This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
 Author's personal copy

Journal of Membrane Science 380 (2011) 223–231

Contents lists available at ScienceDirect

Journal of Membrane Science

journal homepage: www.elsevier.com/locate/memsci

Palladium surface modified La0.6 Sr0.4 Co0.2 Fe0.8 O3−ı hollow fibres for oxygen
separation
Christelle Yacou a , Jaka Sunarso a , Chun X.C. Lin a , Simon Smart a , Shaomin Liu b , João C. Diniz da Costa a,∗
a
The University of Queensland FIMLab – Films and Inorganic Membrane Laboratory, School of Chemical Engineering, Brisbane, Qld 4072, Australia
b
Department of Chemical Engineering, Curtin University of Technology, Perth WA 6845, Australia

a r t i c l e i n f o a b s t r a c t

Article history: La0.6 Sr0.4 Co0.2 Fe0.8 O3−ı (LSCF) hollow fibres were prepared by a phase inversion/sintering method using
Received 29 April 2011 polyetherimide as a binder. In order to overcome surface exchange kinetics limitation, LSCF hollow fibres
Received in revised form 30 June 2011 were coated with ∼200 nm palladium (Pd) nanoparticles. The O2 flux of best performing membranes
Accepted 2 July 2011
increased by up to 350% in comparison to unmodified LSCF hollow fibres. Optimal enhancement was
Available online 8 July 2011
achieved with a single Pd coating. Additional coatings resulted in reduced O2 fluxes, thus counter acting
the beneficial spill-over effect of the catalyst. Long term stability testing in atmospheric air at 850 ◦ C
Keywords:
showed that a LSCF membrane modified with a single Pd coating continually outperformed a pure LSCF
Oxygen separation
Hollow fibres
hollow fibre for over 400 h, though the level of enhancement was reduced over time. A dramatic reduction
Perovskites in performance of more than 45% occurred within the first 24 h of testing, which was attributed to the
LSCF coalescence and aggregation of Pd catalyst particles to ∼1000 nm size at the LSCF grain boundaries. This
Pd surface modification greatly reduced the available area for the oxygen species to spill-over onto the LSCF surface and thus
reduced the overall O2 flux.
© 2011 Elsevier B.V. All rights reserved.

1. Introduction and perovskite membranes have shown great promise, particularly

Separation of oxygen (O2 ) from air for industrial use is a big
business producing nearly 100 million tonnes of O2 each year [1]
and this market is predicted to expand rapidly in the near future
because virtually all large scale carbon feedstock clean energy
technologies will require O2 as a feed [2]. Currently, cryogenic dis-
tillation is the process of choice for air separation, as it offers the
minimal deployment risk of those technologies currently employed
on an industrial scale. However, the cryogenic process has a high
energy cost as it operates at very low temperatures and high pres-
sures. With the recent developments in ceramic ionic transport
membranes, it is envisaged that this technology can replace cryo-
genic air separation on a commercial scale. In this technological
race to commercialise ceramic membranes for air separation, an
array of ceramic materials, based on perovskites (ABO3 ), fluorites
(AO2 ), brownmillerites (A2 B2 O5 ), the Ruddlesden–Popper series
(An+1 Bn O3n+1 ), and Sr4 Fe6−x Cox O13 compounds, have been studied
by the research community over the last 20 years.
The parameters of principal importance in technology deliv-
ery are production and quality capabilities. In principle, defect-free
ceramic membranes allow for O2 transport via ionic diffusion only,
thus delivering 100% purity. Production is related to oxygen fluxes
∗ Corresponding author. Tel.: +61 7 3365 6960; fax: +61 7 3365 4199.
E-mail address: j.dacosta@uq.edu.au (J.C. Diniz da Costa).
barium strontium cobalt iron oxides (BSCF) and lanthanum stron-
tium cobalt iron oxides (LSCF) which have been extensively studied.
Major improvements were realised by the reduction of the mem-
brane thickness from discs (1 mm) to hollow fibres (0.25 mm). The
major advantage here is that the O2 flux is inversely proportional
to the membrane thickness. Hence, reducing the thickness of the
membrane translates to a reduction in the oxygen ionic bulk diffu-
sion resistance. As a result, BSCF hollow fibre geometries operating
at high temperatures in excess of 900 ◦ C have demonstrated large
O2 fluxes of 5 ml cm−2 min−1 [3,4]. Recently Leo et al. [5] reported
even higher O2 fluxes of 9.5 ml cm−2 min−1 by changing the con-
ventional polyethersulfone binder in the hollow fibre synthesis
process to a sulfur-free polymer. Variations in the dopants and com-
position have also led to high membrane permeance values such as
BaCoFeZr [6] and BaBiScCo [7] which delivered oxygen fluxes of 7.3
and 11.4 ml cm−2 min−1 , respectively. Most available literature on
LSCF hollow fibres have been largely devoted to its preparation and
performance aspects [8–11] including the transport layer modifi-
cation via acid etching or methane activation. Nevertheless, LSCF
O2 fluxes remain low compared to BSCF fluxes, generally in the
range of 0.1–2.5 ml cm−2 min−1 depending upon membrane syn-
thesis, surface modification and experimental conditions as listed
in Table 1.
Secondary parameters for consideration deal with the process
efficiencies and operating costs associated with the technology.
Ceramic membranes operate efficiently at high temperatures,

