Scholarly article on topic 'Oxygen nonstoichiometry, defect equilibria, and thermodynamic characterization of LaMnO3 perovskites with Ca/Sr A-site and Al B-site doping'

Oxygen nonstoichiometry, defect equilibria, and thermodynamic characterization of LaMnO3 perovskites with Ca/Sr A-site and Al B-site doping Academic research paper on "Chemical engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Acta Materialia
OECD Field of science
Keywords
{"Solar fuels" / Perovskites / "Oxygen nonstoichiometry" / Thermochemical / "Defect chemistry"}

Abstract of research paper on Chemical engineering, author of scientific article — M. Takacs, M. Hoes, M. Caduff, T. Cooper, J.R. Scheffe, et al.

Abstract This work encompasses the thermodynamic characterization of four doped lanthanum manganites, namely La0.6 A 0.4Mn1−yAlyO3 (A = Ca, Sr and y = 0, 0.4), all showed to be promising redox materials for the solar thermochemical splitting of H2O and CO2 to H2 and CO. We present oxygen nonstoichiometry measurements in the temperature range T = 1573 K–1773 K and oxygen partial pressure range p O2 = 4.5066 × 10−2 bar–9.9 × 10−5 bar. For a given T and p O2, oxygen nonstoichiometry is found to be higher when replacing the divalent dopant Sr in La0.6Sr0.4MnO3 by the divalent Ca but also increases significantly when additionally doping 40 mol-% Al to the Mn-site. La0.6Ca0.4Mn0.6Al0.4O3 revealed the highest mass specific oxygen release, 0.290 mol O2 per kg metal oxide at T = 1773 K and p O2 = 2.360 × 10−3 bar and 0.039 mol kg−1 at T = 1573 K and p O2 = 4.5066 × 10−2 bar. It is shown that the chemical defect equilibrium of all four perovskites can be accurately described by the two simultaneous redox couples Mn4+/Mn3+ and Mn3+/Mn2+. Thermodynamic properties, namely partial molar enthalpy, entropy and Gibbs free energy are consequently extracted from the defect models. Partial molar enthalpy decreases with increasing oxygen nonstoichiometry for the Al-doped perovskites whereas the opposite trend is observed for the others. The enthalpy falls within the range 260–300 kJ mol−1 for all the materials. Equilibrium hydrogen yields upon oxidation with H2O are determined as a function of redox conditions. Although reduction extents of the perovskites are greater compared to CeO2, oxidation with H2O and CO2 is thermodynamically less favorable. This leads to lower mass specific fuel productivity compared to CeO2 under most conditions relevant for solar thermochemical cycles.

Academic research paper on topic "Oxygen nonstoichiometry, defect equilibria, and thermodynamic characterization of LaMnO3 perovskites with Ca/Sr A-site and Al B-site doping"

Contents lists available at ScienceDirect

Acta Materialia

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

Full length article

Oxygen nonstoichiometry, defect equilibria, and thermodynamic characterization of LaMnO3 perovskites with Ca/Sr A-site and Al B-site doping

CrossMark

M. Takacs a, M. Hoes a, M. Caduff a, T. Cooper a, J.R. Scheffe b' *, A. Steinfeld a'

a Department of Mechanical and Process Engineering, ETH Zurich, 8092 Zurich, Switzerland b Department of Mechanical and Aerospace Engineering, University of Florida, Gainesville, FL 32611-6250, USA

ARTICLE INFO

Article history: Received 24 August 2015 Received in revised form 14 October 2015 Accepted 16 October 2015 Available online xxx

Keywords: Solar fuels Perovskites

Oxygen nonstoichiometry Thermochemical Defect chemistry

ABSTRACT

This work encompasses the thermodynamic characterization of four doped lanthanum manganites, namely La0.6A0.4Mn1_yAlyO3 (A = Ca, Sr and y = 0, 0.4), all showed to be promising redox materials for the solar thermochemical splitting of H2O and CO2 to H2 and CO. We present oxygen nonstoichiometry measurements in the temperature range T = 1573 K—1773 K and oxygen partial pressure range pO2 = 4.5066 x 10"2 bar—9.9 x 10"5 bar. For a given T and pO2, oxygen nonstoichiometry is found to be higher when replacing the divalent dopant Sr in La0.6Sr0.4MnO3 by the divalent Ca but also increases significantly when additionally doping 40 mol-% Al to the Mn-site. La0.6Ca04Mn0.6Al04O3 revealed the highest mass specific oxygen release, 0.290 mol O2 per kg metal oxide at T = 1773 K and pO2 = 2.360 x 10"3 bar and 0.039 mol kg"1 at T = 1573 K and pO2 = 4.5066 x 10"2 bar. It is shown that the chemical defect equilibrium of all four perovskites can be accurately described by the two simultaneous redox couples Mn4+/Mn3+ and Mn3+/Mn2+. Thermodynamic properties, namely partial molar enthalpy, entropy and Gibbs free energy are consequently extracted from the defect models. Partial molar enthalpy decreases with increasing oxygen nonstoichiometry for the Al-doped perovskites whereas the opposite trend is observed for the others. The enthalpy falls within the range 260—300 kJ mol"1 for all the materials. Equilibrium hydrogen yields upon oxidation with H2O are determined as a function of redox conditions. Although reduction extents of the perovskites are greater compared to CeO2, oxidation with H2O and CO2 is thermodynamically less favorable. This leads to lower mass specific fuel productivity compared to CeO2 under most conditions relevant for solar thermochemical cycles.

© 2015 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction

Nonstoichiometric ceria (CeO2) is currently considered to be a state-of-the-art redox material for solar-thermochemical splitting of H2O and CO2 to H2 and CO (syngas) because of its rapid oxidation and reduction kinetics and its morphological stability over a range of temperatures and reduction extents [1]. Compared to other metal oxide systems (e.g. volatile ZnO [2] and non-volatile ferrites [3,4]), ceria shows relatively low fuel productivity per unit mass of metal oxide [5—7]. It has been shown that reduction extents of ceria can be increased by introducing 4+ valence dopants such as Zr4+ [8—12] and Hf4+ [8,13] into the ceria lattice. However,

* Corresponding authors. E-mail addresses: jscheffe@ufl.edu (J.R. Scheffe), aldo.steinfeld@ethz.ch (A. Steinfeld).

http://dx.doi.org/10.1016/j.actamat.2015.10.026

1359-6454/© 2015 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

thermodynamic calculations for Zr4+-doped ceria [7,12] showed that oxidation with H2O is less favorable compared to pure ceria. This results in a lower theoretical solar-to-fuel energy conversion efficiency for Zr4+-doped ceria because of larger temperature swings between the redox steps and/or excess oxidant gas [7]. Lower conversion efficiency was also reported for other ceria dopants such as Gd3+, Y3+, Sm3+, Ca2+ and Sr2+ [6].

