26
swkhani wish you were in my lab =)04:26
azonenbergswkhan: i'm pretty sure it can be synthesized from HF and ammonia *confirms*04:26
swkhanwow it's expensive04:27
azonenbergin trace metal grade? Probably04:27
swkhanwell i should read the paper first04:27
swkhani'm easily not the smartest girl in the EE department, but i'm saddened that i am probably the most knowledgeable in my lab of the next batch of grad students. i don't know very much and i can't learn much from my peers =\04:28
swkhanso if you were in my lab... =)04:28
azonenbergfor some stuff intelligence matters less than the level of applied knowledge04:28
azonenberglet's se...  ammonium fluoride, copper sulfate, ascorbic acid, sodium potassium tartrate, and methanol04:29
azonenbergand ascobic or fumaric acid is key to adhesion? hmm04:29
azonenbergthis process is selective to silicon04:33
azonenbergthat could be problematic if you want to deposit over oxide04:33
swkhanmm, in fact i'm sure that will be problematic04:33
swkhani guess i should be looking at copper deposition on glass. but we have so many wafers and not much glass04:33
swkhani get too caught up with the mindset the grad student has: make do with what you got04:33
swkhansometimes "what you got" could be expanded by making a trip to ace or home depot04:33
swkhanfor pretty cheap04:33
azonenbergFigure out what you want to do first04:33
azonenbergthen look at how other people have done it04:34
azonenbergthen see how you can do something of the sort04:34
swkhanthat makes sense04:36
swkhannow i'm curious about metal silicide layers04:38
swkhanwhy do they form? what does that even mean?04:38
swkhani know silicon is weird. it has different behavior than carbon (for example carbon likes to bond with itself but silicon likes to bond with oxygen)04:38
swkhanit has to do with the 3p orbitals and the size of overlap but i am not sure why.04:39
swkhanbut i get sidetracked easily04:39
swkhangoal: get copper on silicon04:39
swkhanthis might work04:39
swkhanbtw, NH4F is dangerous when heated04:39
azonenbergcopper on silicon, or copper on dielectric over silicon?04:39
swkhanso luckily they did this at room temperature04:39
azonenbergAnd fluoride compounds in general are dangerous04:39
swkhanah, yes, i want it to be copper on dielectric04:39
azonenbergit decomposes to ammonia + HF, right?04:39
azonenbergHF can definitely get nasty :P04:40
azonenbergThough in the 3% concentration i use it's not that bad04:40
azonenbergOf course when you say "not that bad" compared to 49% HF you arent saying much :P04:42
swkhancan i make copper sulfate from my copper conductive tape and H2SO4?04:42
azonenbergswkhan: copper conductive tape is copper + adhesive04:43
swkhanhere's my plan so far. i have some copper conductive tape and i think i have some sulfuric acid.04:43
azonenbergyou dont want adhesive04:43
azonenbergpure copper is easy to get hold of04:43
swkhani want to clean the adhesive with methanol and acetone04:43
azonenbergbuy wire from radioshack :P04:43
swkhani also have wire04:43
azonenbergUse wire04:43
swkhanbut we have a lot more conductive tape than wire04:43
swkhanand i heard the wire is expensive04:43
azonenbergUm, lol04:43
azonenberghow much wire do you want?04:43
azonenberga mile?04:43
swkhanwell i want to use this process from now on to make my bottom electrode at least04:44
swkhani should check that this works on glass first04:44
azonenbergi am nearly certain that the process in the paper you just linked does not04:44
azonenbergit was selective for Si over Si3N404:44
swkhanreading this right now: http://www.jstor.org/stable/9282404:45
azonenbergIf you're looking for CuSO4 http://unitednuclear.com/index.php?main_page=product_info&cPath=16_17_69&products_id=13104:45
swkhani know i can buy it04:45
swkhanbut we have loads of copper tape and sulfuric acid04:45
azonenbergcopper tape is expensive though04:45
azonenbergby comparison04:45
azonenbergCuSO4 - $6 for 2 ounces or $23 for 8 ounces04:46
swkhanoh wow04:46
azonenbergfive nines purity copper, $20 for one pound04:46
swkhanthey also didn't say in that paper what currents they used04:46
azonenberg$43.25 for 1/2" copper double sided tape04:47
azonenbergyou could buy two pounds of pure copper or one pound of CuSO4 for that price04:47
