Perovskite-structured AMn: XB1- xO3 (A = Ca or Ba; B = Fe or Ni) redox catalysts for partial oxidation of methane

Article abstract of DOI:10.1039/c5cy02186c

Methane is one of the simplest and most abundant organic compounds on earth. Methane reforming offers the versatility to produce value-added fuels and chemicals. Unlike typical reforming processes, which suffer from efficiency losses associated with endothermic reforming reactions and/or oxygen/steam generation, chemical looping reforming (CLR) utilizes the redox properties of an oxygen carrier, a.k.a. a redox catalyst, to partially oxidize methane into syngas with its lattice oxygen. The reduced redox catalyst is subsequently regenerated with air, providing in situ air separation with minimal energy penalty. The performance of the CLR process is highly dependent on the redox catalyst. In the current study, CLR performances and the underlying mechanisms for perovskite-structured redox catalysts with a general formula of AMn_xB_1-xO₃ (A = Ca or Ba; B = Fe or Ni) are reported. CaMnO₃ and BaMnO₃ perovskites are worth investigating due to their desirable redox properties and relatively low cost. BaMn_xB_1-xO₃ (B = Fe or Ni) based redox catalysts are shown to be more selective and coke-resistant to methane partial oxidation when compared to CaMn_xB_1-xO₃ (B = Fe or Ni) based redox catalysts. While undoped BaMnO₃ exhibited high selectivity towards syngas, addition of B-site dopants such as Ni or Fe leads to higher oxygen carrying capacity without significantly impacting the coke resistance of the redox catalysts. In contrast, nickel doped CaMnO₃ is significantly more prone to coke formation. As a low cost and environmentally benign option, a BaMn_xFe_1-xO₃ based redox catalyst was tested for CLR operations in a fluidized bed reactor. >95% syngas selectivity was observed with no signs of deactivation over 20 cycles.

Full text of DOI:10.1039/c5cy02186c

Catalysis
Science &
Technology
View Article Online
View Journal
PAPER
Perovskite-structured AMn B O (A = Ca or Ba;
x 1−x 3
B = Fe or Ni) redox catalysts for partial oxidation
of methane
Cite this: DOI: 10.1039/c5cy02186c
Amit Mishra, Nathan Galinsky, Feng He, Erik E. Santiso and Fanxing Li*
Methane is one of the simplest and most abundant organic compounds on earth. Methane reforming of-
fers the versatility to produce value-added fuels and chemicals. Unlike typical reforming processes, which
suffer from efficiency losses associated with endothermic reforming reactions and/or oxygen/steam gener-
ation, chemical looping reforming (CLR) utilizes the redox properties of an oxygen carrier, a.k.a. a redox
catalyst, to partially oxidize methane into syngas with its lattice oxygen. The reduced redox catalyst is sub-
sequently regenerated with air, providing in situ air separation with minimal energy penalty. The perfor-
mance of the CLR process is highly dependent on the redox catalyst. In the current study, CLR perfor-
mances and the underlying mechanisms for perovskite-structured redox catalysts with a general formula
of AMn
x
B
1−x
O
3
(A = Ca or Ba; B = Fe or Ni) are reported. CaMnO
3
and BaMnO
3
perovskites are worth in-
(B = Fe or Ni) based
vestigating due to their desirable redox properties and relatively low cost. BaMn
x
B
1−x
O
3
redox catalysts are shown to be more selective and coke-resistant to methane partial oxidation when
compared to CaMn (B = Fe or Ni) based redox catalysts. While undoped BaMnO exhibited high se-
B O
x 1−x 3
3
Received 15th December 2015,
Accepted 4th February 2016
lectivity towards syngas, addition of B-site dopants such as Ni or Fe leads to higher oxygen carrying capac-
ity without significantly impacting the coke resistance of the redox catalysts. In contrast, nickel doped
CaMnO
BaMn
selectivity was observed with no signs of deactivation over 20 cycles.
3
is significantly more prone to coke formation. As a low cost and environmentally benign option, a
DOI: 10.1039/c5cy02186c
x
Fe1−xO based redox catalyst was tested for CLR operations in a fluidized bed reactor. >95% syngas
3
www.rsc.org/catalysis
oxygen from the air into its lattice. In a subsequent step, the
1
. Introduction
active lattice oxygen in the redox catalyst is released for meth-
ane partial oxidation and syngas generation. Using the redox
catalyst as the reaction intermediate, the two-step cyclic pro-
cess provides in situ air separation with minimal energy pen-
alty. In addition, safety issues encountered with a CH /O
4 2
mixture in the conventional methane partial oxidation
schemes are avoided.
Methane is the simplest and one of the most abundant or-
ganic compounds on earth. As an alternative for combustion
based approaches, methane reforming offers the advantage of
producing value-added fuels and chemicals such as hydrogen,
hydrocarbons, and alcohols.
reforming processes, methane is converted to syngas in the
presence of gaseous oxidants such as steam, oxygen, and/or
carbon dioxide. These processes, however, suffer from effi-
ciency losses due to endothermic steam/CO reforming reac-
tions as well as energy consumption for steam or oxygen
1
2–7
In conventional methane
A critical factor affecting the effectiveness of the CLR
scheme is the performance of the redox catalyst. Satisfactory
redox catalysts possess high redox activity, long-term chemi-
cal and physical stabilities, and excellent syngas selectivity.
In addition, they should be cost-effective, environmentally be-
nign, and highly resistant towards coke formation. Lewis &
Gilliland explored the use of CuO based redox catalysts for
syngas generation nearly 60 years ago. The non-selective na-
2
7–12
generation.