0376-7388/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.memsci.2011.07.008
 Author's personal copy
224 C. Yacou et al. / Journal of Membrane Science 380 (2011) 223–231
Table 1
Comparison of reported oxygen fluxes through La0.6 Sr0.4 Co0.2 Fe0.8 O3−ı hollow fibres membranes.
Hollow fibre Transport layer Operating temperature
thickness (cm) modification range (◦ C)
0.2 540–950
0.3 800–900
0.2 700–900
0.2 CH4 activation 858
0.6 CH4 activation 800–1000
0.6 Acid modification 800–1050
0.6 Acid etching 800–1050
0.8 Porous LSCF modified 700–1000
traditionally in excess of 900 ◦ C, and it is questionable as to whether
efficiency improvements can be realised at these extreme operating
conditions [19]. Indeed heating air to high temperatures, where the
recovery rate for oxygen may be 10% or lower compared to the total
air volume, will attract considerable energy costs. Hence, process
analysis dictates the optimal design will be realised by reducing
the operating temperature whilst maintaining reasonably high O2
fluxes. This can be achieved by anchoring catalysts to the mem-
brane surface. Improvements of additional O2 flux were reported
for BSCF hollow fibres, surface modified with platinum (Pt) [20]
and silver (Ag) [21] catalysts, as well as LSCF hollow fibres, modified
with Pt [22] and Ag [18] particles. Recently, Leo et al. [5] also demon-
strated that BSCF coated with palladium (Pd) nanoparticles allowed
O2 fluxes to significantly increase, by a factor of 10 below 700 ◦ C,
thus reducing operating temperature requirements by 250 ◦ C. The
major advantage here is the spill-over effect of the catalyst which
allows for a higher O2 absorption and faster surface diffusion of
atomic oxygen species to the membrane surface, thus overcom-
ing the surface exchange kinetics limitation and translating into
higher O2 fluxes. In addition, palladium has superior thermal sta-
bility than silver as its melting point of 1554.9 ◦ C is higher than
silver at 961.8 ◦ C [23].
A final, but no less important, parameter for technology deliv-
ery is the long-term operational performance, which in this case
is related to the thermal stability of the ceramic membrane mate-
rial. This is an area of weakness for BSCF membranes, as prolonged
exposure to temperatures below 900 ◦ C results in the cubic BSCF
structure transforming into the hexagonal phase, thus reducing
oxygen permeation [24]. By contrast, LSCF membranes have been
proven to be stable for operation for over 3000 h at 800 ◦ C [25].
Although LSCF oxygen fluxes are much lower that BSCF, the tech-
nological race may point to the former as the preferred option,
particularly when operating and replacement costs are imperative
to predict the commercialisation prospects of the technology.
In this work, we investigate the performance of LSCF hollow
fibre membranes coated with Pd nanoparticles. LSCF was chosen
due to its well-publicised thermal stability, whilst Pd has deliv-
ered the best shift in O2 production of those catalysts investigated.
We also study how Pd loading on the LSCF surface aids O2 flux
improvement at temperatures ranging from 700 to 950 ◦ C, and with
several driving forces created by varying the sweep gas flow rates
in the permeate stream. We are also interested in the long term
operational stability of Pd coated LSCF hollow fibres. Hence, the
membranes are further tested for over 400 h using atmospheric
air rather than synthetic air, to emulate commercial air separation
process units.
2. Experimental
2.1. Preparation methods and characterisation
La0.6 Sr0.4 Co0.2 Fe0.8 O3−ı (herein after called LSCF) powders
were prepared by a combined EDTA–citrate complexing sol–gel
Atmosphere used O2 flux (ml cm−2 min−1 ) Ref.
low pO2 /high pO2
He/air 0.09–0.04 [15]
Ar/air 0.1–1.2 [14]
Ar/air 0.1–0.8 [13]
CH4 /air 0.26–0.8 [15]
He/air 0.06–0.725 [12]
He/air 0.35–2.5 [16]
He/air 0.5–2.5 [17]
He/air 0.08–1.48 [18]
process. Stoichiometric quantities of La(NO3 )3 ·6H2 O (99.0+%,
Ajax Finechem, Australia), Sr(NO3 )2 (99.0+%, Sigma–Aldrich),
Co(NO3 )2 ·6H2 O (98.0+%, Sigma–Aldrich) and Fe(NO3 )2 ·9H2O
(99.0+%, Ajax Finechem) were added to the EDTA (99.4+%, Ajax
Finechem, Australia), anhydrous citric acid (99.5%, Fluka) and NH3
aqueous solution (28%, Ajax Finechem, Australia) to control the pH
at 6 (avoiding precipitation) under heating and stirring followed
by the addition of anhydrous citric acid resulting in a violet aque-
ous solution. The molar ratio of all metal nitrates:EDTA:citric acid
is kept at 1:1.1:2. The solution was heated at 80 ◦ C for 6–8 h under
continuous stirring to remove water until a viscous gel was formed.
The gel was heated further by a hot plate until they dried. The dried
gel was then pretreated on the furnace at 700 ◦ C for 12 h (heat-
ing rate = 5 ◦ C min−1 ) to remove residual carbon and form a solid
precursor. The powders were then ball-milled in a planetary mill
(Pulverisette 5, Fritsch) at 500 rpm for 2 h to obtain particle size of
less than 100 ␮m.
A phase inversion, wet spinning technique was employed to fab-
ricate green LSCF hollow fibres as described elsewhere [26–28],
though modified with the use of a sulfur-free polymeric binder
[5]. Briefly, the milled LSCF powder was added to a mixture of
1-methyl-2-pyrrolidinone (NMP) [synthesis grade, Sigma Life Sci-
ence] and polyetherimide (PEI) [(SABIC Innovative Plastics, Saudi
Arabia)] in a mass ratio 6.71:4:1 and stirred at high speed for 2 h
to ensure a uniform mixture. A small amount, 0.5–1% by mass,
of polyvinylpyrrolidone (PVP) [M.W. 1,300,000; Alfa Aesar] was
added to adjust the viscosity of the mixture. To form LSCF–NMP–PEI
mixture into the required hollow fibre geometry, a tube-in-orifice
spinneret with orifice diameter/inner diameter of 2.5 mm/0.8 mm
was used. Tap water was employed as the internal and external
coagulant. The extruded hollow fibres were dried and cut onto short
lengths and sintered at 1350 ◦ C for 12 h (heating rate = 1 ◦ C min−1 )
to obtain gas tight membranes.
The surface modification process of the sintered hollow fibres
was adapted from a previous work published by Leo et al. [5]. A
palladium precursor solution (PdNO3 at 0.1 M) was firstly care-
fully coated on both inner and outer surfaces of the membrane
followed by a drying step at 60 ◦ C for 2 h. An equivalent amount
of N2 H4 (7% by volume in water), was then applied to the inner
and outer surfaces of the membrane to induce the chemical
reduction of Pd2+ to Pd0 particles based on the following equa-
tion 2Pd2+ + N2 H4 → 2Pd0 + 4H+ + N2 [29,30]. The fibres were then
immediately heat treated in air at 700 ◦ C for 10 min to remove
residual N2 H4 and other by-products. The fibres were then cooled
to room temperature at 1 ◦ C min−1 . This Pd surface modification
process was repeated up to three times to establish optimal cov-
erage. Regarding the nomenclature used in this work, Pd modified
hollow fibres were labelled LSCF–Pd × y, where y represents the
number of catalyst coatings. For comparison purposes the original
La0.6 Sr0.4 Co0.8 Fe0.2 O3−ı membranes without catalyst deposition
were also investigated and simply labelled LSCF.
The morphological features of the prepared hollow fibres were
examined by a Philips XL-30 Scanning Electron Microscope (SEM).
 Author's personal copy