swkhanpoint made and conceded =)04:48
azonenbergSo while using copper tape might be justifiable if you needed it RIGHT NOW04:48
azonenbergif you can afford to wait a little bit, go buy the stuff :P04:48
azonenberg in fact, copper sulfate is a common enough material some other lab on your campus will probably have some04:49
swkhanit's nearly 10 pm. i want to try something before the end of tonight. i am desperate enough to use the vitamin supplements my labmate left on her desk =p04:49
swkhanwe're in a weird building04:49
swkhaneveryone does astrobiology04:49
swkhanwe're one of a few labs that does cvd stuff04:49
swkhanfor an EE lab04:49
swkhanwe're treated like second class citizens at nasa... for now04:49
swkhani'll get my real nasa badge soon enough. in the mean time i have my lowly "special building for kiddos" badge -_-04:50
swkhanrandom person on the internet says: "Silicon dioxide is a molecule (neutral), not an ion. You can only electroplate electrically charged ions but not molecules which do not have an electrical charge."04:52
swkhanthat makes sense04:52
azonenbergThat's depositiong, though04:52
azonenbergyou can only deposit ions04:52
azonenbergYou can deposit on a molecule by electroplating04:52
azonenbergwith the caveat that said molecule mus be conductive04:52
azonenbergfor example you can deposit onto a doped silicon surface or a metal04:53
azonenbergbut not onto an insulator04:53
azonenbergIf you are depositing onto an insulator i can almost guarantee you will need some sort of PVD04:53
swkhancan i somehow ionize SiO2?04:54
azonenbergAre you trying to deposit SiO2 onto something?04:54
azonenbergOr deposit something on SiO204:54
azonenbergIn the second case, ionizing wont help you04:54
azonenbergbasically what would happen if you tried is that as soon as a few ions hit the surface the SiO2 would charge up04:55
azonenbergand become neutral04:55
azonenbergthe bond is covalent and not ionic so you wont be able to do anything of the sort04:55
swkhanoh yeah it is covalent =\04:57
azonenbergelectroless chemical plating is possible but i know little about it04:57
azonenbergDo you have access to high vacuum anywhere?04:57
azonenberg1E-6 torr range04:57
swkhanno i don't04:58
swkhani have access to ~5E-1 torr range04:58
azonenbergNot low enough for evaporation but enough for sputtering04:58
swkhanah it's the cvd reactor though04:59
azonenbergthing is, sputteirring makes up for that by being more complex to set up :P04:59
swkhanso i can't really change it much04:59
swkhanand i've heard about metals can cause deep level traps04:59
azonenberglol i see04:59
azonenbergsorry to say i think you're pretty much out of luck at this point04:59
azonenbergif you're willing to wait for some common materials to arrive there are options05:00
azonenbergfor example, if you had access to a sputter coater of some sort, even a tiny one for SEM sample prep05:00
swkhannope =\05:00
azonenbergyou could deposit enough of a conductive layer on your sample to electroplate over that05:00
swkhanhmm so i'm trying to consider the advantages of each05:01
swkhanpure sputtering: very fast, no chemicals, costs me about $15/hour maybe $7.50/hour at night05:02
swkhansputter seed layer, electroplate: very cheap with materials i have (effectively $0 cost), i get to learn something new05:02
swkhanpure electroplating: free, i get to learn a lot of new things05:02
swkhanoops cost shouldn't have been there. i'm getting sleepier and sloppier =\05:03
swkhanpure sputtering: minor disadvantage of cost and currently down so can't really do it05:03
azonenbergpure electroplating, disadvantagee: can't deposit on nonconductive materials :P05:05
swkhanyes =(05:05
swkhanwell as far as i know now there may be a method around it somehow...05:05
azonenbergelectroless copper plating is possible05:05
swkhani really want to try the berkeley paper's recommendations without as many additives. i don't have rochelle's salt or ascorbic acid05:05
azonenbergbut i suspect that it wont be doable with materials you have in house05:05
swkhanbut i do have methanol05:05
azonenbergThe ascobic acid (or fumaric acid) was crticial for adhesion05:05
azonenbergmethanol was almost unnecessary05:05
