Improved Thermochemical Energy Storage Behavior of Manganese Oxide by Molybdenum Doping

To improve the thermochemical energy storage (TCS) behavior of Mn2O3, several Mn–Mo oxides with varying amounts of MoO3 (0–30 wt%) were prepared by a precipitation method. The physico-chemical properties of the solids were studied by N2 adsorption–desorption, X-ray diffraction (XRD), scanning electron microscopy (SEM), and H2-temperature-programmed reduction (TPR), while their TCS behavior was determined by thermogravimetric analysis coupled with differential scanning calorimetry (TGA-DSC). Apart from Mn2O3 and MoO3 phases, XRD revealed a mixed MnMoO4 phase for MoO3 loadings equal or higher than 1.5 wt%. All samples showed a well-formed coral-like surface morphology, particularly those solids with low MoO3 contents. This coral morphology was progressively decorated with compact and Mo-enriched MnMoO4 particles as the MoO3 content increased. TPR revealed that the redox behavior of Mn2O3 was significantly altered upon addition of Mo. The TCS behavior of Mn2O3 (mostly oxidation kinetics and redox cyclability) was enhanced by addition of low amounts of Mo (0.6 and 1.5% MoO3) without significantly increasing the reduction temperature of the solids. The coral morphology (which facilitated oxygen diffusion) and a smoother transition from the reduced to oxidized phase were suggested to be responsible for this improved TCS behavior. The samples containing 0.6 and 1.5 wt% of MoO3 showed outstanding cyclability after 45 consecutive reduction–oxidation cycles at high temperatures (600–1000 °C). These materials could potentially reach absorption efficiencies higher than 90% at concentration capacity values typical of concentrated solar power plants.


Introduction
Global energy consumption is increasing by the ever-growing demand of energy from developing countries, and renewable energy sources are becoming more important in covering this demand [1]. Among the renewable energy technologies available today, concentrated solar power (CSP), with an installed generation capacity of 4.4 GW in 2014 [2], is particularly relevant. As is common to renewable energies, the discontinuous nature of solar energy makes necessary the development of storage technologies to overcome the negative effects of transitory weather conditions, to modulate low-to-high energy gaps, and to achieve continuous 24 h electricity production [3,4]. In the case of CSP, the development of efficient thermal energy storage (TES) systems is particularly important. Three different TES approaches based on sensible, latent, and thermochemical heat are currently available. Among them, thermochemical energy storage (TCS, i.e., the utilization of a reversible chemical process to store thermal energy) holds great promise for future CSP plants owing to its higher energy density and discharge temperatures compared to sensible-and latent heat-based approaches. While high discharge temperatures allow increasing the efficiency of steam turbines, high energy densities are important for reducing the size of the storage tank facilities and, therefore, decreasing the capital expenses of the plant [3,5,6]. TCS allows solar thermal energy to be stored (and subsequently released) in form of chemical energy by promoting a high-temperature reversible chemical process. Thus, a highly endothermic reaction is used for the charging step, while the reverse exothermic process allows releasing energy in the discharge step while closing the chemical loop for a new cycle. As chemical bonds are broken and formed during the process, large amounts of energy are potentially absorbed/released within each cycle [7,8].
A good number of solids including hydrides, hydroxides, carbonates, and oxides have been tested as TCS materials. Among them, metal oxides are particularly interesting because of their high reaction temperatures and the utilization of oxygen as a reactant, which opens the possibility of using air as a working fluid and thus simplify gas storage installations [5,6,9]. General Atomics conducted a comprehensive analysis on the main TCS technologies with potential to be used in CSP plants, covering a large number of metal oxides [10]. This study identified cobalt (II,III) oxide (Co 3 O 4 ) as the most promising TCS material in virtue of its superior energy density, high conversion temperature, and good cyclability. Despite its demonstrated thermal stability [11], Co 3 O 4 is costly and toxic and, therefore, less harmful and cheaper alternatives are required. In this sense, other oxides such as CuO have shown superior energy density and reaction temperatures [11]. However, the reduction temperature of CuO is close to the melting point of its reduced form (Cu 2 O), which increases the risk of sintering and loss of activity [11][12][13]. Barium oxide (BaO), despite having a reaction temperature well suited for superheated steam cycles, suffers from poor thermal stability and high sensitivity to water and CO 2 traces (leading to inactive hydroxide or carbonate materials, respectively) [6,11,[14][15][16].
Manganese (III) oxide (Mn 2 O 3 ) has been widely studied as a TCS material [17] in virtue of the following charge and discharge processes: Despite its low cost and abundancy, Mn 2 O 3 suffers from a number of drawbacks. Thus, the reduced form of Mn 2 O 3 (Mn 3 O 4 ) has been reported to show very low re-oxidation rates, which leads to incomplete regeneration of the original oxide and poor cycling behavior [10,17]. These limitations can be overcome by enhancing the redox behavior of Mn via formation of a mixed oxide [13,[18][19][20]. Among the large number of combinations tested, Mn-Fe mixed oxides have shown good oxidation behaviors and faster re-oxidation rates compared to Mn 2 O 3 . However, these improvements are achieved at the expense of increasing the reduction (charge) temperature of the resulting mixed oxides, which complicates the CSP process [8,10,13,21,22]. To overcome this issue, we suggest herein, for the first time, the utilization of Mo in replacement of Fe to improve the TCS properties of Mn 2 O 3 without significantly increasing the reduction temperature of the solids. Mo was selected because it possesses very similar size (ionic radius of Mo 6+ : 73 pm) compared to Mn 3+ (72 pm), thereby facilitating the substitution process. In fact, Mn and Mo have been previously described to form stable mixed-oxide phases such as MoMnO 4 and Mo 3 Mn 2 O 8 [23,24], and Mo 6+ has been reported to substitute Mn 3+ by forming six-fold octahedral coordination compounds [25][26][27]. Having this in mind, we herein analyzed the suitability of Mo-doped Mn 2 O 3 as TCS materials. Several Mo-doped manganese oxides with varying loadings of Mo were prepared by a precipitation method. The textural, structural, morphological, redox, and thermochemical properties of these materials were characterized by N 2 adsorption-desorption, X-ray diffraction (XRD), scanning electron microscopy (SEM), H 2 -temperature-programmed reduction (TPR), and thermogravimetric analysis coupled with differential scanning calorimetry (TGA-DSC), respectively. were dissolved in 250 mL of distilled water and stirred for 30 min. A 2.5 M NaOH aqueous solution was subsequently added dropwise until pH = 12 was reached. This pH was maintained for 4 h while stirring, and the as-obtained precipitate was subsequently filtered and cleaned three times with 500 mL of Milli-Q ® water. The as-obtained solid was subsequently dried at 100 • C overnight, ground to fine powders with an agate mortar, calcined at 750 • C under static air for 4 h (10 • C/min heating rate), ground again, and finally stored for analysis.
In addition, the 100% MoO 3 sample was prepared by directly calcining (NH 4 ) 6 Mo 7 O 24 · 4H 2 O at 350 • C for 4 h. The calcined sample was ground in an agate mortar. This material was not characterized in full since it sublimates at 600 • C [28].

