Co, Fe, and Mn in La-perovskite oxides for low temperature thermochemical CO2 conversion

Article abstract of DOI:10.1016/j.cattod.2019.04.028

Perovskite oxides of the form ABO₃ have shown substantial promise in reverse water gas shift chemical looping (RWGS-CL) process for low temperature thermochemical CO₂ conversion to CO. Transition metals on the ‘B’ site of these perovskite oxides hold the key to tuning the material properties essential for efficient CO₂ conversion. The role of Co, Fe and Mn in LaBO₃ has been investigated through a combined theoretical and experimental approach. Intrinsic oxygen vacancy formation characteristics of these materials and the electronic charge distribution were explored via ab-initio density functional theory (DFT) simulations, while the microscale properties like crystallite size and CO₂ conversion yield were probed experimentally. Through this multiscale study, the material properties that govern the stable and enhanced CO₂ conversion phenomenon by Fe rich perovskites as opposed to Co and Mn rich phases are differentiated.

Full text of DOI:10.1016/j.cattod.2019.04.028

Accepted Manuscript
Title: Co, Fe, and Mn in La-perovskite oxides for low
temperature thermochemical CO2 conversion
Authors: Adela E. Ramos, Debtanu Maiti, Yolanda A. Daza,
John N. Kuhn, Venkat R. Bhethanabotla
PII:
S0920-5861(18)31506-2
DOI:
Reference:
https://doi.org/10.1016/j.cattod.2019.04.028
CATTOD 12123
To appear in:
Catalysis Today
Received date:
Revised date:
Accepted date:
1 November 2018
19 February 2019
7 April 2019
Please cite this article as: Ramos AE, Maiti D, Daza YA, Kuhn JN, Bhethanabotla
VR, Co, Fe, and Mn in La-perovskite oxides for low temperature thermochemical CO2
conversion, Catalysis Today (2019), https://doi.org/10.1016/j.cattod.2019.04.028
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Co, Fe, and Mn in La-perovskite oxides for low temperature thermochemical CO2
conversion
Adela E. Ramos, Debtanu Maiti, Yolanda A. Daza, John N. Kuhn*, Venkat R.
Bhethanabotla*
Department of Chemical & Biomedical Engineering, University of South Florida, Tampa,
Florida 33620, United States
*Email: jnkuhn@usf.edu; bhethana@usf.edu
Graphical Abstract
Highlights
• Low temperature (600°C) CO2 conversion via reverse water gas shift chemical looping
• Stable operation for several conversion cycles – promising for industrial application
• Perovskite oxides with three transition metals in B-site demonstrate CO2 conversion
• Role of transition metals (Fe, Co, Mn) explored through multiscale approach
• Preferential surface segregation of Co and Fe found favorable for CO2 conversion
Abstract
Perovskite oxides of the form ABO3 have shown substantial promise in reverse water gas
shift chemical looping (RWGS-CL) process for low temperature thermochemical CO2
1

conversion to CO. Transition metals on the ‘B’ site of these perovskite oxides hold the
key to tuning the material properties essential for efficient CO2 conversion. The role of
Co, Fe and Mn in LaBO3 has been investigated through a combined theoretical and
experimental approach. Intrinsic oxygen vacancy formation characteristics of these
materials and the electronic charge distribution were explored via ab-initio density
functional theory (DFT) simulations, while the microscale properties like crystallite size
and CO2 conversion yield were probed experimentally. Through this multiscale study, the
material properties that govern the stable and enhanced CO2 conversion phenomenon
by Fe rich perovskites as opposed to Co and Mn rich phases are differentiated.
Keywords: carbon dioxide conversion; carbon monoxide production; reverse water gas
shift chemical looping; RWGS-CL; perovskites.
1. Introduction
Despite $20 billion in investment [1], Carbon Capture & Storage (CCS) is still an
emerging technology, which is limited by technical challenges to lower costs and risks in
order to achieve viability. Alternatively, Carbon Capture & Utilization/Use (CCU) is an
attractive alternative option due to its closed carbon cycle, but feasibility is currently
lacking due to low performance metrics (low CO2 conversion rates on catalyst amount
and/or energy input bases) for various conversion technologies. Due to reliance on fossil
2

fuels and the investments in CCS, a vast amount of CO2 is or could be available. An
estimated 16.67%, 3.73% and 0.28% of total CO2 emissions, respectively, could be
reduced if CO2 was the sole carbon feedstock for gasoline, methanol, and oil derived
chemicals, assuming carbon-free energy sources [2]. However, CO2 recycling at
reasonable scales is limited [3].
Processes transforming CO2 by first thermochemically converting CO2 to CO
followed by CO hydrogenation to liquid fuels (i.e., methanol, diesel, jet fuel) via Fischer-
Tropsch Synthesis (FTS) were found to have higher potential solar energy recovery than
technologies involving biomass, algae, or direct photosynthetic conversion of CO2 to fuel
[4]. CO2 can also be converted by thermochemical, electrochemical, and photochemical
approaches. Thermochemical processes include solar thermochemical cycles [5-7], and
reverse water gas shift (RWGS) catalytic reaction [2-4], and chemical looping [8-12]. In
the RWGS-chemical looping (RWGS-CL) approach, a metal oxide (MOx) or mixed metal
oxides such as perovskite (ABO3) and spinel (AB2O4) oxides are chemically reduced with
hydrogen and subsequently oxidized with CO2, producing CO as the material re-oxidizes.
As the reduction and oxidation occurs in cycles, this stoichiometric reaction avoids
thermodynamic restraints of the catalytic reaction, provides inherent separation of
products, and avoids formation of methane since the carbon and hydrogen species are
never fed simultaneously. As a result, the RWGS-CL process is predicted to be more
energy efficient than its catalytic counterpart [13]. Since the reduction is thermochemical,
as opposed to only thermal in solar thermochemical cycles, operating temperatures of
RWGS-CL are lower and range from 400 to 950 °C [8-11], and even isothermal operation
has been reported on certain oxides [12, 14]. The selection of materials to balance the
3