swkhanoh right that's right i got that backwards05:05
swkhani thought fumaric was not necessary05:05
azonenbergIf you had more time and budget i know of an electroless gold plating solution05:06
azonenbergthat could be followed by electroplating of copper05:06
azonenbergyou spin coat the stuff, then bake in air to decompose the precursors and form a few tens of nm of metallic gold05:06
swkhantime is not too big of an issue. budget is05:06
swkhanshiny =D05:07
azonenberg$500 for 4 ounces of the solution :P05:07
azonenbergemulsitone goldfilm05:07
swkhanoh ya that stuff05:07
swkhanmaybe i can look around for stuff like ascorbic acid05:08
azonenberg*checks if the MSDS has any info on composition* nope05:08
azonenbergand i dont think that will help05:08
swkhani want to figure out whatever "hydrogen scavenger" is05:08
azonenbergthe point is, it wont help because your surface isnt silicon05:09
azonenbergthe process was selective for deposition on silicon only05:09
azonenbergand electroplating requires a conductive surface05:09
azonenbergYou need to figure out how to deposit a conductive layer of some sort over the dielectric05:09
azonenbergeither by itself, thick enough to use as-is05:09
azonenbergor as a seed layer for electroplating05:09
azonenbergi dont see any alternatives05:09
azonenbergelectroless nickel is an option that works on nonconductive materials05:10
azonenbergyou could then follow it with plating of copper05:10
azonenbergor just use the nickel as-is05:11
swkhancan i use nickels (like the coins) as nickel sources?05:11
azonenbergUnited states nickels do not contain any Ni05:11
azonenbergsince about the 1950s05:11
swkhanoh lol05:11
azonenbergcanadian ones do still afaik05:11
azonenbergUSA are all ZnCu alloy05:11
swkhanor even pennies... they are electroplated copper on zinc right?05:12
azonenbergI dont know how the copper is deposited, but yes05:12
azonenbergthey're Cu plating over Zn05:12
swkhanhmm, do you think that's a cheap form of copper?05:13
swkhanor do i run the risk of getting too much zinc off of it?05:13
swkhani seem to remember zinc really likes to be sacrificed05:13
swkhanit has a higher electrochemical potential so it's easy to get it to reduce05:14
azonenbergYou will get a ton of zinc in the solution and the plating is very thin05:15
azonenbergyou want cheap copper?05:15
azonenbergFind broken electronics05:16
azonenbergor scrap AC power cords05:16
swkhanelectricity is free here05:16
swkhanmy time is somewhat free05:16
swkhanour research group's budget is next to nothing =(05:16
swkhanor at least that's how we're told to perceive it05:16
swkhanthen magically i am told every once in a while to "find a way to spend $6000 by friday" on a thursday evening05:16
azonenbergI've done filament evaporaion of copper on Si by getting the copper out of a power cord05:17
swkhanyou're awesome =D05:17
azonenbergfrom a broken inkjet printer someone left on the loading dock and marked "trash"05:17
azonenbergyou'd be amazed at the resources that can be obtained by dumpster diving05:17
swkhani once made a pulse chopper using spare components from a power supply that was chucked in the e-waste bin by the economics department05:18
azonenbergYou have dedicated e-waste bins there?05:18
azonenbergThere's your copper source05:18
swkhanwell that was at ucsc05:19
swkhanat nasa we have nothing about bureacracy bins05:19
swkhanso point being if i need to refine something using electricity, time, and especially if it doesn't need me to be watching it... that's perfect05:20
azonenbergCopper from power cords is easily pure enough to plate with05:21
swkhanif i want to get this done at least on silicon tonight, i'm going to probably end up using the copper wire05:25
azonenbergSilicon, sure05:25
swkhanbut ascorbic acid =\05:25
azonenbergBut it wont be insulated05:25
swkhanthat's okay for now05:25
azonenbergAnd what would you do then?05:25
azonenbergdont you need to pattern it?05:25
swkhani might be able to get away with this...05:26
swkhanbottom contact on the whole sample05:26
azonenbergyou never did mention what you are trying to build in the end05:26
swkhan(not necessarily a full wafer but whatever)05:26
swkhanthen deposit a metal oxide and then probe it with a IV curve tracer05:26
