pubs.acs.org/EF Article Operando Synchrotron X‑ray Diffraction in Calcium Batteries: Insights into the Redox Activity of 1D Ca3CoMO6 (M = Co and Mn) A.P. Black, D. Monti, C. Frontera, D. S. Tchitchekova, R. G. Houdeville, F. Fauth, and M.R. Palacin* Cite This: Energy Fuels 2021, 35, 10898−10907 Read Online ACCESS Metrics & More Article Recommendations * sı Supporting Information ABSTRACT: 1D Ca3Co2−zMzO6 (M = Co z = 0, M = Mn z = 1, and M = Fe z = 0.4) were prepared and tested electrochemically. While the iron-containing phase was not found to be active, the iron- and manganese-containing phases were found to be potentially interesting as positive electrode materials for calcium metal-based high-energy battery technologies and were inves- tigated by operando synchrotron X-ray diffraction. Results indicate that electrochemically driven calcium deintercalation from the crystal structure (ca. 0.7 mol per formula unit) takes place upon oxidation in both cases. The oxidized phases have incommensurate modulated crystal structures with the space group R 3m(00γ)0s and a = 9.127(1) Å, c1 = 2.4226(3) Å and c2 = 4.1857(3) Å, and γ = 0.579 (M = Co) and a = 9.217(1) Å, c1 = 4.9076(4) Å and c2 = 4.3387(4) Å, and γ = 1.139 (M = Mn), which exhibit differences due to the presence of manganese and Mn/Co ordering. The degree of calcium re-intercalation within the structure was found to be extremely limited, if any. Complementary experiments carried out in lithium cells did not show any reversibility either, thus pointing at intrinsic structural/migration constraints in the oxidized phase rather than slow kinetics of high desolvation energies associated with divalent ion charge carriers. ■ INTRODUCTION The quest for high-energy-density battery technologies has such tools needs to be considered as, for instance, X-ray photoelectron spectroscopy or other surface-sensitive techni- revived research on metal anode-based concepts, not only for ques are not trustworthy to assess bulk behavior. Finally, the case of lithium, for which dendrite growth has to date been sample modification during handling/measurement may also a technical showstopper, but also for other more abundant lead to biased results. In this sense, operando diffraction is the electropositive metals such as sodium, potassium, magnesium, most appealing. Besides being nondestructive, it enables calcium, or aluminum. Among these, the first two represent a capturing metastable intermediates and ascertaining the more straightforward option, as the know-how gained in the influence of operation conditions on the redox mechanism. study of lithium-based chemistries can be most valuable. In Nonetheless, the cell geometry must be compatible with the contrast, multivalent concepts come with intrinsic significant experimental setup while at the same time being electrochemi- differences, which make their research more complex in nature with previous lithium-based know-how being less useful. Yet, cally efficient, which can be problematic in the case of new theoretical energy density coupled to an expected reduced cost technologies due to the lack of standard cell components. The makes multivalent chemistries appealing.1−3 use of synchrotron radiation improves data collection statistics Despite the standard reduction potential for Ca2+/Ca being and angular resolution. To the best of our knowledge, −2.87 versus normal hydrogen electrode, research on calcium synchrotron operando diffraction in calcium cells has only batteries has been relatively limited until 2015.4 Moreover, in been reported for graphite electrodes at low potentials using the absence of standards, even the development of reliable activated carbon (AC) as the counter electrode (CE) and experimental protocols has required a significant effort.5 These Ca(TFSI)2 in tetraglyme as the electrolyte and was helpful in are especially critical in the quest for new positive electrode assessing staging.10 materials, as the operation of calcium as the negative electrode can be plagued with a number of issues which can affect the results of positive electrode testing. For instance, electrolytes Received: May 1, 2021 enabling calcium plating and striping are limited and not Revised: June 4, 2021 optimized.6−9 Hence, electrochemical response per se is not Published: June 16, 2021 enough to assess the redox activity and the use of additional advanced materials characterization techniques is compulsory. In addition to that, the length scale that can be probed with © 2021 American Chemical Society https://doi.org/10.1021/acs.energyfuels.1c01343 10898 Energy Fuels 2021, 35, 10898−10907 Energy & Fuels pubs.acs.org/EF Article Looking for positive electrode materials for calcium-based With the aim of fully assessing the reaction mechanism of batteries operating at high potential, intercalation compounds Ca3Co2O6 at room temperature and getting further insights appear as an appealing option. Indeed, topotactic insertion of a into the reversibility and/or formation of alternative phases, we guest species into a host is possible for several metal ions. The performed operando synchrotron diffraction studies using a host compounds should exhibit an open framework of modified coin-cell setup24 and an alternative electrochemical interconnected sites through which the intercalated ion can configuration enabling operation at room temperature while diffuse and an electronic band structure able to reversibly avoiding the use of calcium metal CEs.25 The influence of accept/donate electrons. When looking for compounds able to crystal chemistry within the activity and reversibility of the operate as positive electrode materials, hosts containing system was further explored by broadening the study to other transition metals are the most suitable, which is the reason isostructural phases containing alternative transition metals why these