Materials Characterization
TPR experiments (ca. 45 mg of sample) were performed on a Micromeritics Autochem II within a temperature range of 60-900 • C (10 • C/min heating rate) under flowing (50 mL/min) 10%H 2 /Ar. The specific surface areas of the samples were obtained by adsorption-desorption of N 2 at 77 K on a Micromeritics ASAP 2020 following the multipoint Brunauer-Emmett-Teller (BET) method. The samples were previously degassed at 300 • C under vacuum for 100 min. Powder XRD analyses were carried out on a Rigaku UltimaIV diffractometer employing CuKα radiation and a scanning rate of 2 • /min. SEM observations were carried out on a Hitachi S5200 microscope, and samples were used without any further modification. Inductively coupled plasma atomic emission spectroscopy (ICP-AES) was used to determine the actual oxide content of the different samples. This measurement was made on a Varian Vista AX spectrometer, and samples were previously dissolved in 5 mL of aqua regia (HNO 3 :HCl, 1:3 in volume).

Thermal Cyclability
The TCS behavior of the samples upon repetitive thermal cycles was studied by TGA-DSC on a TA Instruments SDT Q-600 device. The samples (18-22 mg) were placed on 90 µL alumina crucibles. At least 10 redox cycles were performed over each sample under flowing air (50 mL/min, oxidation) at temperatures ranging from 600 to 1000 • C.

Thermodynamic Study
We considered the following process to describe the thermochemical behavior of Mn-Mo mixed oxides: The process was theoretically studied using the HSC Chemistry 6.1 software @Outotec Research Oy. The dependence of the Gibbs free energy of reaction (∆G r T ) with the temperature was studied. This energy was calculated from the standard enthalpies (∆H • f ) and entropies (∆S • f ) of formation for the pure oxides (Mn 2 O 3 and MoO 3 ) and the temperature dependence of its specific heat (Cp(T)). The data for the mixed oxide samples were obtained by interpolating pure oxides at their corresponding content (Figure 1), and the results are summarized in Table 1. As shown in Table 1 and Figure 1, the reduction temperature increased significantly with the amount of Mo in the mixed oxide. This theoretical evaluation assumed the formation of a mixed oxide phase having intermediate physical-chemical properties mixed as compared to the pure manganese or molybdenum oxides.

Physical-Chemical Characterization
The morphology, composition, and crystal structure of the samples were characterized by a number of techniques. As shown in Table 1, the actual Mo contents of the samples (determined by ICP-AES) were very similar to the nominal values. All the samples showed low BET surface areas and low porosity, probably as a result of the high calcination temperature used herein (750 °C). These samples are expected to undergo sintering during thermal cycling, decreasing the surface area below the detection limit of the adsorption device. Figure 2a shows the XRD patterns of all the oxides under study. The diffractograms revealed the presence of a cubic Mn2O3 crystalline phase (ICDD: 00-002-0896) for all the samples except for 100% MoO3. The XRD pattern of this cubic Mn2O3 crystalline phase

