lekernel | azonenberg, what's the difference between epitaxy and evaporation/LPCVD? | 11:00 |
---|---|---|
azonenberg | lekernel: I'm not too fami.liar with the details of epitaxy but it forms a single-crystal coating | 12:12 |
azonenberg | The others form a polycrystalline coating from a vapor, the difference is the source of the vapor | 12:12 |
azonenberg | evporation is purely physical (heated chunk of the material) | 12:12 |
azonenberg | CVD (and variants like PECVD, LPCVD, MOCVD) pump in reactive gases that combine or decompose in some way | 12:13 |
azonenberg | And sputtering is also a physical process but there's an inert gas present to do the sputter | 12:13 |
azonenberg | Reactive sputtering is a hybrid - vapor is formed by normal sputtering processes but before it hits your sample it reacts with another gas | 12:14 |
azonenberg | for example using an Ar / N mix to sputter Ti | 12:14 |
azonenberg | the Ar knocks Ti atoms off the target, which then react with N to form a film of TiN | 12:15 |
azonenberg | Epitaxy, on the other hand, you are depositing either the same material as your substrate or something with a compatible crystal lattice | 12:16 |
azonenberg | in either case it forms a single-crystal coating where the atoms are locked to the lattice of your substrate | 12:16 |
azonenberg | Another important difference is the pressure | 12:20 |
azonenberg | Sputtering has an inert gas present and CVD has source gases, so both happen at relatively low vacuum (few hundred Pa ish) | 12:21 |
azonenberg | Mean free path is short and you get a pretty even coating since the atoms bump into each other and are moving pretty randomly | 12:21 |
azonenberg | Evaporation is done at much low pressure (1E-6 torr ish) | 12:21 |
azonenberg | Mean free path is comparable or greater than the distance from the source to your sample | 12:21 |
azonenberg | Which means atoms evaporating from your source travel in pretty much a straight line (not hitting any gas molecules etc) before hitting the sample | 12:22 |
azonenberg | As a result the coating has some of the characteristics of spray painting - bumps in the surface cast shadows, edges perpendicular to the source are coated thinly if at all, etc | 12:23 |
azonenberg | which is bad for some processes (filling vias for example) but very good for others (lift-off) | 12:23 |
lekernel | so, epitaxy is a particular case of evaporation? | 12:25 |
azonenberg | The source of the vapor for epitaxy can vary | 12:28 |
azonenberg | I'm not familiar with the usual sources | 12:28 |
azonenberg | this isnt a process i've considered using either for my fab, or needed for anything at work etc | 12:29 |
azonenberg | The other processes i've actually studied in some detail | 12:29 |
azonenberg | In any case i'm heading up to campus now, i've got work to do | 12:29 |
azonenberg | And a 10:00 (local time) appointment with a SEM :) | 12:30 |
azonenberg | Expect pictures this evening | 12:30 |
azonenberg_work | lekernel: back | 13:17 |
azonenberg_work | On the SEM now taking pics | 17:05 |
azonenberg_work | probe tips are sharper than i thought | 17:06 |
azonenberg_work | 200nm ish dia flat tip | 17:06 |
bart416 | :D | 17:08 |
bart416 | azonenberg, btw did you think about the destructive stm test? | 17:08 |
lekernel | epitaxy might be interesting as well - think of microwave transistors, diodes, etc. though I don't know the difficulty level to have equivalent performance with the commercial ones that need an interview to get a quote for | 17:12 |
azonenberg_work | lol | 18:00 |
azonenberg_work | worth looking int but prob difficult | 18:00 |
bart416 | lekernel, any ideas for scraping a single layer of atoms of a sample? | 18:42 |
lekernel | er... maybe with those tunnel microscopes? | 18:42 |
bart416 | Yeah, that's the thing | 18:48 |
bart416 | It's too slow to remove the layer from a large surface | 18:48 |