are used in the ubiquitous Li-ion batteries. Within which could act as redox centers. Indeed, these exhibit this scenario, we turned our attention toward Ca3Co2O6. This somewhat different cell parameters and hence different compound is an end member of the general A1+y(Ay′B1−y)O3 intercolumn spacings, which may have an impact on the (0 ≤ y ≤ 1/2) family of compounds11 related to the 2H (y = barriers for the migration of calcium ions. The only suitable 0) hexagonal perovskite with y = 1/2, A = Ca, and A′ = B = phases already reported contain iron or manganese and were Co. Its crystal structure exhibits columns of face-sharing A′O6 previously investigated for the interest in their magnetic trigonal prisms and BO6 octahedra (in a ratio that is given by properties. Among these, the maximum amount of iron in the y) running along the c direction.12−17 This framework is similar structure is achieved for Ca3Co1.6Fe0.4O6, in which Fe(III) to that of TiS3 where the columns consist only of trigonal occupies trigonal prismatic sites26 and has a similar unit-cell prisms, and for which lithium intercalation ability has been volume (741.6 Å3) compared to Ca3Co2O6 (740.9 Å3). For the known for long, albeit at too low potential to present a case of manganese, Ca3CoMnO6 has been reported,27 in which practical interest.18 In the case of Ca3Co2O6, high redox Mn(IV) occupies the octahedral sites and exhibits a larger cell volume (764.1 Å3), resulting in a larger intercolumn space operation potential can be expected related to the oxidation of within the structure. ■ Co(III) to Co(IV) concomitant to electrochemical calcium de- insertion from the structure. Moreover, a high theoretical EXPERIMENTAL SECTION capacity (160 mA h/g) can be calculated considering the extraction of 1 mol of Ca2+ per formula unit to yield Synthesis. Ca3Co2O6 was prepared by the Pechini method.28,29 30 Ca2Co2O6. This hypothetical fully oxidized phase might mL of a 1 M solution of Ca(NO3)2 (99% tetrahydrate, Sigma-Aldrich) and 20 mL of a 1 M solution of Co(NO3)2 (98%, hexahydrate, Sigma- correspond to CaCoO3, the 2H polytype of the hexagonal Aldrich) in water were mixed with 10.5 g of citric acid (C6H8O7, 99%, perovskites, which has only been reported with larger alkali Sigma-Aldrich) and put together in a glass vial to which 6.25 g of earth ions (Sr, Ba) to date.19 ethylene glycol (C2H6O2, 99%, Sigma-Aldrich) was added (corre- In spite of such attractive performance expectations,20 sponding to 3:2:5:10 molar ratios for Ca/Co/C6H8O7/C2H6O2). The experiments carried out under the conditions where calcium solution was subsequently heated at 80 °C for 4 h and thereafter metal plating/stripping is enabled at the CE at 100 °C indicate stirred overnight at room temperature to promote water evaporation, that even if the electrochemical extraction of calcium takes which induced enhanced viscosity of the mixture and the evolution of nitrous oxide gas. Once the gel formed dried, it was placed in an place, formation of the 2H polytype is not achieved and alumina crucible and heated in air first at 600 °C for 6 h and then at reversibility is very limited, if any.21 Density functional theory 950 °C for 20 h and subsequently cooled to room temperature. (DFT) calculations for Ca2Co2O6 yield 0.9 eV energy barriers The synthesis of Ca3Co2‑xFexO6 was attempted by both the Pechini for the migration of calcium ions along two of the three method and a solid-state reaction. In the first case, 30 mL of a 1 M simplest pathways considered.22 Upon calcium de-intercala- solution of Ca(NO3)2 (99% tetrahydrate), 10 mL of a 1 M solution of tion, trigonal prisms are expected to evolve toward distorted Co(NO3)2 (98%, hexahydrate), and 10 mL of a solution of Fe(NO3)3 octahedra, which would hinder calcium re-intercalation due to (98%, hexahydrate, Sigma-Aldrich) in water were mixed with 10.5 g of a crystal field energy splitting. Moreover, calculated activation citric acid (C6H8O7, 99%, Sigma-Aldrich) and put together in a glass vial to which 6.25 g of ethylene glycol (C2H6O2, 99%, Sigma-Aldrich) barriers higher than 2 eV are expected for a hypothetical fully was added (corresponding to 3:1:1:5:10 molar ratios for Ca/Co/Fe/ oxidized 2H Ca2Co2O6 exhibiting the hexagonal perovskite C6H8O7/C2H6O2). Thermal treatments analogous to those enabling structure.22 Alternative DFT studies do also point at Ca3Co2O6 the synthesis of Ca3Co2O6 resulted in a mixture of Ca2Fe2O5, becoming gradually less stable upon calcium extraction upon Ca3Co2O6, and/or Ca3Co2−xFexO6, and CaO despite a previous oxidation despite its chemical decomposition being likely report suggesting the indexation of an analogous diffraction pattern kinetically limited.23 Full oxidation to Ca2Co2O6 has not been with a single triclinic unit cell with no structural model provided.30 achieved experimentally at 100 °C, but the extraction of ca. 0.7 For the solid-state route, a stoichiometric mixture of CaCO3, Fe2O3, and Co3O4 was successively heated at 850, 950, and 1000 °C for 24 h mol of calcium ions from the crystal lattice has been proved, at in each step, with intermediate regrinding, and single-phase a potential consistent with the expected values derived from Ca3FexCo2−xO6 was achieved for x ≤ 0.4. DFT calculations.22,23 Upon oxidation, a Ca2.3Co2O6 phase Ca3CoMnO6 was obtained both using a solid-state route27 (from with an incommensurate modulated structure was formed, CaCO3, Co3O4, and MnO2 after successive treatments in air at 850 °C which does neither exhibit the composition nor the crystal for 20 h and 1200 °C for 24 h with intermediate regrinding) and the structure for which migration barriers were calculated. Partial Pechini method31 (from a 3:1:1:5:10 molar ratio of Ca/Co/Mn/ oxidation was also observed at room temperature using sodium C6H8O7/C2H6O2 using Ca(NO3)2, Co(NO3)2, Mn(NO3)2, citric acid, and