reduction by hydrogen, a strong reducing agent, and oxidation by CO2, a weak oxidizing
agent, is the key in achieving both steps at the same temperature.
Due to their redox properties including non-stoichiometric amounts of oxygen and
their ability of accommodating multiple substitutions of A and B site metals
simultaneously, perovskite oxides (ABO3, where A is a rare earth or alkaline earth metal
and B is a transition metal) are ideal material platforms to design new formulations for
thermochemical and chemical looping processes. In thermochemical cycles, LaxSr1-
xMnyAl1-yO3 outperformed the state-of-the art ceria, producing six times more CO at the
same conditions [15]. In addition, La0.6Sr0.4Cr0.8Co0.2O3 had comparable results to ceria
though at 300 °C lower temperature [16] and La0.6Ca0.4Fe0.4Mn0.6O3 yielded rates
significantly improved compared to ceria [5]. Similarly, selection of elements on the A-
and B-sites yielded improved rates for CO2 conversion and/or lower operation
temperatures in the isothermal RWGS-CL process compared to initial results [12, 14]. In
these cases, the composition of the perovskite oxide phase was controlled to adjust the
redox properties, often quantified in terms of the energy of oxygen vacancy formation [12,
17-19], of the materials under the operation conditions.
There are two primary strategies for tuning the oxygen vacancy formation energy
by composition. First, partial substitution of a divalent atom (e.g., Sr) on the normal
trivalent A-site forces the transition metal on the B-site to increase its valence charge
and/or cause oxygen vacancies to form. Second, population of the B-site with transition
metals of various redox abilities dictates the degree of charge balance by its oxidation or
formation of oxygen vacancies. As indicated above, these two approaches are often
examined simultaneously and more than two metals can be added on each site [20].
4

Alternatively, in this work, the oxygen vacancy formation energy is tuned by B-site
composition through various Fe:Mn:Co ratios while the A-site is fixed with a single
element (La). The goals are to determine the effect of composition on the redox properties
and evaluate if the B-site doping can compensate for the reducibility lost by eliminating
Sr from the A-site. The results indicate it is possible and an isothermal RWGS-CL process
can be achieved, similar to previous studies [12, 14] on Sr doped La perovskite oxides.
To aid in the evaluation, quantum scale calculations of the electronic structure of the
material was conducted to support the microscale experimental results. The combined
multiscale investigation of these perovskite oxides elucidates the trail forward to intricate
structure-function understanding and design of advanced multifunctional materials of the
future.
2. Experimental
2.1 Material synthesis
LaCoxFeyMn1-x-yO3 perovskites were synthesized using the Pechini Method [21],
as done previously [10, 11, 14, 22]. Three samples (Table S1) were synthesized varying
the molar compositions of the metals on the B site, while the A-site was kept as 100% La.
The specific formulations are LaCo0.5Fe0.25Mn0.25O3 (X=0.50 and Y =0.25, named Co50),
LaCo0.25Fe0.50Mn0.25O3 (Y=0.50 and X=0.25, named Fe50), and LaCo0.25Fe0.25Mn0.50O3
(X=Y=0.25, named Mn50). For each synthesis, an aqueous solution of citric acid (Aldrich,
ACS grade ≥99.5%) and stoichiometric amounts, as denoted by the desired composition
of the perovskite oxide, of the corresponding precursors La(NO3)3 (Alfa Aesar, 99.9%),
Fe(NO3)3 (Aldrich, ≥ 98%), Co(NO3)3 (Acros Organics, 99%), MnCO3 (Alfa Aesar, 99.9%)
5

were simultaneously stirred and heated to 60 °C for 2 h. Subsequently, ethylene glycol
(grade ≥99.9; 40 mL) was added. Per literature [23], the ratio of La : B-site : Citric Acid :
Ethylene Glycol was 1:1:10:40. For the poly-esterification process, the solution was
heated to 90 °C and then stirred for 7 h. The sol-gel was transferred to a crucible and
heated to 450 °C for 2 h, with a ramp rate of 20 °C/min. The resulting solid was crushed
and calcined at 1200 °C for 6 h.
2.2 Material characterization
X-ray powder diffraction (XRD) patterns were collected on a Bruker X-ray
diffractometer with Cu Kα radiation (λ = 1.54 Å). 2θ angles were measured from 20 to
100° at a step size of 0.0102° with collection time per step of 1.2 s. Lattice parameters
were calculated by solving the least squares of 2θ, using Bragg’s Law and the plane
spacing equation. X’Pert HighScore Plus software was used to identify the crystal
structure and secondary phases. Miller indices (hkl) were obtained from reference
patterns. The prominent peak at ~32°, was used to calculate the crystallite size using the
Scherrer equation with a shape factor of 0.9.
X-ray photo-electron spectroscopy (XPS) analysis was performed on fresh and
post-experiment samples with a Thermo Scientific K-Alpha instrument using Al Kα
radiation. The pressure in the analysis chamber was on the order of 8e-8 mbar. XPS flood
gun supplies Ar ions (99.997%) with a spot size of 400 µm. The pass energy for spectra
acquisition was 154.494 eV. The adventitious C1s peak was located at 284.85 eV.
Scanning electron microscopy (SEM) was performed with a Hitachi S-800 instrument
6