C. Yacou et al. / Journal of Membrane Science 380 (2011) 223–231 225

Fig. 1. Hollow fibre permeation testing setup.

Energy Dispersive X-ray (EDX) and quantitative composition anal-
ysis were performed on a JEOL JSM-6460 LA low vacuum analytical
SEM equipped with an integrated JEOL Hyper mini-cup, 133 eV
resolution, SiLi crystal, ultra thin window (UTW) and EDX Spec-
trometer. All samples were coated with carbon to reduce charging.
Additional structural data were collected from X-ray diffraction
(XRD) patterns of the equivalent membranes powders using a
Bruker D8 Advance diffractometer with a Cu K␣ radiation (40 kV,
20 mA,  = 1.5409 Å).
2.2. Oxygen permeation measurements
The performance of the LSCF hollow fibre membranes was inves-
tigated in an oxygen permeation cell schematically shown in Fig. 1.

Fig. 2. SEM micrographs of the (a) inner, (b) outer surface of the original LSCF, and (c) inner surface, (d) outer surface, (e) cross-section of LSCF–Pd × 1 membrane and (f) EDX
spectra of identified areas.
 Author's personal copy
226 C. Yacou et al. / Journal of Membrane Science 380 (2011) 223–231
Fig. 3. XRD patterns of the equivalent membranes powder corresponding to LSCF–Pd × 0 to LSCF–Pd × 3.
Membranes, approximately 80 mm in length, were suspended in
a split hinge tube carbolite furnace with a constant temperature
zone of 10 cm. Quartz tubes (diameter of 18 mm) were attached to
both ends of the LSCF fibres and sealed with a silver-based sealant.
The seals were verified as gas-tight if nitrogen was not detected
when the permeate stream was tested at room temperature. Nev-
ertheless, a very small amount of nitrogen was detected by gas
chromatography for temperatures in excess of 600 ◦ C, probably as
a result of thermal expansion of the hollow fibres causing negligi-
ble leaks at the silver paste and membrane interface. For instance,
the oxygen leak rate was 12% at 950 ◦ C relative to the total oxy-
gen flow rate. Permeation testing was conducted by passing argon
(BOC, purity > 99.998%) as a sweep gas through the permeate side
of the membrane (inner shell of the hollow fibre) and by varying
the temperature of the furnace. The argon gas flow rate was varied
from 50 to 200 ml min−1 and the operating temperature between
700 and 950 ◦ C. The permeate stream was fed directly to the gas
chromatograph (Shimadzu GC-2014 equipped with a 5 Å molecu-
lar sieve column) for analysis. The total flow rate of the permeate
stream was measured using a bubble flow meter. In this work, we
report only the oxygen flux derived from ionic transport. There-
fore, the contribution of molecular oxygen flux from leakage was
subtracted from the overall flux values by measuring the nitrogen
concentration. The effective membrane area (A) and oxygen per-
meation flux (JO2 ) was calculated using Eqs. (1) and (2), where L,
do and di is the length, outside and inside diameter of fibre (mm),
respectively; FT is the total flow rate of permeate stream (ml min−1 ).
L(do − di )
A=
ln(do /di )
 21
 F
T
JO2 = % O2 − % N2 · (2)
79 A
3. Results and discussion
3.1. Microstructure and morphology analysis
SEM micrographs of the synthesized and subsequently Pd mod-
ified hollow fibre membranes are depicted in Fig. 2. The inner
and outer surfaces of the LSCF material display a homogenous
microstructure with no apparent pores or cavities. The LSCF parti-
cles in the size range of ∼5–10 ␮m were tightly connected to each
other, thus indicating that the grains coalesced during the sintering
process and gave rise to a dense and smooth surface layer. White
spherical spots assigned to the Pd particles in Fig. 2c and d were
detected on both the inner and outer surfaces of the modified mem-
brane. These figures demonstrate that the Pd particles, which had
an average size of approximately 200 nm, were relatively well dis-
persed over the surface without any obvious aggregation. However,
the Pd deposition was not homogeneous over the entire mem-
brane surface, with several areas containing lower than average
numbers of Pd particles identified and confirmed by EDX analy-
sis (example spectra are shown in Fig. 2f). Of particular interest,
increasing the number of Pd coatings did little to overcome the het-
erogeneous deposition of catalyst particles. Finally, the external and