ethylene glycol with thermal treatments under synthetic air CE and 1 M NaPF6 in ethylene carbonate (EC)/DMC as the at 600 °C for 6 h and 1000 °C for 20 h). electrolyte.21 In contrast, low capacity and no changes in the Characterization. Scanning electron microscopy micrographs diffraction pattern of Ca3Co2O6 were reported for cells with were acquired using an FEI Quanta 200 FEG microscope under high calcium or tin metal CEs and Ca(TFSI)2-based electrolytes vacuum operating at 20 kV. X-ray powder diffraction patterns were upon oxidation to 4 V.23 acquired in a Bruker D8 Advance A25 diffractometer in a Debye− 10899 https://doi.org/10.1021/acs.energyfuels.1c01343 Energy Fuels 2021, 35, 10898−10907 Energy & Fuels pubs.acs.org/EF Article Scherrer configuration equipped with a Mo Kα1 radiation source (λ = to conduct operando experiments on a Bruker D8 Advance A25 0.7093 Å) and a Johansson monochromator. Synchrotron X-ray diffractometer in a Bragg−Brentano configuration equipped with a Cu diffraction (SXRD) patterns were collected on MSPD beamline32 Kα1,2 radiation source (λ = 1.5406 and 1.54443 Å). The XRD patterns (ALBA synchrotron, Cerdanyola del Vallès, Spain) using the position- were collected in the 10 ≤ 2θ ≤ 50° range in 0.02° steps and an sensitive detector MYTHEN and λ = 0.6200 Å. In both cases, the integration time of 2 h. samples, either as-prepared or recovered after running electrochemical experiments and dismantling cells inside an Ar-filled glovebox, were embedded in a 0.5 mm diameter borosilicate glass capillary and spun during data collection to ensure proper powder averaging. ■ RESULTS AND DISCUSSION Synthesis and Electrochemical Characterization. Pure Electrochemical Experiments. Tape-casted electrodes were Ca3Co2O6, Ca2CoMnO6, and Ca3Co2.6Fe0.4O6 were obtained prepared from the mixture of as-synthesized powder samples with using the conditions reported in the Experimental Section. carbon black (Super P, Timcal, Switzerland) to enhance the electronic Figure 1 depicts the corresponding XRD patterns and conductivity and polyvinylidene fluoride (Arkema) as the binder in a weight ratio of 80:10:10. The mixture was further dispersed in N- methyl-2-pyrrolidone (Aldrich, ≥99.9%), ground in a ball mill, casted on an 18 μm thick aluminum foil (Goodfellow, 99%) with a blade gap of 300 μm, and vacuum-dried at 80 °C for 24 h. Punched ca. 1 cm2 electrodes were tested in three-electrode Swagelok cells in different configurations (loading ca. 2 mg/cm2). Galvanostatic cycling with potential limitation (GCPL) tests were carried out at C/100 (1 mol of electrons exchanged in 100 h) at room temperature with sodium counter and reference electrodes and 1 M NaPF6 (Strem Chemicals, 99%) in EC/PC/DMC (0.45:0.45:0.1) as the electrolyte (all solvents, Aldrich, anhydrous, 99.0%).33 The interest of performing such experiments was to assess whether the electrochemical extraction of calcium was feasible using a more standard cell configuration and to provide inputs for comparison with alternative setups. Ca metal was also used as counter and pseudoreference electrodes, with 0.3 M Ca(BF4)2 dissolved in a 1:1 mixture of EC (Aldrich, anhydrous, 99.0%) and propylene carbonate (PC, Aldrich anhydrous, 99.0%) as the electrolyte, prepared from Ca(BF4)2·H2O salt (Alfa Aesar). Prior to electrolyte preparation, Ca(BF4)2 was dried under vacuum in a Büchi oven at 80 °C for 2 days. Once prepared, the electrolyte was further dried inside an Ar-filled glovebox with less than 1 ppm of O2 and H2O by heating at 90 °C overnight to achieve a water content lower than 80 ppm, as determined by Karl Fischer analysis. Cells were tested in the galvanostatic mode at C/200 (defined here as the reaction of 1 mol of Ca2+ in 200 h and equivalent to 1 mol of electrons exchanged in 100 h) on a Bio-Logic VMP3 potentiostat Figure 1. Representative scanning electron micrographs for samples either at room temperature, 100, or 115 °C. The stability of the active resulting from Pechini synthesis ((a) corresponding to materials under the experimental conditions was ascertained by Ca3Co1.6Fe0.4O6) and the solid-state reaction ((b), Ca2CoMnO6) ensuring that their XRD patterns were unmodified after keeping cells and (c) XRD patterns (λ Mo ) of as-prepared compounds at open-circuit potential for 3 h at the temperature of testing. Ca3Co1.6Fe0.4O6 SS, Ca3Co2O6 P, Ca3CoMnO6 SS, and Ca3CoMnO6 Powder electrodes were prepared by simply mixing active materials P (SS denotes solid-state synthesis and P stands for the Pechini with carbon black (Super P, Timcal, Switzerland) in a weight ratio of method). 75:25, respectively. Operando electrochemical tests were performed in two types of in situ cells: drilled 2032 coin cells with a 4−5 mm glass window24 and Leriche-type cells.34 An aluminum foil (3 μm, representative scanning electron micrographs. As expected, Goodfellow) was placed on the positive electrode side to protect the Pechini method delivered more homogeneous and smaller the window and ensure good electric contact. One glass fiber filter (<5 μm) particles than solid-state reactions. disk (Whatman, GE Healthcare, 420 μm thick) was used as a Rietveld refinements indicate that the prepared compounds separator. Electrochemical experiments were carried out in the are free from impurities (except for Ca3CoMnO6 prepared by galvanostatic mode at room temperature and C/50 (defined as the the solid-state reaction, in which some low-intensity impurity reaction of 1 mol of Ca2+ in 50 h) using the same electrolyte as described above. AC cloth (Kynol, ACC-509220) was used as the CE. peaks are clearly visible in the pattern corresponding to 3% The advantages of using AC include high reversibility, operation CaO as deduced from the refinement). Since previous neutron window matching the stability of most electrolytes, versatility in terms powder diffraction studies