Physical-Chemical Characterization
The morphology, composition, and crystal structure of the samples were characterized by a number of techniques. As shown in Table 1, the actual Mo contents of the samples (determined by ICP-AES) were very similar to the nominal values. All the samples showed low BET surface areas and low porosity, probably as a result of the high calcination temperature used herein (750 °C). These samples are expected to undergo sintering during thermal cycling, decreasing the surface area below the detection limit of the adsorption device. Figure 2a shows the XRD patterns of all the oxides under study. The diffractograms revealed the presence of a cubic Mn2O3 crystalline phase (ICDD: 00-002-0896) for all the samples except for 100% MoO3. The XRD pattern of this cubic Mn2O3 crystalline phase :0%MoO 3 ;

Physical-Chemical Characterization
The morphology, composition, and crystal structure of the samples were characterized by a number of techniques. As shown in Table 1, the actual Mo contents of the samples (determined by ICP-AES) were very similar to the nominal values. All the samples showed low BET surface areas and low porosity, probably as a result of the high calcination temperature used herein (750 °C). These samples are expected to undergo sintering during thermal cycling, decreasing the surface area below the detection limit of the adsorption device. Figure 2a shows the XRD patterns of all the oxides under study. The diffractograms revealed the presence of a cubic Mn2O3 crystalline phase (ICDD: 00-002-0896) for all the samples except for 100% MoO3. The XRD pattern of this cubic Mn2O3 crystalline phase

Physical-Chemical Characterization
The morphology, composition, and crystal structure of the samples were characterized by a number of techniques. As shown in Table 1, the actual Mo contents of the samples (determined by ICP-AES) were very similar to the nominal values. All the samples showed low BET surface areas and low porosity, probably as a result of the high calcination temperature used herein (750 °C). These samples are expected to undergo sintering during thermal cycling, decreasing the surface area below the detection limit of the adsorption device. Figure 2a shows the XRD patterns of all the oxides under study. The diffractograms revealed the presence of a cubic Mn2O3 crystalline phase (ICDD: 00-002-0896) for all the samples except for 100% MoO3. The XRD pattern of this cubic Mn2O3 crystalline phase

Physical-Chemical Characterization
The morphology, composition, and crystal structure of the samples were characterized by a number of techniques. As shown in Table 1, the actual Mo contents of the samples (determined by ICP-AES) were very similar to the nominal values. All the samples showed low BET surface areas and low porosity, probably as a result of the high calcination temperature used herein (750 °C). These samples are expected to undergo sintering during thermal cycling, decreasing the surface area below the detection limit of the adsorption device. Figure 2a shows the XRD patterns of all the oxides under study. The diffractograms revealed the presence of a cubic Mn2O3 crystalline phase (ICDD: 00-002-0896) for all the samples except for 100% MoO3. The XRD pattern of this cubic Mn2O3 crystalline phase

Physical-Chemical Characterization
The morphology, composition, and crystal structure of the samples were characterized by a number of techniques. As shown in Table 1, the actual Mo contents of the samples (determined by ICP-AES) were very similar to the nominal values. All the samples showed low BET surface areas and low porosity, probably as a result of the high calcination temperature used herein (750 °C). These samples are expected to undergo sintering during thermal cycling, decreasing the surface area below the detection limit of the adsorption device. Figure 2a shows the XRD patterns of all the oxides under study. The diffractograms revealed the presence of a cubic Mn2O3 crystalline phase (ICDD: 00-002-0896) for all the samples except for 100% MoO3. The XRD pattern of this cubic Mn2O3 crystalline phase : 12%MoO 3 ;

Physical-Chemical Characterization
The morphology, composition, and crystal structure of the samples were characterized by a number of techniques. As shown in Table 1, the actual Mo contents of the samples (determined by ICP-AES) were very similar to the nominal values. All the samples showed low BET surface areas and low porosity, probably as a result of the high calcination temperature used herein (750 °C). These samples are expected to undergo sintering during thermal cycling, decreasing the surface area below the detection limit of the adsorption device. Figure 2a shows the XRD patterns of all the oxides under study. The diffractograms revealed the presence of a cubic Mn2O3 crystalline phase (ICDD: 00-002-0896) for all the samples except for 100% MoO3. The XRD pattern of this cubic Mn2O3 crystalline phase : 31%MoO 3 ;

Physical-Chemical Characterization
The morphology, composition, and crystal structure of the samples were characterized by a number of techniques. As shown in Table 1, the actual Mo contents of the samples (determined by ICP-AES) were very similar to the nominal values. All the samples showed low BET surface areas and low porosity, probably as a result of the high calcination temperature used herein (750 °C). These samples are expected to undergo sintering during thermal cycling, decreasing the surface area below the detection limit of the adsorption device. Figure 2a shows the XRD patterns of all the oxides under study. The diffractograms revealed the presence of a cubic Mn2O3 crystalline phase (ICDD: 00-002-0896) for all the samples except for 100% MoO3. The XRD pattern of this cubic Mn2O3 crystalline phase : 100%MoO 3 .