swkhanthe probe could serve as the second contact05:26
swkhanso i don't even need to pattern just yet. i want to get the proof of concept that the metal oxide does something useful05:27
swkhana postdoc in our group was saying that he worries that when you have the probe make contact two things happen05:27
swkhan1. you scratch the surface [unless you use contact mode AFM (STM really)]05:27
swkhan2. the pressure of the probe affects the IV relationship05:28
swkhanwhat are your thoughts on 2. especially?05:28
azonenbergDo you have access to a wirebonder?05:28
azonenbergYou can get around both potentially05:28
azonenbergbut the methods for doing so introduce unique problems of their own05:28
azonenbergExplain the structure a bit more?05:29
swkhanthe structure?05:29
azonenbergof what you're trying to build05:29
swkhansilicon (maybe with a thin native oxide) with copper on it (bottom contact) then metal oxide on top of the bottom contact then i can use the tungsten probe of my IV curve tracer to serve as the top contact05:30
azonenbergAnd what is the desired device?05:31
swkhanhow thick is the SiO2 layer on top of Si (100) if i haven't done anything special for it05:31
azonenbergVery thin05:31
azonenbergNot enough to be an effective insulator05:31
swkhanwhatever i can get. capacitor, variable resistor05:31
azonenbergi think its on the order of 10A05:31
swkhanokay so when people say they "thermally grew" silicon dioxide... how do they do it? what's going on when they have it in some sort of chamber?05:31
azonenbergDry or wet?05:31
swkhanwith lots of heat i mean05:32
azonenbergThe basic process is the same05:32
swkhan10A wow05:32
azonenbergStick wafer in oven05:32
azonenbergheat to 1200C or thereabouts05:32
azonenberghold for an hour or so and cool slowly05:32
swkhanany idea why that magically grows SiO2?05:32
swkhanah okay reading05:32
azonenbergSi + O2 = SiO2 :P05:32
swkhani feel obligated to repay your knowledge with something that might help you as well :p05:32
swkhanit's interesting so far05:32
azonenbergthis is the same reaction that forms native oxide05:32
swkhani didn't know how useful metal silicides were05:33
azonenbergyou're just accelerating it dramatically05:33
azonenbergby heating05:33
swkhanthat makes sense. so you break the bonds between O2 using enough energy05:33
swkhanbut then... hmm i am not sure how or why SiO2 would form05:33
swkhanor even why that reaction happens in the absence of heat05:33
swkhansome stuff for me (and maybe you?) to read i guess05:35
Action: azonenberg linked you to that a few mins ago :p05:35
azonenberghttp://www.ece.gatech.edu/research/labs/vc/theory/oxide.html is another nice resource05:35
azonenbergif yo uwant an excruciatingly detailed model, http://pubs.acs.org/doi/abs/10.1021/ie00081a024 might be good05:36
azonenbergi havent read it05:36
swkhanwow wow wow http://en.wikipedia.org/wiki/Andrew_Grove05:37
swkhanhe's so inspirational05:37
swkhanazonenberg: oh lol i'm sorry i didn't see that05:37
azonenbergAnd if all you want to do is compute growth rates given parameters (or vice versa) http://www.cleanroom.byu.edu/OxideThickCalc.phtml05:38
azonenbergor http://www.cleanroom.byu.edu/OxideTimeCalc.phtml05:39
azonenbergThe BYU cleanroom website has a lot of useful material in general05:40
azonenbergi've spent a nontrivial amonut of time looking through it reading about everything from wet etch processes to stepper alignment marks05:41
bart416azonenberg, still alive over there?12:43
azonenbergbart416: yep17:29
azonenbergdecently heavy rain but not too much wind atm17:29
bart416azonenberg, the LHC made its first victim it seems18:01
bart416supersymmetry is pretty much debunked18:02
azonenbergsomeone fall off a ladder and bang his head on an electromagnet?18:02
azonenbergah, nice18:02
bart416That piece of crap is finally gone18:03
bart416And due to that booze and cigars for everybody!18:05
bart416Though you might not want to smoke the cigars18:06
bart416They're cuban so you'll probably get arrested for having them in the US18:06
bart416And not joking, I do actually have a box of Cohiba Siglo IV 's :P18:08
bart416But those are for special occasions18:08
bart416Like after beating the crap out of a hard exam18:08