Perovskite oxides have recently been proposed as promising alternative reactive intermediates for solar thermochemical H2O/ CO2 splitting because of potentially increasing the energy conversion efficiency by lowering the reduction temperature or increasing the mass specific fuel yield [14—23]. The two-step solar-thermo-chemical cycle based on a generic perovskite (ABO3) is represented by

dred - dox

dred - dox

dred - df

-ABOa_dred + 2O2

3-dred

dred - do

dred - dox

ABO3-dred + CO2

dred - dox

ABO3_ dox + CO

where Eq. (1) represents the high-temperature endothermic reduction and Eq. (2) the lower temperature exothermic oxidation with H2O (a) and CO2 (b). dred and dox represent the oxygen non-stoichiometry after reduction and oxidation, respectively, whereas the difference dred - dox is the molar amount of fuel produced per mole of metal oxide. Scheffe et al. [14] considered strontium-doped lanthanum manganites La1-xSrxMnO3 (x = 0.30, 0.35, 0.40) and reported reduction extents of La06Sr0.4MnO3 to be nearly 6.5 times larger than those of ceria at 1600 K and twice larger at 1800 K, both for pO2 = 10-5 bar. However, it was shown that their theoretical solar-to-fuel energy conversion efficiency is lower compared to that of ceria because of the thermodynamically less favorable oxidation with CO2 and H2O. Based on measurements of Mizusaki et al. [15], Yang et al. [16] concluded higher reduction extents and, consequently, higher fuel yields with increasing x for La1-xSrxMnO3 (0 < x < 0.5), while H2O-to-H2 conversion rate decreased with increasing x. This led to the conclusion that intermediate doping levels maybe optimal for solar-to-fuel energy conversion. McDaniel et al. [17] showed even higher CO and H2 yields when additionally doping Al to the B-site of La1-xSrxMnO3 while maintaining fast oxidation rates with CO2 and H2O. Dey et al. [18] showed increasing reduction extents and fuel productivity when replacing the divalent A-site dopant Sr in La1-xSrxMnO3 with Ca and obtained best results for x = 0.5. In another recent work [19], they investigated two series of perovskite oxides, Ln0.5Sr0.5MnO3 and Ln0.5Ca0.5MnO3 (Ln = La, Nd, Sm, Gd, Dy, and Y) and concluded highest O2 release for the manganite with the smallest A-site cation radius (in this case Y). A comparison between LaxSr1-xMO3 (M = Mn, Co, Fe) and BaxSr1-x(Co,Fe)O3 [20] showed best results for the Mn-containing perovskites. In a very recent study by Cooper et al. [21], Sr and Ca A-site doped lanthanum-manganites, with and without B-site doping of Al, were examined for redox performance and compared to the state-of-the-art material ceria. La0.6Sr0.4Mn0.6Al0.4O3 and La0.6Ca0.4Mn0.6Al0.4O3 reach four to nine times higher reduction extents compared to ceria in the temperature range T = 1473 K—1673 K, while maintaining fast oxidation kinetics with CO2. Bork et al. [22] reported that La06Sr0.4Cr1-xCoxO3 with the optimal dopant concentration x = 0.2 can split up to 25 times more CO2 when cycling at T = 1073 K—1473 K compared to ceria or exhibits similar reduction extents (5 = 0.034) at 300 K lower temperatures (1473 K instead of 1773 K). On the other hand, cyclability redox studies by Galvez et al. [23] revealed that the chemical stability of Ca, Sr and Al-doped La—Mn perovskites is detrimentally affected by sintering and by the formation and eventual segregation of a carbonate phase during oxidation by CO2.

Some of the conclusions drawn for the perovskites discussed above are mainly based on qualitative reduction experiments under an inert flow of low pO2 and oxidation under a relatively high flow of CO2 and/or H2O in a thermogravimetric analyzer [18—20,22] and/ or in an electrically heated furnace coupled to a gas analysis [17,18]. Such experiments with large excess of CO2 and/or H2O can result in misleading conclusions, as indicated by various thermodynamic analyses [14,16,21]. In the work of Yang et al. [16], the amount of H2O needed to oxidize La1-xSrxMnO3 (0 < x < 0.4) to a certain dox,

and hence produce a fixed amount of H2 in a closed system with variable volume was calculated from thermodynamic data. Similar calculations were performed by Scheffe et al. [14] and Cooper et al. [21], however, there, the initial amount of CO2 and/or H2O was fixed and the fuel yield was predicted as a function of oxidation temperature. Such thermodynamic fuel yield calculations allow for an accurate determination of the material's potential to efficiently split CO2 and/or H2O.

In this work, we build on the recent work of Cooper et al. [21] and report detailed oxygen nonstoichiometry measurements of Lao.6Ao.4Mn1-yAlyO3 (A = Ca, Sr and y = 0, 0.4) over a wide temperature range T = 1573 K—1773 K and oxygen partial pressure range pO2 = 4.5066 x 10-2 bar—9.9 x 10-5 bar. The refinement of the nonstoichiometry measurements allows the development of more appropriate defect models to describe the defect chemical equilibria and to extract finer trends in partial molar thermodynamic properties (DhO, DsO, DgO). From such data we determine equilibrium hydrogen yields and evaluate the potential of these lanthanum-manganites to be used as reactive intermediates in solar thermochemical redox cycles.

2. Experimental section

2.1. Sample preparation and characterization

La0.6Sr0.4MnO3 (LSM40), La0.6Ca0.4MnO3 (LCM40), La0.6Sr0.4M-n0.6Al04O3 (LSMA) and La0.6Ca0.4Mn0.6Al0.4O3 (LCMA) perovskite powders were synthesized by sol—gel method as described by Scheffe et al. [8]. The corresponding metal nitrates (see Table 1 in electronic supplementary information1 (ESI)) and anhydrous citric acid (Sigma—Aldrich, catalog nr. 251275) in aqueous solution were used to carry out the synthesis. The ratio of the metal cations to the citric acid was 1:1.5. The aqueous solution was slowly heated up to 573 K to perform the pyrolysis. Afterwards, powders were calcined at 1273 K under air for 5 h. Dense cylindrical pellets were obtained by uniaxially cold-pressing the powder at 5 tons followed by sintering at 1773 K under air atmosphere for 24 h. The approximate dimensions after sintering were 6.4—6.9 mm diameter and 1—2 mm height and the mass of the pellets was ~250 mg (LSM40), ~290 mg (LCM40), ~150 mg (LSMA) and ~270 mg (LCMA). Dopant concentrations were measured by inductively coupled plasma-atomic emission spectroscopy (ICP-OES) and deviated by less than 4% from their nominal composition for LSMA and LCMA. Powder X-ray diffraction (XRD) was performed in the Bragg Brentano geometry using Cu Ka radiation (Philips, PANalytical/X'Pert MPD/DY636, l = 1.5406 A, 20 = 20—100°, 0.01° s-1 scan rate, 45 kV/20 mA output). Scanning electron microscopy (SEM) of the dense pellets was conducted on a TM-1000Microscope (Hitachi, 15 kV accelerating voltage). ICP-OES analysis, XRD patterns and SEM images are shown in ESI.

2.2. Thermal analysis

Oxygen nonstoichiometry was measured using a thermogravi-metric analyzer (TGA, Setaram Setsys Evolution). Pellets were suspended to the balance with a custom-made platinum hook to ensure good exposure to the purge gas and eliminate gas diffusion limitations. The pO2 of the sweep gas was controlled by mixing Ar (Messer, Argon 4.6) with O2/Ar mixtures (Messer, 5% O2 5.0 in Ar 5.0 and 0.1% O2 5.0 in Ar 5.0). Gases were mixed with electronic mass

1 Electronic supplementary information (ESI) available: List of metal nitrates used for sample preparation, ICP-OES analysis, XRD patterns, SEM images, detailed derivation of defect model and additional results.

Table 1

Linear fitting parameters of lnKi versus 1000/T (cf. Eq. (13)) of all four perovskites represented by the enthalpy Ah- and entropy As- of the two single defect reactions.