have shown that both Mn and Fe of ions that can be adsorbed, and operation temperature. Yet, its replace Co in the octahedral position,27,28 we have presumed relatively low specific capacity (35 mA h/g at 25 mA/g, between 0 this ordering of transition metals to take place. All patterns and −1.4 V vs Ag pseudoreference electrode)25 requires careful cell have been successfully refined using the R3̅c space group and balancing to ensure largely oversizing the capacity of the active the structure involving the alternation of face-sharing MO6 material tested. Therefore, coin cells were loaded with 2 mg of active octahedra CoO6 trigonal prisms (M = Fe and Mn). material, 0.3 mL of electrolyte, and 3 AC fabric discs of 10 mm in Occupancies at octahedral and prismatic sites, constrained by diameter, reaching a total mass of 37 mg at the CE side. Leriche-type the overall stoichiometry and the full occupancy of both sites, cells were loaded with 6 mg of active material and 3 AC fabric discs of 14 mm diameter and a total mass of 66 mg. In the case of the drilled are consistent with the model used. Refined cell parameters coin cells, operando measurements were conducted at ALBA also agree with those reported in the literature: a = 9.0771(2) synchrotron on the powder diffraction station of the MSPD beamline Å and c = 10.3804(2) Å (Ca3Co2O6); a = 9.1296(2) Å and c = with patterns being collected in the ≈2.3 ≤ 2θ ≤ 48° range in 0.006° 10.5810(2) Å (Ca3CoMnO6, solid-state reaction); a = steps and an integration time of 172 s. The Leriche-type cell was used 9.1295(2) Å and c = 10.5822(2) Å (Ca3CoMnO6, Pechini 10900 https://doi.org/10.1021/acs.energyfuels.1c01343 Energy Fuels 2021, 35, 10898−10907 Energy & Fuels pubs.acs.org/EF Article method); and a = 9.0770(2) Å and c = 10.3803(2) Å capacity of 267 mA h/g being achieved at 4.4 V. This behavior (Ca3Co1.6Fe0.4O6). is similar to the one of Ca3Co2O6, with most likely some The feasibility of electrochemical oxidation on iron- and electrochemical capacity being related to electrolyte decom- manganese-containing compounds was tested at room temper- position at high potential. Analogous experiments attempting ature in cells using sodium counter and reference electrodes to oxidize Ca3CoMnO6 were carried out in calcium cells using and sodium-based electrolytes, in conditions which had AC as the CE and a silver wire as the pseudoreference previously enabled the oxidation of Ca3Co2O6 concomitant electrode. Figure 2B depicts the characteristic potential versus to Ca2+ extraction.21 Figure 2A depicts the potential versus capacity profile during a single oxidation/reduction galvano- capacity profile of Ca3CoMnO6 composite electrodes upon static cycle using 0.3 M Ca(BF4)2 in EC/PC as the electrolyte oxidation in sodium cells. The potential increases rapidly to 3.9 (54 ppm H2O) at the C/50 rate. The oxidation evolves V versus Na+/Na and then a sloping region is observed, with a through a pseudoplateau centered around 1.5 V versus AC with an associated capacity of 100 mA h/g. Further oxidation involves a linear increase in potential up to 2.1 V (160 mA h/ g). Upon reduction, the potential decreases rapidly down to −1.2 V versus AC and a sloping region is further observed down to −2 V versus AC (corresponding to a capacity of 120 mA h/g). Figure 2C depicts the XRD patterns of the (a−f) samples shown in Figure 2A oxidized to different extents in sodium cells and an additional pattern (g) oxidized in a calcium cell (200 mA h/g) at room temperature in 0.3 M Ca(BF4)2 in EC/PC using Ca metal as counter and reference electrodes. Upon oxidation, patterns gradually evolve with peaks at Q = 1.37, 2.23, and 2.37 Å−1 of the pristine phase decreasing in intensity with their position remaining unchanged and new (strong) peaks appearing at 2.95 and 1.36 Å−1, together with some tiny peaks in the region of Q ∼ 2 Å−1. Thus, regardless of the CE and electrolyte used, the changes in the diffraction pattern observed upon oxidation are the same and hence attributed to electrochemical calcium extraction from the crystal structure to form Ca3−xMnCoO6, similar to what had been observed for Ca3Co2O6.21 Attempts to oxidize Ca3Co1.6Fe0.4O6 were unsuccessful, as no changes were observed in the XRD patterns despite a plateau being observed at a high potential (3.9 V vs Ca2+/Ca) with a capacity close to 200 mA h/g, which was thus attributed to electrolyte decomposition. In this case, one must conclude that any electrochemical activity for this compound, if existing, takes place beyond the electrochemical stability window of the electrolyte used. Operando Diffraction Studies. Operando synchrotron diffraction experiments involving materials for calcium batteries are challenging. Beyond the general requirement of compatibility of all the components that are integrated in the in situ cells in terms of corrosion stability, the sluggish diffusion rates of the calcium ion into the crystal host material impose a limitation in terms of C-rate and hence unavoidably long experiments. Finally, the lack of electrolytes enabling calcium plating/stripping at room temperature and a fast C-rate which are also stable at high potentials (>4 V vs Ca2+/Ca) constitutes a showstopper for these materials. In the present study, and in order to enhance the efficiency of the full experiment, a setup enabling the alternative sequential measurement of four cells was used (see Figure 3) inspired in ref 24. Operando experiments aiming at getting further insights into Figure 2. Potential vs capacity profiles for Ca2CoMnO6 composite the redox activity of 1D Ca3CoMO6 (M = Co and Mn) were electrodes oxidized in cells at room temperature (a) using sodium carried out using 0.3 M Ca(BF4)2 in EC/PC as the electrolyte reference and counter electrodes and (b) Ca(BF4)2 in the EC/PC (54 ppm H2O) and consisted of a single oxidation/reduction electrolyte at RT at C/50. (c) Evolution of the ex situ XRD patterns (λMo) of Ca2CoMnO6 when oxidized in electrochemical cells at room cycle at C/50 within an electrochemical potential window temperature vs sodium CEs (1 M NaPF6 in EC/PC/DMC as the between 2.5 and −2.5 V versus AC. Since the main objective electrolyte), color codes are related to the top figure for patterns (a− was to assess the reversibility of the process, if any, the switch f). Pattern g corresponds to an electrode fully oxidized in a calcium between oxidation and reduction steps was made manually cell. once there was clear evidence of significant calcium extraction. 