internal diameters of the modified fibres were measured from the
SEM micrographs at 2030 ± 1 ␮m and 1815 ± 1 ␮m, respectively
(Fig. 2e). It must be emphasized that the entire cross-section of
the fibre was not fully dense, as the presence of pores are apparent.
(1)
However, these pores are not interconnected and do not provide
 Author's personal copy
C. Yacou et al. / Journal of Membrane Science 380 (2011) 223–231
Fig. 4. Oxygen flux isotherms of LSCF and Pd modified hollow fibres as a function sweep gas flow rate.
a percolation pathway, as the hollow fibres proved to be gas-tight
during air leak measurements performed at room temperature.
In order to confirm that the perovskite membranes employed in
this work had retained their original crystalline structure after the
catalyst deposition, XRD patterns were collected from the equiva-
lent membrane powders (Fig. 3). Strong diffraction peaks assigned
to the rhombohedral phase of La0.6 Sr0.4 Co0.2 Fe0.8 O3−ı (2 1 1 0 = 33◦ ,
2 2 0 2 = 40.64◦ , 2 0 2 4 = 47.07◦ , 2 2 1 4 = 58.5◦ , 2 2 0 8 = 68.5◦ ,
2 2 1 8 = 78.03◦ ) were detected as reported elsewhere [31], showing
the structural homogeneity of the prepared materials. The XRD
patterns also reveal a broad weak peak at 2 = 34.2◦ which can be
attributed to the main reflection (1 1 0) of PdO [32]. This peak was
only detectable in the LSCF–Pd × 2 and LSCF–Pd × 3 samples due
to the low Pd content in the LSCF–Pd × 1 sample. The intensity of
this peak is emphasized for the LSCF–Pd × 3 sample, indicating
that sequential deposition of catalyst particles did not damage or
change the LSCF phase, whilst also revealing their nanocrystalline
nature. The presence of this phase is in agreement with the
experimental procedure used in this work and attributed to the
oxidation of Pd◦ to PdO in the temperature range of 500–600 ◦ C
[33,34].
3.2. Oxygen permeation measurements
Fig. 4 displays the O2 fluxes isotherms as a function of the
sweep gas rate for the LSCF and Pd modified hollow fibres. The O2
fluxes increased with temperature for all membranes, indicating
the effect of temperature in enhancing ionic transport and sur-
face exchange kinetics. For instance, the unmodified hollow fibre
exhibited O2 fluxes between 0.11 and 0.50 ml cm−2 min−1 as the
227
temperature was raised from 700 ◦ C to 950 ◦ C, respectively, thus
agreeing with previously published work [8,10]. By comparison,
all Pd modified LSCF hollow fibres exhibited higher O2 fluxes over
the entire temperature range, reaching a maximum value close
to 2.0 ml cm−2 min−1 at 950 ◦ C. These results clearly indicate that
the Pd catalyst coating procedure used in this work was beneficial
in improving O2 fluxes. Further, these significant enhancements
in O2 fluxes clearly demonstrate that the kinetics of the surface
reaction is the limiting factor in oxygen transport through the
LSCF hollow fibre. The Pd catalyst particles acted as O2 adsorp-
tion promoters, and helped facilitate O2 transport to the membrane
surface.
Fig. 4 also shows that the sweep gas flow rate plays an impor-
tant role in the ability of the membrane system to separate O2 from
air. In principle, increasing the sweep gas flow rate decreases the
O2 partial pressure in the permeate stream. As a result, the driving
force associated with the O2 partial pressure difference between
the feed and permeate stream should increase, thus translating in
improved O2 fluxes. However, this is not entirely the case here as
there are two distinct O2 transport regimes apparent in Fig. 4. For
temperatures equal to or below 800 ◦ C, the O2 flux remained almost
constant for all membranes, suggesting that bulk ionic diffusion
and/or surface exchange kinetics limited O2 transport independent
of the sweep gas flow rate used. It is possible that at these relatively
lower temperatures, the major transport resistance is associated
with the surface exchange kinetics at the air feed side of the mem-
brane, responsible for transferring gaseous O2 into ionic oxygen in
the solid perovskite lattice. The change of O2 partial pressure in
the permeate stream by the sweep gas was therefore not affected
by the gas atmosphere surrounding the membrane surface in the
 Author's personal copy

