Conversation with zack112358 at Thu 13 Dec 2007 09:01:09 PM CST on kanzure (aim)

(2007-12-13 21:01:09) kanzure: hey, any idea how to make flash memory?
(2007-12-13 21:01:17) zack112358: Nope. Why/
(2007-12-13 21:01:28) zack112358: s/\//?/
(2007-12-13 21:01:56) kanzure: I've been tracking down more information on semiconductor manufacturing techniques. It's reasonable to be able to make circuits via lithography in my own home; so I'm interested in figuring out how to make flash ROM or something, a quick project to see if I can make a circuit to store memory.
(2007-12-13 21:02:18) kanzure: Reasonable as in, 10 mm size (maybe)
(2007-12-13 21:02:26) zack112358: "Reasonable to make circuits via lithography in my own home…?"
(2007-12-13 21:02:38) zack112358: Interesting.
(2007-12-13 21:03:02) zack112358: I'd be curious to know how you're doing the doping steps.
(2007-12-13 21:03:22) kanzure: Doping sounds like one of the easier steps
(2007-12-13 21:03:26) kanzure: as long as I can acquire the chemical washes
(2007-12-13 21:03:28) kanzure: yes?
(2007-12-13 21:03:57) zack112358: Good luck finding a good furnace?
(2007-12-13 21:04:06) kanzure: Well, I'm going to be ignoring that part for the moment
(2007-12-13 21:04:25) kanzure: But I do have plans for that - did I show you the cool project where someboy was able to get up to 1k Fahrenheit by channeling the sun?
(2007-12-13 21:04:27) zack112358: It's called thermal diffusion for a reason, sir.
(2007-12-13 21:04:44) kanzure: That's nearly enough heat to get sand into silicon
(2007-12-13 21:04:47) zack112358: I do not know that the same doping processes work at large scale.
(2007-12-13 21:04:50) kanzure: Unless you refer to another step in the process?
(2007-12-13 21:04:57) zack112358: Yeah, doping.
(2007-12-13 21:04:59) zack112358: Needs heat.
(2007-12-13 21:05:18) zack112358: I'm also curious as to how you'll be doing layer deposition.
(2007-12-13 21:05:33) zack112358: Actually, I'm rather curious as to how you'll be doing… just about all of it.
(2007-12-13 21:05:50) zack112358: I mean, are you going to acquire CVD hardware?
(2007-12-13 21:05:50) kanzure: Layer deposition seems ok with me
(2007-12-13 21:06:04) kanzure: Chemical vapor deposition is for making the wafers, right?
(2007-12-13 21:06:07) kanzure: Hold on, back to layer deposition
(2007-12-13 21:06:08) zack112358: No.
(2007-12-13 21:06:15) zack112358: It's for layer deposition.
(2007-12-13 21:06:32) zack112358: The wafers are sliced out of a monosilicon ingot.
(2007-12-13 21:06:36) kanzure: The lithography part: make some big, bulky masks. Shine a laser through it. Get the condenser and reducer lenses in the right place. They don't have to be all that great, but enough. Then shine the laser through on to the photoresist layer on top of the wafer.
(2007-12-13 21:06:39) zack112358: Usually Czochralski or Float Zone method.
(2007-12-13 21:06:42) kanzure: yes
(2007-12-13 21:06:45) kanzure: "ingot pulling "etc.
(2007-12-13 21:07:38) zack112358: I mean, the lithography part is the easiest. The layer deposition is probably among the hardest, to my mind.
(2007-12-13 21:07:57) zack112358: I mean, where do you even get silane?
(2007-12-13 21:08:01) kanzure: I always thought that layer deposition would just mean processing the wafer after a lithographic run
(2007-12-13 21:08:11) zack112358: Uh, no.
(2007-12-13 21:08:22) zack112358: It is how you process a wafer before a lithographic run.
(2007-12-13 21:08:35) kanzure: by applying a liquid layer of, say, photoresist ?
(2007-12-13 21:08:36) zack112358: Also, "just processing" is a dodge. The whole thing is "just processing."
(2007-12-13 21:09:12) zack112358: Okay. You need to put down a layer of pure polycrystalline silicon (because there's no way you're going to pull off epitaxy on a hobbyist's budget)
(2007-12-13 21:09:37) zack112358: How do you plan to do this?
(2007-12-13 21:09:39) kanzure: And this would be the "base" that you put on just before you go through a run with the photolithography laser, right?
