#homecmos IRC log for Wednesday, 2012-11-21

Syncazonenberg: my prof told me to try regular photoresist for pcbs10:48
Syncit should work pretty good10:48
azonenbergregular resist?10:48
azonenbergdefine regular10:48
Syncas in the stuff you buy in the electronics store to coat your pcbs10:50
azonenbergas opposed to cmos grade?10:50
azonenbergit'd be cheaper for sure10:51
Syncyeah, he said for our structure sizes and general conditions it is not neccessary to get professional resist10:55
azonenbergwhat are yo utrying to do ?10:55
azonenbergi've done 5 micron features with pcb resist :p10:56
Syncnothing special, just had a chat with him after the lecture10:57
azonenbergbasically, pcb resist isnt trace metal grade10:58
azonenbergbut for anything non-cmos it should be fine10:58
Synctrying to do cmos will require some serious effort10:58
Syncthe worst enemies are Ni and Au10:58
azonenbergHow about K and Na :P10:58
SyncNi diffuses through Si at room temp11:01
Syncthat caused some headaches here11:01
azonenbergor, heaven forbid, Li11:01
azonenbergSi(Li) EDS detectors cant even be turned on at above cryogenic temps :P11:02
Syncyeah but you don't have those on your tweezers :P11:02
azonenbergwait, you were using Ni-plated tweezers to handle chips11:02
azonenbergand that caused problems? o_O11:02
Syncit ruined the wafer11:02
azonenbergDid you do any high-temp processing on it?11:03
Syncwe were wondering why our HEMTs were so slow11:03
Syncthey were just sitting around11:03
azonenbergi was going to say, it would make sense if you had heated it11:03
azonenbergbut at room temp, wow11:03
Syncjust from the edge handling11:03
azonenbergThe entire wafer?11:03
azonenbergor just dies near it11:03
Syncit was distributed through the whole wafer11:03
azonenbergover how long11:04
Syncnot equally but still11:04
azonenbergdiffuison over that much of a distance must take a while at room temp through solids, right?11:04
SyncI suspect a few weeks11:04
azonenbergAnd how did you trace it to Ni contamination?11:05
azonenbergits a heavy metal, i wouldnt expect it to diffuse that fast11:05
Syncwe looked at every process step11:05
azonenbergNa or K i could understand11:05
Syncbecause it just did not make sense that the devices were so slow11:05
Syncso it HAD to be metal contamination11:05
Syncbut we could not see why11:05
azonenbergSo more to the point, when you changed tweezers11:06
Syncand then we did one wafer being anal no metal came in contact with11:06
azonenbergdid the problem go away?11:06
Syncand it worked11:06
azonenbergAnd you're sure it was Ni11:06
azonenbergrather than, say, K contamination on that particular pair of tweezers?11:07
Syncthose were Ni plated stainless tweezers only used to handle wafers11:07
Syncin our flowbench11:08
azonenbergDid you do any testing on those tweezers, i mean11:08
azonenbergto confirm11:08
Syncwe put them in the ultrasonic cleaner and tried again11:08
Syncand it still failed11:08
azonenbergHmm, interesting11:08
azonenbergbecause i'd believe it was metal contamination11:08
Syncso we are pretty confident that it is ni11:08
azonenbergbut i'm skeptical that it's the actual tweezer metal11:08
azonenbergas opposed to a contaminant11:08
azonenbergIs Ni known to diffuse at room temp?11:09
azonenbergit just seems counterintuitive that something so heavy would diffuse from momentary contact at room temperature with no high-temp processing11:09
azonenbergif you had a metal chip on the wafer and put it in a furnace i could understand11:09
azonenbergthrough a whole wafer no less ,rather than just the dies near where it was handled11:10
Syncit was strange11:10
Syncwe did not really investigate after we found the issue11:11
Syncit seems the selfdiffusion rate is around 10^-8m²/s11:11
Syncit was a really strange problem11:12
azonenbergi mean for B and P i'm used to hearing of tens of minutes to hour diffusions at 1200C11:12
azonenbergand thats for a few microns11:12
azonenbergi wouldnt expect horizontal diffusion through several cm of Si to take place at room temp at any appreciable rate11:13
