1
- 2 
Click
here to download a PDF version
of this Bulletin
The benefits of using BoronPlus
and PhosPlus planar dopant sources in the production of semiconductor
devices are well known. To obtain the best results, however,
consideration must be given to a number of key processing
parameters. One of these parameters is the thermal response
of the sources when heated to and cooled from the deposition
temperature. This bulletin suggests techniques for insertion,
ramping and withdrawal of the dopant sources. Using these
suggestions, process engineers should be able to achieve the
following:
- Reduce or eliminate source warpage
- Improve sheet resistivity uniformity
- Increase source lifetime
- Reduce thermal gradients across the
silicon wafer and across the dopant source
- Increase product yields
Many factors can affect the results of
a deposition cycle. Of particular importance is the furnace
temperature profile. A very steep temperature gradient normally
occurs between the front of the flat zone and the mouth of
the diffusion tube. If the carrier is rapidly pushed or pulled
through this portion of the diffusion tube, a large temperature
gradient will form across the radius of the sources. If this
gradient is too high, the stresses that develop can warp or
break either the silicon wafer or the dopant source. The gradient
can also introduce slip planes in the silicon wafer which
usually result in decreased product yields.
To minimize these temperature gradients
and to improve the performance of the BoronPlus and PhosPlus
sources, the insertion, ramping and withdrawal parameters
outlined in Table I are suggested. These parameters should
be considered as starting points in the development of the
optimum deposition cycle for a given process.
Insertion and Withdrawal Temperatures:
The insertion and withdrawal temperatures given in Table I
will minimize warpage and are the maximum temperatures that
should be used. Lower temperatures are certainly acceptable,
however, since they will not harm the sources and they could
have other beneficial effects on the silicon (less damage)
and final device properties (higher minority carrier lifetime).
Generally, the difference between the deposition temperature
and the insertion/withdrawal temperature should be greater
than 100°C.
The insertion and withdrawal rates given
in Table I are the maximum rates that should be used when
the temperature gradient in the diffusion tube is about 50°C/inch.
If the insertion rate is too fast, breakage of the sources
could occur. If the withdrawal rate is too fast, the sources
could warp. The withdrawal rate is probably too fast if the
center of the source is observed to be red when the boat is
exiting from the diffusion tube.
Pulling the boat into an ampule (elephant)
is also recommended. The ampule decreases temperature gradients
across the sources and the silicon wafers when the boat is
out of the furnace, and it minimizes the amount of room air,
which often contains various concentrations of particulates,
from flowing between the sources and the silicon during cooling.
Many of the modern furnaces exhibit temperature
gradients which are much steeper than 50°C/inch. Consequently,
slower push/pull rates through the temperature gradient may
be necessary to prevent warpage of the sources and to minimize
silicon damage. Since these furnaces are usually controlled
by microprocessors, the diffusion engineer could consider
a program which has a rapid push to the beginning of the temperature
gradient, a slow push through the gradient, and then a rapid
push into the hot zone.
The boatload of sources should be allowed
to come to thermal equilibrium at the insertion temperature
after reaching the hot zone. This will help to insure that
both ends of the boat will reach the deposition temperature
at the same time after the furnace ramp has been completed.
The result will be an improved doping uniformity across the
boat.
The minimum recommended stabilization
times given in Table I increase with increasing source diameter
because of the greater mass of material which must be heated.
The diffusion engineer may find that even longer times are
necessary if a large number of sources are used and/or if
the furnace recovery time is slow.
1
- 2
|