Therefore, novel process schemes that can re-
duce such energy penalties are highly desirable from both en-
vironmental and process economics viewpoints. Compared to
conventional reforming processes, the chemical looping
reforming (CLR) strategy offers a unique approach for meth-
ane partial oxidation in the absence of a gaseous oxidant.
In such a process, a redox-active metal oxide, a.k.a. a redox
catalyst or oxygen carrier, is used to incorporate gaseous
1
7
ture of CuO, however, resulted in poor syngas selectivity. It
was not until the early 2000s that the CLR concept was re-
1
3–16
1
8
investigated in detail. Early studies found Pt doped Ce
2 3
O
to be an effective redox catalyst for methane partial oxida-
1
9,20
tion.
A number of other CeO containing redox catalysts,
2
, SiO , and MgO supports and
2
including those with Al
O
2 3
Department of Chemical and Biomolecular Engineering, North Carolina State
University, 911 Partners Way, Raleigh, North Carolina, USA. E-mail: fli5@ncsu.edu
mixed cerium oxides with Fe, Mn, Cu, and Ni, were also
This journal is © The Royal Society of Chemistry 2016
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
2
1
found to be active and selective.
Compared to ceria
BaIJNO ) , MnIJNO ) ·4H O, and NiIJNO ) ·6H O (Sigma Al-
3
2
3 2
2
3 2
2
containing redox catalysts, first row transition metal oxides
offer the promise of being less expensive with higher oxygen
drich), were dissolved in deionized water. Citric acid (CA,
Sigma Aldrich) was then added (CA: total cations = 2.5), and
the solution was heated at 50 °C for 30 minutes. This results
in a mixed-metal citric acid complex with good dispersion of
metal ions. Ethylene glycol (EG, Sigma Aldrich) was then
added (EG : CA ratio of 1.5) at 80 °C until a gel forms. This
gel was dried in an oven at slightly above boiling temperature
to remove any moisture. Finally, the sample was sintered at
1200 °C for proper phase formation. The samples used in the
current study are listed in Table 1.
1
4,15,21–28
carrying capacities.
Pröll et al. demonstrated the feasibility of the CLR concept.
Using NiO based redox catalysts,
2
2
Syngas yield in excess of 60% was achieved in a 120 kWth cir-
culating fluidized bed (CFB) reactor. The key challenge with
respect to NiO based redox catalysts, however, resides in their
toxicity and lack of coke resistance. Mn and Fe based redox
7
,12,14,15,21,23,24,26–28
catalysts have also been explored.
eral, supported Mn and Fe oxides are less selective than their
NiO counterparts. He et al. reported a composite La0.8Sr0.2FeO
In gen-
3
supported iron oxide for syngas and hydrogen generation in
a two-step redox process with high activity and exceptional
2.2 Redox testing
2
9,30
H
2
yield.
The syngas selectivity for methane partial oxida-
Initial reactivity studies were conducted on a SETARAM
SETSYS Evolution Thermal Gravimetric Analyzer (TGA) with
reactant gases delivered from a computer controlled gas-
mixing panel. The initial tests were conducted at 900 °C. Re-
duction and oxidation steps were performed in 10% methane
tion, however, was limited to approximately 70%. In their re-
cent studies, Shafiefarhood et al. and Neal et al. have investi-
31–34
2 3 x 3
gated an Fe O @La Sr1−xFeO core–shell redox catalyst,
which achieved 94% syngas selectivity with minimal coke for-
mation. Such a high selectivity was realized through tailoring
the surface composition of the redox catalyst and careful con-
trol of reduction and regeneration conditions.
(
5.0 grade) and 10% oxygen (extra dry grade), respectively,
with helium (5.0 grade) as the inert gas. The total gas flow
−1
rate during these steps was 200 mL min to remove external
In addition to first-row transition metal oxides, perov-
mass transfer resistance. Short term redox studies were
conducted over 10 cycles with oxidation and reduction steps
of 20 minutes followed by 5 minutes of purge. Based on short
term redox studies, promising samples were used for long
term redox studies and fixed bed redox studies. The long
term redox studies used 10 minute reduction and oxidation
steps, as such duration was deemed appropriate based on
the short term redox studies. 20 mg of the redox catalyst with
a particle size below 150 microns was used for both short
and long term studies. A quadrupole mass spectrometer
3
skites with a general formula of ABO offer many unique
properties that are potentially suitable for CLR applications.
2
−
These properties include high oxygen anion (O ) mobility,
structural stability, and tunable redox properties. For in-
stance, pure and A-site doped LaFeO and LaMnO have been
3
3
5–40
3
investigated for methane reforming purposes.
These re-
dox catalysts exhibit high syngas selectivity. However, the use
of a rare-earth metal (La) as the primary A-site material as
well as poor coke resistance limits their feasibility for com-
mercial applications. In the current study, we report the CLR
performances and underlying mechanisms for perovskite ma-
(
QMS, MKS Cirrus II) was used to measure the products of
the reduction and oxidation steps. A set of calibrations was
applied to the signals obtained from the QMS. The produc-
tion of CO, CO , and H was calculated by integrating the cal-
x 1−x 3
terials with a general formula of AMn B O (A = Ca or Ba;
B = Fe or Ni). CaMnO and BaMnO are both perovskites that
3
3
2
2
are worth investigating because they possess reasonable oxy-
gen carrying capacity and present relatively low cost. Our
2 2 2
ibrated signals of CO, CO , and H . CO and CO produced
during regeneration of the redox catalyst were used to quan-
studies indicate that CaMnO
exhibit low selectivity. Doping increases selectivity but de-
creases coke resistance of CaMnO based redox catalysts. In
the case of nickel doping, this is attributed to surface Ni. By
comparison, catalysts with a general formula of BaMnxB O3
3
based redox catalysts generally
tify coke formation. H O was calculated using a hydrogen
2
mass balance. Selectivity was calculated using the total
amount of CO produced during the reforming step.