Defect reaction Extracted fitting parameter LSM40 LCM40 LSMA LCMA

Eq. (4) (Ki) Dh'1 (kJmol"1) 256.12 257.44 324.07 307.35

As'j (J тоГ1 K"1) 98.60 101.54 133.05 127.36

Eq. (8) (K2) Dh'2 (kJ mol"1) 314.52 314.09 229.59 261.88

As2 (1 mol"1 K"1) 96.82 106.52 61.16 82.35

flow controllers (Brooks, Model 5850TR, accuracy ±1%) with a constant total flow rate of 200 ml/min. The gas species and concentrations at the outlet were monitored by mass spectrometry (Pfeiffer Vacuum, OmniStar GSD 320). Temperature was varied between 1573 K and 1773 K and pO2 between 4.5066 x 10-2 bar and 9.9 x 10-5 bar. In all measurements, the sample mass (ms) was equilibrated at a constant T and pO2. Following each equilibrium measurement, the pO2 was rapidly changed by adapting the O2/Ar gas mixture, resulting in a temporal weight change of the sample due to evolving or uptaking of oxygen until a new equilibrium state was reached. To correct for buoyancy, blank runs were performed with inert Al2O3 sintered pellets of same dimensions.

3. Results

3.1. Oxygen nonstoichiometry

The thermogravimetric measurements of LCM40, LSM40, LSMA and LCMA are shown in Fig. 1(a) and (b). Fig. 1(a) shows the reduction and oxidation runs within the higher pO2 measurement range (4.5066 x 10-2 bar to 2.387 x 10-3 bar) whereas Fig. 1(b) shows the runs within the lower pO2 range (9.15 x 10-4 bar—9.9 x 10-5 bar). The samples were heated to different set point temperatures followed by isothermal reduction and oxidation by stepwise changing pO2. As seen, within the temperature and pO2 range investigated, LCMA shows the highest reduction extents followed by LSMA, LCM40 and LSM40. The measurement of LCM40 within the lower pO2 range is shown in ESI because of different relaxation time scales. Oxygen non-stoichiometry 5 is calculated according to:

same n'O2 but at 285 K lower temperatures. The higher reduction extent of LSMA compared to LSM40 is in agreement with measurements of McDaniel et al. [17] and Cooper et al. [21]. The higher reduction extents of the Ca-doped perovskites compared to the Sr-doped are in agreement with the study of Dey et al. [19], attributed to the smaller ionic radii of Ca2+ compared to Sr2+.

3.2. Defect modeling

Several defect models describing the oxygen nonstoichiometry of doped and undoped lanthanum-manganites can be found in literature [14,15,24—29]. For undoped LaMnO3, La and Mn are in the trivalent state. When replacing some of the lanthanum by a divalent dopant (e.g. Sr, Ca) some of the Mn goes from its trivalent to its tetravalent state in order to compensate the charge differences introduced by the divalent dopant. In Kroger-Vink notation [30] this reads as Lax xSrxMnx xMnxO3. It was shown [14,15,24,26,31] that the reduction of Sr-doped LaMnO3 in the oxygen-deficient region can be described in Krooger-Vink notation as

2MnMn + Og = 2MnMn + Vo +102(g)

where tetravalent manganese (MnMn) on manganese lattice sites and oxygen atoms on oxygen lattice sites (OQ) are in equilibrium with gaseous oxygen, trivalent manganese on manganese lattice sites (MnMn) and doubly ionized oxygen vacancies (VO ). By applying the law of mass action, assuming activity coefficients equal 1 and a standard pressure p° = 1 bar, the equilibrium constant for reaction (4) can be defined as [21,27,29].

d = Dms $ Mg

where AmS is the relative weight loss at equilibrium, MS is the molar mass of the sample, and MO the molar mass of atomic oxygen. AmS of the reduction and oxidation runs is calculated relative to the mass at T = 873 K before heating to the set point temperature and after cooling from the set point temperature, respectively. Measured 3 - 5 versus pO2 for T = 1573 K—1773 K is shown in Fig. 2 for LSM40 (a), LCM40 (b), LSMA (c) and LCMA (d). Symbols indicate 3 - 5 values obtained from thermogravimetric (TG) analysis shown in Fig. 1 (a) and (b). Dashed colored and solid black lines indicate defect models used to describe equilibrium data, presented in the next section. Error bars account for AmS s 0 at the stabilization temperature T = 873 K(cf. AmS of LSMA shown in Fig. 1 (b)). As seen in Fig. 2, 5 is highest for LCMA followed by LSMA, LCM40 and LSM40, over the whole measurement range investigated. At T = 1573 K and at pO2 = 2.09 x 10-4 bar, 5 of LCMA is almost 100% higher than LCM40 whereas 5 of LSMA is almost 150% higher than LSM40. At these conditions, LCMA releases about 20% more O2 compared to LSMA (per mole of metal oxide). Compared to ceria [7], higher reduction extents are obtained with all the four perov-skites as seen in Fig. 3, where the mass specific oxygen release n'O2 is shown. For example, n'O2 = 0.15 mol/kg is expected for ceria at T = 1885 K and pO2 = 10-3 bar, while LCMA is expected to reach the

[VO ] [MnMn]

[og] [Mi

;(poi/ pc

where square brackets denote concentrations taken as sublattice site fractions, e.g. for oxygen vacancies:

nVo$ + nOg

Trivalent manganese in Eq. (4) can further disproportionate to tetravalent and divalent manganese (MnMn) according to

MnMn + Mn,

By combining Eq. (7) with Eq. (4), the disproportionation reaction can be written as

MnMn + Og = MnMn + Vo + iü2(g) with the equilibrium constant

[Vo] [MnMn]

[°g] [MnMn]

К poj p°

Fig. 1. Percent mass change as a function of time for all reduction and oxidation runs of LCM40, LSM40, LSMA and LCMA for: (a) temperature range T = 1573—1773 K and O2 partial pressure range pO2 = 4.5066 x 10-2—2.387 x 10-3 bar; and (b) T = 1573—1673 K and P02 = 9.15 x 10-4—9.9 x 10-5 bar.

d1/2(x + y _ 2d - l)(po2/p°)1/4 = (3 - Ô)1/2(2Ô_X + y_ 1)

_ d1/2(x _ 2d)(po2/p°)1/4

po - x\1/2

( x_2 5 )(3 _ Ô)

_ K1/2

Complete derivation of Eq. (12) is shown in ESI. By plotting the measured nonstoichiometry data in the form Y versus X (cf. Eq. (12) and Fig. 5 in ESI), K1 and K2 can be determined by linear regression. -K1/2 equals the Y-intercept and K2 the slope of the linear fit. Fitted equilibrium constants K1 and K2 versus 1000/T are shown in Fig. 4 (a) and (b), respectively. Lines indicate the best linear fits of lnK1 and lnK2 versus 1000/T. The fitting parameters, represented by the enthalpy Dh° and entropy Ds° of the single defect reactions (i = 1 for Eq. (4) and i = 2 for Eq. (8)), were extracted from the linear fits by using Eq. (13). Results are summarized in Table 1 for all four materials.

Error bars shown in Fig. 4 represent ±2s (two times standard deviation) of intercept (K1) and slope (K2) of the linear regression. With the knowledge of K1 and K2, d can be calculated for arbitrary pO2 by solving Eq. (12). d values calculated by using individual data points of Fig. 4 (a) and (b) are indicated by the solid black lines in Fig. 2. d values shown by the colored dashed lines in Fig. 2 are determined by using the inverse linear temperature dependence of lnK1 and lnK2 as shown by the solid lines in Fig. 4 (a) and (b). In general, the measured d values are well described by both fits. Thus, the reduction of all the four pe-rovskites studied in this work can be well described by the reduction of Mn4+ to Mn3+ and Mn2+, at least within the T and pO2 range investigated here. Minor deviations of the two fits (black solid and colored dashed lines), especially towards higher d values, can be attributed to differences between individual data points (symbols) and linear fit in Fig. 4 (a) and (b). For LSM40, at T = 1723 K, 1623 K and 1573 K, best fits were obtained by only considering Eq. (4) (K2 = 0).