(2007-12-13 21:10:00) kanzure: By hand, at first. I think it's OK as long as I keep it all very well decontaminated, and it would be allowed to take as long as I want, right?
(2007-12-13 21:10:00) zack112358: …
(2007-12-13 21:10:09) kanzure: Please excuse me, I am still learning
(2007-12-13 21:10:19) zack112358: "by hand"?
(2007-12-13 21:10:40) kanzure: Don't I need to confirm my methods?
(2007-12-13 21:11:02) zack112358: I think you're missing the point. Are your hands fireproof?
(2007-12-13 21:11:09) zack112358: I mean, the easiest approach involves heating your wafer to a few hundred degrees in a rarified silane atmosphere, I believe…
(2007-12-13 21:11:22) zack112358: Although I don't recall exactly.
(2007-12-13 21:11:25) kanzure: heating it to make a new layer stay?
(2007-12-13 21:11:39) zack112358: Heating it to decompose the silane on the surface, thus depositing silicon.
(2007-12-13 21:11:43) kanzure: ah
(2007-12-13 21:11:51) kanzure: So, I haven't seen that part yet
(2007-12-13 21:11:55) zack112358: How much are you trying to match…
(2007-12-13 21:12:02) zack112358: That's the part that actually involves expensive toys.
(2007-12-13 21:12:18) zack112358: How much are you trying to match the industrial process?
(2007-12-13 21:12:35) kanzure: It's all expensive. The articles I've been reading say that most of the components for billionaire-companies are "40,000 pound components"
(2007-12-13 21:12:43) kanzure: I am not sure, surely I do not need to match anything from the last 30 years
(2007-12-13 21:12:51) kanzure: I mean, expecting 250 nm on my own is ridiculous
(2007-12-13 21:12:58) kanzure: I'm OK with 1 mm, even 10 ;)
(2007-12-13 21:13:02) zack112358: No, I mean in technique, not quality.
(2007-12-13 21:13:12) zack112358: I'm not sure that anyone knows how to do 10mm.
(2007-12-13 21:13:35) kanzure: What did they start with? I heard that Intel and friends started out with doing this stuff in wax and with camera lenses from a shop "down the street" from their garage.
(2007-12-13 21:13:43) kanzure: I haven't explored it enough to know if I want to do a 100% matchup with industrial standards
(2007-12-13 21:13:52) zack112358: My guess would be that the largest you're going to do is not large.
(2007-12-13 21:13:56) kanzure: I'm sure they have lots of redundancy and testing involved that I would have to omit
(2007-12-13 21:14:10) zack112358: I mean, yes, this is possible, right? But you need to figure out how to deposit a layer onto your wafer,
(2007-12-13 21:14:21) zack112358: how to arrange for regions of the silicon to be p- or n- type,
(2007-12-13 21:14:26) kanzure: right
(2007-12-13 21:14:42) kanzure: for a test I'd be happy with just a transistor, that NPN junction
(2007-12-13 21:14:43) zack112358: how to do photoresist/etch (shouldn't be hard)
(2007-12-13 21:15:12) kanzure: yeah, that part would just be a few bins of chemicals or something .. and some way of confirming that the area is sufficiently clean?
(2007-12-13 21:15:16) kanzure: I have no idea how to make a clean room. heh'
(2007-12-13 21:15:26) zack112358: and possibly how to do metallization and glassivation, although I think you probably don't care about glassivation.
(2007-12-13 21:15:51) zack112358: I would suggest lots of air filters, an immaculate room, and running a large-scale process.
(2007-12-13 21:16:06) zack112358: I mean, 500 micrometers seems reasonable.
(2007-12-13 21:17:07) kanzure: sure, sure
(2007-12-13 21:17:15) kanzure: I haven't done any calculations yet, but maybe DVD RW lasers will be sufficient?
(2007-12-13 21:17:18) kanzure: I guess it depends on the photoresist
(2007-12-13 21:17:19) zack112358: And not amazingly particulate-sensitive.
(2007-12-13 21:17:22) kanzure: and I should be focused on the heating process
(2007-12-13 21:17:27) kanzure: for making the layers permanent
(2007-12-13 21:17:28) zack112358: Uh… I'm not sure why you think you want lasers.
(2007-12-13 21:17:32) kanzure: lasers for the lithography?