bart416Or winning an insane contest18:09
Action: azonenberg is not a fan of tobacco18:09
Action: bart416 neither18:09
bart416I actually hate smoking18:09
bart416But cigars it's the idea that counts :P18:09
azonenbergAnd i try not to smoke pot too much either...18:10
azonenbergthe smell of burning carbon composition is kinda irritating18:10
azonenbergto say nothing of the fact that you're destroying a perfectly fine variable resistor from overcurrent18:10
bart416Cuban cigars are quite different from cigarettes though18:14
bart416Also one of the only reasons I'm not on the floor dropping dead >_>18:14
bart416(I'm allergic to regular cigarettes)18:14
azonenbergwhy are they so different? i mean i know they smell a lot different18:15
azonenbergbut its the same stuff burning, right?18:15
bart416Don't think it is18:16
bart416Cigars don't have all the chemical crap added to them18:16
azonenbergthey add stuff to cigarettes?18:17
azonenbergshows how little i know about such things lol18:17
bart416Tobacco in cigarettes is chemically altered18:17
bart416They add chemicals to make them more addictive, at least they used to18:18
bart416Now it's "flavour"18:18
bart416In other words, the same thing, lol18:18
azonenbergto make them more addictive? o_o18:19
lekernelbart416, why is supersymmetry crap?18:19
bart416I failed a course on the subject when I was doing physics :P18:20
lekernelI'm not saying it isn't, I actually do not understand it18:20
lekerneljust wondering about your thoughts18:20
bart416Well, the issue with supersymmetry in my opinion is that it's "too nice"18:20
bart416It's more of a construct than an actual theory if you ask me18:21
Action: azonenberg doesnt know enough about physics to be able to say much about it18:21
bart416I'm not going to try to explain it over the internet18:22
bart416And I woiuldn't try reading wikipedia about it either if their reputation on physics keeps up18:22
Action: azonenberg has never tried reading wp on any physics more advanced than newtonian mechanics18:24
azonenbergi can say their info on semiconductor fab is usually spot on18:24
bart416The text is usually too retarded to read on theoretical physics models18:27
bart416And if you try to correct them they link to some flawed internet page written by a 14 year old who got his hands on a book and tried to post a summary online18:27
berndjthat reminds me, britneyspears.ac actually has some pretty real semiconductor physics18:28
berndjdespite apropos-nothing pics18:28
azonenberg?? lol18:28
berndjtbh i haven't seen too many wikipedia articles that were clearly written by bombastic know-nothings18:30
azonenbergi usually stick to the highly technical pages on rather non-controversial topics18:30
azonenbergand just use them for summaries and rely on journal articles for details18:31
berndjand keeping 1e6+ articles right is a battle against entropy18:31
berndjtook me a while to figure out, for example, why they had silicon's boiling point at 2500K or so18:31
berndjcelsius/fahrenheit confusion in that case18:31
azonenbergThey had a crew going through all of hte phsyical data and fixing it afaik18:32
bart416That's another thing, theoretical physics is SI or CGS unit systems18:32
bart416Not their damned imperial system18:32
azonenbergBoiling point 3538 K,3265 °C,5909 °F18:32
lekerneli'm reading "the road to reality" by Penrose atm, but I still have not reached supersymmetry18:32
azonenbergthat looks reasonable18:32
bart416lekernel, well skip that part18:33
bart416The LHC delivered a final headshot to it :D18:33
bart416Lets see what my periodic table lists as boiling point of silicon18:35
Action: azonenberg doesnt have CRC tables on hand18:36
bart416at 1 bar of pressure18:36
bart416Melting point is 1412°C18:37
azonenbergI see two major values being mentioned online - 2355 and 3265C18:37
azonenbergTrying to find a reliable source for either18:37
berndjhmm, is't SU-8 also a symmetry group?18:37
berndjerr, maybe you write it SU(8)18:38
azonenberg2355 seems correct18:38
azonenberg two different MSDSes list it18:39
azonenbergWhere did the higher number come from?18:39
bart416Melting temperature depends on purity and pressure eh18:39
azonenbergthe thing is, there are two very specific temperatures18:39
