Abstract
A small quantity of high-value, crude, mono-isotopic silane is a prospective
gas for
a small-scale, high-recovery, ultra-high membrane purification process. This is
an unusual application of gas membrane separation for which we provide a
comprehensive analysis of a simple purification model.
The goal is to develop direct analytic expressions for estimating the feasibility
and efficiency of the method, and guide process design; this is only possible for
binary mixtures of silane
in the dilute limit which is a somewhat realistic case. Among the common
impurities in crude silane, methane poses a special membrane separation challenge
since it is chemically similar to silane. Other potential problematic surprises are:
ethylene, diborane and ethane (in this order). Nevertheless, we demonstrate,
theoretically, that a carefully designed membrane system may be able to purify
mono-isotopic, crude silane to electronics-grade level in a reasonable amount of
time and expenses.
We advocate a combination of membrane materials that preferentially reject
heavy impurities based on mobility selectivity, and light impurities based
on solubility selectivity.
We provide estimates for the purification of significant contaminants of
interest. To improve the separation selectivity, it is advantageous to use a
permeate chamber under vacuum, however this also requires greater control of
in-leakage of impurities in the system.
In this study, we suggest cellulose acetate and polydimethylsiloxane as examples of
membrane materials on the basis of limited permeability data found in the open
literature.
We provide estimates on the membrane area needed and priming volume of the
cell enclosure for fabrication purposes when using the suggested membrane materials.
These estimates are largely theoretical in view of the absence of reliable
experimental data for the permeability of silane.
Last but not least, future extension of this work to the non-dilute limit
may apply to the recovery of silane from rejected streams of natural silicon
semi-conductor processes.