Click
here to download a PDF version of this Bulletin
Selecting the proper gas flow rate and
gas composition for a given diffusion system and diffusion
process is critical to the successful performance of the BoronPlus®
and PhosPlus® dopant sources. This bulletin was therefore
prepared to provide the diffusion engineer with the appropriate
general guidelines necessary to select the parameters which
will give the best results for his system. The bulletin also
describes a procedure which eliminates the need for a separate
low temperature oxidation cycle when BoronPlus sources are
used as the dopant source.
The gas flow rate for a given diffusion
system is primarily determined by the diameter of the diffusion
tube. Typical flow rates can be selected from the curve plotted
in Figure 1.
Generally,
the gas flow rate should be as low as possible, but high enough
to prevent air from backstreaming down the diffusion tube.
Exceptionally high flow rates are normally avoided during
the time at the deposition temperature since high rates tend
to produce non-uniform sheet resistivities at the gas inlet
(source) end of the boat. This is because the rapidly-flowing
gas carries the B2O3 or P2O5
away from the diffusion source and down the diffusion tube
before it has a chance to deposit on the silicon wafer located
next to the source.
Backstreaming of room air down the diffusion tube must not
be allowed to occur or non-uniform sheet resistivities will
be observed at the handle (load) end of the boat. This is
because the excess oxygen grows a relatively thick oxide film
on the silicon wafers which masks off the deposited B2O3
or P2O5 from the diffusion source. One
way in which backstreaming can be significantly reduced in
many diffusion systems is to use a high gas flow rate (10-15
l/min) whenever the end cap is off the end of the tube (i.e.,
during insertion and removal of the boat). The flow rate should
then be turned down to the recommended rate when the end cap
is replaced.
A high gas flow rate, however, is not
very effective in preventing backstreaming in large diameter
diffusion tubes (i.e., 200 mm). It has been reported that
the nitrogen tends to flow to the top of the tube and the
air tends to backstream down the bottom of the tube, regardless
of the flow rate. Consequently, sufficient time should be
allowed at the insertion temperature after the tube is closed
so that the nitrogen can purge the tube of backstreamed air
before ramping to the deposition temperature. Also, since
cantilever systems are usually used with large diameter tubes,
care should be taken to ensure all excess areas in the end
cap around the cantilever arms are closed.
Most diffusion systems use nitrogen as
the carrier gas in the diffusion tube. Some diffusion engineers
prefer to use argon gas when depositions are to be made above
about 1050°C because argon is an inert gas and does not
react with the silicon surface to cause nitride pitting.
The
proper use of oxygen with the selected carrier gas can have
a significant effect on silicon surface damage. The amount
of oxygen that must be blended into the carrier gas tends
to increase with increasing deposition temperature. Generally,
some oxygen must be used at temperatures above about 1000°C
while little or no oxygen is required below this temperature.
Figure 2 can be used as a guide to select an oxygen content
which normally minimizes silicon surface damage when depositions
are being made with BoronPlus or PhosPlus sources.
Care must be taken not to use too much
oxygen during the deposition cycle. If this occurs, an oxide
film will grow on the silicon surface that is thick enough
to mask off most of the B2O3 or P2O5
, resulting in non-uniform doping of the silicon wafer.
Some phosphorus emitter diffusion
processes done with gas-type dopant sources often include
an oxidation of the silicon surface immediately following
the phosphorus deposition. This can also be done with the
PhosPlus solid source system if the oxidation is done with
dry oxygen. Dry oxygen can be used with the PhosPlus sources
at any temperature since oxygen has no effect on the sources.
However, the oxidation cannot be done at the deposition temperature
with wet oxygen or with steam since the moisture will quickly
deplete the PhosPlus sources of available P2O5
.