equipped with an EDAX-Phoenix Energy Dispersive X-Ray Spectrometry (EDS) system.
Images were taken with an acceleration voltage of 25 kV.
2.3 Temperature-Programmed experiments
An MKS Cirrus Mass Spectrometer (MS) was used to evaluate the reduction and
oxidation capabilities of the perovskite oxides. For each experiment, approximately 75 mg
of sample was packed in between high-temperature glass wool in a quartz U-tube reactor,
positioned inside a furnace. All experiments used He (Airgas, 99.999%) as the carrier
gas, 50 sccm of total flow, and, if applicable, the furnace temperature ramp rate was 10
°C/min. All gases were purchased from Airgas in the highest purity available. Alicat mass
flow controllers were used to control gas flow. Mass to charge (m/z) of signals 2, 18, 28,
and 44 were monitored in the MS corresponding to H2, H2O, CO, and CO2, respectively.
For CO quantification, CO2 fragmentation contribution to m/z = 28 was subtracted.
Temperature programmed reduction (TPR-H2) was performed to determine the
reduction temperature of the perovskites during the 1st step of the RWGS-CL process.
10% H2 (Airgas, 99.999%)/He was flowed at room temperature, then the sample was
heated to 950 °C. Water formed was quantified and correlated to the extent of reduction
of the material.
Temperature programmed oxidation (TPO-CO2) experiments were performed to
identify CO2 conversion temperature on the reduced perovskites. After reduction of the
perovskites at 550 °C under 10% H2/He for 30 min, the samples were cooled to room
temperature. Then 10% CO2 (Airgas, 99.999%)/He was flowed until stable. The sample
was heated to 950 °C. The amount of CO formed is representative of the CO2-conversion
capabilities of the reduced materials
7

2.4 Isothermal RWGS-CL
These experiments were largely conducted similarly to the temperature-
programmed experiments and used the same reactor system. First, the perovskites were
heated under 20% O2/He to eliminate the adsorbed CO2 and, then, were cooled to room
temperature. After this, samples were heated to 600 °C, upon signal stabilization, were
reduced for 20 min in 10% H2/He, flushed for 20 min under He, then 10% CO2 /He was
flowed for 20 min, followed by another 20 min He flushing. This process was repeated
four more times for a total of five cycles.
2.5 Density Functional Theory Calculations
Density functional theory (DFT) based free energies were computed using the
Vienna Ab-initio simulation package (VASP), version 5.3.3 [24-28]. Generalized gradient
approximation of the electron densities and Perdew-Burke-Ernzerhof (GGA-PBE)[29]
variant of exchange correlation were implemented and the plane wave basis cutoff energy
was set to 600 eV. The Brillouin zone of sixty-atom supercell was sampled with a 4 x 4 x 3
Monkhorst Pack k-point mesh [30]] and the energy computed converged to 1 μeV.
Different configurations of sixty-atom crystal structure of Co50, Fe50, and Mn50 were
tested and the most stable configuration was used for further computations. Bulk oxygen
vacancy formation energy (Evac) was calculated following Eq. 1, where 퐸퐴퐵푂 represents
3
the energy of the pure stoichiometric perovskite structure, 퐸퐴퐵푂
is the energy of the
3−훿
oxygen vacant perovskite, 퐸푂 is the energy of molecular oxygen, n is the number of unit
2
cells, and 훿 is the oxygen vacancy extent. A correction factor of 1.36 eV was applied to
퐸푂 to account for PBE functional oxygen over-binding error [31].
2
8

퐸푣푎푐 = 퐸퐴퐵푂 + 훿 × 푛2 퐸푂 − 퐸퐴퐵푂
(1)
2
3
3−훿
Evac was computed for each active site on the sixty-atom supercell of the three
compositions. The optimum Evac range of 3.36 ± 0.5 eV determined by Maiti et al. was
used to evaluate oxygen vacancy and CO2 conversion capability [12].
Oxidation state of each atom was calculated using the Bader charge analysis
algorithm developed by the Henkelman group [32, 33], which integrates the electronic
density enclosed in the Bader volume to obtain the charge of an atom. VESTA was used
to visualize electron density plots with an isosurface level of 0.0107 e a0-3 [34].
3. Results and Discussion
Co50, Fe50, and Mn50 samples synthesized via Pechini method achieved an
orthorhombic structure (ref# 01-089-2471) (Fig. 1a), which is a main phase found in
multiple lanthanum rich perovskites [10, 35]. Markers in Fig. 1b indicated the minor
presence of cubic Fe3O4 (ref # 01-085-1436) in Fe50 and Mn50. Presence of Fe3O4 on
the XRD patterns could be explained due to excess Fe-precursor on the synthesis or
caused by migration of Fe out of the perovskite structure, which would result in a B-site
deficient material. EDS elemental mapping of Fe50 sample (Fig. 2) shows all the
elements are well distributed on the bulk structure. Since the Fe50 sample was the best
performing one and we used the same synthesis procedure, only one sample was tested
for the composition distribution. Co50 and Mn50 have similar morphology to that of Fe50.
9

a
b
Figure 1. X-ray diffraction patterns of as-synthesized Co50, Fe50, and Mn50 samples.
(a) Results for full scan and (b) Narrow range indicating Fe3O4 (●) (ref # 01-085-1436)
Fe50 and Mn50 samples.
La
Co
O
Fe
Mn
Figure 2. Energy dispersive spectroscopy elemental mapping of Fe50
10