azonenbergnot a range18:39
azonenbergbeing mentioned18:39
bart416Crystaline structure...18:41
bart416If the binding energy of the crystal is higher you need more energy to melt it18:41
azonenbergsilicon only has one crystal structure though18:42
azonenbergdiamond cubic18:42
bart416Also http://img153.imageshack.us/img153/6559/imag0503.jpg18:42
bart416Binding energy isn't always the same I think18:42
bart416Dunno, crystals = chemistry18:42
bart416mhhh, it really does come down to the crystals it says here in this shitty book18:44
bart416monocrystaline is harder to melt than non-monocrystaline18:44
azonenbergBut monocrystalline boils at 2355 as far as i can see18:45
azonenbergAnd once melted, BP is the same18:45
azonenbergregardless of whether it came from a crystal or not18:45
azonenbergI'm trying to find a single source for 3265 that isnt wikipedia or something they're citing18:45
azonenbergor something that cited them*18:46
bart416The 3265 looks like a typo18:46
bart416The small variances in melting are down to the way the crystal is composed it seems18:46
bart416Melting point, wikipedian technology, lol18:46
berndjazonenberg, my copy of _Chemistry of the Elements_ has MP=1420C, BP=~3280C19:11
azonenbergSo we have multiple reliable sources for each19:11
azonenbergMaybe depends on purity?19:12
berndjwhat if that MSDS was lazily copied from wikipedia during the time wikipedia was borked?19:13
berndjhttp://en.wikipedia.org/w/index.php?title=Template%3AInfobox_silicon&action=historysubmit&diff=435185046&oldid=421589811  <-- there's my fix to the infobox19:13
berndjsome previous editor made it 2355, so there's a link there19:14
bart416btw azonenberg, looked on another MSDS20:32
bart416e) Melting/freezing point Melting point/range: 1.410 °C - lit.20:32
bart416f) Initial boiling point and20:32
bart416boiling range20:32
bart4162.355 °C - lit20:32
azonenbergVery interesting20:32
azonenbergwell, as i dont plan to boil silicon any time soon, i'll live with the situation :p20:32
bart416boiling silicium, what could possibly go wrong :')20:32
berndjif i knew where my periodic table was that predated wikipedia, i'd look it up there20:35
berndjbut then what about wikipedia's table of vapour pressures?  those seem consistent with the higher boiling point20:35
azonenbergYeah, after a bit more reading the higher value looks right20:36
bart416Just buy a good periodic table20:38
bart416I think sigma aldrich sells their own ones azonenberg20:38
bart416Maybe piggy back it on another order azonenberg?20:38
bart416Well yeah, I have two; one from taminco that has a nice list with a redox scale that's pretty complete20:43
bart416and acid constants20:43
bart416And then another one from where I worked for a while that lists pretty much every single isotope known to mankind that's stable enough to be worth mentioning :D20:43
azonenberglol nice20:46
bart416I'm going to guess azonenberg lost power20:57
bart416wb azonenberg21:41
azonenbergISim decided to drive my machine into swap lol21:56
azonenbergSo hard that wpa_supplicant's latency increased beyond the allowable limit21:56
azonenbergand the wifi dropped :P21:56
azonenbergi dont know why it decided it needed >3GB of ram21:57
azonenbergbut it was using 70%+ of my memory lol21:57
azonenbergi now have a killall command on an SSH session in another machine21:58
azonenbergready to go at a moment's notice if things go bad21:58
Last message repeated 1 time(s).22:14
azonenbergOh, and the worst part is that it should have aborted22:18
azonenbergthere was a syntax error in one of the verilog files22:18
azonenbergsomehow it synthesized *something*22:18
azonenbergthat, well, blew up :p22:18
azonenbergLol, found the bug22:29
azonenbergan initial block was turned into an always22:30
azonenbergsimulator went into an infinite loop collecting more and more waveform data without advancing the timer at all22:30
berndjyou implemented ECRAY!22:32
azonenbergwoo my ted pella order gets here on thursday23:35
azonenbergcopper pellets for evaporation, a selection of tungsten filaments, some more general lab gloves, and diamond polishing compound for cross section prep23:35
azonenbergi needed somethign midway between 60nm colloidal silica and 1500 grit sandpaper :p23:36