Physical-Chemical Characterization
The morphology, composition, and crystal structure of the samples were characterized by a number of techniques. As shown in Table 1, the actual Mo contents of the samples (determined by ICP-AES) were very similar to the nominal values. All the samples showed low BET surface areas and low porosity, probably as a result of the high calcination temperature used herein (750 • C). These samples are expected to undergo sintering during thermal cycling, decreasing the surface area below the detection limit of the adsorption device. Figure 2a shows the XRD patterns of all the oxides under study. The diffractograms revealed the presence of a cubic Mn 2 O 3 crystalline phase (ICDD: 00-002-0896) for all the samples except for 100% MoO 3 . The XRD pattern of this cubic Mn 2 O 3 crystalline phase was characterized by an intense diffraction peak at 32.9 • . As shown in Figure 2b, the addition of small amounts of Mo (0.6% MoO 3 ) resulted in a very slight shifting of this peak to higher diffraction angles. However, no gradual angle shifting was observed as the amount of Mo increased above 0.6% MoO 3 . Given the very similar ionic radii of Mn 3+ and Mo 6+ ions was characterized by an intense diffraction peak at 32.9°. As shown in Figure 2b, the addition of small amounts of Mo (0.6% MoO3) resulted in a very slight shifting of this peak to higher diffraction angles. However, no gradual angle shifting was observed as the amount of Mo increased above 0.6% MoO3. Given the very similar ionic radii of Mn 3+ and Mo 6+ ions (72 vs. 73 pm), no significant changes in the lattice parameter of Mn2O3 are expected upon Mo doping, thereby resulting in negligible diffraction angle shifts. At higher Mo loadings (8.7-31% MoO3), a fraction of the Mo added ended up forming a new monoclinic MnMoO4 crystal phase (ICDD: 00-050-1287) with the main diffraction peak at 25.7°. The intensity of this peak increased with the Mo loading. The 100% MoO3 sample showed a very different XRD pattern, characterized by a main peak at 27.4°. This pattern was ascribed to a MoO3 crystalline phase (ICDD: 00-001-0706). The diffraction peaks characteristics of this phase were not found for the rest of the samples, allowing us to rule out the presence of MoO3 in these materials. Unlike MoO3, the Mn-Mo mixed oxide phases are thermally stable at 750 °C [28]. No characteristic peaks corresponding to Mn2Mo3O8 were revealed by XRD [23]. The morphology of the materials was observed by SEM, and the most representative micrographs are shown in Figure 3a. Remarkably, all the samples except 30% MoO3 The morphology of the materials was observed by SEM, and the most representative micrographs are shown in Figure 3a. Remarkably, all the samples except 30% MoO 3 showed a well-formed coral-like morphology. This coral-like morphology is believed to be key in facilitating oxygen diffusion during the redox process and in preventing excessive sintering of the material at high temperatures. However, as the amount of Mo increased in the samples, the micrographs revealed the presence of compact secondary particles (encircled in red in Figure 3a), which was more evident for the 8.7, 12, and 31% MoO 3 materials. We carried out energy-dispersive X-ray spectroscopy (EDS) analyses on our samples by performing elemental measurements at different spots of the SEM pictures (indicated by numbers in Figure 3a). The EDS spectra obtained for each of these spots are shown in Figure 3b. These analyses we aimed to confirm the presence of Mo in our samples. These analyses were performed over zones with coral-like morphologies. At these zones, we found Mo contents relatively close to the nominal loadings in all cases except for the 12% MoO 3 sample, which showed an unexpectedly low value. EDS also allowed us to compare the composition of compact secondary particles with that of the coral-like zones. As shown for the 1.5% MoO 3 sample (spots 3 and 4 in Figure 3a,b), the compact particle analyzed was enriched in Mo as compared to the coral-like zone, which showed a composition more similar to the nominal loading. This result seems to support the hypothesis that compact particles might correspond to MnMoO 4 , in line with XRD results. However, we note that these compositional analyses based on EDS are not fully representative of the totality of the samples and have a certain variability depending on the spot analyzed. showed a well-formed coral-like morphology. This coral-like morphology is believed to be key in facilitating oxygen diffusion during the redox process and in preventing excessive sintering of the material at high temperatures. However, as the amount of Mo increased in the samples, the micrographs revealed the presence of compact secondary particles (encircled in red in Figure 3a), which was more evident for the 8.7, 12, and 31% MoO3 materials. We carried out energy-dispersive X-ray spectroscopy (EDS) analyses on our samples by performing elemental measurements at different spots of the SEM pictures (indicated by numbers in Figure 3a). The EDS spectra obtained for each of these spots are shown in Figure 3b. These analyses we aimed to confirm the presence of Mo in our samples. These analyses were performed over zones with coral-like morphologies. At these zones, we found Mo contents relatively close to the nominal loadings in all cases except for the 12% MoO3 sample, which showed an unexpectedly low value. EDS also allowed us to compare the composition of compact secondary particles with that of the coral-like zones. As shown for the 1.5% MoO3 sample (spots 3 and 4 in Figure 3a,b), the compact particle analyzed was enriched in Mo as compared to the coral-like zone, which showed a composition more similar to the nominal loading. This result seems to support the hypothesis that compact particles might correspond to MnMoO4, in line with XRD results. However, we note that these compositional analyses based on EDS are not fully representative of the totality of the samples and have a certain variability depending on the spot analyzed.  The TPR profiles of the samples under study are shown in Figure 4. The 0% MoO3 sample showed a reduction profile similar to that previously reported for Mn2O3, with two reduction peaks at 250 and 400 °C [29,30]. Considering the amount of hydrogen consumed, these peaks can be attributed to the successive reduction of Mn 3+ to Mn( 3+ , 2+ ) and finally to Mn 2+ . Interestingly, the addition of low amounts of Mo (0.6 and 1.5% MoO3) changed dramatically the reduction profile of Mn2O3 from a double-peak to a single reduction peak centered at ca. 500 °C. This dramatic change in the reduction profile suggests that Mo ions could be forming part of the structure of Mn2O3, strongly affecting the redox behavior of Mn 3+ . The replacement of Mn 3+ with Mo 6+ is expected to generate a charge imbalance in the Mn2O3 lattice, given the different oxidation states of the exchangeable ions. Thus, each Mo +6 ion introduced will generate a surplus of positive charges in the The TPR profiles of the samples under study are shown in Figure 4. The 0% MoO 3 sample showed a reduction profile similar to that previously reported for Mn 2 O 3 , with two reduction peaks at 250 and 400 • C [29,30]. Considering the amount of hydrogen consumed, these peaks can be attributed to the successive reduction of Mn 3+ to Mn( 3+ , 2+ ) and finally to Mn 2+ . Interestingly, the addition of low amounts of Mo (0.6 and 1.5% MoO 3 ) changed dramatically the reduction profile of Mn 2 O 3 from a double-peak to a single reduction peak centered at ca. 500 • C. This dramatic change in the reduction profile suggests that Mo ions could be forming part of the structure of Mn 2 O 3 , strongly affecting the redox behavior of Mn 3+ . The replacement of Mn 3+ with Mo 6+ is expected to generate a charge imbalance in the Mn 2 O 3 lattice, given the different oxidation states of the exchangeable ions. Thus, each Mo +6 ion introduced will generate a surplus of positive charges in the lattice. As a result, the Mn 2 O 3 lattice is expected to generate certain crystalline defects (i.e., one Mn 3+ vacancy for each Mo 6+ ion exchanged) to offset the excess of positive charges. These cation vacancies can produce significant changes in the properties of Mn 2 O 3 even with relatively low doping amounts. Thus, Mn 3+ cation vacancies generated upon Mo +6 doping have been previously reported to result in significant changes in the morphology and redox behavior of manganese oxide materials [27]. As the amount of Mo increases in the samples (8.7-31% MoO 3 ), new broad reduction peaks appeared at temperatures higher than 600 • C. These broad reduction peaks could be attributed to the reduction of the mixed phase MnMoO 4 , which was present in these samples according to the XRD patterns. Similar TPR profiles were reported by Cadus et al. for Mn 2 O 3 -supported molybdenum materials [31]. The 100% MoO 3 sample showed a broad reduction signal between 550 and 700 • C, the temperature at which sublimation was observed and the experiment was interrupted. lattice. As a result, the Mn2O3 lattice is expected to generate certain crystalline defects (i.e., one Mn 3+ vacancy for each Mo 6+ ion exchanged) to offset the excess of positive charges. These cation vacancies can produce significant changes in the properties of Mn2O3 even with relatively low doping amounts. Thus, Mn 3+ cation vacancies generated upon Mo +6 doping have been previously reported to result in significant changes in the morphology and redox behavior of manganese oxide materials [27]. As the amount of Mo increases in the samples (8.7-31% MoO3), new broad reduction peaks appeared at temperatures higher than 600 °C. These broad reduction peaks could be attributed to the reduction of the mixed phase MnMoO4, which was present in these samples according to the XRD patterns. Similar TPR profiles were reported by Cadus et al. for Mn2O3-supported molybdenum materials [31]. The 100% MoO3 sample showed a broad reduction signal between 550 and 700 °C, the temperature at which sublimation was observed and the experiment was interrupted.