The introduction of the three metals in the B-site increases the number of sites with
different Evac values, caused by the presence of multiple valence states and a variety of
metal-oxygen bonds. Oxidation states of the transition metals (as obtained through Bader
calculation) exhibit charge variation within the structure, thereby creating metal-oxygen
bonds of varying strengths. Similarly, doping of Sr2+ in a LaMnO3 perovskite has been
reported to introduce a mixture Mn3+ and Mn4+ states, where Mn3+-O bonds have different
ionic strength than Mn4+-O bonds [15, 17, 36]. Site-weighted average oxygen vacancy
formation energy (Evac) is thus an appropriate descriptor of the vacancy formation
characteristics of these perovskite oxides. The Evac values (Fig. 3a) calculated for all three
perovskite compositions (Co50, Fe50, Mn50) fall within the optimum range of 3.36 ± 0.5
eV [12] predicting high oxygen vacancy formation and CO2 conversion capabilities. The
theoretical trend of Evac matches with the experimental water formation (indicating oxygen
vacancy formation).
a
b
Figure 3. (a) DFT-calculated Evac results correlated to experimental amount of oxygen
vacancies formed. (b) Average net charge on each transition metal
11

The net charge (Fig. 3b) for Co50 is higher for each constituent transition metal
than for Fe50 and Mn50, indicating higher charge localization than the Fe50 and Mn50
structures. This distribution of charge can be visualized in Fig. S1, where Mn50 and Fe50
exhibit slightly higher electron density between M-O bonds than Co50. The increased
electron density exhibited in the interatomic regions are representative of stronger bonds
that require higher energy for oxygen removal.
In general, temperature programmed reduction experiments of perovskite oxides
exhibit two reduction areas, the lowest temperature peak is representative of oxygen
vacancy formation on the oxides, whereas the highest temperature one, corresponds to
the decomposition of the oxygen vacant perovskites to its base oxides. TPR-H2
experiments (Fig. 4a) show Co50 starts reducing at 270 °C, whereas Fe50 and Mn50
start forming vacancies at 400 °C, indicating that the Co-rich perovskite (Co50) is more
easily reducible than the other two samples, agreeing with the theoretical trends of Co50
exhibiting lower Evac than Fe50 and Mn50 (Fig. 3a), and several other computational
studies predicting easier oxygen vacancy formation over Co-rich oxides [12]. This
favorable reduction of Co also causes Co50 to start decomposing to base oxides at ~ 100
°C earlier than Fe50 and Mn50 (second peak in the TPR-H2 profiles). Mn50 seems to
have an additional transition in the low-temperature area, likely due to Mn undergoing
more oxidation states transitions than Fe or Co. At ~550 °C, all the samples have
undergone the formation of oxygen vacancies (passed the low-temperature reduction
area) but have not transitioned to the base oxides (before the high-temperature reduction
area). Therefore, this temperature was chosen as the reduction temperature for the
subsequent oxidation (TPO-CO2) experiment. Choosing a temperature below the high
12

temperature reduction area would contribute to keeping the material in a single crystalline
phase when undergoing RWGL-CL cycles.
a
b
Figure 4. TPR-H2 and TPO-CO2 profiles. (a) TPR-H2: water formation as a mark of
oxygen vacancy formation (m/z=18 signal) for different reaction temperatures. The dotted
line represents the reduction temperature chosen for TPO-CO2 experiment (b) TPO-CO2:
CO formation (m/z=28 signal) as a function of reaction temperatures.
Integrating the area under the curve in Figure 4.b (Table 1) reveals that Co50
converted more CO2 to CO than Mn50, Fe50 and a reference material La0.75Sr0.25FeO3
(LSF) [14]. Both Fe50 and Co50 exhibit two CO2 conversion regions. Similar Co-based
perovskites have shown two or more re-oxidation areas after their reduction, suggesting
the reduced perovskite underwent several phase transitions during their re-oxidation with
CO2, these result is in agreement with the profiles of reduced oxides in Figure 6, or could
also be a consequence of the changes of oxidation state of the metals on the B-site of
13

the perovskite, suggesting that co-existence of certain combinations of oxidation states
of the metals on the B-site are more favored to re-oxidize than others. The onset CO2-
conversion temperature between Co50 and Fe50 was similar (and lower than Mn50),
which indicates the reduced state of Mn50 is less favored to re-oxidize with CO2 as the
other samples. It is to be noted that Fe50 produces more CO at lower temperatures
(below 600 °C) than Co50, which makes Fe50 appealing for its possibility of
demonstrating an isothermal RWGS-CL cycle. The reduction profiles in Figure 4a and the
200 °C difference in the peak CO formation temperatures between Co50 and Fe50 in
figure 4b, suggest that Co50 is more favored to reduce at low temperatures but less
favored to re-oxidize with CO2 when compared to and Fe50.
Fe50, on the other hand, converted 1420 μmol CO/g-perovskite at a peak
temperature of 570 °C. The onset decomposition temperature is higher at 700 °C, making
an isothermal RWGS-CL process feasible. These results agree with our previous study
with LaSrFe perovskites, indicating that the redox abilities of the material are highly
dependent on the predominant B-site metal.
Table 1. Water formation (oxygen vacancy generation) and CO in μmol/g-perovskite
calculated from integrating the area under the curve from Fig. 4a and 4b, respectively,
and peak CO2 conversion temperature.
H2O formation
(μmol/g-
perovskite)
Sample
Name
CO formation
(μmol/g-perovskite)
Peak
Temperature (°C)
Oxidation
Mn50
Fe50
Co50
1130
1790
2170
390
660
570
815
1420
1780
14