10901 https://doi.org/10.1021/acs.energyfuels.1c01343 Energy Fuels 2021, 35, 10898−10907 Energy & Fuels pubs.acs.org/EF Article Figure 3. Sample holder with four coin cells to be simultaneously monitored operando, single coin-cell image, and an exploded view of a coin cell. Since C/50 is a rather slow rate, in order to optimize the diffraction patterns taken ex situ after electrochemical experi- beamtime, some cells were cycled ex situ and placed under the ments were carried out at 100 °C and using Ca metal as the beam to be monitored only when they reached a certain state CE and indicate that the electrochemical extraction of calcium of charge. This also enabled us to assess the reproducibility of from the crystal structure is irreversible in the experimental the redox behavior both under and outside the beam. conditions used and the capacity recorded is only related to In the case of Ca3Co2O6, different cells were monitored side reactions involving the electrolyte. operando starting at different states of charge. Cell A was Figure 5a displays the characteristic potential versus capacity launched to cover the initial oxidation up to a specific capacity profiles for Ca3CoMnO6 electrodes during galvanostatic testing of 52 mA h/g. Cell B was initially oxidized out of the beam at C/50 in the course of the operando experiment. The red (160 mA h/g, equivalent to 1 mol of Ca2+ extracted if this was curve corresponds to the oxidation step, which evolves through the only redox process taking place, see the upper abscissa axis a pseudoplateau around 1.3 V versus AC [equivalent to 4.5 V in Figure 4a) and placed under the beam for further oxidation vs Ca2+/Ca (see the right axis in Figure 5a)], up to a capacity up to 213 mA h/g, followed by a reduction step. The of about 133 mA h/g that would correspond to a nominal electrochemical curve evolves through a pseudoplateau extraction of 0.82 mol of Ca2+ from the crystal structure (top centered at about 1.5 V versus AC upon oxidation, and a abscissa axis in Figure 5a). The reduction step is denoted in more sloping trend upon reduction, centered at ca. −1.2 V green for the section monitored under the beam (operando) to versus AC. Selected in situ SXRD patterns for those cells a capacity of 112 mA h/g and in blue beyond that value, with a displayed in Figure 4b−f are in full agreement with the results final diffraction pattern taken after extended reduction to a from previous experiments using the same electrolyte but capacity of 325 mA h/g (black dot). Oxidation evolves through carried out at 100 °C and using calcium metal CEs.21 The a pseudoplateau centered at about 1.3 V versus AC. The (300) and (100) Bragg reflections, at Q = 2.396 and 1.383 Å−1, reduction process proceeds through a sloping region down to respectively, progressively decrease in intensity, whereas new −3.8 V versus AC and ca. 260 mA h/g, after which noisy peaks grow at Q = 2.386 and 1.377 Å−1. The latter are ascribed behavior is observed, probably due to bubbling related to to Ca3−xCo2O6 with x ∼ 0.7, which exhibits an incommensu- electrolyte decomposition.35 rate modulated crystal structure [space group R 3m(00γ)0s, a The SXRD patterns collected throughout the experiment are = 9.127(1) Å, c1 = 2.4226(3) Å and c2 = 4.1857(3) Å, and γ = shown in Figure 5b. Zoom-in images (c−f) highlight the 0.579(2)]. The Ca content in this phase is straightforwardly regions where the evolution of the phase composition is clearly related to the propagation vector, determined by the position observed, with red patterns corresponding to oxidation and of the tiny peak appearing at Q ∼ 2 Å−1 among others.12,21 The green patterns to reduction. Upon oxidation of Ca3CoMnO6, a patterns depicted in Figure 4b−f correspond to pristine progressive decrease in the intensity of the peaks at Q = 1.365 Ca3Co2O6, intermediate, and final stage of oxidation as well as and 2.363 Å−1 occurs at the expense of new peaks appearing at the intermediate and final stage of reduction. From the Q = 1.375 and 2.382 Å−1 [zoom-in images (d) and (e)] and evolution of patterns represented in Figure 4e, a progressive also at Q = 2.03, 2.09, 3.00, and 3.30 Å−1, which gradually grow intensity increase for the (100)′ reflection at Q = 1.377 Å−1 as upon oxidation. These are consistent with the formation of a a function of oxidation time can be inferred. The patterns new Ca3−xCoMnO6 phase upon oxidation, in agreement with observed at the end of oxidation are consistent with the the results of ex situ experiments (Figure 2c). electrodes containing a mixture of both oxidized Ca3−xCo2O6 The patterns corresponding to Ca3−xCoMnO6 bear great and pristine Ca3Co2O6 phases. No significant evolution of the similarity with those corresponding to Ca3−xCo2O6 (x ∼ 0.7). peaks, neither in position nor in intensity, is seen when the Peaks (100) and (300) progressively decrease, while reflections polarity of the cell is reversed. Indeed, Figure 4f depicts the attributed to Ca3−xCoMnO6 appear at slightly lower Q-values patterns taken upon reduction and no changes are evidenced in (Figure 5C,D) than for Ca3−xCo2O6. In addition to those, the region of (300)′ (Q = 2.386 Å−1) throughout the process other low-intensity peaks appear, which