Measured oxygen nonstoichiometry of the four perovskites investigated in this work was modeled as a function of temperature and pO2 by fitting the equilibrium constants K1 and K2 to the experimental data. A defect cluster model was not considered because it was reported to be limited to lower pO2 values [26]. This is further supported by the very good fit of the proposed defect model to the experimental data. Site balances for manganese and oxygen

nMnMn + nMnM„ + n no* + nvo = 3

and charge neutrality

2nVo + nMn„„ = x + n

allow K1 and K2 to be written in terms of d, pO2 and one sub-lattice site fraction. The last unknown site fraction can be eliminated by combining K1 and K2 to one expression shown in a linear form by

3.3. Thermodynamic properties

Assuming unity activity for the solid and ideal gas behavior for O2, the standard partial molar Gibbs free energy, equivalent to the Gibbs free energy change of an infinitesimal reaction of Eq. (1), can be written as [21].

DlO(5, T) = _2RT ln(pojp)

DgO can be additionally related to the standard partial molar enthalpy (DhO) and entropy (DsO) according to

DgO = Dho _ T DsO

By combining Eqs. (14) and (15) and assuming temperature independent partial molar enthalpy and entropy, DhO and DsO as a function of d (defined per 1 mol of oxygen vacancies created in the lattice or equivalent to half a mole of O2 released to the surrounding gas atmosphere) are obtained by determining the slope and

Fig. 2. Measured oxygen nonstoichiometry (symbols) of LSM40 (a), LCM40 (b), LSMA (c) and LCMA (d) versus pO2 for T = 1573 K—1773 K. Colored dashed and black solid lines indicate defect models used to describe experimental results. Black solid lines are calculated based on individual defect equilibrium constants whereas the colored dashed lines are calculated by using the inverse temperature dependence of lnK and lnK2 (cf. Fig. 4 and Eq. (13)).

intercept of -ln(pO2/p°) versus 1/T for a constant 5,

-ln (po2/ p°

2Dhc RT

Constant 5 values are obtained by evaluating the defect models in the temperature range T = 1573 K—1773 K. Fig. 5 shows -ln(pO2/p°) versus 1/T of LCM40 for the temperature range T = 1573 K—1773 K and oxygen nonstoichiometry range 5 = 0.0103—0.0675. The error bars correspond to deviations between measured and fitted pO2 (obtained from fitted 5) shown in Fig. 2, whereas the symbols correspond to the fitted pO2 shown by the black lines in Fig. 2. Similar plots for the other perovskites are shown in the ESI. All data points are well represented by a linear fit (R2 > 0.93) indicating generally temperature independent DhO and DsO for all the perovskites within the temperature range investigated.

DhO and DsO as a function of 5 are shown in Fig. 6 and Fig. 7,

respectively, for all four perovskites investigated. Symbols represent DhO and DsO values calculated from ô values shown by the solid black lines of Fig. 2(a—d). Dashed lines represent modeled values calculated based on only the colored dashed lines of Fig. 2(a—d). Error bars were calculated from ± 2s of slope (for DhO) and intercept (for DsO ) of the linear regression of -ln(pO2/p° ) versus 1/T (cf. Fig. 5 and Eq. (16)). The agreement between measured and modeled DhO and DsO values is quite good except for LSM40, where the discrepancy can be attributed to larger deviations between defect model and measured ô values (cf. Fig. 2(a)). Interestingly, DhO of the Al-doped perovskites (LSMA and LCMA) decreases with increasing ô whereas DhO of the two non-Al-doped perovskites (LSM40 and LCM40) slightly increases with increasing ô. DhO values of the Ca-doped perovskites are slightly higher compared to their equivalent Sr-doped ones. DhO values of LSM40 shown in this work are consistent with the values calculated by Yang et al. [16] using nonstoichiometry measurements of Mizusaki et al. [15], however, they observed a small decrease in enthalpy with increasing ô. The

5=const

Fig. 3. Mass specific oxygen evolution for LSM40, LCM40, LSMA, LCMA and ceria [7] versus temperature at pO2 = 10-3 bar.

trend of increasing enthalpy with d was also observed by Cooper et al. [21], although their values are lower than those reported here. In contrast, DhO values of LCMA and LSMA measured by Cooper et al. [21] increased with d and therefore values differ by up to 100 kJ mol-1 at low d (»0.02). Measurements for LCM40 are_in good agreement. Compared to literature data of ceria [7], DhO values obtained for the perovskites are around 150 kJ mol-1 lower at d » 0.07. This result is expected because of the much higher reduction extents (or lower reduction temperature) of the perovskites compared to ceria. Although DhO of the Ca-doped perov-skites is higher compared to that of Sr-doped ones, reduction extents are larger. This can be explained by the higher DsO (discussed below) which results in a lower DgO for the Ca-doped materials.

The DS¿ values of the Al-doped perovskites (LCMA and LSMA) show a much steeper decrease with increasing 5 compared to the non-Al-doped ones (LCM40 and LSM40). DsO is highest for LCMA followed by LSMA, LCM40 and LSM40, at least for 5 < 0.75. Entropy values of LSM40 and their dependence on 5 reported by Yang et al. [16] are similar to this work. As°O values measured by Cooper et al. [21 ] are comparable to the measurements shown in this work, except for LSM40 which are higher in this work. Cooper et al. [21 ] show highest DsO for LCM40 (out of LCMA, LSMA, LCM40 and LSM40) whereas here, highest DsO was measured for LCMA (for 5 < 0.125). Deviations in the partial molar thermodynamic properties from the work of Cooper et al. [21 ] might be explained by differences in the measured oxygen non-stoichiometry. The increase in DhO and DsO when adding Al to the B-site of La1-xCaxMnO3 is in agreement with the measurements of Tanasescu et al. [32] (at least for 5 < 0.02). Compared to ceria [7], DsO values of the perovskites are in general more than 50 J mol"1 K"1 lower.

The partial molar Gibbs free energy DgO as a function of T and 5 was calculated from modeled DhO and DsO values (cf. dashed lines in Figs. 6 and 7) according to Eq. (15), where it was assumed that DhO and DsO are independent of T. DgO of the perovskites and ceria [7] are shown in Fig. 8 for 5 = 0.10, relevant for the metal oxide reduction reaction, and negative DgO for 5 = 0.01, relevant for the oxidation reaction with H2O or CO2. For comparison, the corresponding values for ceria were calculated: DgO(5 = 0.10) with DhO = 405 kJ mol"1 and

= 160 J mol"

and DgO(5 = 0.01) with

Fig. 4. Equilibrium constants versus inverse temperature for LSM40, LCM40, LSMA and LCMA: (a) Ki; and (b) K2. Lines indicate linear fits of lnK to 1000/T.