Upon reduction, Co50 and Mn50 formed small amounts of metallic Co and Mn
respectively (Fig. 5), as seen previously likely due to the high reducibility of Co and the
multiple oxidation states Mn can attain. Metallic Mn was also noted on the reduced Fe50.
These reduced metals, were not seen on the XRD profiles of the materials after oxidation
with CO2, indicating that upon reduction, a small amount of B-site metals migrated out of
the original perovskite structured, and re-incorporated back into the structure after re-
oxidation with CO2. Similarly, Fe3O4 disappears from the XRD on the reduced Fe50, and
reappears on the XRD profile after CO2 conversion.
For all the materials, the XRD-profiles shifted to lower 2-theta values indicating an
increase in lattice parameters due to their reduction and thermal expansion. In previous
studies, it has been shown that reducing La1−YSrYCoO3 perovskites into their high-
temperature reduction area, can lead to sufficient Co migrating out of the perovskite
structure, thus causing its decomposition into a Ruddlesden−Popper perovskite
(La2−YSrYCoO4) and metallic Co [10], noticeable by a diffraction line at ~32 2θ° (1 2θ°
lower than the main perovskite line at 33 2θ°). In this study, none of the samples exhibits
a Ruddlesden-Popper diffraction line at ~32 2θ°, which shows the original perovskite
phase was maintained in all cases. Nonetheless, though minor, the presence of small
amounts of metallic Co, were enough to cause different re-oxidation profiles on the
reduced samples, as seen in Figure 4b.
15

Figure 5. X-ray diffraction pattern for Co50, Fe50, and Mn50 after TPR-H2 and TPO-
CO2. For each pair, the post reduction profile is on the bottom, and post TPO-CO2 on the
top. Other observed phases include: Mn (◊), Co (×) and Fe3O4 (●).
Crystallite size after TPR (Table 2) is lower for Co50 followed by Fe50 and Mn50,
the same trend as CO2 generation, suggesting lower crystallite size after reduction
enhances CO2 conversion. After oxidation, Mn50’s crystallite size increased by 60%. This
increase in crystallite size during the oxidation step hindered the reaction, generating only
390 μmol CO/g-perovskite.
16

Table 2. Volume and crystallite size calculated from XRD
Sample VaFRESH (Å3) VaPOST-TPO (Å3) LbFRESH (Å) LbPOST-TPR (Å) LbPOST-TPO (Å)
Name
Mn50
Fe50
Co50
239
239
236
238
241
237
489
356
418
765
373
372
365
308
243
a V represents cell volume; b L represents crystallite size
XPS analysis on fresh and post-TPO samples display the change in valence state
after oxidation and the possible effects after multiple cycles. La 3d5/2 and La 3d3/2 peaks
(Fig. S2) binding energy correspond to 833.7 eV and 838 eV, representative of La2O3
peaks [37]. The difference between 3d5/2 multiplet peaks (~4.3 eV) confirms the trivalent
state [38]. All the post-TPO samples showed shifts to lower binding energies that marks
mild reduction of lanthanum oxidation states due to oxygen vacancy formation.
Figure 6a. shows the Co-2p x-ray photoelectron spectra. Since both Co-2p3/2 and
Co-2p1/2 exhibit similar B.E. for CoO and Co3O4 [39], the oxidation states for Co are
identified mainly by the position of Co-2p satellite features. Co50 samples have more
pronounced satellite features due to the presence higher Co content in the sample. Close
observation suggests post TPO Co50 sample has more Co(+2) states as the satellite
peak (~805 eV) shifts to the lower binding energies (B.E.). The Co-2p1/2 satellite peak for
Fe50 and Mn50 are also shifted to the lower B.E. than that of Co50, thereby revealing
lower oxidation states of Co ions in Fe and Mn rich samples. This is also predicted from
the DFT-calculated results (Fig. 3b) where we observe the same trend of lower Co
oxidation states for Fe50 and Mn50 samples.
17

Figure 6. X-ray photo-electron spectra (a) Co-2p, (b) Fe-2p, (c) Mn-2p, and (d) O-1s for
fresh (black) and post-TPO (grey) samples.
Fe-2p3/2 peak and satellites features for FeO and Fe2O3 are used to identify
oxidation states. The bump that ranges from 715 eV to 720 eV (Fig. 6b) is indicative of
Fe2+ (715.4 eV) and Fe3+ (719.3 eV) [39]. After TPO, the main 2p3/2 peak shifts to lower
binding energy for all materials, indicating Fe2+ becomes predominant after regeneration.
This reduction of metal ions after TPO is attributed to mild presence of oxygen vacancies
after TPO [14], thus resulting in decreased oxidation states of the metal ions. The broad
peaks for both Mn-2p3/2 and Mn-2p1/2 peaks (Fig. 6c) indicate there is presence of mixed
valence states Mn3+/Mn4+. 2p3/2 peaks positioned at 641.8 eV and 642.4 eV are assigned
18

