Pore network simulation and preparation

techniques of nanoporous SiC membranes

Ryan Mourhatch, Muhammad Sahimi
and Theodore T. Tsotsis

Outlines
Motivations and objectives
Membrane preparation methods
CVD/CVI Technique
Continuum membrane formation model
Pore network simulation
Conclusion
Motivation
In recent years, there has been considerable effort
to produce high-temperature resistant membranes.
Polymeric membranes have proven, so far,
unsatisfactory in terms of performance and/or
material stability in high-temperature applications,
particularly in the presence of steam.
Silicon carbide is a promising material that is
known to be able to withstand high temperatures
and corrosive environments.
Motivation (cont.)
Properties that make SiC attractive as high-temperature
membrane material:
High chemical stability
High thermal conductivity
Strength at high temperatures
Low thermal expansion
High thermal shock resistance
High temperature resistance for abrasion
Potential Applications of SiC Membranes:
Hot Gas Filtration and Clean-Up
Hot Gas Separation
Liquid Filtration
Catalytic Membrane Reactors
Motivation (cont.)
Objective
To prepare hydrothermally stable, nanoporous SiC
membranes to be used in the separation of high-
temperature, hydrogen-containing mixtures.

Longer-term, (the objective is) to apply such
membranes in catalytic membrane reactors in the
production of hydrogen through the steam
reforming or water gas shift reactions.
Current Status of Research Effort
A better understanding is needed of the CVD/CVI reaction and transport
processes that lead to the formation of SiC membranes.
A new reactor was built which allows for better monitoring and control of
the CVD/CVI process.
Models were developed to describe the preparation of asymmetric,
microporous SiC membranes by the CVD/CVI process.
A three-dimensional network is used to represent the membrane's pore
space, in which the effective pore radii are distributed according to a
given PSD.
The models are currently being used for optimizing the CVD/CVI
preparation conditions for membranes of different shapes and sizes.
Preparation Methods
CVD/CVI techniques
Pyrolysis of pre-ceramic precursors
Preparation Technique for SiC Tubes
! SiC powder (0.6 m)+ resin + boron carbide +
acetone.
! Drying at room temperature for 3 days.
! Pressing of prepared powders in our home-made
die.
! Sintering of green tubes at 1800
o
C for 3 h
! Ultrasonication of tubes in acetone.
! Burning potential contaminants at 450
o
C for 2 h
in air.
Sequence of Events during CVD
1) Reactants diffuse through boundary
layer
2) They are adsorbed on the surface of
substrate
3) Chemical reaction takes place
4) Gaseous by-products are desorbed
5) They diffuse across the boundary
layer and are carried out by the main
flow
Membrane Preparation by CVD
Deposit supports with TPS at
700-750 C for various times
until He permeance levels off.
Anneal membranes at 1,000
C for 2 h for further
conversion to SiC
Ar and He permeances andselectivity
Ar and He permeances as a function of Deposition Time Ideal selectivity as a function of Deposition Time
Continuum membrane formation models.
Governing
Equations
Where f():
For !0.3
For >0.3
@ r=R
1
and t!t
TP

Boundary Conditions:
@ r=R
1
and t>t
TP

@ r=R
2
and t>0
Tube Side
Relative Porosity Change Along the Membrane
Thickness
Ar Permeance vs. Deposition Time
Membrane Pore Network Model
! A three-dimensional network is used to represent the
membrane's pore space, in which the effective pore radii are
distributed according to a given PSD.
! The connectivity of the pores, as well as the broadness of the
PSD, are varied in order to study their effect on the transport
and separation processes.
! The Maxwell-Stefan equations are used for describing the
pore-level transport processes, which include Knudsen and
hindered diffusion.
Membrane Pore Network Model, cont.
Two types of PSD are used. The first
resembles the experimental PSD (we
varied the r
a
):
Three pore size distributions
generated by Eq. above, with r
a
=5A,
(continuous curve), 7A (dotted
curve), and 9A (dashed-dotted
curve).
We also used a Gaussian PSD:
We varied the variance !
2
in order to study
the effect of the PSD's broadness on the results.
Diffusion Mechanisms
Knudsen diffusion, which is dominant when the collisions
between the molecules and the pores' walls are important
Hindered or configurational diffusion, which is significant
when the pore size is small enough
viscous flow contribution.
Total Flux:
Where within each pore!
Monte Carlo simulation
Discretized governing equations within
each pore using the finite-difference
(FD) approximation:
For two ends of each pore we use
the one-sided FD discretization
! For r
min
<r
p
<R
Ar
, only He contributes to the rate of gas transport.
! For r
p
>R
Ar
both gases diffuse. Therefore, as far as Ar is concerned,
there is a percolation effect at work, namely, a fraction q of the
pores exists, given by,
for which when q>p
c
, no transport of Ar can take place in the pore
network.
Percolation Effect
Percolation Effect
Comparison of the single- and mixed-gas permeances and their dependence on the average pore size.
Connectivty=6
Single- and mixed-gas permeances
Selectivity vs. Average Pore Size
Hindered and Knudsen Diffusion
Contributions
Dependence of the gases' permeances in their mixture on the average
pore size, when only hindered diffusion is the meachism of gas transport.
Connectivity Effect
Conclusions
SiC nanoporous membranes are prepared by a CVD/CVI.
Models were developed to describe the CVD/CVI process leading to
membrane formation.
The model provides good agreement with the experimental data. They are
currently utilized for optimizing membrane preparation conditions.
A pore network model of membrane transport was also developed.
It provides a fundamental understanding on how the characteristics of the
membrane pore structure determine its transport properties.
The network model provides a means of determining, from first principles,
the transport and reaction parameters of the continuum membrane
formation models.
Acknowledgment

The support of National Science Foundation and the U.S. Department of
Energy is gratefully acknowledged.