(2007-12-13 21:17:36) kanzure: There's two ways, that I know of
(2007-12-13 21:17:37) kanzure: (1) lasers
(2007-12-13 21:17:39) kanzure: (2) EM beams
(2007-12-13 21:17:40) zack112358: Yeah, why?
(2007-12-13 21:17:43) kanzure: erm, E beams
(2007-12-13 21:17:44) kanzure: not EM beams :)
(2007-12-13 21:17:50) zack112358: (0) a lamp
(2007-12-13 21:17:55) kanzure: that's enough?
(2007-12-13 21:18:16) zack112358: I mean, the reason lasers are nifty is that you can pull amazing precision stunts with them.
(2007-12-13 21:18:25) zack112358: If you're doing 500 micrometer,
(2007-12-13 21:18:31) zack112358: which is, what, 1 mm wide traces,
(2007-12-13 21:18:35) zack112358: you don't care.
(2007-12-13 21:18:37) kanzure: I guess I'm asking if the wattage on the lamp is enough
(2007-12-13 21:18:48) kanzure: which depends on the photoresist's properties
(2007-12-13 21:19:55) zack112358: Well, for the price of a spectacularly shitty laser you should be able to make or buy a few kilowatts of arclamp.
(2007-12-13 21:20:00) zack112358: Is my guess, anyway.
(2007-12-13 21:20:05) kanzure: So, the chamber for heating
(2007-12-13 21:20:11) kanzure: It sounds like I need only a few hundred Fahrenheit?
(2007-12-13 21:20:14) kanzure: That's a basic oven, yes?
(2007-12-13 21:20:15) zack112358: Also, I think you need little energy.
(2007-12-13 21:20:22) zack112358: Heating at what step?
(2007-12-13 21:20:39) kanzure: the heating in a rarified silane atmosphere
(2007-12-13 21:20:52) kanzure: which step is that?
(2007-12-13 21:20:55) zack112358: Heating is a part of a number of steps. You need to decide what processes seem appropriate.
(2007-12-13 21:20:56) kanzure: the deposition?
(2007-12-13 21:21:08) zack112358: Yes.
(2007-12-13 21:21:27) kanzure: I'd need something more than an oven, you don't want to just open up so much silane in your face
(2007-12-13 21:21:32) zack112358: Basically, you want to not do CVD if you can, but I don't know of any other way.
(2007-12-13 21:21:47) zack112358: Well, okay, first of all, if you fuck up with silane, you die.
(2007-12-13 21:22:03) zack112358: It's pyrophoric down to 20 celsius.
(2007-12-13 21:22:06) kanzure: http://en.wikibooks.org/wiki/Microtechnology Other methods:

Overview
Oxidation
Chemical Vapor Deposition (CVD)
Physical Vapor Deposition (PVD)
Epitaxial Growth
Electrochemical Methods
Spinning
Surface Functionalization
(2007-12-13 21:22:23) zack112358: Ooh, you might try electrochem. Don't know how well that works.
(2007-12-13 21:22:36) kanzure: Would that be just a few electrodes in a liquid solution?
(2007-12-13 21:22:41) kanzure: and pass a current through it?
(2007-12-13 21:22:43) zack112358: Hopefully?
(2007-12-13 21:22:45) zack112358: Don't know.
(2007-12-13 21:22:46) kanzure: heh
(2007-12-13 21:22:51) zack112358: Of course, your wafer is, you know, insulative.
(2007-12-13 21:22:55) zack112358: Could make that hard.
(2007-12-13 21:22:57) kanzure: http://en.wikibooks.org/wiki/Microtechnology/Additive_Processes#Electrochemical_Methods
(2007-12-13 21:23:04) kanzure: electroplating, electroless depositions
(2007-12-13 21:23:12) kanzure: http://en.wikibooks.org/wiki/Microtechnology/Additive_Processes#Electrochemical_Methods
(2007-12-13 21:23:18) kanzure: oops, I meant: http://en.wikipedia.org/wiki/Electroplating
(2007-12-13 21:24:14) kanzure: I could have easy success with electroplating, since electrolysis is common throughout many chemistry classes, suggesting lots of publications on the topic
(2007-12-13 21:24:18) zack112358: I don't know whether or not you'll have any luck deposition silicon in solution.
(2007-12-13 21:24:26) zack112358: Well, see again that your wafer is insulative.
(2007-12-13 21:24:27) kanzure: have to find the right solution
(2007-12-13 21:24:48) zack112358: The problem is that you're building an electric circuit. Which, by definition,
(2007-12-13 21:24:52) zack112358: has non-conductive bits.