to Mn3+ and Mn4+ respectively [39]. After CO2 conversion, peaks shift towards lower
binding energy indicating Mn3+ predominance over Mn4+.
O-1s spectra for the perovskite oxides depicts a major O1s peak (around ~529.5
eV) corresponding to M-O bonds [38, 40]. The post TPO samples show a shift of this
peak towards lower B.E. This is attributed to minor reduction of the oxidation states of the
metals as observed for Co-2p, Mn-2p and Fe-2p spectra as well (Figs. 6a-c). The peak
around 531 eV marks the presence of carbonate [40] as also observed around 289 eV of
C-1s spectra (Fig. S4). The minor peak around 533 eV can be due to the presence of
surface adsorbed water as well as oxygen vacancies [40, 41]. The increased intensity for
post TPO samples justifies the presence of both as increased oxygen vacancies are more
probable after RWGS-CL experiments and oxygen deficient surfaces are more prone to
adsorb moisture.
As shown in Table 3, the Co/La, Fe/La, and Mn/La atomic ratios, for as synthesized
and post-TPO samples are compared in terms of the near surface (XPS) and near bulk
(EDS) techniques. Fresh surface atomic ratios for Co50 and Mn50 show the surface
composition obtained from XPS are similar to the bulk composition obtained from EDS.
For Fe50, cobalt is promoted to the surface to almost the same ratio. This means similar
amounts of Co-O and Fe-O bonds are available on the surface, allowing Fe50 to create
similar amount of oxygen vacancies as Co50. Co50 creates more oxygen vacancies
because it has more Co-O bonds available on the surface which are easier to break than
Fe-O and Mn-O.
19

Table 3. Atomic ratio of Co50, Fe50, Mn50 fresh Co-2p, Fe-2p, Mn-2p and La-3d XPS
and EDS.
XPS
Co/La
0.64
0.58
0.64
0.90
0.84
0.26
0.12
EDS
Fe/La
0.41
0.51
0.67
1.23
1.32
0.35
0.73
Mn/La
0.30
0.27
0.42
0.45
0.39
0.69
0.66
Co/La
Fe/La
Mn/La
0.56
0.26
0.21
Co50 fresh
-
-
-
Co50 post TPO
Fe50 fresh
0.44
0.78
0.31
-
-
-
Fe50 post TPO
Fe50 5 cycles
Mn50 fresh
-
-
-
0.33
-
0.31
-
0.51
-
Mn50 post TPO
The stability of Fe50 was tested over five isothermal RWGS-CL cycles (Fig. 7) at
600 °C. Consistent to previous results [11, 14], oxygen atoms removed from the sample
in the first reduction includes adsorbed oxygen which is linked to the pre-treatment in
conditions simulated air (20% O2/He). After the 1st cycle, H2O formation per cycle
significantly decreases to ~1200 umol of H2O/ g perovskite, and remains relatively
constant (between 1100 and 1200 umol H2O/ g perovskite, corresponding to ~0.27 mol
H2O / mol perovskite) after the final cycle.
20

Figure 7. Five cycle experiment on Fe50 showing water (oxygen vacancies) and carbon
monoxide generation in μmol/g-perovskite.
Reaching a stable formation of CO throughout the cycles (~ 900 umol CO/g
perovskite, corresponding to ~ 0.22 mol CO/mol perovskite) suggests that the formation
of CO is limited by the amount of oxygen vacancies the material is able to form at the
reducing conditions, but maximized in this particular oxide. Earlier studies showed that
the amount of oxygen vacancies forming decreased slightly with each cycle, gradually
approaching the amount of CO generated [14]. This trend suggested oxygen vacancies
formed initially were not sufficiently regenerated on the initial cycles and oxygen shuttling
occurred at substoichiometric levels. In this case, oxygen shuttling occurred from ~ delta
values from ~2.6 to 2.8 during the final cycles. Since CO2 is a weaker oxidizing agent
21

than H2 is a reducing agent at this conditions for this material, the cycling occurred at
substoichiometric amounts of oxygen. Oxygen shuttling at substoichiometric oxygen
amounts suppresses reduction since the sample is partially reduced at the beginning of
the reduction step after a couple cycles. Maintaining similar oxygen vacancy formation
and CO generation (as exhibited by Fe50 on Figure 7) is important to avoid decomposition
of the material [14] without periodically (or at least less frequently) oxidizing the material
with a stronger oxidizing agent such as air to regenerate the perovskite oxide structure.
This material maintained its crystalline structure as observed by XRD (Fig. S3) and SEM
(Fig. 8) indicated the lack of morphology change for Fe50 after five RWGS-CL cycles.
Multiple valence states were present after TPO, as observed by XPS (Fig. 6), which
allows for stable CO generation. Due to this behavior, and due to Fe50 being among the
highest performance materials yielding 900 μmol CO/g-perovskite, Fe50 is a promising
material for CO2 conversion by thermochemical methods.
Fresh
RWGS-CL
Figure 8. Scanning electron microscopy of Fe50 before and after isothermal RWGS-CL
cycles
22

4. Conclusion
The simultaneous doping of three transition metals (Co, Fe, and Mn) in the B-site
of the perovskite oxide structure was studied computationally and experimentally. Density
functional theory-based calculations depict the reluctance of the Mn rich sample (Mn50)
to exhibit oxygen vacancies. The Co rich sample (Co50), although promoting easy oxygen
vacancy formation, was conducive to decomposition under RWGS-CL operation. In
contrast, the higher Fe composition induced reduction and oxidation at similar
temperatures, making it an ideal candidate material for isothermal RWGS-CL process.
Isothermal RWGS-CL cycles at 600 °C revealed Fe50 is capable of steady CO generation
without decomposition. Experimental results reveal multiple valence states on the
surface, enhancing CO2 conversion further than reported materials with two cations.
Higher CO2 conversion was achieved over oxygen deficient samples with smaller
crystallite size (higher exposed surface are for CO2 conversion reactions). This was
observed with Mn50, where the crystallite size increased during oxidation and resulted in
lower CO2 conversion than expected, while the decreased Co50 crystallite size enhanced
its CO2 conversion performance.
Acknowledgements
This study was funded by National Science Foundation through CBET-1335817, IIP-
1743623, and CHE-1531590 grants. Authors would also like to acknowledge funding from
the Venturewell E-Team and NASA through the Florida Space Grant Consortium. Y.A.D.
acknowledges the Florida Education Fund for the McKnight Dissertation Fellowship.
23