Redox Cycling Behavior
The redox behavior of the samples was studied by monitoring the losses and gains of weight after consecutive reduction-oxidation cycles, respectively ( Figure 5). Thus, a weight loss was indicative of reduction (O2 release, reduction, charge), whereas a weight gain was associated to oxidation (O2 uptake, oxidation, discharge). As shown in Figure 5, pure Mn2O3 showed the poorest redox cyclability among the materials studied herein, with very low re-oxidation rates (i.e., slow weight gains). Both the oxidation kinetics and the thermal cyclability improved significantly upon addition of Mo to Mn2O3. Thus, the Mn-Mo samples showed complete reduction at temperatures below 1000 °C. As shown in Table 2, the weight loss of the samples decreased with the Mo content. We calculated the theoretical weight change considering only the actual amount of manganese in the mixed oxide (determined by ICP-AES, Table 1) and included those values in Table 2. Interestingly, the deviation from the theoretical value increased with the amount of Mo. Thus, the 8.7% MoO3, 12% MoO3, and 31% MoO3 samples showed weight losses above their theoretical value, thereby suggesting that the MnMoO4 phase identified by XRD (Figure 2) possessed some redox activity, as suggested by the TPR profiles ( Figure 4).

Redox Cycling Behavior
The redox behavior of the samples was studied by monitoring the losses and gains of weight after consecutive reduction-oxidation cycles, respectively ( Figure 5). Thus, a weight loss was indicative of reduction (O 2 release, reduction, charge), whereas a weight gain was associated to oxidation (O 2 uptake, oxidation, discharge). As shown in Figure 5, pure Mn 2 O 3 showed the poorest redox cyclability among the materials studied herein, with very low re-oxidation rates (i.e., slow weight gains). Both the oxidation kinetics and the thermal cyclability improved significantly upon addition of Mo to Mn 2 O 3 . Thus, the Mn-Mo samples showed complete reduction at temperatures below 1000 • C. As shown in Table 2, the weight loss of the samples decreased with the Mo content. We calculated the theoretical weight change considering only the actual amount of manganese in the mixed oxide (determined by ICP-AES, Table 1) and included those values in Table 2. Interestingly, the deviation from the theoretical value increased with the amount of Mo. Thus, the 8.7% MoO 3 , 12% MoO 3 , and 31% MoO 3 samples showed weight losses above their theoretical value, thereby suggesting that the MnMoO 4 phase identified by XRD (Figure 2) possessed some redox activity, as suggested by the TPR profiles ( Figure 4).  The 0.6% MoO3 and 1.5% MoO3 samples showed the best long-term redox cyclability among the samples studied herein, with no noticeable loss of reactivity after 10 consecutive redox cycles. These samples showed square reduction and oxidation profiles, which are indicative of a stable redox behavior. The addition of Mo above 1.5% MoO3 did not enhance the oxidation rate, and a progressive loss of cyclability was observed as the amount of Mo increased in the samples. Thus, additional long-term cyclability tests were performed on the 0.6% MoO3 and 1.5% MoO3 samples. Figure 6 shows the weight losses of the reduction process for 45 consecutive redox cycles. After the first two cycles, both samples showed excellent cyclability, with no significant changes in the weight losses. The optimum long-term redox behavior of these samples could be ascribed to their coral-like morphology (Figure 3). This morphology allows oxygen diffusion through the solid, thereby facilitating the reduction and oxidation processes. This coral morphology has been previously associated with optimum redox behavior in Mn-containing materials [32]. Interestingly, these samples showed lower surface areas than the rest of oxides (Table  1), which reveals that this parameter does not determine the redox behavior of these materials. The fact that pure Mn2O3 (0% MoO3) also showed a well-formed coral-like morphology and a poor oxidation rates is a strong indicator that Mo necessarily plays a role in improving the oxidation behavior of the 0.6% MoO3 and 1.5% MoO3 samples. For these samples, the shift of the main diffraction peak of Mn2O3 to higher angles (Figure 2) suggests that Mo 6+ ions (73 pm) were able to replace Mn 3+ ions (72 pm) in the Mn2O3 lattice,  The 0.6% MoO 3 and 1.5% MoO 3 samples showed the best long-term redox cyclability among the samples studied herein, with no noticeable loss of reactivity after 10 consecutive redox cycles. These samples showed square reduction and oxidation profiles, which are indicative of a stable redox behavior. The addition of Mo above 1.5% MoO 3 did not enhance the oxidation rate, and a progressive loss of cyclability was observed as the amount of Mo increased in the samples. Thus, additional long-term cyclability tests were performed on the 0.6% MoO 3 and 1.5% MoO 3 samples. Figure 6 shows the weight losses of the reduction process for 45 consecutive redox cycles. After the first two cycles, both samples showed excellent cyclability, with no significant changes in the weight losses. The optimum long-term redox behavior of these samples could be ascribed to their corallike morphology (Figure 3). This morphology allows oxygen diffusion through the solid, thereby facilitating the reduction and oxidation processes. This coral morphology has been previously associated with optimum redox behavior in Mn-containing materials [32]. Interestingly, these samples showed lower surface areas than the rest of oxides (Table 1), which reveals that this parameter does not determine the redox behavior of these materials. The fact that pure Mn 2 O 3 (0% MoO 3 ) also showed a well-formed coral-like morphology and a poor oxidation rates is a strong indicator that Mo necessarily plays a role in improving the oxidation behavior of the 0.6% MoO 3 and 1.5% MoO 3 samples. For these samples, the shift of the main diffraction peak of Mn 2 O 3 to higher angles (Figure 2) suggests that Mo 6+ ions (73 pm) were able to replace Mn 3+ ions (72 pm) in the Mn 2 O 3 lattice, likely occupying six-fold octahedral coordination sites [25][26][27]. The poor oxidation behavior of pure Mn 2 O 3 has been previously ascribed to the significant structural mismatches existing between the reduced (Mn 3 O 4 , tetragonal or cubic spinels) and the oxidized (Mn 2 O 3 , cubic) phases [8]. Thus, the transition from tetragonal Mn 3 O 4 to cubic Mn 2 O 3 involves large structural changes (e.g., lattice shrinking) and atomic rearrangements, which ultimately results in a sluggish oxidation process. The addition of Fe to Mn 2 O 3 has been reported to improve the oxidation kinetics by promoting the formation of a cubic Mn 3 O 4 phase, thereby allowing a smoother transition (from cubic to cubic) from the reduced to the oxidized materials [33]. The addition of low amounts of Mo (0.6% MoO 3 and 1.5% MoO 3 ) could