azonenbergas opposed to if you were diffuisng in through the surface from contamination in a bath etc11:13
Syncplz wait, I have to start my process11:13
azonenbergDoes your lab have a rule about only CMOS-grade resist on the coaters?11:14
azonenbergthe cleanroom here does11:14
SyncI don't think so11:15
SyncI suspect that there was some thermal process after that.11:18
azonenbergafter you handled them?11:18
Syncjust reading some protocols again11:18
azonenbergAnd was there an RCA clean first? :P11:19
azonenbergif not, there's your problem :P11:19
Syncwe always hf dip, priahna, ozone and hf dip wafers before every process11:19
azonenbergis piranha good at trace metals?11:20
azonenbergi'm used to using RCA-2 for that11:20
Syncwe ususally have no trace metal on the surfaces11:20
azonenbergwell you did this time11:21
azonenbergwhich is why i'm asking11:21
Syncwe used wafers from the same carrier11:22
Syncand the same prep11:23
Syncand after we changed the tweezers it was fine11:23
Syncthe wafers had some heating done to them in the 400-600°C range but not all of them where we made lifetime measurements11:30
azonenbergand were they cleaned before that?11:30
azonenbergif not, there's your problem11:30
Syncand in those that were kept at room temperature it was less pronounced11:30
Syncyes they have gone through our regular process11:30
azonenbergRT diffusion i would expect little to none, 600C is significant11:30
azonenbergno, i mean during the lifetime measurements11:31
azonenbergwere they cleaned before going in the oven?11:31
azonenbergYour clean sucks :P11:31
Syncso how does that explain the bad lifetime on the wafers that were just stored for further processing?11:33
Syncsomething strange was going on obv.11:33
soul-dso no NERD_N_130F in mail yet :( would expected this revision in stock since it was used as example11:37
azonenbergsoul-d: Lol11:37
soul-dno complaint information either11:37
soul-dcan't even complain11:37
azonenbergYeah, the resellers have a habit of taking a long time11:37
azonenbergand if you go to the vendor you have to wait for the 9-month factory lead time plus many years of firmware updates11:38
azonenbergwhy cant they just ship with full firmware images from the factor???11:38
soul-dyou might want to include  an age  code :P11:39
soul-d_xxJ  would be comon on chips to11:39
azonenbergsoul-d: standard JEDEC date code will be included in the packaging information section11:40
Action: azonenberg is date code 9033 :P11:40
soul-danhow   http://imgur.com/jnpEC   would expected beter then 30 mA  but i should probably focus on  measureing the resitance maybe  put in  a coil of few can's  and get rid of paint of it first11:42
azonenbergwhat is this apparatus?11:42
soul-dtarget is generating oxigen11:42
azonenbergi see a red bull can and a hose, that can't be good11:42
soul-douter can is   -11:42
Syncwat. pressure just went through the roof in thermal cleaning11:42
soul-dadn the can is the sacrificial  one11:43
soul-dshould gerate oxigen if im correct :P11:43
soul-douter can should generate  hydrogen   and be protected     a bit  but did read metal can have hydrogen somthign failure and  become brittle11:44
Synchydrogen embrittlement11:44
Syncbut that is not going to be a problem there11:44
Syncmostly pronounced when welding11:45
soul-dand i want to oxdize copper :P   to cuprous oxide  or somthing  so don't realy expect much  since i doubt reaction happens quick but  formula din't mention heat or something11:47
soul-dnow i only have to figure out  how to increase current to reasonably test it11:52
Synchow about just heating copper?11:53
soul-dno thats the black oxide11:53
soul-di want the red one  cuprous (i) oxide11:53
soul-done is CuO  other is Cu2O11:55
Syncah yes true11:57
soul-dim doing it to learn an do some chemistry ofcourse   i could probable buy most stuff cheaply11:58
Syncgoddamn machine13:22
--- Thu Nov 22 201200:00

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