228 C. Yacou et al. / Journal of Membrane Science 380 (2011) 223–231

feed stream. However, the effect of the sweep gas flow rate was
clearly noticeable for temperatures above 800 ◦ C, similar to find-
ings published elsewhere [18]. Surface reactions begin enhancing
the overall O2 permeation rate above 800 ◦ C, and the rate of oxygen
ions recombining into O2 molecules at the inner surface of the hol-
low fibre became similar to the transport rate of the O2 molecules
into the convective sweep gas flow [7]. Thus, increasing the sweep
gas flow rate also increased the O2 partial pressure difference across
the membrane and the O2 flux accordingly. Of course, once the
transport of the O2 molecules into the convective sweep gas flow
exceeded the rate of ionic recombination, or the rate of oxygen ion
supply prior to ionic recombination, then further increases in the
sweep gas flow rate were ineffectual, as evidenced by the plateau-
ing of the O2 flux in Fig. 4 for the Pd modified membranes at higher
sweep gas flow rates.
Fig. 5a shows the relative improvement in O2 flux as a func-
tion of the number of Pd coatings. Optimal improvements were
observed for a single Pd coating which led to a significant increase
of O2 fluxes by two orders of magnitude, up to 350% as compared
to the pure LSCF hollow fibre. On the other hand, it was unusual
to observe that the relative improvement in O2 flux decreased as
the number of catalyst loadings increased. It would be anticipated
that by increasing the catalyst loading on the membrane surface,
O2 fluxes would increase accordingly. One major distinction here is
the non-homogeneity of the catalyst coverage of the LSCF surface
(Fig. 2). In addition, Fig. 6 exhibits the SEM micrographs, clearly
showing a process leading to the aggregation of Pd particles as the
number of Pd coats increase from rather than dispersing on the LSCF
membrane surface. This aggregation and growth of catalyst parti-
cles was detrimental as it reduced the surface area available for the
oxygen surface exchange reactions per Pd particle volume. Never-
theless, the remarkable relative improvement in O2 flux observed
in this work using a Pd catalyst is significantly higher than previ-
ous work, which included flux increases for surface modified LSCF
membranes of 128% (Ag catalyst) [18], and for surface modified
BSCF membranes of 25% (Pt catalyst) [20], 93% (Ag catalyst) [21]
and 105% (Pd catalyst) [5].
Interestingly, the relative improvement in O2 flux peaked at
different temperatures depending upon the number of Pd coat- Fig. 5. (a) Percentage improvement of the O2 flux through LSCF–Pd × 1, LSCF–Pd × 2
ings. For instance, Fig. 5a shows the optimal improvement was and LSCF–Pd × 3 compare to the original LSCF membrane, as a function of the oper-
ating temperature, and (b) Arrhenius diagram of oxygen fluxes. Sweep gas flow rates
reached at 850 ◦ C (LSCF–Pd × 1), 900 ◦ C (LSCF–Pd × 2) and 950 ◦ C at 200 ml min−1 .
(LSCF–Pd × 3). In principle, the Pd particle operates as a catalyst
which overcomes the rate determining surface reaction exchange
as depicted in the Arrhenius diagram in Fig. 5b. Therefore, the oxy- for the oxygen fluxes reduced from 87 to 75 kJ mol−1 , from pure to
gen fluxes significantly increase for a single Pd coated membrane, a single Pd coated LSCF hollow fibre, respectively. For temperatures
aided by the spill-over of oxygen from Pd particle to the surface in excess of 850 ◦ C when a single Pd coat is applied, the relative
of the LSCF hollow fibre. Hence, the rate of oxygen permeation increment of oxygen flux to the base case (e.g. pure LSCF) observed
is determined by the rate of the supply of oxygen ions to the in Fig. 5a is no longer significant, and actually reduces at 950 ◦ C.
membrane surface in tandem with diffusion rates, and the rate of These results suggest that the activity of oxygen chemisorption
oxygen ion recombination at the permeate side. The positive effect on Pd particle started decaying at high temperatures in excess of
of the Pd catalyst is observed as the apparent energy of activation 800 ◦ C, thus reducing the oxygen spill-over effect of the Pd catalyst.