bart416 | I suggested it somewhat unserious yesterday to use a STM to scan a sample layer by layer | 18:50 |
bart416 | But the more I think about it the more it makes sense | 18:50 |
bart416 | An STM could break a crystal latice | 18:50 |
bart416 | It'd take several millenia to do that for a large surface :p | 18:54 |
lekernel | what do you need atomic precision for? | 18:58 |
lekernel | especially on a large surface ... | 18:58 |
bart416 | I want to scan layer per layer with a stm :P | 18:59 |
bart416 | Somebody in ##electronics had a nice idea | 18:59 |
bart416 | Using a pulsed laser with a kerr lens | 18:59 |
lekernel | mh, maybe the stm isn't the best tool for that | 18:59 |
lekernel | or maybe you could make it faster by increasing the voltage and the distance between the probe and the sample? (just guessing here) | 19:03 |
bart416 | Maybe | 19:08 |
azonenberg_work | Pics coming later today | 19:48 |
azonenberg_work | Some interesting stuff but the guy had to leave before he could show me how to use EDS | 19:49 |
azonenberg_work | So i have to come back tomorrow | 19:49 |
bart416 | EDS as in what sense? | 20:01 |
azonenberg_work | energy dispersive spectroscopy | 20:03 |
bart416 | ah | 20:03 |
bart416 | That should show interesting results | 20:04 |
azonenberg_work | using x-rays emitted from the sample to determine what elements are present | 20:04 |
azonenberg_work | Which will let me tell whether the stuff i'm looking at is hardmask (tantalum based), photoresist (organic), or a mixture (some of each) | 20:04 |
bart416 | If you're willing to do some math you should be able to get a pretty good formula for it | 20:04 |
azonenberg_work | http://i.imgur.com/R5o3m.jpg is a quick shot of what i did today (low res, original is 2x in both dimensions) | 20:04 |
azonenberg_work | This is an etched die (F7) seen at 2000x | 20:05 |
azonenberg_work | The thin layer on top is hardmask residue (not removed completely by liftoff), the silicon can be seen extending outwards below it to the etched surface in the background | 20:05 |
azonenberg_work | Diagonal surfaces are <111> planes, the background is a <100> plane | 20:06 |
azonenberg_work | And the microscopy tools have the software to identify most of the peaks in the chart | 20:07 |
azonenberg_work | I just dont know how to use it or how to install the detector (it's not installed by default) | 20:07 |
bart416 | heh | 20:07 |
bart416 | read the manual! | 20:07 |
azonenberg_work | iirc you need to physically remove the backscatter detector to install this one | 20:07 |
azonenberg_work | keeping the everhart-thornley secondary electron detector for aiming | 20:07 |
azonenberg_work | (and imaging) | 20:07 |
berndj | if you forego atom-size res, could you push target atoms around faster than normal with an STM? | 21:13 |
berndj | just thinking of that destructive test procedure you guys were talking about | 21:13 |
berndj | (ab)using the STM probe tip as a mechanical cutting tool, in effect | 21:13 |
bart416 | berndj, yeah but a good stm top doesn't come cheap :( | 21:21 |
bart416 | *tip | 21:22 |
berndj | oh - i was assuming DIY STM with broken lightbulb filament probe! | 21:22 |
bart416 | Those won't do well for atom moving I fear | 21:23 |
berndj | no? | 21:23 |
berndj | do you need different / better tips for atom moving than for imaging? | 21:25 |
bart416 | well, moving an atom is a lot harder | 21:29 |
bart416 | Precision becomes more important | 21:30 |
azonenberg | http://i.imgur.com/DljuL.jpg | 22:41 |
azonenberg | Colorized version of another etched die (G3) | 22:41 |
azonenberg | green is unetched Si, blue is etched down by around five microns | 22:42 |
azonenberg | yellow is hardmask that wasn't completely stripped, plus possibly other surface contamination | 22:42 |
--- Thu Jul 7 2011 | 00:00 |
Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!