Chemistry Reference
In-Depth Information
induced by the heavy doping.
207
According to this model, the etch rate decreases at a
concentration corresponding to a doping level of at which the material is
degenerated and the Fermi level moves into the valence band making silicon behave
like a metal. As a result of degeneracy the space charge layer on the silicon surface
shrinks to a value on the order of one atomic layer. The electrons injected into the con-
duction band by the oxidation steps are no longer confined to the silicon surface but
rather penetrate into the bulk where they have a high probability of recombining with
holes. As a result, these holes are no longer available for the reduction of water as part
of the etch reactions. The fourth-power decrease of etch rate is then explained by the
fact that for the dissolution of one silicon atom, four charges are required.
The electron deficiency model implies that etch stop should not occur on
n
-Si
where there are abundant electrons in the conduction band. Unable to explain the etch
rate reduction on phosphorus- or germanium-doped
n
-Si, Seidel
et al.
207
suggested that
the etch rate reduction observed on phosphorus-doped
n
-Si is caused by a different
mechanism. Also, the electron deficiency model assumes that electrochemical reactions
are responsible for the etching process in alkaline solutions. However, experimental
results indicate that the etching process is mainly of chemical nature as described in
detail in Chapter 5.
The passivation model, proposed by Palik
et al
.,
151,269
attributed the etch rate
reduction to the easier formation of an oxide film on highly doped silicon which pas-
sivates the surface. This model is supported by a number of experimental observations.
First, the difference between passivation potential and OCP decreases with increasing
doping concentration (Fig. 5.42) implying easier passivation for highly doped materi-
als. Second, ellipsometric measurements of the samples during etching in KOH solu-
tions indicate the existence of a surface phase on highly doped materials but not on
lightly doped materials. This coincides with the
i-
curves of highly doped material,
which shows no current peak, implying that an oxide film already exists near the OCP.
Due to the relatively slow etch rate of oxides in alkaline solutions, boron oxide and
hydroxides, once formed, tend to stay on the surface and would thus passivate the
surface and prevent further dissolution of the silicon. The fact that EDP exhibits
more effective etch stop than KOH at high boron doping is another indication of the
possible passivation mechanism, because EDP etches oxides at a much slower rate
than does KOH. The passivation model can also explain the lack of etch rate reduc-
tion in the
V
system because silicon oxide etches at high rates in HF-based
solutions.
Palik
et al
.
151,269
suggested that the tendency to grow an oxide layer on highly
doped materials is due to the strain in the silicon surface which enhances the Si-B bond
such that a borate glass is formed. The atomic concentration of boron-doped silicon at
etch stop levels corresponds to an average separation between boron atoms of 20-
25
and is near the solid solubility limit for boron substitutionally intro-
duced into the silicon lattice. Silicon doped with boron is under tension as the smaller
boron atom enters the lattice substitutionally, thereby creating a local tensile stress field.
At high boron concentrations the tensile force becomes so large that it is more ener-
getically favorable for the excess boron to enter interstitial sites. It is proposed that the
strong B-Si bond tends to bind the lattice more rigidly, therefore increasing the energy
required to remove a silicon atom high enough to stop etching. Formation of Si-X
Å