(2007-12-13 21:25:01) zack112358: So I think electroplating is out, unless you're very clever.
(2007-12-13 21:25:43) kanzure: I always thought that it would be as simple as a chemical wash, and the introduced chemicals would wash away everything that wasn't heated by the lamp/laser/ebeam
(2007-12-13 21:25:43) zack112358: The rest of the processes are fairly nontrivial, as they're all gas-phase, I believe.
(2007-12-13 21:25:53) zack112358: Nope.
(2007-12-13 21:26:03) zack112358: The light is not heating.
(2007-12-13 21:26:11) kanzure: Couldn't it be that simple? Find something that, with enough energy, would bond (energy transfer via the lamp)
(2007-12-13 21:26:17) zack112358: The modern process puts down a resist that destroys itself under light.
(2007-12-13 21:26:36) zack112358: If you can find a reasonable way to heat something to a few thousand degrees locally, more power to you.
(2007-12-13 21:26:42) kanzure: hah
(2007-12-13 21:26:53) zack112358: You realise that silicon is incredibly heat resistant.
(2007-12-13 21:27:15) kanzure: I was thinking of the opposite of what the modern process is doing: put down a resist that bonds under light, and everything else can be washed away with a common "supplement" that reacts with it, maybe making a powder that I can later wash away
(2007-12-13 21:27:23) kanzure: that would require some very creative chemical engineering, I suspect
(2007-12-13 21:27:37) zack112358: I mean, that's the litho process. Do the modern litho process, it's trivial.
(2007-12-13 21:27:45) zack112358: You still need to figure out how to deposit silicon.
(2007-12-13 21:28:08) kanzure: hmm
(2007-12-13 21:28:16) kanzure: Gas-phase sounds like a good idea right about now ;)
(2007-12-13 21:28:17) zack112358: I mean, at that scale, you can have it machined. The actual patterning is not going to be hard.
(2007-12-13 21:28:38) kanzure: CNC machined, perhaps
(2007-12-13 21:28:41) zack112358: Yeah.
(2007-12-13 21:29:32) zack112358: But layer deposition… you'll have to look at CVD processes, I guess, if you're going to do gas-phase. PVD would be nastily hard, methinks.
(2007-12-13 21:29:42) zack112358: I don't even want to know what the boiling point of silicon is.
(2007-12-13 21:29:49) kanzure: It's somewhere up at 1200 Fahrenheit
(2007-12-13 21:30:07) kanzure: oops
(2007-12-13 21:30:10) kanzure: That's melting point
(2007-12-13 21:30:14) zack112358: Nope.
(2007-12-13 21:30:18) zack112358: That's not even melting.
(2007-12-13 21:30:23) zack112358: Melting at 2577 F
(2007-12-13 21:30:32) zack112358: Boiling at 5909 F
(2007-12-13 21:30:34) zack112358: says Wiki.
(2007-12-13 21:30:43) kanzure: Oh shit
(2007-12-13 21:30:47) zack112358: *nods*
(2007-12-13 21:30:53) zack112358: Silicon is insane.
(2007-12-13 21:31:04) kanzure: I have no idea how to get that hot.
(2007-12-13 21:31:12) zack112358: I mean, the Czochralski process? Destroys the crucible.
(2007-12-13 21:31:19) kanzure: * I have no ideas on how to get up to those temperatures
(2007-12-13 21:31:27) kanzure: It destroys the crucible??
(2007-12-13 21:31:36) zack112358: It's thrown out and replaced at each step. The heating is RF because any heating coil couldn't take it.
(2007-12-13 21:32:01) zack112358: That's why you want to do CVD… there are gas-phase silicon-containing chemicals that decompose at reachable temperatures.
(2007-12-13 21:32:05) kanzure: I've seen some silicon heating furnace plans that have electrodes *inside* and it's most definitely not RF
(2007-12-13 21:32:05) zack112358: Silane only needs… hmm.
(2007-12-13 21:32:09) kanzure: Would you recommend RF for silicon heating?
(2007-12-13 21:32:30) zack112358: RF heating is in the "if you make a tiny mistake everyone dies" category.
(2007-12-13 21:32:47) kanzure: I'd start with other RF projects, methinks
(2007-12-13 21:32:54) kanzure: like .. directed antennae ;)
(2007-12-13 21:32:59) zack112358: I'd recommend not doing it.
