>
>    I  try to dope n+ into the Si trench.
>    The trench opening is ~1um, and depth is 20um.
>

This is an interesting problem. The two methods I know of doping may both
work. The first is using a solid source in a diffusion furnace. Basically
you can by these solid source wafer-like substrates that fit into a slot in
a wafer boat. These solid sources diffuse their dopant into air when heated
up. Some of this dopant lands on silicon wafers. This method should work
reasonably well.

Another technique is to use a spin-on dopant. These are typically similar to
spin on glasses. I'm not sure if there is any solid content in these
solutions. If so, the limiting factor will be the particle size. It seems as
long as the particle size is <= 0.1 μm (100 nm), you should be able to get
dopant into the trench. To make sure the dopant gets into the trench, you
could either (a) dispense a large amount of solution on a wafer and allow it
to fill everything and then spin or (b) do a two step spin, the first step
at very low speeds, the second at normal speeds.

In both cases you'll need some sort of patterned diffusion barrier to only
dope the areas you're interested. A typical barrier is a 1 μm thermal oxide
(also called a field oxide or FOX).

Bill Eaton, Ph.D.
Materials & Analysis Manager
NP Photonics
5706 Corsa Avenue, Suite 100
Westlake Village, CA  91362
Voice:  (818) 991-7044 x211
eFax:   (503) 214-5559
mailto://bill@npphotonics.com
www.parvenutech.com