DhO = 480 kJ mol-1 and DsO = 260 J mol-1 K-1 [7] and additionally shown in Fig. 8. Dashed lines represent the Gibbs free energy change of H2 oxidation: H2 + 0.502 = H2O (-DrGH2O, blue) and of CO oxidation: CO + 0.502 = C02 (-DrGC0, reel), obtained from NIST-JANAF thermochemical tables. The reduction reaction of the metal oxide is at equilibrium at pO2 = 1 bar and d = 0.10 at the temperature where Dg0(d = 0.10) = 0. The oxidation with H2O/CO2 to d = 0.01 is thermodynamically favorable at temperatures where -Dgo(d = 0.01) = -DrG^o or = -DrGCO , respectively. From Fig. 8 it can be concluded that at pO2 = 1 bar, LCMA reduction to d = 0.10 is favorable at T = 2070 K and at even higher temperatures for the other perovskites and ceria. The oxidation of LSM40 and LCM40 with H2O and CO2 to d = 0.01 is not favorable at T > 400 K, whereas the oxidation of the Al-doped perovskites with H2O is favorable at 420 K (LSMA) and 405 K (LCMA). Their oxidation with CO2 is thermodynami-cally unfavorable at T > 400 K. The oxidation of ceria to d = 0.01 with H2O and CO2 is favorable at much higher temperature (T = 1050 K) compared to the oxidation of the perovskites. At T = 1050 K, the oxidation with H2O and CO2 is identically favorable. The favorable oxidation the rmodynamics of ceria can be explained by its relatively high DhO and DsO values. The high DsO results in a relatively high DgO value at lower temperatures and a relatively low DgO at higher temperatures. Out of the perovskites, LCMA shows the best thermodynamic performance because of the highest DsO and additionally a decreasing DhO with increasing d.

3.4. Water splitting

In this work we have decided to focus on H2O splitting, but the thermodynamics of CO2 splitting are qualitatively very similar and give similar insights into the performance of candidate redox materials [6,21]. Thermodynamically, the water splitting reaction can be described by the simultaneous oxidation of the perovskite with O2 and the splitting of H2O into H2 and 1/2O2, as shown by Eqs. (17) and (18), respectively.

-1-ABO3 d d + !O2 = -1-ABO3 d (17)

5red - 5ox 3-5red 2 2 5red - <W 3-dox ^ '

H2O = H2 + 2O2 (18)

The chemical equilibrium is described by the two simultaneous reaction equilibria

DrGox(5ox, Tox) = -Dgc(5ox, TK) = 2RTox ln(po2/p°)

DrG?,2o(Tox) = -RTox ln K

-RToxlnfpHpC/2p°-1/2' \ pH2O

where pO2, pH2 and pH2O denotes the equilibrium partial pressures of O2, H2 and H2O, respectively and Kw is the reaction equilibrium constant of the H2 oxidation reaction. Assigning reaction coordinates X1 to Eq. (17) and £2 to Eq. (18) and assuming that reactions proceed in a closed-system with variable volume, the equilibrium gas-phase composition may be found according to Ref. [33].

nH2O = nH2O,i - £2 nH2 = nH2,i + £2

nO2 = nO2,i - 2 x1 + 2 x2

ntot(g) = nO2 + nH2 + nH2O dox = dred " ?1

where the subscript i denotes initial molar amount of the gas species. The partial pressure of gas phase species g is defined by

Fig. 6. Standard partial molar enthalpy DhO for LSM40, LCM40, LSMA and LCMA as a function of d. Symbols represent measured values calculated based on d values shown by the black solid lines in Fig. 2 (a—d) whereas dashed lines represent modeled values calculated based on d values shown by the colored dashed lines in Fig. 2 (a—d).

pg = ^ptot ntot

where ptot denotes the closed system total pressure. For given nH2O,i, nH2,i (=0), nO2,i (=0), Tox, dred and ptot, the equilibrium gas phase composition (nH2O, nH2, nO2), the equilibrium partial pressures (pH2O, pH2, pO2) and the reaction coordinates (X1, £2) can be determined by simultaneously solving Eqs. (19)—(22). Gas phase amounts of H, O and OH were assumed to be negligible at T < 1600 K [34]. Fig. 9 (a) shows £1 = dred - dox, the H2 yield (black dashed lines), and nH2, the equilibrium amount of H2, (red solid

Fig. 5. -ln(pO2/p°) versus 1/T for LCM40 for the temperature range T = 1573 K-1773 К and oxygen nonstoichiometry range d = 0.0103-0.0675.

Fig. 7. Standard partial molar entropy DsO for LSM40, LCM40, LSMA and LCMA as a function of d. Symbols represent measured values calculated based on d values shown by the black solid lines in Fig. 2(a—d) whereas dashed lines represent modeled values calculated based on d values shown by the colored dashed lines in Fig. 2 (a—d).

lines) versus temperature upon oxidation of LCMA (dred = 0.127, calculated at Tred = 1673 K and po2,red = 10~4 bar) at ptot = 1 bar for nH2o,i = 1000 mol (circles), nH2O,i = 100 mol (triangles), nH2O,i = 1 mol (squares) and nH2Oi/infinity (+symbols). For nH2O,i = 1000 mol and T < 1200 K, ?1 z nH2, meaning that most of the O2 produced by Eq. (18) is consumed by the perovskite and therefore the amount of H2 from thermolysis of the excess H2O has a negligible contribution to the total equilibrium amount of H2. At T > 1200 K, nH2 > X1, meaning that a significant amount of the excess H2O splits into H2 and O2 whereas this amount of O2 is not consumed by the perovskite. When cooling down the equilibrium gas mixture to room temperature, nH2 minus X1 moles of H2 may recombine with the gas-phase O2 to H2O. Therefore, the minimum amount of H2 that can be used for further processing at lower temperatures (e.g. converting to liquid fuels) is X1, indicated by the black dashed lines and colored grey area in Fig. 9 (a). It is shown that H2 yields increase with decreasing temperature and increasing nH2O,i. For comparison, the equilibrium amount of H2 for thermolysis only (X1 = 0) as a function of temperature is shown in Fig. 9 (a) (blue dotted lines). As expected, towards higher temperatures, nH2 for thermolysis only converges towards nH2 upon oxidation of the perovskite. This is expected because towards higher temperatures the amount of gas-phase oxygen consumed by the perovskite decreases (decreasing X1 and increasing dox). Equilibrium amounts of H2 are always higher in the case of perovskite oxidation compared to thermolysis only. Fig. 9 (b) shows the corresponding equilibrium partial pressures pH2 (red solid lines) and pO2 (green solid lines) upon oxidation of LCMA. By increasing nH2O,j, pH2 decreases because of the higher dilution of H2 in H2O, which ultimately results in an increase in pO2 (cf. Eq. (20)). Because dox decreases with increasing pO2 at constant temperature (cf. Fig. 2), H2 yields (X1) increase with increasing nH2O,i. However, it should be kept in mind that the H2 concentration in the product gas will decrease when

Fig. 8. Standard partial molar Gibbs free energy Ag"O for LSM40, LCM40, LSMA, LCMA and CeO2 [7] for d = 0.10, relevant for the reduction, and negative DgO for d = 0.01, relevant for the oxidation reaction. Dashed lines represent the Gibbs free energy change of the H2 oxidation reaction: H2 + 0.5O2 = H2O (-DrGHjO, blue) and of the CO oxidation reaction: CO + 0.5O2 = CO2 (-DrGCOj, red). The metal oxide reduction reaction is at equilibrium in 1 bar pO2 at d = 0.10 at the temperature where DgO(d = 0.10) = 0. The oxidation with H2O/CO2 to d = 0.01 is thermodynamically favorable at temperatures where -Ag'O(d = 0.01) = -ArG"HlO or = -DrGCO2, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

flowing large amounts of H2O. The + symbols shown in Fig. 9(b) represent the minimum pH2 and maximum pO2 that could theoretically be obtained at ptot = 1 bar and are equal to the equilibrium partial pressures of the thermolysis reaction. Therefore, + symbols in Fig. 9(a) show the maximum H2 yield upon oxidation of LCMA with H2O (dred = 0.127) at 1 bar. For an open system, this would imply a perfectly purged reactive structure or running the oxidation for an infinite long time, whereas for a closed system, this would require an infinitely amount of H2O. The results for nH2O,i/infinity are in agreement with the thermogravimetric analysis of Cooper et al. [21] where they experimentally showed that CO production upon oxidation of LCMA with CO2 significantly decreases with increasing temperature and that the material cannot be fully oxidized at T > 1323 K under pure CO2.