are not consistent with (the last pattern corresponds to a capacity of 325 mA h/g). the incommensurate structure of Ca3−xCo2O6. This does not Such results are fully consistent with those derived from come as a surprise because despite pristine compounds being 10902 https://doi.org/10.1021/acs.energyfuels.1c01343 Energy Fuels 2021, 35, 10898−10907 Energy & Fuels pubs.acs.org/EF Article Figure 4. Potential vs capacity profiles from GCPL experiments on Ca3Co2O6//AC cells using Ca(BF4)2 in EC/PC as the electrolyte at Figure 5. Potential vs capacity profiles from GCPL experiments on RT and C/50 rate (a), corresponding SXRD patterns (b) and zoom- Ca3CoMnO6//AC cells using Ca(BF4)2 in EC/PC as the electrolyte in images (c,d), selected patterns correspond to the first and last at RT and C/50 (a), corresponding SXRD patterns (b) and zoom-in pattern of cell (A) in black and light red, last oxidation and last images (c,d), selected patterns correspond to pristine and last pattern reduction of cell (B) in red and dark green, and last ex situ pattern cell in oxidation in black and red, last reduction in dark green, and last ex (B) in blue. (e,f) Zoom-in images of reflections 100 and 300 of in situ situ pattern of cell (C) in blue. (e,f) Zoom-in images of reflections SXRD patterns of cells (A,B), respectively. 100 and 300 of in situ SXRD patterns, respectively. isostructural, the incommensurate structure of Ca3−xCo2O6, (when varying the Ca content) from octahedral to prismatic, with only one Co crystallographic position,12,13 is not but it is unlikely that Mn would occupy a prismatic position. compatible with the Mn/Co ordering present in According to neutron powder diffraction (and our own Ca3−xCoMnO6. Indeed, in the incommensurate structure of refinements), Mn−O bond distances are close to 1.905 Å28 Ca3−xCo2O6, any transition metal position is suitable to switch and consistent with Mn4+, far from the bond distances 10903 https://doi.org/10.1021/acs.energyfuels.1c01343 Energy Fuels 2021, 35, 10898−10907 Energy & Fuels pubs.acs.org/EF Article Figure 6. (a) Crystal structure model for Ca2.28MnCoO6 (Co is placed at the center of the dark blue polyhedra and Mn at the center of magenta octahedra, and oxygen atoms are represented by small red spheres and Ca by light blue, large spheres) showing the proposed incommensurate modulation. (b) Rietveld refinement of the crystal structure based on the diffraction pattern using the R3m(00γ)0s structure plus a remaining fraction of the pristine Ca3CoMnO6 phase and a small signal coming from the Al foil. 10904 https://doi.org/10.1021/acs.energyfuels.1c01343 Energy Fuels 2021, 35, 10898−10907 Energy & Fuels pubs.acs.org/EF Article observed in prisms (>2.1 Å). Thus, an alternative incom- mensurate crystal structure model (Figure 6a) has been built, which is compatible with the Co/Mn order while keeping Mn in the octahedral environment and is detailed in the Supporting Information. The Rietveld refinement of the last pattern collected upon oxidation using this model is plotted in Figure 6b (background has been subtracted for clarity). The refined values of the cell parameters are a = 9.217(1) Å, c1 = 4.9076(4) Å, and c2 = 4.339(3) Å (γ = 1.139) for the incommensurate phase [48(2) wt %] and a = 9.150(1) Å and c = 10.596(1) Å for the pristine phase [52(2) wt %]. The calcium content can be deduced from the value of the propagation vector and is consistent with Ca2.28CoMnO6. At the end of oxidation, Ca2.28CoMnO6 represents about 48(2) wt % and is hence equivalent to 0.35 mol of Ca2+ extracted per formula unit. This value is significantly smaller than the one estimated from the electrochemical capacity (0.82), which indicates that partial contribution to the current comes from a parasitic redox process, most likely related to electrolyte decomposition. The reasons for the extraction of a lower amount of calcium for this compound are not clear at this stage. It could be speculated that the ordering of Co and Mn in Ca3CoMnO6 makes calcium deinsertion entropically less favorable with respect to that of Ca3Co2O6, as the position of pyramidal environments is restrained to Co sites. Moreover, the migration mechanism for calcium ions (see the Supporting Information) implies the simultaneous rotation of two oxygen triads versus one single rotation in the case of Ca3Co2O6. Figure 7. Potential vs capacity profiles from the GCPL experiment on the Ca3CoMnO6//AC cell using 1 M LiPF6 in EC/DMC as the Rietveld refinements of the pattern taken at the end of electrolyte at RT and C/50 (a) and corresponding selected XRD reduction indicate that despite a slight growth of (100) and patterns (b). (300) peaks at Q = 1.374 and 2.381 Å−1, the amount of Ca2.28CoMnO6 is still 42(3)wt %, which would point at only 0.04 mol of Ca2+ ions being reinserted in the crystal structure consistent with that observed in calcium cells, reaching a stage upon reduction. Moreover, this result must be taken with in which both pristine and oxidized phases still coexist for a caution as the precise region of the sample where the capacity of 160 mA h/g. Upon reduction, no changes in the measurement is done could not be exactly the same after patterns are evidenced, which forces us to conclude that the relocating the cell under the beam. capacity observed is related to side reactions involving the To further assess whether the limited reversibility observed, electrolyte. These results point at the irreversibility of the if any, is related to the sluggish difusion of Ca2+ ions within the redox process being related to intrinsic structural/migration oxidized phase or the strong Ca2+−solvent interactions, constraints rather than to slow kinetics or high desolvation causing high desolvation energies which may hinder calcium energies. Calcium ion rearrangement in the oxidized phase reintercalation, or related to electrolyte