Additional redox testing was also conducted on a 4 mm ID
U-tube reactor loaded with 100 mg of catalyst to simulate a
fixed bed. These tests were conducted at 900 °C with 100
3
1−x
(B = Fe or Ni) are determined to be suitable for partial oxida-
tion of methane. In these catalysts, the presence of Fe and Ni
dopants is found to enhance the oxygen carrying capacity of
the redox catalysts without negatively affecting their coke re-
Table 1 Redox catalysts
sistance. Such redox catalysts are compared to CaMn
B = Fe or Ni) based redox catalysts, which exhibit relatively
low syngas selectivity and coke resistance.
x 1−x 3
B O
(
Catalyst name
Composition
BaMnO
BM
3
BMN25
BMN50
BMF25
BMF50
CM
BaMn0.75Ni0.25
BaMn0.5Ni0.5
BaMn0.75Fe0.25
BaMn0.5Fe0.5O
3
O
3
3
O
2
. Experimental
O
3
2
.1 Redox catalyst preparation
CaMnO
3
A sol–gel method was used to prepare the aforementioned pe-
rovskites. Stoichiometric amounts of metal nitrates, e.g.
CMN
CMF
CaMn0.8Ni0.2
CaMn0.8Fe0.2
O
O
3
3
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2016

View Article Online
Catalysis Science & Technology
Paper
sccm total gas flow rate. Reduction and oxidation were
performed in 10% methane (5.0 grade) and 10% oxygen (ex-
tra dry grade), respectively, balanced by helium (5.0 grade) as
the inert gas. The reduction time used was 20 minutes, and
the oxidation time was 10 minutes. These times were deter-
mined to be sufficient for the completion of syngas produc-
tion and removal of any carbon species.
areas of as-prepared BaMn Ni O , BaMnO , and CaMnO
0
.5 0.5
−1
3
3
3
2
are 0.85, 0.76, and 1.22 m g , respectively. XRD spectra of
as-prepared samples are shown in Fig. 1. Since the crystal
structures for most of the doped samples are not available
in the database, comparing XRD patterns of doped BaMnO3
3 3 3
and CaMnO redox catalysts with pure BaMnO and CaMnO
can provide initial evidence of the doping effect. Table 2
presents the primary and secondary phases determined by
fitting the XRD spectra using the HighScore© software. Sec-
ondary phases are those associated with peaks that cannot
be fully accounted for by the primary peaks, although they
often provide worse fit with XRD spectra than the primary
phases. We note that the XRD spectra for most doped sam-
ples investigated in the current study are not included in
the database. Therefore, less than perfect fittings are often
Fluidized bed redox experiments were conducted at 800 °C
−
1
using a reducing gas of 3.2 mL min methane (5.0 grade)
−
1
and oxidizing gas of 6.8 mL min H O during cycles 1–15
2
−
1
and 4.7 mL min H O during cycles 16–20. The carrier gas
for both reducing and oxidizing steps was 200 mL min
2
−
1
2
N .
2
.3 Sample characterization
Several characterization techniques were used to analyze the
surface, structural, and reactive properties of the redox cata-
lysts. X-ray powder diffraction (XRD) was used to identify the
crystalline phases present in each sample. The XRD pattern
was obtained using a Rigaku SmartLab X-ray diffractometer
observed. Iron doped BaMnO redox catalysts have the best
3
match with BaMn0.7Fe0.3O . This indicates that iron dopants
3
successfully integrated into the BaMnO
3
parent structure. In
the case of nickel doped BaMnO , most of the peaks were
3
3
associated with BaMnO . Unidentified peaks were also seen
with CuK (λ = 0.1542) radiation operating at 40 kV and 44 mA.
α
at 2θ ≈ 28° and 31°. They are likely to result from Ni doping
Scanning was conducted by using a step-wise approach with
a step size of 0.1° over the 2θ range of 20–80° and a scan
time of 4.5 s at each step. HighScore Plus© software was used
for phase matching. Surface areas of the samples were deter-
mined using a Micromeritics ASAP 2020 physi-chemisorption
system with krypton as the adsorption gas.
X-Ray photoelectron spectroscopy (XPS) was used to ana-
lyze surface compositions and oxidation states of fresh and
reduced samples. Reduced samples were prepared using a 2
cycle redox experiment with hydrogen as the reducing gas
in the BaMnO parent structure since the crystal structure of
3
the doped sample is not available in the database. The
slight shift of 2θ angles when compared to undoped
BaMnO also indicates the incorporation of a fraction of Ni
3
cation into the perovskite structure. However, some of the
remaining peaks, albeit insignificant, were associated with
NiO, suggesting that Ni cations were not fully integrated
into the nickel doped redox catalysts. This, however, does
not affect the use of the resulting oxide as a redox catalyst
for CLR.
(5.0 grade). Methane was not used as carbon deposition can
affect surface measurements. The XPS system consists of a
Thermo-Fisher Alpha 110 hemispherical energy analyzer, a
Thermo-Fischer XR3, a 300 W dual-anode X-ray source, and a
chamber with a base pressure of 1 × 10 torr. Survey spectra
used a pass of 100 eV, and narrow scan spectra are obtained
with a pass energy of 20 eV. A Mg anode was used for all
analyses.
3.2. Material screening
Initial characterization of the redox catalysts was conducted
by exposing them for 10 redox cycles using the TGA in differ-
ential bed mode. Since the bulk and surface properties of the
redox catalysts can change during redox reactions, redox cata-
lysts can exhibit multiple reaction stages when reacted with
methane, as reported by Neal et al.