The influence of ptot on the H2 yield upon oxidation of LCMA (dred = 0.127 at Tred = 1673 K and pO2,red = 10~4 bar) for nH2O,i = 100 mol is shown in Fig. 10(a) and the corresponding equilibrium partial pressures are shown in Fig. 10(b). Labeling is similar to Fig. 9, but here circles represent results for ptot = 0.01 bar, triangles ptot = 1 bar and squares ptot = 100 bar. It is shown that under most conditions (ptot > 0.01 bar and T < 1300 K), the total pressure does not significantly affect the equilibrium H2 yield, except for higher temperatures and lower ptot, where X1 s nH2; here the H2 yield slightly increases with increasing ptot. As long as X1 z nH2, implying that thermolysis of the excess steam does not significantly affect the H2 equilibrium amount, 1 mol of H2 is produced per mole of H2O, whereas the O2 is consumed by the perovskite. Therefore the number of moles in the gas phase stays constant and therefore the overall reaction is not influenced by a change in ptot. When thermolysis becomes significant, the number of moles in the equilibrium gas phase increases and therefore a change in ptot shifts the chemical equilibrium. Indeed, this is shown in Fig. 10(b), where pO2 does not significantly change by increasing or decreasing ptot, at least for ptot > 0.01 bar and T < 1300 K pH2 increases with increasing ptot, but the ratio of pH2O/pH2 is constant which leads to a constant pO2 according to Eq. (20). For high temperatures and large nH2O,i, where pO2 converges to pO2,thermolysis and pH2 to pH2,thermolysis, higher H2 yields could theoretically be obtained at higher ptot. However, such operation conditions (large amount of nH2O,i or running oxidation for a very long time) might not be efficient for a solar thermochemical process.

Fig. 11(a) shows the theoretical H2 yield at ptot = 1 bar, assuming reaction in a closed system with variable volume and constant pressure, versus temperature for LCMA, LSMA, LCM40, LSM40 and CeO2 [7]. Results are shown for nH2O,j = 1 mol per mole metal oxide (squares) and nH2O,i = 1000 mol (circles). The nonstoichiometry before oxidation (dred) was calculated for the reduction conditions Tred = 1673 K and pO2 = 10-4 bar and additionally for Tred = 1873 K and pO2 = 10~4 bar for CeO2 (open symbols). dred for each material is indicated by the solid horizontal lines and represent the maximum H2 yield that could be obtained when oxidizing to dox = 0. For nH2O,i = 1 mol, all the perovskites show only minor H2 production and reach oxidation extents of less than 42% for T > 600 K. However, CeO2 oxidizes by more than 85% as high as T = 1200 K; its maximum H2 yield (dred = 0.018) is relatively low under the reduction conditions considered for the perovskites however. By increasing nH2O,j to 1000 mol, all materials can be oxidized at higher temperatures because of the higher pO2, resulting in H2 yields higher than 80% at T = 1200 K for all materials. In summary, all perovskites investigated within this work only produce more H2 compared to ceria under a large excess of oxidant gas and/or at relatively low oxidation temperatures. Both scenarios imply additional energy penalties, for example to heat oxidant gas, separate reaction products and to overcome a high temperature difference between the low temperature oxidation and high temperature

reduction (large sensible heat penalty). The finding of higher H2 yield for CeO2 (at least for T < 750 K) compared to the perovskites under relatively small nH2O,i is in good agreement with the calculations of Scheffe et al. [14] for LSM30 and LSM40 and Cooper et al. [21] for LCM40, LCMA and LSMA. The finding of higher fuel yields with increasing amount of oxidant gas is in agreement with literature studies on doped ceria [6,7] and perovskites [14,21]. For such a high amount of oxidant gas (e.g. nH2O>i = 1000 mol per mole of metal oxide), highest H2 yield is reached with LCMA, the perovskite showing the highest reduction extent.

These results point out that qualitative oxidation experiments

under large CO2 and/or H2O excess in a thermogravimetric analyzer or equivalent test setup can result in misleading predictions of a material's potential to efficiently split CO2 and/or H2O in a solar reactor because of flowing relatively large amounts of oxidant gas for a long time and therefore attaining a high pO2. The high pO2 results then in a low oxygen nonstoichiometry after oxidation and therefore a high fuel yield. When assuming a total pressure of 1 bar and T = 1100 K, a maximum pO2 z 7.5 x 10-7 bar could be attained when flowing H2O only (equal to the equilibrium pO2 of the thermolysis reaction). However, the assumption of such a high pO2 for the equilibrium of the oxidation reaction with H2O would imply a perfectly purged material structure or an infinitely long oxidation

Fig. 9. (a) H2 yield (£ 1 = dred - dox) upon oxidation of LCMA (black dashed lines), total equilibrium amount of H2 (nH2) (red solid lines) and equilibrium amount of H2 considering direct thermolysis only (blue dotted lines) versus temperature. (b) Corresponding equilibrium partial pressures of H2 (red solid lines), of O2 (green solid lines) upon oxidation of LCMA. In (a) and (b), squares represents results for nH2cy = 1 mol, triangles for nH2Oi = 100 mol, circles for nH2Oi = 1000 mol and + symbols for nH2O i / infinity, all at ptot = 1 bar. pH2 and pO2 shown by the + symbols (nH2O,i/infinity) represents the minimum pH2 and maximum pO2 and are equal to the equilibrium partial pressures of the H2O thermolysis reaction. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 10. (a) H2 yield (£1 = dred - dox) upon oxidation of LCMA (black dashed lines) as a function of temperature and 3 different operating pressures. Total equilibrium yields of H2 (nH2) (red solid lines) and equilibrium amount of H2 considering direct thermolysis only (blue dotted lines) are also shown. (b) Corresponding equilibrium partial pressures of H2 (red solid lines), of O2 (green solid lines) upon oxidation of LCMA. Blue and black dotted lines indicate the equilibrium partial pressure of H2 and O2 for thermolysis only, which is equivalent to the partial pressures when nH2O i / infinity. In (a) and (b), squares represents results for ptot = 100 bar, triangles for ptot = 1 bar and circles for ptot = 0.01 bar, all for nH2O,i = 100 mol. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

reaction when considering an open reaction system, or an infinite amount of H2O when considering a closed system. For solar reactor systems, typical maximum pO2 during oxidation are much lower because it is inefficient to run oxidation until completion [35] and/ or to heat large amounts of excess CO2 or H2O. For example, by limiting the minimum H2 concentration in the product gas to 1% (at T = 298 K), a maximum pO2 of only about 1.6 x 10~14 bar could be obtained at T = 1100 K. This is more than seven orders of magnitude lower compared to the pO2 when flowing steam only. By looking at oxygen nonstoichiometry measurements of La05Sr05MnO3 [15] it can be seen that at T = 1100 K and pO2 z 7.5 x 10~7 bar (pure steam), its oxygen nonstoichiometry is dox z 0, whereas at

pO2 z 1.6 x 10~14 bar, dox z 0.17. This means that by doing experiments in a thermogravimetric analyzer (or equivalent test setup) with a large amount of H2O for a relatively long reaction time, H2 yield gets over-predicted by around 0.17 mol H2 per mole of La0.5Sr0.5MnO3 compared to a solar reactor where the minimum outlet concentration of H2 was fixed to 1%. Therefore extracted fuel yields from simple Ar/CO2 or Ar/H2O cycling experiments performed in a thermogravimetric analyzer (or equivalent setup) should be considered as maximum fuel yields attainable but do not necessarily represent a material's fuel productivity in a solar reactor.