Additional thanks goes to research computing at USF for computational resources, Yajing
Chang for SEM and EDS, and Dr. Anne J. Meier for XPS measurements.
References
[1] T. Maverick, A Carbon Capture and Storage Breakthrough, Wall St. Daily, last
accessed 8/19/2016 (2014).
[2] Y.A. Daza, J.N. Kuhn, CO2 conversion by reverse water gas shift catalysis:
Comparison of catalysts, mechanisms and their consequences for CO2 conversion to
liquid fuels, RSC Advances, 6 (2016) 49675-49691.
[3] G.A. Olah, The Role of Catalysis in Replacing Oil by Renewable Methanol Using
Carbon dioxide Capture and Recycling (CCR), Catalysis Letters, 143 (2013) 983.
[4] D.S. Mallapragada, N.R. Singh, V. Curteanu, R. Agrawal, Sun-to-Fuel Assessment of
Routes for Fixing CO2 as Liquid Fuel, Industrial & Engineering Chemistry Research, 52
(2013) 5136−5144.
[5] W.C. Chueh, C. Falter, M. Abbott, D. Scipio, P. Furler, S.M. Haile, A. Steinfeld, High-
Flux Solar-Driven Thermochemical Dissociation of CO2 and H2O Using Nonstoichiometric
Ceria, Science, 330 (2010) 1797-1801.
[6] J.R. Scheffe, D. Weibel, A. Steinfeld, Lanthanum-Strontium-Manganese Perovskites
as Redox Materials for Solar Thermochemical Splitting of H2O and CO2, Energy & Fuels,
27 (2013) 4250-4257.
[7] P. Furler, J.R. Scheffe, A. Steinfeld, Syngas production by simultaneous spliting of
H2O and CO2 via ceria redox reactions in a high-temperature solar reactor, Energy &
Environmental Science, 5 (2012) 6098-6103.
[8] V.V. Galvita, H. Poelman, V. Bliznuk, C. Detavernier, G.B. Marin, CeO2‑Modified
Fe2O3 for CO2 Utilization via Chemical Looping, Ind. Eng. Chem. Res., 52 (2013)
8416−8426.
[9] N.V.R.A. Dharanipragada, L.C. Buelens, H. Buelens, E. De Grave, V.V. Galvita, G.B.
Marin, Mg–Fe–Al–O for advanced CO2 to CO conversion: carbon monoxide yield vs.
oxygen storage capacity, Journal of Mater. Chem. A, 3 (2015) 16251–16262.
[10] Y.A. Daza, R.A. Kent, M.M. Yung, J.N. Kuhn, Carbon Dioxide Conversion by Reverse
Water−Gas Shift Chemical Looping on Perovskite-Type Oxides, Industrial & Engineering
Chemistry Research, 53 (2014) 5828−5837.
[11] B.J. Hare, D. Maiti, Y.A. Daza, V.R. Bhethanabotla, J.N. Kuhn, Enhanced CO2
Conversion to CO by Silica-Supported Perovskite Oxides at Low Temperatures, ACS
Catalysis, 8 (2018) 3021-3029.
[12] D. Maiti, B.J. Hare, Y.A. Daza, A.E. Ramos, J.N. Kuhn, V.R. Bhethanabotla, Earth
Abundant Perovskite Oxides for Low Temperature Thermochemical CO2 Conversion,
Energy & Environmental Science, 11 (2018) 648-659.
[13] M. Wenzel, L. Rihko-Strukmann, K. Sundmacher, Thermodynamic Analysis and
Optimization of RWGS Processes for Solar Syngas Production from CO2, AICHE Journal,
63 (2016) 15-22.
24