decomposition, sublattice may hinder mobility in Ca3−xCoMnO6, as suggested resulting in the formation of some blocking films on the for Ca3−xCo2O6.22 Indeed, oxidation of Ca3CoO6 and surface of the particles, an additional operando experiment was Ca3CoMnO6 does not follow a topotactic mechanism and it conducted in lithium cells. In this case, standard electrolytes is reasonable to assume that the phase transformation that known to be stable at high potentials are used. The charge gives rise to the formation of the incommensurate modulated compensation mechanism during the oxidation of structure induces changes in the Ca migration pathways, with Ca3CoMnO6 should also be related to the extraction of the intercolumn space not making any significant difference. calcium ions from the crystal structure, while upon reduction, the lithium ions present in the electrolyte (much more mobile than calcium ions) might be inserted in the crystal structure. ■ CONCLUSIONS The operando synchrotron diffraction experiments indicate that Figure 7a displays the characteristic potential versus capacity the extraction of Ca2+ from K4CdCl6-type Ca3CoMO6 (M = profiles of Ca3CoMnO6 electrodes during operando galvano- Co, Fe, or Mn) structures is achieved at room temperature for static oxidation and reduction at C/50 with 1 M LiPF6 in EC/ M = Co and Mn only. Different incommensurate modulated DMC as the electrolyte and using AC as the CE. The potential structures are formed upon oxidation for M = Co and M = Mn, versus capacity profiles upon oxidation in lithium and calcium as a result of Mn/Co ordering, with the propagation vector cells are similar, while the reduction takes place at a slightly enabling the direct estimation of the calcium content, which is higher potential for the case of lithium, which may be related similar for both oxidized phases (ca. 2.3 mol per formula unit). to the different ionic conductivities of the electrolytes inducing The reversibility of the redox process was found to be different cell polarizations. Representative operando diffraction extremely limited, if any. Further operando experiments patterns taken in lithium cells are depicted in Figure 7b. Upon performed in lithium cells also resulted in the electrochemical the oxidation of Ca3CoMnO6, the evolution observed is fully extraction of calcium from the crystal structure upon oxidation, 10905 https://doi.org/10.1021/acs.energyfuels.1c01343 Energy Fuels 2021, 35, 10898−10907 Energy & Fuels pubs.acs.org/EF Article but intercalation of Li+ in Ca3−xCoMnO6 upon reduction was (2) Canepa, P.; Sai Gautam, G.; Hannah, D. C.; Malik, R.; Liu, M.; not found to be feasible either. Thus, the main factor behind Gallagher, K. G.; Persson, K. A.; Ceder, G. Odyssey of Multivalent the absence of redox reversibility seems to be related to Cathode Materials: Open Questions and Future Challenges. Chem. intrinsic structural/migration constraints rather than to slow Rev. 2017, 117, 4287−4341. kinetics or high desolvation energies commonly associated (3) Ponrouch, A.; Bitenc, J.; Dominko, R.; Lindahl, N.; Johansson, P.; Palacin, M. R. Multivalent rechargeable batteries. Energy Storage with divalent ion charge carriers. Mater. 2019, 20, 253−262. Finally, although still not conventionally used in the (4) Elena Arroyo-de Dompablo, M.; Ponrouch, A.; Johansson, P.; emerging field of calcium batteries, we believe that the Palacin, M. R. Achievements, Challenges, and Prospects of Calcium experimental protocols reported herein can be generally Batteries. Chem. Rev. 2020, 120, 6331. extended to the screening of new multivalent ion positive (5) Dugas, R.; Forero-Saboya, J. D.; Ponrouch, A. Methods and electrode materials and be a valuable tool to unambiguously Protocols for Reliable Electrochemical Testing in Post-Li Batteries assess reactivity and reversibility. ■ (Na, K, Mg, and Ca). Chem. Mater. 2019, 31, 8613−8628. (6) Ponrouch, A.; Frontera, C.; Bardé, F.; Palacín, M. R. Towards a ASSOCIATED CONTENT Calcium-Based Rechargeable Battery. Nat. Mater. 2016, 15, 169−172. * sı Supporting Information (7) Wang, D.; Gao, X.; Chen, Y.; Jin, L.; Kuss, C.; Bruce, P. G. The Supporting Information is available free of charge at Plating and Stripping Calcium in an Organic Electrolyte. Nat. Mater. 2018, 17, 16−20. https://pubs.acs.org/doi/10.1021/acs.energyfuels.1c01343. (8) Li, Z.; Fuhr, O.; Fichtner, M.; Zhao-Karger, Z. Towards stable Proposed description of the incommensurate structure and efficient electrolytes for room-temperature rechargeable calcium of Ca3−xCoMnO6 (PDF) batteries. Energy Environ. Sci. 2019, 12, 3496−3501. ■ (9) Shyamsunder, A.; Blanc, L. E.; Assoud, A.; Nazar, L. F. Reversible Calcium Plating and Stripping at Room Temperature AUTHOR INFORMATION Using a Borate Salt. ACS Energy Lett. 2019, 4, 2271−2276. Corresponding Author (10) Richard Prabakar, S. J.; Ikhe, A. B.; Park, W. B.; Chung, K. C.; M.R. Palacin − Institut de Ciència de Materials de Barcelona, Park, H.; Kim, K. J.; Ahn, D.; Kwak, J. S.; Sohn, K. S.; Pyo, M. ICMAB-CSIC, 08193 Bellaterra, Spain; orcid.org/0000- Graphite as a long-life Ca2+-intercalation anode and its implementa- 0001-7351-2005; Email: rosa.palacin@icmab.es tion for rocking-chair type calcium-ion batteries. Adv. Sci. 2019, 6, 1902129. Authors (11) PerezMato, J. M.; Zakhour-Nakhl, M.; Weill, F.; Darriet, J. A.P. Black − Institut de Ciència de Materials de Barcelona, Structure of composites A1+x(A’xB1‑x)O3 related to the 2H hexagonal ICMAB-CSIC, 08193 Bellaterra, Spain perovskite: relation between composition and modulation. J. Mater. D. Monti − Institut de Ciència de Materials de Barcelona, Chem. 1999, 9, 2795−2807. ICMAB-CSIC, 08193 Bellaterra, Spain (12) Fjellvåg, H.; Gulbrandsen, E.; Aasland, S.; Olsen, A.; Hauback, C. Frontera − Institut de Ciència de Materials de Barcelona, B. C. Crystal structure and possible charge ordering in one- ICMAB-CSIC, 08193 Bellaterra, Spain; orcid.org/0000- dimensional Ca3Co2O6. J. Solid State Chem. 1996, 124, 190−194. 