−
9
Hydrogen (5.0 grade) TPR was conducted using a TGA (TA
SDT Q600) with 20% hydrogen balance helium or argon (5.0
grade) at a total flow rate of 200 sccm and a ramp rate of 10
31,33,34
Fig. 2 shows prod-
uct gas concentration profiles as a function of time for both
As can be seen,
CaMn0.8Ni0.2
O
3
3
and BaMn0.75Ni0.25O .
−
1
°
C min to 900 °C. Temperature programmed desorption
CaMn0.8Ni0.2O3 exhibited a reaction scheme that is notably
different that of the BaMn0.75Ni0.25O3 redox catalyst. For
CaMn0.8Ni0.2O , concurrent production of CO and CO was
3 2
first observed as the redox catalyst is reduced. This was
followed by significant coke formation as evidenced by a sig-
2 2
nificant increase in the H /(CO + CO ) ratio and sample
(
TPD) was conducted using the same TGA in either helium or
−
1
argon (5.0 grade) at a ramping rate of 5 °C min to 900 °C
from 100 °C. The weight and differential thermogravimetric
(DTG) results of the redox catalyst were obtained to deter-
mine the amount of oxygen release. The sample was pre-
treated with 20% oxygen at 900 °C as well as during cooling
to 100 °C.
weight gain shown by the TGA. Coke formation was also con-
firmed by the significant amount of CO formed during the
2
oxidation stage, as shown in Fig. 2a. In comparison,
BaMn0.75Ni0.25O exhibited high CO selectivity at the begin-
ning of the reaction, as evidenced by the small CO2 peak.
3
2
3
. Results and discussion
3
.1. Fresh sample characterization
This is followed by an “auto-activation” region with high syn-
Resulting from the high temperature sintering step, all sam-
ples exhibited low BET surface areas. For instance, surface
gas selectivity. Unlike CaMn0.8Ni0.2O , coke formation was
3
minimal during the methane–BaMn0.75Ni0.25O reaction.
3
This journal is © The Royal Society of Chemistry 2016
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Fig. 1 XRD spectra of (a) CMF (i), CMN (ii) and CM (iii); (b) BMF25 (i), BMF50 (ii) and BM (iii); (c) BMN25 (i) and BMN50 (ii).
Fig. 3a summarizes the selectivity to CO (SCO), defined as
for all the redox catalysts. It can be
structure. For instance, CaMnO samples with Ni and Fe dop-
ants exhibited over 20% higher selectivity when compared
with the undoped sample. This can be explained by the for-
mation of a reduced Ni or Fe phase on the surface of the
3
CaMnO oxides. The presence of metallic Ni or Fe phases,
3
,
seen that BaMnO
3 3
doped samples outperformed CaMnO
doped samples. This suggests that a BaMnO parent structure
3
which are known to be active for methane activation, can
4
promoted selective oxidation of CH to CO. In addition, the
lead to improved syngas selectivity. A similar effect has been
effect of dopant on syngas selectivity was not significant for
BaMnO based redox catalysts. For instance, BaMnO shows
42
3
reported for a Pt doped CaTiO catalyst for CO oxidation.
3
3
Iron doped BaMnO , however, showed the highest selectivity
3
a CO selectivity of 90%. All nickel doped samples exhibited a
selectivity of 93% ± 0.5%. In comparison, iron doped samples
exhibited a selectivity of 95.2% ± 0.5%. The insignificant
effect of Ni/Fe doping can be explained provided that the cat-
alyst surface is enriched with A-site metal cations. Such a
(
∼95%) and exhibited properties that are desirable over a
pure BaMnO
3
catalyst. Fig. 3b summarizes the yield of CO,
expressed as
. Even though the selectivity
phenomenon was reported for BaZrO , where preferential
3
4
x 1−x 3
of the BaMn B O redox catalysts remained relatively
1
surface aggregation of Ba was observed. Further discussion
of dopant effects on surface and bulk properties of the redox
catalyst is provided in section 3.3. It is also noted that the
presence of Ni and Fe dopants had more pronounced effects
unchanged with Ni or Fe doping, the yield of CO increased
with doping. The increase in CO yield resulted from an in-
creased amount of active lattice oxygen from the doped cata-
lyst. In fact, Ni and Fe dopants can destabilize the BaMnO3
parent structure, thereby increasing the amount of usable lat-
tice oxygen. Iron doping, in particular, exhibited this effect:
BaMn0.75Fe0.25O and BaMn0.5Fe0.5O showed the highest CO
3
on the selectivity of redox catalysts with a CaMnO parent
3
3
Table 2 Phase compositions identified in as-prepared redox catalysts
yields.
Besides CO selectivity, resistance towards coke formation
represents another important performance parameter for re-
dox catalysts. The amount of coke formation is determined
Catalyst
Primary phase
Secondary phase
BMN25
BMN50
BMF25
BMF50
BM
CMN
CMF
CM
BaMnO
BaMnO
3
NiO
NiO
None
2 4
BaFe O
None
NiO
None
None
3
BaMn0.7Fe0.3
BaMn0.7Fe0.3
O
O
3
2
by integrating the amounts of CO and CO formed during
3
the oxidation of the reduced redox catalyst. Since coke forma-
BaMnO
3
31
tion tends to occur upon depletion of active lattice oxygen,
CaMnO
CaMnO
CaMnO
3
3
3
each catalyst is reduced for 20 minutes to allow complete re-
duction and to determine the maximum coke formation.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2016

View Article Online
Catalysis Science & Technology
Paper
3 3
Fig. 2 Product distributions from typical redox cycles of (a) CaMn0.8Ni0.2O and (b) BaMn0.75Ni0.25O .
4 2
Fig. 3 (a) CO selectivity and (b) yield for the redox catalysts (cycle 10 of the short term redox study, 10% CH and 10% O at 900 °C).