Boron silicide is the thin, metallic compound
that forms under the deposited glassy film during a boron
deposition cycle. Synonymous names are boron-rich phase, boron-silicon
phase and silicon stain. The phase is insoluble in HF, and
it can be detected after the HF etch by the hydrophilic (wetting)
silicon surface compared to the normal hydrophobic (non-wetting)
surface of undoped silicon wafers.
Boron silicide is beneficial to silicon
processing because it produces uniform sheet resistivities
in the doped silicon slices and because it can be used as
a limited source of boron in certain drive cycles. However,
this phase is removed most of the time before subsequent processing
steps are taken so that it does not become a source of additional
problems later.
The Traditional LTO Cycle:
The most common method of removing the boron silicide phase
is to use the low temperature oxidation (LTO) cycle. This
technique involves stripping the glass from the doped silicon
slices and then re-inserting them into the diffusion furnace
without the sources being present. Holding the silicon slices
at 800°C for 20-30 min in steam or in wet oxygen is usually
sufficient time to oxidize a thin layer of this phase. The
new oxide layer is then etched off the silicon surface in
dilute HF before continuing the processing of the silicon
wafers. The LTO cycle tends to raise the sheet resistivity,
but it does not normally destroy the uniformity of the doped
silicon slices.
The In-situ LTO Cycle:
An important variation of the above LTO cycle that can be
used in the presence of the BoronPlus sources is to start
oxidizing the silicon with pure oxygen either during the cool-down
portion of the deposition cycle or after the boat has reached
the removal temperature [1]. Fig. 3 shows
schematically how the in-situ LTO eliminates a processing
step during a boron deposition cycle.
Although
cooling in oxygen has eliminated the need for a subsequent
separate LTO process, it can affect the uniformity of the
doped silicon wafers. This is because the oxygen must diffuse
through the deposited glass before it can oxidize the boron
silicide, and any non-uniformity in the thickness of that
glass will affect the rate of the oxidation of the silicon
surface below it. Experience has shown that this process usually
works very well for high temperature isolation and pnp emitter
processes, but it is not always acceptable for low temperature
base diffusions where uniformity is much more important.
Wet oxygen or steam cannot be used in
this modified LTO process since the moisture will quickly
deplete the BoronPlus sources of available B2O3.
However, dry oxygen can be used at any temperature, since
oxygen has no effect on the sources. In this case, the next
run will be the same as the one that has just been completed.
Uncontrolled amounts of moisture entering
the diffusion tube can cause silicon surface damage, non-uniform
doping of the silicon, etc. However, when moisture is formed
in the diffusion tube at very low levels and in a controlled
manner in the presence of GS-126 BoronPlus sources as described
in Product Bulletin 310, uniformity of doping can be improved
and excellent devices can be produced. The work done at the
University of California, Berkeley showed how the technique
can be used to produce p-type source/drain diffusions in a
CMOS IC that are equivalent to those produced with ion implantation
[2].
Optimum doping results are obtained from
the BoronPlus and PhosPlus sources when the proper carrier
gas flow rates and carrier gas oxygen concentrations are selected.
The ability to use the sources in the presence of oxygen provides
greater flexibility during the deposition process since oxygen
has no effect on their performance. This aspect is particularly
useful with the BoronPlus sources, where they may be used
during an in-situ LTO cycle to oxidize the boron-silicon phase.
The in-situ LTO completely eliminates the need for a separate
LTO step and can result in significant savings.
For more information on this Product Bulletin
or on the BoronPlus and PhosPlus dopant sources, contact the
Planar Dopants Team.
References:
1. J.E. Rapp, “The Planar Diffusion
Technique”, Semicon Technology Asia 1998/9, Nordica
International.3.F Block B, Quarry Bay, Hong Kong, p. 33.
2. R. Alley, P.K. Ko and K.Voros, “Characterization
of the BoronPlus Planar Dopant Source Moisture Enhanced Process”,
Thesis at University of California, Berkeley, Sept 18, 1986.
Memorandum No. UCB/ERL M86/75.
|