4. Conclusions

Fig. 11. (a) Calculated H2 yield upon oxidation of LCMA, LSMA, LCM40, LSM40 and CeO2 [7] versus temperature for nH2O i = 1 mol (squares) and nH2O i = 1000 mol (circles) at ptot = 1 bar. (b) Calculated H2 yield upon oxidation of LCMA, LSMA, LCM40, LSM40 and CeO2 versus nH2Oi for T = 1200 K at ptot = 1 bar. In (a) and (b), the oxygen nonstoichiometry before oxidation (dred) is shown by the solid lines and was determined for Tred = 1673 K and pO2 = 10~4 bar for all materials and additionally for Tred = 1873 K and pO2 = 10~4 bar for CeO2 (open symbols). H2 yields close to their maximum are not shown because partial molar thermodynamic data was only calculated for d > 0.001.

Oxygen nonstoichiometry measurements of La0.6A0.4Mn1_yAlyO3 (A = Ca, Sr and y = 0, 0.4) in the temperature range T = 1573 K—1773 K and oxygen partial pressure range pO2 = 4.5066 x 10~2 bar—9.9 x 10~5 bar revealed that the highest reduction extents are obtained using La0.6Ca0.4Mn0.6Al0.4O3 (LCMA). Compared to the state-of-the-art material ceria, LCMA releases more than 500% more O2 per unit mass of redox material at T = 1700 K and pO2 = 10~3 bar. It releases 0.15 mol O2 per kg of redox material at 285 K lower reduction temperature than ceria (1885 K for CeO2 and 1600 K for LCMA) at pO2 = 10~3 bar. It was found that oxygen nonstoichiometry increases when replacing the divalent dopant Sr in La0.6Sr0.4MnO3 with Ca and additionally significantly increases when doping 40 mol-% Al on the Mn-site. The oxygen nonstoichiometry of all perovskites investigated was accurately modeled by a chemical defect model considering the reduction of Mn4+ to Mn3+ in combination with a disproportionate reaction of Mn3+ to Mn4+ and Mn2+. From the defect models, partial molar thermodynamic properties (DhO, DsO, DgO) were extracted. When doping Al to the Mn-site of La0 6Sr04MnO3 and La06Ca04MnO3 , DhO changes its trend of increasing DhO with d to decreasing DhO with d. This has an impact on the performance because a low enthalpy value towards higher d is desired for a favorable reduction and a high enthalpy value at low d i s desired for a favorable oxidation with H2O and CO2. In general DhO and DsO of the perovskites are significantly lower compared to CeO2, leading to a thermodynamically more favorable reduction step, but less favorable oxidation (fuel production) step. Indeed, closed system calculations indicate that H2 or CO yields of CeO2 can only be surpassed using high amounts of excess H2O/CO2 and/or at much lower oxidation temperatures; both approaches imply additional energy penalties to heat excess H2O/CO2 and/or to overcome larger temperature differences between oxidation and reduction steps. Therefore, compared to the perovskites investigated in this work, a material with a higher DhO and higher As°O would be desired, yielding a higher Ag°O at lower temperatures (relevant for oxidation) and a lower DgO at higher temperatures (relevant for reduction). A strong decrease of DhO with d would also be favorable.

Acknowledgments

The financial support by the European Research Council under the European Union's ERC Advanced Grant (SUNFUELS - no. 320541) is gratefully acknowledged. We thank Miriam Ezbiri and Viola Becattini for the sample preparation and characterization.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.actamat.2015.10.026.

Nomenclature

DrGco2

DgO DhO

LCM40 LCMA LSM40 LSMA

Mo Ms ms

Dms ng

ng,i nO2'

ptot p°

DsO Ds+

dox 5red

standard Gibbs free energy change of the CO2 formation reaction (kJ mol-1)

standard Gibbs free energy change of the H2O formation reaction (kJ mol-1)

standard Gibbs free energy change of the perovskite

oxidation reaction with O2 (kJ mol-1)

standard partial molar Gibbs free energy (kJ mol-1)

standard partial molar enthalpy (kJ mol-1)

enthalpy of Mn4+ to Mn3+ defect reaction (kJ mol-1)

enthalpy of Mn4+ to Mn2+ defect reaction (kJ mol-1)

Mn4+ to Mn3+ defect reaction equilibrium constant (-)

Mn4+ to Mn2+ defect reaction equilibrium constant (-)

H2O dissociation equilibrium constant (-)

La0.6Ca0.4MnO3

La0.6Ca0.4Mn0.6Al0.4O3

La06Sr04MnO3

La0.6Sr0.4Mn0.6Al0.4O3

divalent manganese on manganese lattice site

trivalent manganese on manganese lattice site

tetravalent manganese on manganese lattice site

molar mass of O (g mol-1)

molar mass of reactive sample (g mol-1)

mass of reactive sample (mg)

relative mass change of reactive sample (-)

equilibrium molar amount of species g (mol)

initial molar amount of species g (mol)

mass specific oxygen release (mol kg-1)

oxygen atom on oxygen lattice site

partial pressure of gas g (bar)

system pressure (bar)

standard pressure (bar)

universal gas constant (J mol-1 K-1)

standard partial molar entropy (J mol-1 K-1)

entropy of Mn4+ to Mn3+ defect reaction (J mol-1 K-1)

entropy of Mn4+ to Mn2+ defect reaction (J mol-1 K-1)

temperature (K)

doubly ionized oxygen vacancy

A-site molar dopant concentration (-)

B-site molar dopant concentration (-)

degree of oxygen nonstoichiometry (-)

degree of oxygen nonstoichiometry after oxidation (-)

degree of oxygen nonstoichiometry after reduction (-)

reaction coordinate of perovskite oxidation with O2

reaction coordinate of water dissociation

References

[1] W.C. Chueh, S.M. Haile, A thermochemical study of ceria: exploiting an old material for new modes of energy conversion and CO2 mitigation, Philos. Trans. R. Soc. A Math. Phys. Eng. Sci. 368 (2010) 3269—3294.

[2] C. Perkins, P.R. Lichty, A.W. Weimer, Thermal ZnO dissociation in a rapid aerosol reactor as part of a solar hydrogen production cycle, Int. J. Hydrogen Energy 33 (2008) 499—510.

[3] M.D. Allendorf, R.B. Diver, N.P. Siegel, J.E. Miller, Two-step water splitting using mixed-metal ferrites: thermodynamic analysis and characterization of synthesized materials, Energy Fuels 22 (2008) 4115—4124.

[4] T. Kodama, Y. Nakamuro, T. Mizuno, A two-step thermochemical water splitting by iron-oxide on stabilized zirconia, J. Sol. Energy Eng. 128 (2006) 3—7.

[5] J.R. Scheffe, A. Steinfeld, Oxygen exchange materials for solar thermochemical splitting of H2O and CO2: a review, Mater. Today 17 (2014) 341 —348.

[6] J.R. Scheffe, A. Steinfeld, Thermodynamic analysis of cerium-based oxides for solar thermochemical fuel production, Energy Fuels 26 (2012) 1928—1936.