Fig. 4 summarizes the amount of coke formation for the re-
dox catalysts investigated in the current study. When com-
cation may be enriched on the surface of the redox catalyst
for the BaMn system. Overall, BaMn0.75Fe0.25O
3
x 1−x
B
O
3
pared to MgAl
2
O
4
supported Fe and Ni oxides which exhibit
exhibited the highest coke resistance, forming merely 4 mg
of coke on each gram of the redox catalyst.
−
1
43
>430 mg g catalyst coke formation, all the perovskite
based redox catalysts exhibited significantly improved coke
resistance. The dopant effect on catalyst coke resistance
exhibited similar trends as that for CO selectivity, i.e. Ni or
Fe dopants have an insignificant effect on coke resistance for
BaMn B O based redox catalysts, whereas Ni doping in a
x 1−x 3
CaMnO3 parent structure leads to significantly weakened
coke resistance. This again indicates that the A-site metal
3
.3. Surface composition effects
The presence of reduced Ni on redox catalyst surfaces is
anticipated to enhance the catalyst's activity for methane acti-
vation and CO selectivity. In the meantime, it can lead to in-
creased coke formation. Such effects are apparent for redox
catalysts with a CaMnO parent structure (Fig. 3 and 4). In
3
3
contrast, Ni doping in a BaMnO parent structure exhibited
an insignificant impact on CO selectivity and coke formation.
As discussed earlier, this may point to a significantly differ-
ent surface metal composition when compared to the bulk,
i.e. the Ni dopant is significantly underrepresented on the
surface. Compositions of the outermost layers for many
mixed oxides have been reported to deviate from its bulk
4
4
stoichiometry. To investigate this phenomenon, near sur-
face compositions of as-prepared BaMn0.5Ni0.5O3 (fresh
3
BMN50) and CaMn0.75Ni0.25O (fresh CMN25) and reduced
BaMn0.5Ni0.5O3 (reduced BMN50) and CaMn0.75Ni0.25O3 (re-
duced CMN25) are examined using XPS. While more surface
sensitive techniques such as low-energy ion scattering (LEIS)
are desirable, XPS provided a convenient approach to deter-
mine elemental compositions within 1–3 nm from the
Fig. 4 Coke formation for the redox catalysts (cycle 10 of the short
term redox study, 10% CH and 10% O at 900 °C).
4 2
This journal is © The Royal Society of Chemistry 2016
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
surface. Surface deviation is shown in Fig. 5 and is calculated
as follows:
As can be seen, the surface is over-expressed with A-site
cations for all samples. In comparison, Ni is under-
represented by an SD below 0.5 for all samples. While the as-
prepared samples have similar degrees of surface deviation
3
(SD), reduced CaMn0.75Ni0.25O exhibited five times the sur-
face deviation for Ni when compared to BaMn0.5Ni0.5O . In
contrast, barium and oxygen were enriched by an SD of 1.22
in BaMn0.5Ni0.5O . This indicated that the surface of doped
3
3
−1
Fig.
6
DTG of
H
2
(20%) TPR (10° min
3
, and CaMn0.8Ni0.2O .
) of redox catalysts
BaMn0.75Ni0.25
O
3
, BaMnO
3
BaMnO samples, when reduced, is practically covered with
3
BaO. This is consistent with the observation that Ni and Fe
doping had insignificant effects on CO selectivity and coke
formation for BaMnO3 based redox catalysts. The dopants,
however, can change the bulk properties of the perovskite,
allowing more lattice oxygen to be released during the redox
cycles.
tightly bonded in such a structure when compared with
CaMnO . This could contribute to higher selectivity observed
3
3
for the BaMnO based redox catalysts. The three peaks of
BaMnO are consistent with Mn changing oxidation states in
3
the following order: 4+ → 3+ → 8/3+ → 2+. The two extra peaks
seen in BaMn0.75Ni0.25O are likely to result from an oxidation
3
state change for Ni either in the perovskite structure or as
separate NiO/Ni phases as indicated by XRD (Fig. 1).
3
.4. Effects of parent structure stability
Another interesting observation is that all redox catalysts with
2
To supplement H -TPR experiments, temperature pro-
a BaMnO parent structure exhibited significantly higher CO
grammed desorption is conducted to measure the amount of
oxygen that can evolve from the redox catalyst under an inert,
non-reducing atmosphere. Shafiefarhood et al. showed that
surface oxygen species contributed to deep oxidation. DTG re-
sults of these TPD experiments are shown in Fig. 7. These ex-
3
selectivity when compared to those with a CaMnO structure.
3
3
4
Recent studies of a La
x
Sr1−xFeO
3
containing redox catalyst in-
dicated that electrophilic surface oxygen species, which may
evolve from lattice oxygen species in the bulk, play an impor-
34
tant role in the non-selective oxidation of methane. There-
fore, the stability of the parent structure, which affects the
evolution of lattice oxygen to electrophilic oxygen species,
can play a role in the selectivity of the redox catalysts. TPR
and TPD experiments are therefore conducted to investigate
periments showed little to no oxygen desorption with BaMnO
3
structured redox catalysts. However, CaMn0.8Ni0.2O showed a
3
noticeably higher amount of oxygen desorption. Since the de-
sorption of gaseous oxygen from a perovskite lattice occurs
through lattice oxygen evolution to surface oxygen species
2
−
−
2−
such relationships. Fig. 6 illustrates the H -TPR results for
O
→ O → O
2
→ O
2
, one would anticipate that the non-
based redox catalyst
2
3
CaMn0.8Ni0.2O , BaMnO
3
3
, and BaMn0.75Ni0.25O . These results
selective oxygen species on the CaMnO
3
indicate that all the active lattice oxygen species are removed
surface would be far more abundant than that on BaMnO3
based redox catalysts. Once again, these findings are consis-
tent with the higher amount of deep oxidation seen in
from CaMn0.8Ni0.2O below 700 °C, where TPR peaks occurred
3
at 425 °C and 575 °C. In contrast, the TPR peak temperatures
for BaMn0.75Ni0.25O and BaMnO appeared around 400–550
CaMnO based redox catalysts.