produce a similar effect in the reduced Mn 3 O 4 phase, facilitating the transition from the reduced to the oxidized phases. The TPR profiles of these samples, with a single reduction peak instead of a double-peak profile for pure Mn 2 O 3 (Figure 4), seem to support this hypothesis. As shown in Figure 5, the addition of Mo in amounts larger than 1.5% MoO 3 did not improve the oxidation behavior of Mn 3 O 4 , with the oxidation rate decreasing as the amount of Mo increased in the samples. The amount of monoclinic phase MnMoO 4 in these samples also increased with the Mo loading (Figure 2), which suggests that this phase was not responsible for improving the oxidation behavior of the solids. likely occupying six-fold octahedral coordination sites [25][26][27]. The poor oxidation behavior of pure Mn2O3 has been previously ascribed to the significant structural mismatches existing between the reduced (Mn3O4, tetragonal or cubic spinels) and the oxidized (Mn2O3, cubic) phases [8]. Thus, the transition from tetragonal Mn3O4 to cubic Mn2O3 involves large structural changes (e.g., lattice shrinking) and atomic rearrangements, which ultimately results in a sluggish oxidation process. The addition of Fe to Mn2O3 has been reported to improve the oxidation kinetics by promoting the formation of a cubic Mn3O4 phase, thereby allowing a smoother transition (from cubic to cubic) from the reduced to the oxidized materials [33]. The addition of low amounts of Mo (0.6% MoO3 and 1.5% MoO3) could produce a similar effect in the reduced Mn3O4 phase, facilitating the transition from the reduced to the oxidized phases. The TPR profiles of these samples, with a single reduction peak instead of a double-peak profile for pure Mn2O3 (Figure 4), seem to support this hypothesis. As shown in Figure 5, the addition of Mo in amounts larger than 1.5% MoO3 did not improve the oxidation behavior of Mn3O4, with the oxidation rate decreasing as the amount of Mo increased in the samples. The amount of monoclinic phase MnMoO4 in these samples also increased with the Mo loading (Figure 2), which suggests that this phase was not responsible for improving the oxidation behavior of the solids. Besides the weight variation (which is directly related with the storage capacity), the TES operational temperatures are also very important in determining the efficiency of thermal storage devices for CSP plants [34]. Thus, since reduction and oxidation take place at different temperatures (i.e., thermal hysteresis) for the Mn-Mo samples, heat is stored and released at different temperatures. This thermal hysteresis is one of the main factors affecting to the charge-discharge exergy efficiency, such that it is desirable that these temperatures are as similar as possible, which would result in higher exergy efficiencies [34]: where Tred and Tox are the temperatures at which reduction and oxidation processes take place, respectively, and Tc is the cold reservoir temperature (298 K). Table 3 summarizes the operational temperatures, the temperature hysteresis, and the charge-discharge exergy efficiency for the 0, 0.6, and 1.5% of MoO3 samples (for the redox cycles described in Besides the weight variation (which is directly related with the storage capacity), the TES operational temperatures are also very important in determining the efficiency of thermal storage devices for CSP plants [34]. Thus, since reduction and oxidation take place at different temperatures (i.e., thermal hysteresis) for the Mn-Mo samples, heat is stored and released at different temperatures. This thermal hysteresis is one of the main factors affecting to the charge-discharge exergy efficiency, such that it is desirable that these temperatures are as similar as possible, which would result in higher exergy efficiencies [34]: likely occupying six-fold octahedral coordination sites [25][26][27]. The poor oxidation behavior of pure Mn2O3 has been previously ascribed to the significant structural mismatches existing between the reduced (Mn3O4, tetragonal or cubic spinels) and the oxidized (Mn2O3, cubic) phases [8]. Thus, the transition from tetragonal Mn3O4 to cubic Mn2O3 involves large structural changes (e.g., lattice shrinking) and atomic rearrangements, which ultimately results in a sluggish oxidation process. The addition of Fe to Mn2O3 has been reported to improve the oxidation kinetics by promoting the formation of a cubic Mn3O4 phase, thereby allowing a smoother transition (from cubic to cubic) from the reduced to the oxidized materials [33]. The addition of low amounts of Mo (0.6% MoO3 and 1.5% MoO3) could produce a similar effect in the reduced Mn3O4 phase, facilitating the transition from the reduced to the oxidized phases. The TPR profiles of these samples, with a single reduction peak instead of a double-peak profile for pure Mn2O3 (Figure 4), seem to support this hypothesis. As shown in Figure 5, the addition of Mo in amounts larger than 1.5% MoO3 did not improve the oxidation behavior of Mn3O4, with the oxidation rate decreasing as the amount of Mo increased in the samples. The amount of monoclinic phase MnMoO4 in these samples also increased with the Mo loading (Figure 2), which suggests that this phase was not responsible for improving the oxidation behavior of the solids. Besides the weight variation (which is directly related with the storage capacity), the TES operational temperatures are also very important in determining the efficiency of thermal storage devices for CSP plants [34]. Thus, since reduction and oxidation take place at different temperatures (i.e., thermal hysteresis) for the Mn-Mo samples, heat is stored and released at different temperatures. This thermal hysteresis is one of the main factors affecting to the charge-discharge exergy efficiency, such that it is desirable that these temperatures are as similar as possible, which would result in higher exergy efficiencies [34]: where Tred and Tox are the temperatures at which reduction and oxidation processes take place, respectively, and Tc is the cold reservoir temperature (298 K). Table 3 summarizes the operational temperatures, the temperature hysteresis, and the charge-discharge exergy efficiency for the 0, 0.6, and 1.5% of MoO3 samples (for the redox cycles described in where T red and T ox are the temperatures at which reduction and oxidation processes take place, respectively, and T c is the cold reservoir temperature (298 K). Table 3 summarizes the operational temperatures, the temperature hysteresis, and the charge-discharge exergy efficiency for the 0, 0.6, and 1.5% of MoO 3 samples (for the redox cycles described in Figure 6). First of all, it is important to remark that the experimental temperatures are slightly higher than the one predicted by the thermodynamic study (Table 1) as result of the critical influence of diffusion and chemical kinetics in the process, which is not considered in the equilibrium calculations [35].

