Semiconductor fabrication question

I got to thinking a bit sideways on this. You can probably achieve almost all the methods involved in early semiconductor processing (this will take you up through early ion implantation techniques) if you partner with a local high school or college. All of them are still in use.

It is not so much what the material you convert to vapor but the nature of the solid condensation process that determines the outcome.

1969, mesa devices, that is pretty out of date.

Not necessarily. You must also consider the crystal pitch (atomic spacing in the crystalline form). If the substrate pitch in not very similar to what you want to deposit, you get seriously many crystal lattice imperfections. And if tempco of expansion is dissimilar, trouble results from thermal cycling.

Sounds like a scheduling issue. Tough luck.

Try this:

also various elements by melting point:

Not perfect nor complete.

It confuses operating temperatures with process temperatures.

Solid substrate dopant carriers is a well established technology. But yes, it does suffer from limited usage lifetime, and the exposure time corrections that result.

You have me stumped on that one. Why would a diffusion oven be so dangerous? How dangerous are we talking about here? I was thinink in the range of 1e-4 torr vacuum with heated material. I guess that could be dangerous, maybe, slightly, I don't know. Perhaps if I'm clumbsy and punture a hole in the steel vacuum chamber.

Thanks, Anon

Actually, in olden times, diffusion ovens used to run near atmospheric pressure. They still run at 10 to 100 torr at times. The issue is not so much the ovens themselves (which often ran at yellow heat ~1300 C) but the often very toxic gasses used for dopants, like arsine.

in

That's true. I've always had the thought of using boron. I won't play with arsine.

I'm still trying to decide the method of creating the vacuum, since I want to build everything myself. No cheating by buying everything except for things I can't build such as motors. Correct any error, but I figured 1e-4 torrs is good enough to diffuse amorphous silicon. That's a mean path of 100cm. Maybe that's not low enough for silicon. Maybe too many air particles will bond with the silicon. I don't know. As far as diffusing metal, 1e-4 torr is plenty low enough since it's not a big deal if some air particles bond with the metal.

Well, the diffusion pumps sound good, but liquid nitrogen to prevent the oil backsplatter is out of question. After some thoughts as to how rotary vane pumps work, it seems two in series would have no problem obtaining 1e-4 torr. Each pump decreases the pressure by 10000 times, and is limited by the leakage of air because of the difference in pressure from out to in. The second stage pump will see a pressure about 8e-2 torr thanks to the first stage pump, and should decrease the pressure to about 8e-6 torr, or at least that's what I think.

Thanks, Anon

Just the same, Wikipedia has a nice starter article on vacuum pumps. One issue is at pressures below about 10^-2 torr rotary vane pumps do not work so good. Nor do you need that low of a pressure for other than ion implanting. And for that you will need an ion pump to get below 10^-6 torr.

1

nk in

d y s

That's true, a first stage rotary vane gets down to about 1e-2 torr. Although I've never read any experiments that show what a *two* stage rotary vane pump will achieve. I'm thinking that the seals in the pump are the limitation, which is relative to the input versus output pressure. As far as the second stage pump is concerned the highest pressure it experiences is about 1e-2 torr. What if the earths pressure was 1e-2 torr. If a single pump at normal pressure can decrease the pressure by 10000 times, then couldn't a pump in 1e-2 pressure also decrease by 10000? I've read about 2 stage rotary vane pumps, but haven't seen any data.

OK, you said 1e-2 is fine, but are you referring to silicon deposition? If the mean path is 100 cm, and it travels at 2cm, then is the odds of striking an air molecule 2% and 2/20 =3D 0.1% for oxygen, no? Oxygen and silicon are highly attrated to each other. If there's

1 oxygen per 1000 silicon atoms in the diode then that some massive heavy doping. It's probably lower than 0.1%, but even 1 out of a million could be a lot.

Thanks, Anon

Actual doping rates in useful semiconductors are well below 0.1% often in the parts per million range and lower. Keep plenty of argon tanks handy.

Please also read:

The part in parenthesis are for disambiguation.

but

st 1

inink in

ould

mbsy

ric

e is

300 t . w

es,

I'll "keep plenty of argon tanks handy."

A WTi alloy is pretty robust. The best diffusion barrier I'm aware of is reactive sputtering of WTi with N2 (a 200nm layer withstands

1000hr@400°C without degrading). I wonder what the proposed structure is supposed to accomplish though. Also I'm not exactly sure about what the OP means with "ohmic contact". "Ohmic" sounds like "non-semiconductor" which sounds like "conductor" which sounds like "metal" which leaves the poly-metal contact the only electrically active interface in his device.

Why sputter poly on top of metal rather than the other way around?

robert