3
3
3
°C, 700 °C, and 825 °C. The higher reduction peaks for
BaMnO based samples indicate that lattice oxygen is more
3
°
−1
Fig. 7 DTG of TPD (5 min ) of redox catalysts BaMn0.75Ni0.25
3
O ,
Fig. 5 Surface deviation of redox catalysts.
BaMnO , and CaMn0.8Ni0.2
3
3
O .
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2016

View Article Online
Catalysis Science & Technology
Paper
Table 3 Change in selectivity, coke, and CO yield for nickel doped redox
catalysts from cycles 5 to 25
to retard coke formation reactions such as Boudouard reac-
tion and methane decomposition.
3
While nickel doped BaMnO redox catalysts exhibit desir-
BMN25 BMN25 BMN50 BMN50
able performance, they may not be applicable industrially
due to their inherit toxicity and relatively high cost. In com-
parison, iron doped BaMnO would be advantageous from
3
Cycle number
CO (%)
Coke (mg g catalyst)
5
25
5
25
S
92.0
6.4
3.2
91.9
6.6
3.1
93.1
7.2
5.2
92.5
9.5
4.8
−1
−1
CO yield (mmol g catalyst)
both cost and environmental sustainability standpoints. It is
therefore desirable to determine the underlying cause for de-
activation of the iron doped redox catalysts. The loss of redox
activity for iron doped BaMnO3 redox catalysts can result
from two possible mechanisms: (i) lack of phase regene-
rability: a redox catalyst underwent an irreversible phase
change during the reduction reaction. One would anticipate a
decrease in redox activity if the reduced phases cannot be
reoxidized to form the original phases of the redox catalyst;
3
.5. Cyclic redox studies
Another important factor for a redox catalyst is its recyclabil-
ity. The recyclability of the catalysts is tested with 10 minute
reduction/oxidation half cycles. Table 3 summarizes the
change in CO selectivity, yield, and coke formation for nickel
doped BaMn1 Ni O redox catalysts from cycles 5 to 25. Both
−x
x 3
(
ii) sintering: severe loss of surface area and pore volume
Ni doped catalysts showed stability, with BaMn0.75Ni0.25O
3
leads to deactivation for oxygen carriers with low ionic and/or
electronic conductivity. To determine the phase regenerability
of the redox catalyst, XRD analysis was performed on
performing exceptionally well: both selectivity and CO yield
showed almost no decline within 25 cycles. BaMn0.5Ni0.5O3
also showed a marginal decline in selectivity and CO yield.
We note that the CO yields shown in Table 3 are slightly dif-
ferent from those in Fig. 3 due to slight differences in operat-
ing conditions.
Long term redox studies were also conducted for iron
doped catalysts, with the results summarized in Table 4. In
contrast to nickel doped catalysts, iron doped catalysts
3 3
BaMn0.5Fe0.5O (Fig. 9). As can be seen, BaMnO was fully reg-
enerated after oxidation. It is therefore concluded that the re-
duced iron doped BaMnO3 redox catalysts can be fully reg-
enerated to their original phases when oxidized. Therefore,
lack of phase regenerability is not likely to be the cause of de-
activation. A further observation of the cycled redox catalyst
sample indicated significant aggregation of the solid particles,
and this indicated sintering and loss of surface area. We note
that the iron based redox catalyst has the highest oxygen car-
rying capacity among all the redox catalyst samples (Fig. 3b).
Therefore, one would anticipate significant heat release dur-
ing the regeneration reaction. The large heat release coupled
with the ineffectiveness for heat removal in a packed bed and
relatively low melting point of reduced iron (∼1200 °C) can
lead to significant sintering of the redox catalyst sample. To
confirm this, we performed a cyclic redox experiment using a
U-tube fixed bed reactor operated in integral bed mode. The
results of the integral bed experiment are shown in Fig. 10.
The results indicated that the iron doped catalysts maintained
high performance as redox catalysts under fixed bed opera-
tions, with the exception of BMF25 having slightly lower selec-
tivity. However, deactivation remained with both selectivity
3
showed signs of deactivation. Both BaMn0.75Fe0.25O and
BaMn0.5Fe0.5O showed a decrease in selectivity and CO yield.
3
This is a seen as a result of the reaction slowing down with
cycling and can be illustrated in Fig. 8. The notable decrease
in coke formation can also be explained by the slower reduc-
tion kinetics, i.e. the presence of active lattice oxygen tends
Table 4 Change in selectivity, coke, and CO yield for iron doped redox
catalysts from cycles 5 to 25
BMF25 BMF25 BMF50 BMF50
Cycle number
CO (%)
Coke (mg g catalyst)
5
25
5
25
86.9
1.3
S
88.2
1.2
2.7
76.4
1.2
0.9
95.1
10.5
6.8
−1
−1
CO yield (mmol g catalyst)
3.63
Fig. 8 (a) Gaseous product profile during the methane conversion stage at the 5th cycle and (b) 25th cycle (10% CH
BaMn0.5Fe0.5
4 2
and 10% O at 900 °C) for
3
O .
This journal is © The Royal Society of Chemistry 2016
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
and CO yield decreasing. It is apparent that the reduction
rates became slower as evidenced by the decrease in CO yield
from cycle 2 to cycle 5. In addition, cycle 2 showed comple-
tion of reaction stage 3 (the highly selective reaction region),
while cycle 5 does not (Fig. 11). Inspection of the post-
experiment samples again confirmed sintering of the redox
catalyst, which is the likely cause of deactivation.