0002-0091-4756 (13) Darriet, J.; Subramanian, M. A. Structural relationships between D. S. Tchitchekova − Institut de Ciència de Materials de compounds based on the stacking of mixed layers related to hexagonal perovskite-type structures. J. Mater. Chem. 1995, 5, 543−552. Barcelona, ICMAB-CSIC, 08193 Bellaterra, Spain; (14) Dussarrat, C.; Grasset, F.; Darriet, J. Crystallographic shears in orcid.org/0000-0003-0214-4115 hexagonal perovskites and K2NiF4-type structure - structural relation- R. G. Houdeville − CELLSALBA Synchrotron, 08290 ships between BaRuO3-Ba7Ir6O19(Cs7Cu6F19)-Cs4Cu3F10 and related- Cerdanyola del Vallès, Spain compounds. Eur. J. Solid State Inorg. Chem. 1995, 32, 557−576. F. Fauth − CELLSALBA Synchrotron, 08290 Cerdanyola (15) Boulahya, K.; Parras, M.; González-Calbet, J. M. New del Vallès, Spain; orcid.org/0000-0001-9465-3106 commensurate phases in the family (A3Co2O6)α(A3Co3O9)β (A=Ca, Complete contact information is available at: Sr, Ba). Chem. Mater. 2000, 12, 25−32. https://pubs.acs.org/10.1021/acs.energyfuels.1c01343 (16) Boulahya, K.; Parras, M.; González-Calbet, J. M. The An+2BnB’O3n+3 family (B=B’=Co=: Ordered intergrowth between 2H-BaCoO3 and Ca3Co2O6 structures. J. Solid State Chem. 1999, 145, Notes 116−127. The authors declare no competing financial interest. ■ (17) Boulahya, K.; Parras, M.; González-Calbet, J. M.; Vegas, A. A new orthorhombic Ba 8 Co 7 O21 phase: polymorphism in the ACKNOWLEDGMENTS (Ba3Co2O6)α(Ba3Co3O9)β system. J. Solid State Chem. 2000, 151, The authors are grateful to Alexandre Ponrouch and 77−84. ALISTORE-ERI members for helpful discussions and to (18) Murphy, D. W.; Trumbore, F. A. The Chemistry of TiS3 and ALBA synchrotron for beamtime (proposal 2019073634). NbSe3 Cathodes. J. Electrochem. Soc. 1976, 123, 960−964. ICMAB-CSIC members thank the Spanish Agencia Estatal de (19) Yamaura, K.; Zandbergen, H. W.; Abe, K.; Cava, R. J. Synthesis Investigación Severo Ochoa Programme for Centres of and properties of the structural one-dimensional cobalt oxide Ba1‑xSrxCoO3 (0≤x≤0.5). J. Solid State Chem. 1999, 146, 96−102. Excellence in R&D (CEX2019-000917-S). D.M. acknowledges (20) Monti, D.; Ponrouch, A.; Araujo, R. B.; Barde, F.; Johansson, the EC for a H2020 MSCA-IF grant (contract number P.; Palacín, M. R. Multivalent Batteries - Prospects for High Energy 743439). R.G.H. acknowledges the EC for a H2020 MSCA- Density: Ca Batteries. Front. Chem. 2019, 7, 79. DOC-FAM grant (contract number 754397). ■ (21) Tchitchekova, D. S.; Frontera, C.; Ponrouch, A.; Krich, C.; Bardé, F.; Palacín, M. R. Electrochemical Calcium Extraction from REFERENCES 1D-Ca3Co2O6. Dalton Trans. 2018, 47, 11298−11302. (1) Muldoon, J.; Bucur, C. B.; Gregory, T. Quest for Nonaqueous (22) Torres, A.; Bardé, F.; Arroyo-de Dompablo, M. E. Evaluation of Multivalent Secondary Batteries: Magnesium and Beyond. Chem. Rev. Cobalt Oxides for Calcium Battery Cathode Applications. Solid State 2014, 114, 11683−11720. Ionics 2019, 340, 115004. 10906 https://doi.org/10.1021/acs.energyfuels.1c01343 Energy Fuels 2021, 35, 10898−10907 Energy & Fuels pubs.acs.org/EF Article (23) Park, H.; Cui, Y.; Kim, S.; Vaughey, J. T.; Zapol, P. Ca Cobaltites as potential cathode materials for rechargeable Ca-ion batteries: theory and experiment. J. Phys. Chem. C 2020, 124, 5902. (24) Herklotz, M.; Weiss, J.; Ahrens, E.; Yavuz, M.; Mereacre, L.; Kiziltas-Yavuz, N.; Dräger, C.; Ehrenberg, H.; Eckert, J.; Fauth, F.; Giebeler, L.; Knapp, M. A novel high-throughput setup for in situ powder diffraction of coin cell batteries. J. Appl. Crystallogr. 2016, 49, 340−345. (25) Verrelli, R.; Black, A.; Dugas, R.; Tchitchekova, D.; Ponrouch, A.; Palacin, M. R. Steps towards the use of TiS2 electrodes in Ca batteries. J. Electrochem. Soc. 2020, 167, 070532. (26) Jain, A.; Singh, S.; Yusuf, S. M. Structural and magnetic properties of spin chain compounds Ca3Co2‑xFexO6. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 74, 174419. (27) Zubkov, V. G.; Bazuev, G. V.; Tyutyunnik, A. P.; Berger, I. F. Synthesis, crystal structure and magnetic properties of quasi-one- dimensional oxides Ca3CuMnO6 and Ca3Co1+xMn1‑xO6. J. Solid State Chem. 2001, 160, 293−301. (28) Tran, H.; Mehta, T.; Zeller, M.; Jarman, R. H. Synthesis and characterization of mixed phases in the Ca−Co−O system using the Pechini method. Mater. Res. Bull. 2013, 48, 2450−2456. (29) Pechini, M. P. Method of preparing lead and alkaline earth titanates and niobates and coating method using the same to form a capacitor. U.S. Patent 3,330,697 A 1963. (30) Nowik, I.; Jain, A.; Yusuf, M. S.; Yakhmi, J. W. Effect of Fe substitution on the magnetic ordering in Ca3(Co1‑xFex)2O6. Phys. Rev. B: Condens. Matter Mater. Phys. 2008, 77, 054403. (31) Ouyang, Z. W.; Xia, N. M.; Wu, Y. Y.; Sheng, S. S.; Chen, J.; Xia, Z. C.; Li, L. Short-range ferromagnetic correlations in the spin- chain compound Ca3CoMnO6. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 84, 054435. (32) Fauth, F.; Peral, I.; Popescu, C.; Knapp, M. The new Material Science Powder Diffraction beamline at ALBA Synchrotron. Powder Diffr. 2013, 28, S360−S370. (33) Ponrouch, A.; Dedryvère, R.; Monti, D.; Demet, A. E.; Ateba Mba, J. M.; Croguennec, L.; Masquelier, C.; Johansson, P.; Palacín, M. R. Towards high energy density sodium ion batteries through electrolyte optimization. Energy Environ. Sci. 2013, 6, 2361. (34) Leriche, J. B.; Hamelet, S.; Shu, J.; et al. An electrochemical cell for operando study of lithium batteries using synchrotron radiation. J. Electrochem. Soc. 2010, 157, A606. (35) Xu, K. Nonaqueous liquid electrolytes for lithium-based rechargeable batteries. Chem. Rev. 2004, 104, 4303−4418. 10907 https://doi.org/10.1021/acs.energyfuels.1c01343 Energy Fuels 2021, 35, 10898−10907