Fig. 6. SEM micrographs of the surface of Pd modified LSCF membranes.

 Author's personal copy

C. Yacou et al. / Journal of Membrane Science 380 (2011) 223–231 229

Time (hours)
0 100 200 300 400

LSCF-0×Pd
0.4 LSCF-1×Pd

Oxygen flux (ml.cm .min )

-1
0.3

-2
0.2

0.1

Log O2 flux improvement

Fig. 7. Schematic of the effect of Pd coated LSCF membranes. O2 flux improvement (%)
100

A similar trend was also observed for silver coated BSCF hollow
fibres reported elsewhere [21].
The increase in the number of coats was accompanied by Pd par-
ticle aggregation shown in Fig. 6, which led to a diminished ratio
of the available spill-over area to the surface area occupied by the
catalyst itself. Consequently, the diffusion of oxygen from the cata-
lyst particles to the LSCF surface, particularly at low temperatures,
affected the overall oxygen flux as observed in Fig. 5b. Hence, the
reduction of effective area for oxygen surface access resulted in an
increase of the apparent energy of activation for oxygen fluxes to
values ∼120 kJ mol−1 , much higher than those calculated for the
pure and single Pd coat LSCF hollow fibres. The optimal improve-
ment of O2 fluxes observed at 900 and 950 ◦ C for the two and three
catalyst coatings in Fig. 5a respectively, could then be explained
by the combination of the increased surface exchange kinetics of
the LSCF surface at higher temperatures together with improved
bulk diffusion, though accompanied by the reduced spill-over effect
from the larger Pd catalyst particles. This is schematically shown in
Fig. 7 and based on the SEM micrographs in Fig. 6. As the Pd par-
ticles increase in size, they reduce the LSCF surface area accessible
to ionic diffusion via spill-over, or via direct access of O2 molecules
from the gas phase to the surface of the membrane. Thus, whilst the
supply of oxygen ions from the LSCF surface exchange reaction is
generally lower than from the catalytic Pd particles, the faster kinet-
ics of this reaction counteracts the diminishing spill-over effect of
the larger Pd particles at higher temperatures. It must be borne in
mind that the improvements in O2 fluxes in Fig. 5a are relative to
(%)
10
0 100 200 300 400
Time (hours)
Fig. 8. Time-dependence of O2 flux through original LSCF and LSCF–Pd × 1 hollow
fibres (T = 850 ◦ C, sweep gas flow rate = 100 ml min−1 ).
the performance of a pure LSCF membrane. In principle, the ionic
flux in LSCF materials will increase as a function of temperature.
The relative increments observed should be considered in the light
of trade-offs between the catalyst activity, catalyst particle size,
particle distribution on the membrane surface, and accessible area
for surface ionic diffusion.
3.3. Long term oxygen permeation testing
In order to evaluate the long term operational stability of the
prepared hollow fibres, the O2 flux was evaluated over 410 h of
operation at 850 ◦ C for the pure LSCF hollow fibre (LSCF–Pd × 0)
and the surface modified (LSCF–Pd × 1), as shown in Fig. 8. During
the first 24 h of operation, LSCF–Pd × 0 exhibits an initial reduction
in O2 flux of 5.8%. After 24 h, the change in O2 flux stabilized, result-
ing in a constant rate of reduction of approximately 0.05% per hour.
These long term results are in good agreement with Schlehuber et
al. [25] who reported a fairly stable performance for a similar LSCF