Fluidized bed testing. Investigation of a fluidized bed ap-
plication will gauge the effect of sintering and determine the
applicability of iron doped BaMnO redox catalysts. Sintering
3
is expected to be minimized because of the excellent gas–
solid heat transfer in fluidized beds. To further minimize
sintering, the sample is partially regenerated with steam to
reduce the rate and degree of regeneration for the redox
Fig. 9 XRD patterns of (i) BMF50 after the 10th cycle reduction and (ii)
the oxidized phases after the 10th cycle reduction.
4 2
Fig. 10 (a) CO selectivity, CO yield, and (b) coke formation during fixed bed experiments (10% CH and 10% O at 900 °C).
3 3 3 3 4
Fig. 11 (a) Cycle 2 of BaMn0.75Fe0.25O , (b) cycle 5 of BaMn0.75Fe0.25O , (c) cycle 2 of BaMn0.5Fe0.5O , and (d) cycle 5 of BaMn0.5Fe0.5O (10% CH
and 10% O at 900 °C) during fixed bed experiments.
2
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2016

View Article Online
Catalysis Science & Technology
Paper
Among BaMnxB O based redox catalysts, BaMn0.5Fe0.5O3
1−x
3
exhibits the highest syngas selectivity and oxygen carrying ca-
pacity. The catalyst, however, deactivates over multiple redox
cycles in fixed bed testing. It was determined that sintering
of the redox catalyst, resulting from the lack of heat dissipa-
tion in the fixed bed, is the cause of deactivation. However,
high syngas selectivity, methane conversion, and long term
stability were achieved in a fluidized bed using steam as the
x 1−x 3
oxidant. It is therefore concluded that BaMn B O based re-
dox catalysts, especially those with Fe as B-site dopants, can
potentially be effective for chemical looping reforming of
methane. However, care must be taken to prevent excessive
sintering of the redox catalysts during redox operations.
Fig. 12 CO selectivity and methane conversion during fluidized bed
experiments.
Acknowledgements
catalysts. This helps to lower the heat released during the re-
generation stage. As can be seen from Fig. 12, the redox cata-
lyst maintained excellent recyclability over 20 cycles. CO se-
lectivity as high as 95% was achieved with average methane
conversion exceeding 95%. The high recyclability of the
This work was supported by the U.S. Department of Energy
(Award No. FE0011247). The authors acknowledge the North
Carolina State University for the use of their Analytical Instru-
mentation Facility (AIF), which is supported by the State of
North Carolina and the National Science Foundation, the
Guangzhou Institute of Energy Conversion, Chinese Academy
of Sciences, and Dr. Fang He and Kun Zhao of the Guangzhou
Institute of Energy Conversion, Chinese Academy of Sciences.
3
BaMn0.5Fe0.5O catalyst in the fluidized bed mode with lim-
ited regeneration again confirms that sintering is likely to be
the primary cause of the deactivation. In addition, a fluidized
bed operating mode is potentially suitable for chemical
looping reforming of methane using iron doped BaMnO3
based redox catalysts. Other approaches to increase the
References
x 3
sintering resistance of BaMn1−xFe O based redox catalysts in-
clude improving its lattice oxygen conductivity through
doping or inclusion of a compatible secondary phase, as
discussed in our previous article.
1 BP Statisical Review of World Energy June 2015, http://www.
bp.com/en/global/corporate/energy-economics/statistical-review-
of-world-energy/natural-gas-review-by-energy-type/natural-gas-
production.html, (accessed December 2015).
4
3
2
G. P. Van der Laan and A. Beenackers, Catal. Rev.: Sci. Eng.,
999, 41, 255–318.
4
. Conclusion
1
The current study investigates redox catalysts with a general
formula of AMnxB O (A = Ca or Ba; B = Fe or Ni) for chemi-
3 P. J. A. Tijm, F. J. Waller and D. M. Brown, Appl. Catal., A,
2001, 221, 275–282.
1−x
3
cal looping reforming of methane. Redox catalysts with a
4 J. P. Lange and P. J. A. Tijm, Chem. Eng. Sci., 1996, 51,
2379–2387.
BaMnO parent structure exhibited considerably higher syn-
3
gas selectivities when compared to a CaMnO3 based redox
5 J. Fox, Catal. Rev.: Sci. Eng., 1993, 35, 169–212.
6 B. C. Enger, R. Lodeng and A. Holmen, Appl. Catal., A,
2008, 346, 1–27.
7 A. P. E. York, T. C. Xiao and M. L. H. Green, Top. Catal.,
2003, 22, 345–358.
8 J. R. Rostrup-Nielsen, J. Sehested and J. K. Nørskov, in Advances
in Catalysis, Academic Press, 2002, vol. 47, pp. 65–139.
9 J. R. H. Ross, Catal. Today, 2005, 100, 151–158.
10 IEA Hydrogen Production and Storage, http://www.iea.org/
publications/freepublications/publication/hydrogen.pdf, (accessed
December 2015).
catalyst. Moreover, BaMn
more coke-resistant to methane partial oxidation than
CaMnxB O . While undoped BaMnO also exhibited high
x 1−x 3
B O based redox catalysts are
1−x
3
3
selectivity towards syngas, addition of B-site dopants leads to
higher oxygen carrying capacity (and hence syngas yield)
without significantly impacting coke resistance. In contrast,
3
nickel doped CaMnO is significantly more prone to coke for-
mation. The higher coke resistance of the BaMnO based re-
3
dox catalysts can be explained by a BaO terminated surface
which minimizes the exposure of metallic Ni or Fe, thereby
inhibiting the methane cracking and Boudouard reactions.
