In my experience, measured oxide thicknesses never match those from
textbooks perfectly.
Here are two reasons; there may be more:

1. There is an initial native oxide thickness before you even before you
push the wafers in.
When you load the wafers into the boat, heat from the furnace warms the
wafers, thickening the native oxide.
Even during pushing the wafers in with dry nitrogen flowing out, there is
probably some oxygen and water in the air near the wafers, which grows the
oxide a little more.

2. A higher oxygen pressure will result in a larger concentration gradient
across the oxide thickness and faster growth during the diffusion-limited
(later; parabolic) portion of the growth time.
The pressure is often not monitored in "atmospheric-pressure" furnaces, but
the pressure is something other than 760 Torr.
If you have 5 lpm O2 flowing in the back and out the front, the pressure
inside the tube is must be higher than atmospheric pressure.
How high depends on the pressure drop out the port in the front.

The good news is that oxide thicknesses tend to be repeatable for the same
furnace under the same conditions, including an HF dip prior to oxidation.

    --Kirt Williams

----- Original Message -----
From: "ÇÇ´óÓÂ" 
To: 
Sent: Thursday, March 30, 2006 6:46 AM
Subject: [mems-talk] Question about the Deal Grove model


> I am doing dry oxidation and use Deal Grove model to predict the thickness
of
> thermal oxide.The oxidation temperature 1000 C and the oxidation time is
100 min.
> As refered from text book, the parabolic rate constant B is 0.0117u2/h,
the linear
> rate constant B/A is 0.165u and the time shift ¦Ó is 0.37h.
> The Deal Grove model predicts a thickness of about 0.092u for a 100 min's
dry
> oxidation on (100) wafer.
>
> In my experiment, the O2 gas flow is 5 l/min, and the measured thickness
is about
> 0.075u. There is large deviation between the predicted and the measured
thickness.
> I wonder if anybody can provide an explanation about this kind of
deviation,
> thanks in advance.