Fig. 9. SEM micrographs of the LSCF–Pd × 1 membrane surfaces (a) before and (b) after O2 permeation measurement (c) EDX spectrum of identified area.
 Author's personal copy
230 C. Yacou et al. / Journal of Membrane Science 380 (2011) 223–231
membrane tested for 3000 h of operation at 800 ◦ C. In contrast, the
LSCF–Pd × 1 membrane displayed a significant loss in O2 flux in the
first 24 h. For instance, the initial oxygen flux of 0.4 ml min−1 cm−2
underwent a considerable reduction of 45.2 ± 0.1%; equivalent to
a rate of reduction of 0.2% per hour. From thereon, the Pd surface
modified membrane became more stable and much slower decline
of 0.047% per hour was observed. Comparatively the Pd surface
modified LSCF membrane gave superior performance throughout
the entire 410 h. However, the improvement in O2 flux offered
by the Pd surface modification reduced from 134% at 0 h to 32%
within the first 24 h, followed by further reduction to 24% after
410 h.
As the long term testing was carried out using atmospheric air,
the small loss of performance observed within the first 24 h of oper-
ation for the pure LSCF hollow fibre could be associated with the
effect of the small concentration of CO2 present. However, degra-
dation as a result of CO2 exposure would not explain the significant
drop in O2 flux measured for the Pd surface modified membrane.
To further investigate this problem, SEM micrographs were taken
after O2 permeation testing and compared with those taken prior to
testing as depicted in Fig. 9. It is observed that the Pd particles (con-
firmed through EDX analysis – Fig. 9c) coalesced and increased in
size from ∼200 nm to ∼1000 nm during testing. These results sug-
gest that the Pd particles did not adhere strongly to the LSCF surface
and were free to coalesce and aggregate at elevated temperatures.
Similar non-adherence effects were also reported for Ag catalysts
on BSCF electrodes [35]. Furthermore, prolonged testing at high
temperatures resulted in the thermodynamically favoured move-
ment of the Pd particles towards the perovskite grain boundaries.
This coalescence and subsequent increased size of the Pd particles
served to counteract their beneficial spill-over effect on the LSCF
surface.
4. Conclusions
This work demonstrates that the O2 flux of LSCF hollow fibres
can be dramatically enhanced through Pd surface modification.
The coating of Pd nanoparticles onto the surface of the LSCF hol-
low fibres overcame the surface exchange reaction kinetics and
enhanced the O2 flux by up to 350%, in comparison to unmodi-
fied LSCF hollow fibres. This represents the largest performance
increase of an LSCF hollow fibre by catalytic surface modifica-
tion in the published literature to date. Optimal enhancement
was achieved with a single Pd coating. Further coatings were
shown to be ineffectual at increasing the O2 flux due to the non-
homogeneous distribution and aggregation of Pd nanoparticles,
thus reducing the spill-over effect of the catalyst. The flow rate
of the sweep gas in the permeate stream was found to improve
O2 fluxes only at temperatures above 850 ◦ C, where the surface
reactions were dominant. Long term stability testing showed that
whilst the Pd modified LSCF membrane continually outperformed
the pure LSCF hollow fibre for over 400 h, the level of enhance-
ment was reduced over time. The reduction in performance was
non-linear with the most dramatic reduction occurring within the
first 24 h, attributed to the coalescence and aggregation of the Pd
nanoparticles at the LSCF grain boundaries.
Acknowledgments
This work was financially supported by the Australian Research
Council (ARC) through Discovery Project program (DP0878849).
The authors gratefully acknowledge Ms. Anya Yago (Centre
for Microscopy and Microanalysis), Mr. Hong-yi Xu (Centre for
Microscopy and Microanalysis) for their contribution in the XRD
and the SEM-EDX analysis, respectively.
References
[1] J. Emsley, Oxygen, Nature’s Building Blocks: An A–Z Guide to the Elements,
Oxford University Press, Oxford, England, UK, 2001.
[2] A. Leo, S. Liu, J.C. Diniz da Costa, Development of mixed conducting mem-
branes for clean coal energy delivery, Int. J. Greenhouse Gas Control 3 (2009)
357–367.
[3] S. Liu, G.R. Gavalas, Oxygen selective ceramic hollow fiber membranes, J.
Membr. Sci. 246 (2005) 103–108.
[4] S. Liu, X. Tan, Z. Shao, J.C. Diniz da Costa, Ba0.5 Sr0.5 Co0.8 Fe0.2 O3−ı ceramic hollow-
fiber membranes for oxygen permeation, AIChE J. 52 (2006) 3452–3461.
[5] A. Leo, S. Smart, S. Liu, J.C. Diniz da Costa, High performance per-
ovskite hollow fibres for oxygen separation, J. Membr. Sci. 368 (2011)
64–68.
[6] T. Schiestel, M. Kilgus, S. Peter, K.J. Caspary, H. Wang, J. Caro, Hollow fibre
perovskite membranes for oxygen separation, J. Membr. Sci. 258 (2005)
1–4.
[7] J. Sunarso, S. Liu, Y.S. Lin, J.C. Diniz da Costa, High performance BaBiScCo oxygen
transport hollow fibres, Energy Environ. Sci. 4 (2011) 2516–2519.