Sample
Cycle Figure 6. Long-term cyclability tests for the 0.6% MoO3 and 1.5% Besides the weight variation (which is directly related TES operational temperatures are also very important in thermal storage devices for CSP plants [34]. Thus, since red at different temperatures (i.e., thermal hysteresis) for the M and released at different temperatures. This thermal hyste affecting to the charge-discharge exergy efficiency, such th peratures are as similar as possible, which would result in where Tred and Tox are the temperatures at which reduction place, respectively, and Tc is the cold reservoir temperatur the operational temperatures, the temperature hysteresis, ergy efficiency for the 0, 0. Regarding the operational temperatures, the reduction temperature (i.e., the temperature at which maximum oxygen release rates are reached) increased with the cycle number, while the oxidation temperature (i.e., the temperature at which maximum oxygen uptake rates are reached) decreased upon thermal cycling for both samples, although this was more evident for the 1.5% MoO 3 sample. This fact implies that the higher the number of cycles the higher the temperature hysteresis, although this parameter seems to stabilize for both materials after the 30th cycle. Additionally, as the change in the operational temperatures is more evident for the 1.5% MoO 3 sample, it presents a higher decrease in the hysteresis temperature as compared to Mn 2 O 3 . Most importantly, the temperature hysteresis was significantly reduced (by 80-30 and 95-55 • C) for the 0.6% MoO 3 and 1.5% MoO 3 samples as compared to Mn 2 O 3 , respectively, thereby enhancing the charge-discharge efficiency of the cycle (by 3.6-2.0 and 4.1-3.0%, respectively).
Additionally, the operational temperatures must be compatible with the working conditions of other parts of the CSP plants such as the solar concentrator (which must concentrate the solar radiation to achieve the required temperatures) and the receiver (where the reactions take place). If we assumed that the solar reactor is a perfectly insulated blackbody receiver, the absorption efficiency is given by Equation (5) [36]: where σ is the Stefan-Boltzmann constant (5.67·10 −11 kW/m 2 ·K 4 ); T reactor is the solar reactor temperature (T red ); I is the direct-normal solar irradiation (taken as 1 kW·m 2 ); C is the concentration ratio which is defined as the solar radiative flux normalized to 1 kW/m 2 and is usually expressed in "suns" units. Table 4 summarizes the absorption efficiency for 0.6 and 1.5% of MoO 3 samples for the different cycles as a function of the concentration ratio of the solar concentrator. Although the temperature hysteresis is different for both materials, the absorption efficiency was almost similar and higher than 90% when the concentration capacity is higher than 2000 suns, which indicates that the operational temperatures of these materials are compatible with current solar technology, and that these materials are promising for large-scale thermochemical heat storage applications. Table 4. Solar receiver absorption efficiencies for the 0.6% and 1.5% MoO 3 samples.