[14] Y.A. Daza, D. Maiti, R.A. Kent, V.R. Bhethanabotla, J.N. Kuhn, Isothermal reverse
water gas shift chemical looping on La0.75Sr0.25Co(1−Y)FeYO3 perovskite-type oxides,
Catal. Tod., 258 (2015) 691-698.
[15] A.H. McDaniel, E.C. Miller, D. Arifin, A. Ambrosini, E.N. Coker, R. O'Hayre, W.C.
Chueh, J. Tong, Sr- and Mn-doped LaAlO3- for solar thermochemical H2 and CO
production, Energy & Environmental Science, 6 (2013) 2024-2028.
[16] A.H. Bork, M. Kubicek, M. Struzik, J.L.M. Rupp, Perovskite La0.6Sr0.4Cr1-xCoxO3-
solid solutions for solar-thermochemical fuel production: strategies to lower the operation
temperature, Journal of Materials Chemistry A, 3 (2015) 15546-15557.
[17] A.M. Deml, V. Stevanović, A.M. Holder, M. Sanders, R. O’Hayre, C.B. Musgrave,
Tunable Oxygen Vacancy Formation Energetics in the Complex Perovskite Oxide SrxLa1–
xMnyAl1–yO3, Chemistry of Materials, 26 (2014) 6595-6602.
[18] A.M. Deml, V. Stevanović, C.L. Muhich, C.B. Musgrave, R. O’Hayre, Oxide enthalpy
of formation and band gap energy as accurate descriptors of oxygen vacancy formation
energetics, Energy & Environmental Science, 7 (2014) 1996-2004.
[19] D. Maiti, Y.A. Daza, M.M. Yung, J.N. Kuhn, V.R. Bhethanabotla, Oxygen vacancy
formation characteristics in the bulk and across different surface terminations of La(1-
x)SrxFe(1-y)CoyO(3-) perovskite oxides for CO2 conversion, Journal of Materials Chemistry
A, 4 (2016) 5137-5148.
[20] N. Lakshminarayanan, H. Choi, J.N. Kuhn, U.S. Ozkan, Effect of additional B-site
transition metal doping on oxygen transport and activation characteristics in
La0.6Sr0.4(Co0.18Fe0.72X0.1)O3−ı (where X= Zn, Ni or Cu) perovskite oxides, Applied
Catalysis B: Environmental, 103 (2011) 318–325.
[21] M. Popa, M. Kakihana, Synthesis of lanthanum cobaltite (LaCoO3) by the
polymerizable complex route, Solid State Ionics, 151 (2002) 251-257.
[22] B.J. Hare, D. Maiti, S. Ramani, A.E. Ramos, V.R. Bhethanabotla, J.N. Kuhn,
Thermochemical conversion of carbon dioxide by reverse water-gas shift chemical
looping using supported perovskite oxides, Catalysis Today, 323 (2019) 225-232.
[23] S. Ivanova, A. Senysshyn, E. Zhecheva, K. Tenchev, V. Nikolov, R. Stoyanova, H.
Fuess, Effect of the synthesis route on the microstructure and the reducibility of LaCoO3,
Journal of Alloys and Compounds, 480 (2009) 279-285.
[24] P. Hohenberg, W. Kohn, Inhomogeneous Electron Gas, Physical Review, 136 (1964)
B864-B871.
[25] W. Kohn, L.J. Sham, Self-Consistent Equations Including Exchange and Correlation
Effects, Physical Review, 140 (1965) A1133-A1138.
[26] G. Kresse, J. Furthmüller, Efficient iterative schemes for ab initio total-energy
calculations using a plane-wave basis set, Physical Review B, 54 (1996) 11169-11186.
[27] G. Kresse, J. Furthmüller, Efficiency of ab-initio total energy calculations for metals
and semiconductors using a plane-wave basis set, Computational Materials Science, 6
(1996) 15-50.
[28] G. Kresse, J. Hafner, Ab initio molecular dynamics for liquid metals, Physical Review
B, 47 (1993) 558-561.
[29] J.P. Perdew, K. Burke, M. Ernzerhof, Generalized Gradient Approximation Made
Simple, Physical Review Letters, 77 (1996) 3865-3868.
[30] H.J. Monkhorst, J.D. Pack, Special points for Brillouin-zone integrations, Physical
Review B, 13 (1976) 5188-5192.
25

[31] L. Wang, T. Maxisch, G. Ceder, Oxidation energies of transition metal oxides within
the GGA + U framework, Physical Review B, 73 (2006) 195107 (195101-195106).
[32] W. Tang, E. Sanville, G. Henkelman, A grid-based Bader analysis algorithm without
lattice bias, J Phys Condens Matter, 21 (2009) 084204-084201 - 084204-084207.
[33] G. Henkelman, A. Arnaldsson, H. Jónsson, A fast and robust algorithm for Bader
decomposition of charge density, Computational Materials Science, 36 (2006) 354-360.
[34] K. Momma, F. Izumi, VESTA 3 for three-dimensional visualization of crystal,
volumetric and morphology data, Journal of Applied Crystallography, 44 (2011) 1272-
1276.
[35] Q. Jiang, J. Tong, G. Zhou, Z. Jiang, Z. Li, C. Li, Thermochemical CO2 splitting
reaction with supported LaxA1-xFeyB1-yO3 (A=Sr, Ce, B=Co, Mn; 0≤x, y≤1) perovskite
oxides, Solar Energy, 103 (2014) 425-437.
[36] A. Demont, S. Abanades, E. Beche, Investigation of Perovskite Structures as
Oxygen-Exchange Redox Materials for Hydrogen Production from Thermochemical Two-
Step Water-Splitting Cycles, The Journal of Physical Chemistry C, 118 (2014) 12682-
12692.
[37] M. O'Connell, A.K. Norman, C.F. Hutteramnn, M.A. Morris, Catalytic oxidation of
lanthanum-transition metal perovskite materials, Catalysis Today, 47 (1999) 123.
[38] H.S. Fogler, Elements of Chemical Reaction Engineering 4th ed. ed., Pearson, New
York, 2006.
[39] M.C. Biesinger, B.P. Payne, A.P. Grosvenor, L.W.M. Lau, A.R. Gerson, R.S.C.
Smart, Resolving surface chemical states in XPS analysis of first row transition metals,
oxides and hydroxides: Cr, Mn, Fe, Co and Ni, Applied Surface Science, 257 (2011) 2717-
2730.
[40] N.A. Merino, B.P. Barbero, P. Eloy, L.E. Cadús, La1−xCaxCoO3 perovskite-type
oxides: Identification of the surface oxygen species by XPS, Applied Surface Science,
253 (2006) 1489-1493.
[41] S. Jaiswar, K.D. Mandal, Evidence of Enhanced Oxygen Vacancy Defects Inducing
Ferromagnetism in Multiferroic CaMn7O12 Manganite with Sintering Time, The Journal of
Physical Chemistry C, 121 (2017) 19586-19601.
26