When compared with a BaMnO3 based redox catalyst,
11 S. C. Tsang, J. B. Claridge and M. L. H. Green, Catal. Today,
1995, 23, 3–15.
CaMnO
3
contains significantly higher amounts of loosely
12 J. Adanez, A. Abad, F. Garcia-Labiano, P. Gayan and L. F. de
Diego, Prog. Energy Combust. Sci., 2012, 38, 215–282.
13 L.-S. Fan, Chemical Looping Systems for Fossil Energy
Conversions, Wiley, 2011.
14 A. Thursfield, A. Murugan, R. Franca and I. S. Metcalfe,
Energy Environ. Sci., 2012, 5, 7421–7459.
bound oxygen species, as indicated by TPR and TPD experi-
ments. The facile evolution of such oxygen species into non-
selective, electrophilic oxygen species under the reaction con-
ditions can explain the lower syngas selectivity for CaMnO3
based redox catalysts.
This journal is © The Royal Society of Chemistry 2016
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
1
5 A. Murugan, A. Thursfield and I. S. Metcalfe, Energy Environ.
Sci., 2011, 4, 4639–4649.
30 F. He and F. Li, Int. J. Hydrogen Energy, 2014, 39,
18092–18102.
1
1
6 M. Ryden, A. Lyngfelt and T. Mattisson, Fuel, 2006, 85, 1631–1641.
7 W. K. Lewis, E. R. Gilliland and W. A. Reed, Ind. Eng. Chem.,
31 L. M. Neal, A. Shafiefarhood and F. Li, ACS Catal., 2014, 4,
3560–3569.
1
949, 41, 1227–1237.
32 A. Shafiefarhood, N. Galinsky, Y. Huang, Y. Chen and F. Li,
ChemCatChem, 2014, 6, 790–799.
1
8 T. Mattisson and A. Lyngfelt, Applications of chemical-looping
combustion with capture of CO , Second Nordic Minisymposium
33 L. Neal, A. Shafiefarhood and F. Li, Appl. Energy, 2015, 157,
391–398.
2
on CO Capture and Storage, Göteborg, Sweden, 2001.
2
1
2
2
2
2
2
2
2
2
2
9 K. Otsuka, Y. Wang, E. Sunada and I. Yamanaka, J. Catal.,
34 A. Shafiefarhood, J. C. Hamill, L. M. Neal and F. Li, Phys.
Chem. Chem. Phys., 2015, 31297–31307.
35 M. Ryden, A. Lyngfelt, T. Mattisson, D. Chen, A. Holmen and
E. Bjorgum, Int. J. Greenhouse Gas Control, 2008, 2, 21–36.
36 F. He, X. Li, K. Zhao, Z. Huang, G. Wei and H. Li, Fuel,
2013, 108, 465–473.
37 A. Evdou, V. Zaspalis and L. Nalbandian, Fuel, 2010, 89,
1265–1273.
38 K. Zhao, F. He, Z. Huang, A. Zheng, H. Li and Z. Zhao, Int. J.
Hydrogen Energy, 2014, 39, 3243–3252.
39 O. Mihai, D. Chen and A. Holmen, Ind. Eng. Chem. Res.,
2011, 50, 2613–2621.
40 D. Xiaoping and Y. Changchun, Chin. J. Catal., 2011, 32,
1411–1417.
41 I. E. Wachs and K. Routray, ACS Catal., 2012, 2, 1235–1246.
42 H. Tanaka, M. Taniguchi, M. Uenishi, N. Kajita, I. Tan, Y.
Nishihata, J. ichiro Mizuki, K. Narita, M. Kimura and K.
Kaneko, Angew. Chem., Int. Ed., 2006, 45, 5998–6002.
43 N. L. Galinsky, A. Shafiefarhood, Y. Chen, L. Neal and F. Li,
Appl. Catal., B, 2015, 164, 371–379.
1
998, 175, 152–160.
0 K. Otsuka, T. Ushiyama and I. Yamanaka, Chem. Lett.,
993, 22, 1517–1520.
1 Q. Zafar, T. Mattisson and B. Gevert, Ind. Eng. Chem. Res.,
005, 44, 3485–3496.
1
2
2 T. Pröll, J. Bolhàr-Nordenkampf, P. Kolbitsch and H.
Hofbauer, Fuel, 2010, 89, 1249–1256.
3 T. Mendiara, J. M. Johansen, R. Utrilla, P. Geraldo, A. D.
Jensen and P. Glarborg, Fuel, 2011, 90, 1049–1060.
4 T. Mendiara, J. M. Johansen, R. Utrilla, A. D. Jensen and P.
Glarborg, Fuel, 2011, 90, 1370–1382.
5 S. Bhavsar, M. Najera and G. Veser, Chem. Eng. Technol.,
2
012, 35, 1281–1290.
6 M. Najera, R. Solunke, T. Gardner and G. Veser, Chem. Eng.
Res. Des., 2011, 89, 1533–1543.
7 J. P. E. Cleeton, C. D. Bohn, C. R. Müller, J. S. Dennis and
S. A. Scott, Int. J. Hydrogen Energy, 2009, 34, 1–12.
8 C. D. Bohn, C. R. Mueller, J. P. Cleeton, A. N. Hayhurst, J. F.
Davidson, S. A. Scott and J. S. Dennis, Ind. Eng. Chem. Res.,
2
008, 47, 7623–7630.
44 S. V. Merzlikin, N. N. Tolkachev, L. E. Briand, T. Strunskus,
C. Wöll, I. E. Wachs and W. Grünert, Angew. Chem., Int. Ed.,
2010, 49, 8037–8041.
2
9 N. L. Galinsky, Y. Huang, A. Shafiefarhood and F. Li, ACS
Sustainable Chem. Eng., 2013, 1, 364–373.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2016