Thickness Uniformity In Dry High Temperature Oxidation Process??

[diffeng]

I'm trying to increase my thickness uniformity in SiO2 deposition for a dry, high temperature oxidation process. My wafer profile is thin in the middle and thicker towards the outsides. The film is thin (~100-50 Angstroms) and the proces temperature is relatively low (~780). It appears to be a heat transfer problem as the outside of the wafer heats faster than the center (higher temperature, faster reaction). Anyone have any suggestions? Any help would be greatly appreciated!!

 

[IRstuff]

Is this for school?

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[diffeng]

Nope, it is for a process at work. Just trying to improve it.

 

[IRstuff]

There are a number of things that might affect oxidation rate:

>> temperature: have you actually measured the temperature profile of the wafer
>> reactants: does your process provide sufficient reactants to prevent depletion of same in the middle of the wafer?
>> access: does your process provide equal access to different parts of the wafer?
>> initial conditions: do your wafer start out uniform with their native oxide and are the same porosity across the wafer?
>> surface prep: is your surface preparation resulting in uniform surface condition?

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[IRstuff]

Have you asked the other process engineers at work?

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[diffeng]

I've talked with some other engineers. They haven't had much success in reducing range in the past. I'm trying to be a hero . Due to the equipment, measuring wafer temperature profile would be difficult (if not impossible). Equipment uses a vertical "boat" of 100+ wafers in a heated tube. I have temperature measurements between vertical zones (i.e. top, top-center, center, etc.) but the wafers themself are somewhat of a mystery. Depletion is not an issue. Uping concentration of reactants did not have much effect. Access is a possibility. Wafers are pretty close so you may experience some blocking and vertical radiation effects. I will try to run an experiment with leaving some slots empty. Porosity should be equal but I will definitely look into that. I forgot to mention that this is not really a growth but rather a deposition. The layer already down is a nitride and does not react. The oxidation is mainly a surface deposition reaction (i.e. does not consume the wafer). I believe we measure surface uniformity prior to the process but will check again. Thanks for your help. Sometimes it is good just to get outside opinions and use the "community" for ideas.

 

[IRstuff]

How long does the boat sit before you start the deposition? Temperature uniformity can be affected.

How fast is the deposition? The slower the deposition, the more uniform; supposedly, anyway, since that's one way to mitigate depletion.

If this is a CVD, then pressure also affects the uniformity; not sure why, myself, I wasn't involved in that part of the process. LP works better than atmospheric. We were able to switch from a planar APCVD to diffusion furnace LPCVD with boats, etc., and got better uniformity to boot.

Nitrides can be a funny, in a bad way, material. We had a nitride film deposition that refused to be uniformly annealed unless it was cut into thin strips, despite the face that the strips were something like 20 microns across, and the film was only 400[&angstrom;] thick. You may need to do sone nonstandard analysis of the nitride layer to determine whether there's a difference in dangling bonds, or whatever.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[BrentShepard]

Greetings

I realize I may be a little late on this (just joined/first post) but I'll assume you can't answer process questions so I'll just think out loud.

if we are talking about a tube operation (diffusion tube)then the item to be oxidized should be loaded in to your hot zone in an N2 environment, and because you are looking for a precise layer, give it 20 to 30 minutes to warm up before turning on O2.

If (in a diffusion tube)you already are loading in N2 then 1 of 3 thing are a problem, 1:tube has a leak (causing ambient O2 to start the process), 2: N2 flow to high (Not likely, but should be looked at), 3: N2 flow to low (not keeping out ambient O2 and more likely).

If this process is being done in something like a kiln then, you get what you get (opening the door/access port/etc allows O2 to come in with the samples).

Hope it helps

Brent

 

[BrentShepard]

Another thought


I had a situation where I had to control a very precise Boron/shallow diffusion, I used an O2 descum tool to give me about 200A on oxide to slow things down.

If you have access to an O2 plasma cleaner, oxide thicknesses can be very well controlled using this method.

Just a thought

Brent