Sample Cycle
Besides the weight variation (which is directly related with the storage capacity TES operational temperatures are also very important in determining the efficien thermal storage devices for CSP plants [34]. Thus, since reduction and oxidation take at different temperatures (i.e., thermal hysteresis) for the Mn-Mo samples, heat is s and released at different temperatures. This thermal hysteresis is one of the main f affecting to the charge-discharge exergy efficiency, such that it is desirable that thes peratures are as similar as possible, which would result in higher exergy efficiencie where Tred and Tox are the temperatures at which reduction and oxidation processe place, respectively, and Tc is the cold reservoir temperature (298 K). Table 3 summ the operational temperatures, the temperature hysteresis, and the charge-dischar ergy efficiency for the 0, 0.

Conclusions
The addition of small amounts of Mo to Mn 2 O 3 was found to improve the TCS behavior (e.g., oxidation kinetics and redox cyclability) of this oxide without significantly increasing its reduction temperature. XRD suggested that most of the Mo added ended up forming a separate monoclinic phase MoMnO 4 , while only a limited fraction of the Mo ions were inserted in the cubic Mn 2 O 3 structure. Despite the limited extent of this substitution, the presence of Mo 6+ ions in the structure changed dramatically the redox behavior of Mn 2 O 3 , as revealed by TPR. SEM observations revealed a coral-like morphology, particularly for samples with low Mo loadings. Mo-enriched compact particles and a partial collapse of the coral-like structure were observed for Mo loadings above 8.7% MoO 3 . The 0.6 and 1.5% MoO 3 samples showed excellent long-term (45 cycles) TCS behavior, with fast oxidation processes and outstanding cyclability. These samples also showed lower temperature hysteresis as compared to Mn 2 O 3 , which resulted in higher exergy charge-discharge efficiencies. The operational temperatures of the Mn-Mo materials prepared herein were compatible with those typical of solar receivers, potentially reaching absorption efficiencies higher than 90% at concentration capacity values typical of CSP plants.