(141c) Oxygen Permeable Ceramic Membranes for Hydrocarbon Conversion Reactors

Because of their various potential applications,
non-porous, selective oxygen permeable ceramic membranes with mixed ionic and
electronic conductivity (MIECs) have received significant interest over the
past decade [1-8].  In particular, the
incorporation of these materials into catalytic reactors for the oxidation of
hydrocarbons is being investigated as a way to eliminate costly air separation
steps and to achieve staged addition of oxygen [1-7].  Previous studies have typically focused on the use of ceramic
MIEC membranes for the partial oxidation of methane and have shown that the
presence of a catalyst near the permeate surface of a membrane can increase
oxygen flux up to ten times by reducing the permeate side oxygen partial
pressure [1,2].  However, currently
attainable oxygen fluxes remain insufficient for widespread industrial-scale
partial oxidation applications.

This work
describes the novel use of a mixed conducting oxygen permeable SrFeCo_0.5O_x
(SFC) membrane for the CO₂ reforming of CH₄ and demonstrates
that oxygen evolved from a membrane in direct contact with a reforming catalyst
can significantly enhance catalyst performance.  Additionally, semiconductor processing techniques have been used
to deposit catalyst particles and thin catalyst films directly on the high pO₂
surfaces (source side) of SFC membranes, and the effect of deposited catalyst
on both membrane oxygen flux and catalyst performance is reported.

Flux and
reaction experiments were conducted in custom-built stainless steel and quartz
CSTR-type reactors with disk-shaped SFC membranes.  For the CO₂ reforming experiments, a traditional
powder catalyst (0.43 wt% Pt/ZrO₂) was dispersed in a thin layer
across the membrane's permeate side to ensure good contact with the membrane
surface.  The baseline reaction set
includes tests with a small amount of co-fed oxygen to examine the effect of
the mode of oxygen introduction.  These
reactions were performed over a stainless steel ?blank? membrane coated with
the same inert BN₃ paint used to coat the interior of the stainless
steel reactor.  No catalyst reduction
was performed prior to any of the reactions.

For deposited
catalyst particle studies, the SFC membranes were polished to a near optical
finish before being patterned with Pt using a bi-layer lift-off
photolithography procedure common to the semiconductor processing
industry.  Unpatterned membranes were
also polished to maintain consistent surface area within a study.

The powder
Pt/ZrO₂ catalyst was chosen for the CO₂ reforming studies
because it exhibits relatively rapid deactivation under the reaction conditions
studied (800°C, 10 mL/min each of CH₄ and CO₂ with
5mL/min of Ar).  As expected, the
catalyst lost most of its activity within 2 hours of operation on the ?blank?
membrane, which is consistent with this catalyst's performance in a packed-bed
reactor configuration.  The activity
trends with 1 mol% oxygen in the feed stream are nearly identical, indicating a
similar rate and extent of catalyst deactivation with and without co-fed oxygen.  As confirmed by the parallel packed-bed
reactor study, a slightly higher steady state methane conversion occurs as a
result of the added oxygen.  However,
replacing the stainless steel blank with an SFC membrane produces much slower
and less extensive deactivation as well as higher initial activity than with
co-fed oxygen.  This observation
supports the conclusion that oxygen alone does not retard the deactivation of
this Pt/ZrO₂ catalyst and suggests that the addition of oxygen via
the membrane is more beneficial than oxygen added to the reactor feed
stream. 

Catalyst
deactivation can result either from loss of catalyst surface area via platinum
particle sintering or from carbon deposition and/or adsorption of other
species.  After six hours of reaction
without oxygen over the blank membrane, 1 mol% oxygen was added temporarily to
the reactor feed.  During this period,
methane conversion was comparable to the conversion with continuously co-fed
oxygen in the feed (1 mol%) at the corresponding reaction time, which confirms
the assumption of similar catalyst deactivation behavior with and without
co-fed oxygen.  After seven hours of
reaction over the blank with continuous co-fed oxygen, the reactant feed was
stopped and the catalyst was exposed to 1% oxygen in argon for 2 hours before
restarting the reactant feed.  The
catalyst showed no significant increase in activity upon resuming the CO₂
reforming reaction, which implies irreversible catalyst deactivation that could
be caused by platinum sintering.  This
hypothesis was supported by similar results from packed-bed reactions, which
included post-run temperature-programmed oxidations (TPOs) that indicated
negligible carbon deposition.  Platinum
particle sintering in the used powder catalyst is currently being evaluated by
transmission electron microscopy.

Flux studies
were performed using air as the oxygen source and argon as the sweep gas.  As described previously [9], oxygen flux was
determined at temperatures between 500 and 800 °C for both an unpatterned
polished membrane and an identical membrane patterned with 5 μm diameter
platinum circles spaced by 3 μm (120 nm deposition thickness), and the
platinum features were observed to increase membrane oxygen flux by
approximately 100% at all temperatures that exhibited flux.  Additionally, the platinum pattern reduced
the temperature at which oxygen permeation could first be detected from 600 °C
to 550 °C.  Subsequent work has been
conducted on the deposition of very thin films (~0.7 nm) of platinum and palladium
on the oxygen source side of SFC membranes including surface roughness analysis
by atomic force microscopy to obtain estimates of the true membrane surface
area.  Oxygen flux and catalyst activity
profile results for these thin film-coated membranes will be presented.

References

[1]        Tsai, C. Y.; Dixon, A. G.; Moser, W. R.; and Ma, Y. H. AIChE
J., 1997, 43 (11A), 2741.

[2]        Balachandran, U.; Dusek, J. T.; Maiya, P. S.; Ma, B.;
Mieville, R. L.; Kleefisch, M. S.; and Udovich, C. A. Catal. Today, 1997, 36,
265.

[3]        Balachandran, U.; Dusek, J. T.; Mieville, R. L.; Poeppel, R.
B.; Kleefisch, M. S.; Pei, S.; Kobylinski, T. P.; Udovich, C. A.; and Bose, A.
C., Appl. Cat. A, 1995, 133, 19.

[4]        Hazbun, E. A., U.S. Pat. 4,791,079, 1988.

[5]        ten Elshof, J. E.; Bouwmeester, H. J. M.; and Verweij,
H.,  Appl. Catal. A, 1995, 130, 195.

[6]        Lin, Y.S.; and Zeng, Y., 
J. Catal., 1996, 164, 220.

[7]        Xu, S.J.; and Thomson, W.J., 
Ind. Eng. Chem. Res., 1998, 37, 1290.

[8]        DiCosimo, R.; Burrington, J.D.; and Grasselli, R.K., U.S.
Pat. 4,571,443, 1986.

[9]        Slade, D.A.; Murphy, S.M.; Nordheden, K.; Stagg-Williams,
S.M. AIChE Annual Mtg, Cincinnati, OH, 2005, 353c.