Why is oxide not used in the CVD process in the manufacture of gate oxide in the NMOS device?

[lufer17]

I know that chemical vapor deposition (CVD) is a vacuum deposition method used to produce high-quality, high-performance solid materials. The process is often used in the manufacture of semiconductors to produce thin films, there are several CVD methods, such as LPCVD, PECVD etc.
Depending on the material used, the temperature and the correct material can penetrate the substrate, creating a thin layer on the substrate used in the process.

My question

Why is oxide not used in the CVD process in the manufacture of gate oxide in the NMOS device?

I can't find the reasons and the main causes.

 

[Dominik Przyborowski]

How you want to achieve SiO2 in non-solid state and use it to deposit on any substrate?

 

[dick_freebird]

A few potential reasons come to mind - consider that I
am not a process engineer, but ate lunch with them
every day for over a decade and made it my business
to know as much as I could about what makes my parts.

CVD, "chemical" vapor deposition, stands to entrain
"stuff" into the deposited film. It's probably
hydrogen-heavy (CVD often uses "-anes" - silane,
borane, other hydrides as these can be vapor or vaporized).
High hydrogen content is bad news in hot carrier / NBTI
as the hydrogen makes the device look good, until it's
knocked loose by hot carriers or radiation and exposes
all the traps it was passivating. This makes the transistor
unreliable (or prone to "aging" as many modern nodes
are - we used to insist that devices were highly stable
to the detriment of raw performance).

Now I expect that high-K gate dielectrics such as HfO2
must be deposited - that, or Hafnium sputtered and
then oxidized. I never worked anywhere near that kind
of modern-sexy.

Grown gate ox from underlying silicon is definitely going
to be more pure.

Wet oxide still puts a lot of hydrogen but at such a high
temp that most of it is driven off in process. Dry O2, forget
about hydrogen. But that's harder to make a pretty interface
out of, with no passivant.

In order to minimize bulk traps the oxide needs to be
as quartz-like as possible. That is, it needs to form the
crystalline bonds. Deposited films tend to form grains
and have no inherent order to them. "Low and slow"
used to be the recipe for high quality gate ox although
some places like Sandia have figured out how to make
very good oxides using dry oxidation (secret sauce).
You want full densification. How to get that from a
deposited film, would be a process development project
I reckon.

Certainly a grown gate ox is the least needy of exotic
equipment and feedstocks. Hot tube and gas flow.

Uniformity is certainly going to be a concern. Control
at the low end of deposited thickness follows from that.
Grown oxides can be well controlled to arbitrarily thin
Tox (down to the room temp native oxide, pretty much).

 

[c_mitra]

SiO2 is a refractory material that is not easy to vaporize.

Silane (SiH4) is already a gas at room temperature. It decomposes at a high temp (>300C; reasonable usable temp is around 600C) and can deposit a film of Si.

That is the chemical part (thermal decomposition).

You can control the rate of deposition by the temp because the decomposition rate depends on the temp.

It is not easy to deposit SiO2 directly because I do not know any suitable vapor that will decompose and produce SiO2 directly. But it is easy to add O2 and that can oxidise the Si atoms on the surface and produce SiO2.

Because of crystal structure mismatch problems, I think thin films of SiO2 are oxygen deficient (closer to SiOx where 1.5>x>1).

 

[timof]

Thermally grown, native silicon oxide (SiO2) has an ultimate quality - minimum number of defects, good stability, good mobility for free carriers (when inversion layer is formed), etc.

It is the invention of thermally grown oxidation, making a high-quality gate oxide dielectric, than made the MOSFET a winner.

This was done by Mohamed Atalla at Bell Labs.

In hafnium based oxide, there is a think SiO2 intermediate layer, on the surface of Si, to keep defects low and quality high.

No other semiconductor material has such a nice native, high-quality, stable oxide, that's why silicon is still in use, in spite of its relatively poor carrier mobility.