[8] K. Li, X. Tan, Y. Liu, Single-step fabrication of ceramic hollow fibers for oxygen
permeation, J. Membr. Sci. 272 (2006) 1–5.
[9] X. Tan, Z. Wang, K. Li, Effects of sintering on the properties of
La0.6 Sr0.4 Co0.2 Fe0.8 O3− perovskite hollow fiber membranes, Ind. Eng. Chem.
Res. 49 (2010) 2895–2901.
[10] B. Zydorczak, Z. Wu, K. Li, Fabrication of ultrathin La0.6 Sr0.4 Co0.2 Fe0.8 O3−
hollow fibre membranes for oxygen permeation, Chem. Eng. Sci. 64 (2009)
4383–4388.
[11] X. Tan, Z. Wang, B. Meng, X. Meng, K. Li, Pilot-scale production of oxygen
from air using perovskite hollow fibre membranes, J. Membr. Sci. 352 (2010)
189–196.
[12] H. Liu, Z. Pang, X. Tan, Z. Shao, J. Sunarso, R. Ding, S. Liu, Enhanced oxygen per-
meation through perovskite hollow fibre membranes by methane activation,
Ceram. Int. 35 (2009) 1435–1439.
[13] X. Tan, K. Li, Oxidative coupling of methane in a perovskite hollow-fiber mem-
brane reactor, Ind. Eng. Chem. Res. 45 (2005) 142–149.
[14] X. Tan, Y. Liu, K. Li, Mixed conducting ceramic hollow-fiber membranes for air
separation, AIChE J. 51 (2005) 1991–2000.
[15] A. Thursfield, I. Metcalfe, Methane oxidation in a mixed ionic–electronic con-
ducting ceramic hollow fibre reactor module, J. Solid State Electrochem. 10
(2006) 604–616.
[16] Z. Wang, H. Liu, X. Tan, Y. Jin, S. Liu, Improvement of the oxygen permeation
through perovskite hollow fibre membranes by surface acid modification, J.
Membr. Sci. 345 (2009) 65–73.
[17] H. Liu, X. Tan, Z. Pang, J.C. Diniz da Costa, G.Q. Lu, S. Liu, Novel dual structured
mixed conducting ceramic hollow fibre membranes, Sep. Purif. Technol. 63
(2008) 243–247.
[18] X. Tan, Z. Wang, H. Liu, S. Liu, Enhancement of oxygen permeation through
La0.6 Sr0.4 Co0.2 Fe0.8 O3− hollow fibre membranes by surface modifications, J.
Membr. Sci. 324 (2008) 128–135.
[19] S. Smart, C.X.C. Lin, L. Ding, K. Thambimuthu, J.C. Diniz da Costa, Ceramic mem-
branes for gas processing in coal gasification, Energy Environ. Sci. 3 (2010)
268–278.
[20] A. Leo, S. Liu, J.C. Diniz da Costa, Z. Shao, Oxygen permeation through perovskite
membranes and the improvement of oxygen flux by surface modification, Sci.
Technol. Adv. Mater. 7 (2006) 819–825.
[21] A. Leo, S. Liu, J.C. Diniz da Costa, The enhancement of oxygen flux on
Ba0.5Sr0.5 Co0.8 Fe0.2 O3− (BSCF) hollow fibers using silver surface modification,
J. Membr. Sci. 340 (2009) 148–153.
[22] A. Thursfield, I.S. Metcalfe, Air separation using a catalytically modified mixed
conducting ceramic hollow fibre membrane module, J. Membr. Sci. 288 (2007)
175–187.
[23] C.R. Hammond, The elements, in: D.R. Lide (Ed.), Handbook of Chemistry and
Physics, CRC Press LLC, Boca Raton, 2002.
[24] S. Švarcová, K. Wiik, J. Tolchard, H.J.M. Bouwmeester, T. Grande, Structural
instability of cubic perovskite Bax Sr1−x Co1−y Fey O3− , Solid State Ionics 178
(35–36) (2008) 1787–1791.
[25] D. Schlehuber, E. Wessel, L. Singheiser, T. Markus, Long-term operation of a
La0.58 Sr0.4 Co0.2 Fe0.8 O3− membrane for oxygen separation, J. Membr. Sci. 351
(2010) 16–20.
[26] S. Liu, X. Tan, K. Li, R. Hughes, Preparation and characterization of
SrCe0.95 Yb0.05 O2.975 hollow fibre membranes, J. Membr. Sci. 193 (2001)
249–260.
[27] X. Tan, S. Liu, K. Li, Preparation and characterization of inorganic hollow fiber
membranes, J. Membr. Sci. 188 (2001) 87–95.
[28] S.P. Deshmukh, K. Li, Effect of ethanol composition in water coagulation bath
on morphology of PVDF hollow fibre membranes, J. Membr. Sci. 150 (1998)
75–85.
[29] L.F. Audrieth, B. Ackerson, The Chemistry of Hydrazine, John Wiley and Sons,
New York, 1951.
[30] A. Furst, R.C. Berlo, S. Hooton, Hydrazine as a reducing agent for
organic compounds (catalytic hydrazine reductions), Chem. Rev. 65 (1965)
51–68.
[31] JCPDS 49-285 file. ICDD PDF-2 Release 2008, Joint Committee on Powder
Diffraction Standards 411 (JCPDS) – International Centre for Diffrac-
tion Data (ICDD), Pennsylvania, 2008, 412 Powder Diffraction File (PDF)
database.
 Author's personal copy

C. Yacou et al. / Journal of Membrane Science 380 (2011) 223–231 231

[32] JCPDS 41-1107 file. ICDD PDF-2 Release 2008, Joint Committee on Powder
Diffraction Standards 411 (JCPDS) – International Centre for Diffraction Data
(ICDD), Pennsylvania, 2008, 412 Powder Diffraction File (PDF) database.
[33] A. Baylet, P. Marecot, D. Duprez, P. Castellazzi, G. Groppi, P. Forzatti, In situ
Raman and in situ XRD analysis of PdO reduction and Pd oxidation supported
on ␥-Al2 O3 catalyst under different atmospheres, Phys. Chem. Chem. Phys. 13
(2011) 4607–4613.
[34] J.J. Chen, E. Ruckenstein, Sintering of palladium on alumina model catalyst in a
hydrogen atmosphere, J. Catal. 69 (1981) 254–273.
[35] W. Zhou, R. Ran, Z. Shao, W. Jin, N. Xu, Effect of a reducing agent for sil-
ver on the electrochemical activity of an Ag/Ba0.5 Sr0.5 Co0.8 Fe0.2 O3− electrode
prepared by electroless deposition technique, J. Power Sources 186 (2009)
244–251.