swkhanyay azonenberg =)23:38
azonenberghopefully by middle of next month i'll have some nice cross section pics to show23:38
swkhanazonenberg: i'm trying to learn solid state physics but i'm getting stuck on a section about the drude model for metals23:45
swkhan"Given that the momentum per electron is p(t) at time t, let us calculate the momentum per electron p(t + dt) an infinitesimal time dt later. An electron taken at random at time t will have a collision before time t + dt, with probability dt/tau, and will therefore survive to time t + dt without suffering a collision with probability 1 - dt/tau."23:45
swkhan"If it experiences no collision, however, it simply evolves under the influence of the force f(t) (due to the spatially uniform electric and/or magnetic fields) and will therefore acquire an additional momentum f(t) dt - O(dt)^2."23:45
swkhanwhere does O(dt)^2 come from? they say it is big O notation23:45
berndjs = ut + at^2/2 by analogy23:47
berndjno, wrong degree23:48
berndjswkhan, in physics you often just add terms of higher degree to make things more accurate - then figure out the coefficients later23:48
swkhantrying to add additional terms of higher degree makes sense23:49
swkhanbut in this situation, i'm not sure where that's coming from23:49
azonenbergswkhan: Are you doing some kind of DSMC?23:49
swkhani don't know what DSMC is23:49
azonenbergdirect simulation monte carlo23:50
swkhandon't know what that is =\ i don't know anything23:50
swkhanbut, no, i'm not. i'll check it out at some point though23:50
berndjinside the metal the force is unlikely to be perfectly isotropic; i imagine that O(dt^2) term captures that23:51
swkhanit's weird that they bring that up23:52
swkhanbecause they try to explain drude's model of metals23:53
swkhanwhich is basically an application of kinetic molecular theory to electrons23:53
azonenbergWhen i first read that section it sounded very much like it23:53
swkhanso electrons are treated as a gas where each of the constituent particles don't interact with each other except for when they mechanically collide into one another23:53
azonenbergYeah, i've seen similar models23:53
azonenbergNot as easy to model as a Fermi gas but not too bad either23:54
swkhani guess for the sake of pushing on i'll concede what berndj said and come bother you all if i find anything interesting23:54
swkhanoh btw azonenberg that berkeley paper on electroplating copper on silicon had a followup paper that was interesting as well23:55
swkhanthey explained why they use HF and follow it up with a methanol dip23:55
azonenbergHF is probably to strip native oxide and make it more conductive23:56
azonenbergmethanol, not sure off the top of my head23:56
swkhanit gets terminated with hydrogens apparently and so the surface is really hydrophobic23:56
azonenbergOh, yeah23:56
swkhanmaking it really difficult to electroplate23:56
azonenbergThe methanol fixes that?23:56
swkhanthey don't know why though23:56
azonenbergThats actually a common sign of when your oxide strip is complete23:56
azonenbergif you pull it out of the HF and water beads up23:56
azonenbergif you're doing a single die in a tray of HF it'll actualyl float to the surface23:57
azonenbergbecause it picks up bubbles from reaction products23:57
azonenbergthe hf pulls back and doesnt want to touch it23:57
swkhananother thing i don't understand is why people use 1% concentrations of HF23:57
azonenbergand eventually it floats23:57
azonenbergi normally use 3%23:57
swkhancan you etch / clean with any old acid?23:57
azonenbergi can buy it over the counter at my local grocery store23:57
azonenbergand it doesnt fume like 49% etc23:57
swkhani have sulfuric and phosphoric and base wise i have sodium hydroxide23:58
swkhansilicon is such a weird and interesting element23:58
swkhanif carbon makes life exist, silicon makes it worth living =)23:58
berndjpart of that is probably because it's one of the most studied elements23:58
berndjrivalled only by iron or uranium23:58
berndjazonenberg, is your rust remover pretty much only HF and hydroxylic acid, or is there other stuff in it?23:59
--- Mon Aug 29 201100:00

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