[7] M. Takacs, J.R. Scheffe, A. Steinfeld, Oxygen nonstoichiometry and thermo-dynamic characterization of Zr doped ceria in the 1573—1773 K temperature range, Phys. Chem. Chem. Phys. 17 (2015) 7813—7822.

[8] J.R. Scheffe, R. Jacot, G.R. Patzke, A. Steinfeld, Synthesis, characterization, and thermochemical redox performance of Hf4+, Zr4+, and Sc3+ doped ceria for

splitting CO2, J. Phys. Chem. C 117 (2013) 24104-24114. S. Abanades, A. Legal, A. Cordier, G. Peraudeau, G. Flamant, A. Julbe, Investigation of reactive cerium-based oxides for H2 production by thermochemical two-step water-splitting, J. Mater. Sci. 45 (2010) 4163-4173. M. Kuhn, S.R. Bishop, J.L.M. Rupp, H.L Tuller, Structural characterization and oxygen nonstoichiometry of ceria-zirconia (Ce1 xZrxO2 5) solid solutions, Acta Mater. 61 (2013) 4277-4288.

F. Call, M. Roeb, M. Schmücker, H. Bru, D. Curulla-Ferre, C. Sattler, R. Pitz-Paal, Thermogravimetric analysis of zirconia-doped ceria for thermochemical production of solar fuel, Am. J. Anal. Chem. 4 (2013) 37. Y. Hao, C.-K. Yang, S.M. Haile, Ceria-zirconia solid solutions (Ce1-xZrxO2 j, x < 0.2) for solar thermochemical water splitting: a thermodynamic study, Chem. Mater. 26 (2014) 6073-6082.

Q.-L Meng, C.-i. Lee, T. Ishihara, H. Kaneko, Y. Tamaura, Reactivity of CeO2-based ceramics for solar hydrogen production via a two-step water-splitting cycle with concentrated solar energy, Int. J. Hydrogen Energy 36 (2011) 13435-13441.

J.R. Scheffe, D. Weibel, A. Steinfeld, Lanthanum-strontium-manganese pe-rovskites as redox materials for solar thermochemical splitting of H2O and CO2, Energy Fuels 27 (2013) 4250-4257.

J. Mizusaki, N. Mori, H. Takai, Y. Yonemura, H. Minamiue, H. Tagawa, M. Dokiya, H. Inaba, K. Naraya, T. Sasamoto, T. Hashimoto, Oxygen non-stoichiometry and defect equilibrium in the perovskite-type oxides La1 xSrxMnO3+d, Solid State Ion. 129 (2000) 163-177.

C.-K. Yang, Y. Yamazaki, A. Aydin, S.M. Haile, Thermodynamic and kinetic assessments of strontium-doped lanthanum manganite perovskites for two-step thermochemical water splitting, J. Mater. Chem. A 2 (2014) 13612-13623.

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 j for solar thermochemical H2 and CO production, Energy Environ. Sci. 6 (2013) 2424-2428. S. Dey, B.S. Naidu, A. Govindaraj, C.N.R. Rao, Noteworthy performance of La1-xCaxMnO3 perovskites in generating H2 and CO by the thermochemical splitting of H2O and CO2, Phys. Chem. Chem. Phys. 17 (2015) 122-125. S. Dey, B.S. Naidu, C.N.R. Rao, Ln0.5A0.5MnO3 (Ln=Lanthanide, A= Ca, Sr) pe-rovskites exhibiting remarkable performance in the thermochemical generation of CO and H2 from CO2 and H2O, Chem. A Eur. J. 21 (2015) 7077-7081. A. Demont, S. Abanades, E. Beche, Investigation of perovskite structures as oxygen-exchange redox materials for hydrogen production from thermochem-ical two-step water-splitting cycles, J. Phys. Chem. C 118 (2014) 12682-12692. T. Cooper, J.R. Scheffe, M.E. Galvez, R. Jacot, G. Patzke, A. Steinfeld, Lanthanum manganite perovskites with Ca/Sr A-site and Al B-site doping as effective oxygen exchange materials for solar thermochemical fuel production, Energy Technol. (2015), http://dx.doi.org/10.1002/ente.201500226. A.H. Bork, M. Kubicek, M. Struzik, J.L.M. Rupp, Perovskite La0.6Sr0.4Cr1-xCoxO3-j solid solutions for solar-thermochemical fuel production: strategies to lower the operation temperature, J. Mater. Chem. A 3 (30) (2015) 15546-15557. M.E. Galvez, R. Jacot, J. Scheffe, T. Cooper, G. Patzke, A. Steinfeld, Physico-chemical changes in Ca, Sr and Al-doped La-Mn-O perovskites upon ther-mochemical splitting of CO2 via redox cycling, Phys. Chem. Chem. Phys. 17 (2015) 6629-6634.

J.H. Kuo, H.U. Anderson, D.M. Sparlin, Oxidation-reduction behavior of undoped and Sr-doped LaMnO3 nonstoichiometry and defect structure, J. Solid State Chem. 83 (1989) 52-60.

J.A.M. Van Roosmalen, E.H.P. Cordfunke, A new defect model to describe the oxygen deficiency in perovskite-type oxides, J. Solid State Chem. 93 (1991) 212-219.

J. Nowotny, M. Rekas, Defect chemistry of (La,Sr)MnO3, J. Am. Ceram. Soc. 81 (1998) 67-80.

M. Oishi, K. Yashiro, K. Sato, J. Mizusaki, T. Kawada, Oxygen nonstoichiometry and defect structure analysis of B-site mixed perovskite-type oxide (La, Sr)(Cr, M)O3 d (M=Ti, Mn and Fe), J. Solid State Chem. 181 (2008) 3177-3184. A.Y. Zuev, D.S. Tsvetkov, Oxygen nonstoichiometry, defect structure and defect-induced expansion of undoped perovskite LaMnO3±j, Solid State Ion. 181 (2010) 557-563.

S. Sengodan, J. Kim, J. Shin, G. Kim, Thermodynamic properties, defect analysis, and electrical conductivity of the Lao.8Sro.2ScxMn1-xO3 j infiltrated into YSZ scaffolds, J. Electrochem. Soc. 158 (2011) B1373-B1379. F.A. Kröger, H.J. Vink, Relations between the concentrations of imperfections in crystalline solids, in: S. Frederick, T. David (Eds.), Solid State Physics, vol. 3, Academic Press, 1956, pp. 307-435.

S. Tanasescu, C. Marinescu, F. Maxim, A. Sofronia, N. Totir, Evaluation of manganese and oxygen content in La0.7Sr03MnO3 j and correlation with the thermodynamic data, J. Solid State Electrochem. 15 (2011) 189-196. S. Tanasescu, F. Maxim, F. Teodorescu, L. Giurgiu, Influence of composition and particle size on spin dynamics and thermodynamic properties of magneto-resistive perovskites, J. Nanosci. Nanotechnol. 8 (2008) 914-923. J.M. Smith, H.C. Van Ness, M.M. Abbott, Introduction to Chemical Engineering Thermodynamics, seventh ed., McGraw-Hill, Boston, 2005. J. Lede, F. Lapicque, J. Villermaux, Production of hydrogen by direct thermal decomposition of water, Int. J. Hydrogen Energy 8 (1983) 675-679.

D. Marxer, P. Furler, J. Scheffe, H. Geerlings, C. Falter, V. Batteiger, A. Sizmann, A. Steinfeld, Demonstration of the entire production chain to renewable kerosene via solar thermochemical splitting of H2O and CO2, Energy Fuels 29 (5) (2015) 3241-3250.