(2007-12-13 21:32:59) kanzure: or maybe RF on lasers?
(2007-12-13 21:33:05) zack112358: See also, dangerous, massively illegal.
(2007-12-13 21:33:13) zack112358: (RF heating)
(2007-12-13 21:33:55) zack112358: Silane only needs 420 C to decompose to silicon and hydrogen.
(2007-12-13 21:34:10) zack112358: So that's, what, 700F wafer temperature?
(2007-12-13 21:34:13) zack112358: That's plausible.
(2007-12-13 21:34:23) zack112358: But see also, high heat in the same process as pyrophoric gases.
(2007-12-13 21:34:33) zack112358: Are you absolutely sure you don't have leaks?
(2007-12-13 21:34:38) zack112358: I mean, what it boils down to is,
(2007-12-13 21:34:46) zack112358: the internet doesn't have a good way to ask it
(2007-12-13 21:35:01) zack112358: "What are good silicon CVD processes for the hobbyist"
(2007-12-13 21:35:13) zack112358: so I would suggest talking to a material scientist.
(2007-12-13 21:35:33) kanzure: yes, clearly
(2007-12-13 21:35:41) kanzure: also, good ways to detect leaks include colored, nontoxic gas, I think
(2007-12-13 21:35:48) kanzure: they use this in aerodynamics and wind tunnels
(2007-12-13 21:36:17) kanzure: you'd have to find a gas that's close enough, of course, otherwise you're just fooling yourself
(2007-12-13 21:37:38) zack112358: I mean, this is definitely a do-outdoors-away-from-burnable-things project.
(2007-12-13 21:37:45) zack112358: I would suggest a very disposable shed.
(2007-12-13 21:37:49) kanzure: haha
(2007-12-13 21:38:01) kanzure: I am not sure if that would be a good idea either
(2007-12-13 21:38:07) zack112358: I mean, the point is, if you screw up handling the silane, run the fuck away, it's all going up in flames.
(2007-12-13 21:38:11) kanzure: because of the heat buildup, I don't know if I can deisgn a cooling system that would be sufficient
(2007-12-13 21:38:12) kanzure: *design
(2007-12-13 21:38:29) zack112358: If you screw up handling the 800F heat, run the fuck away etc.
(2007-12-13 21:38:33) kanzure: yes
(2007-12-13 21:38:36) zack112358: Uh, a big fan?
(2007-12-13 21:39:24) kanzure: That'd work
(2007-12-13 21:39:33) kanzure: I need to figure out where I can find somebody willing to help me out on the design of this thing
(2007-12-13 21:39:37) kanzure: as you said, a material scientist
(2007-12-13 21:39:46) zack112358: Yup.
(2007-12-13 21:39:48) kanzure: but who would have the guts ? I mean, this is kind of deadly
(2007-12-13 21:39:59) zack112358: Well, if you fuck up these processes. It may well be the case
(2007-12-13 21:40:12) zack112358: that there's some relatively safe process that just isn't good for industrial processes.
(2007-12-13 21:40:17) zack112358: For whatever reason.
(2007-12-13 21:40:26) zack112358: But you would have to talk to someone in the field to find out about that.
(2007-12-13 21:41:01) kanzure: Oh!
(2007-12-13 21:41:03) kanzure: I just remembered something
(2007-12-13 21:41:08) kanzure: I was talking with my father about this earlier tonight
(2007-12-13 21:41:08) zack112358: mrrh?
(2007-12-13 21:41:15) kanzure: and he mentioned that he remembrs back in college how they were doing it back then
(2007-12-13 21:41:27) kanzure: heh, they were literally setting out their wafers in the sun to "dry" their etch in and so on
(2007-12-13 21:41:44) kanzure: at the scales they were working at, it worked
(2007-12-13 21:41:58) zack112358: Actually, we have a chip fab here at the university. I should find out how we do it.
(2007-12-13 21:42:00) kanzure: obviously that isn't useful for depositing the atoms that are specifically needed, but still interesting
(2007-12-13 21:42:36) kanzure: It's worth investigating. I've seen a community college setup up in Austin, and it seemed very basic: a few carted machines, no more than ten steps. Of course, I doubt they were pulling their own ingots ...
(2007-12-13 21:42:55) zack112358: I mean, you won't be. You'll want to buy wafers.
(2007-12-13 21:43:23) kanzure: True, the other steps are hard enough for the moment
(2007-12-13 21:43:27) zack112358: Mhmm.