Article pubs.acs.org/JPCC Mechanistic and Kinetic Study of the Electrochemical Charge and Discharge of La2MgNi9 by in Situ Powder Neutron Diffraction Michel Latroche,*,† Fermìn Cuevas,† Wei-Kang Hu,‡ Denis Sheptyakov,§ Roman V. Denys,‡ and Volodymyr A. Yartys‡,∥ † Institut de Chimie et des Matériaux Paris-Est, UPEC−CNRS, UMR 7182, 2 rue Henri Dunant, 94320 Thiais, France Institute for Energy Technology, Department of Energy Systems, 2007 Kjeller, Norway § SINQ-HRPT, Paul Scherrer Institute, 5232 Villigen, Switzerland ∥ Norwegian University of Science and Technology, 7491, Trondheim, Norway ‡ S Supporting Information * ABSTRACT: The intermetallic La2MgNi9 has been investigated as negative electrode material for NiMH battery by means of in situ neutron powder diffraction. This hydrideforming compound exhibits suitable plateau pressures ranging within the practical electrochemical window and leads to significant reversible electrochemical capacities. Charge and discharge of the composite electrode have been performed in beam following various current rates and galvanostatic intermittent titration. From the diffraction data analysis, phase amounts and cell volumes have been extracted, allowing the interpretation of the hydride formation and decomposition. From the evolution of the diffraction line widths, differences are observed between charge and discharge with the possible formation of an intermediate γ phase on charge. The electrode readily responds to current rate variations and does not show any kinetic limitation in the range C/10 and C/5 (C/n: full capacity C in n hours). This material shows excellent properties regarding electrochemical storage of energy. ■ strains induced by the significant volume change between the intermetallic and its hydride. However, despite all these efforts to design materials with enhanced properties, the use of LaNi5-type compounds remains limited by their low capacities, around 320 mAh g−1. This is mainly due to the heavy rare earths that lead to high molar mass. To overcome this latter point, Kohno et al.12 developed materials for which part of the heavy rare earths can be replaced by light element, such as magnesium. Indeed, it exists in several phases in the Ni-rich part of the binary La−Ni diagram. Between 75% and 79.2% of nickel, line compounds LaNi3, La2Ni7, and La5Ni19 can be described as a stacking along the c axis of [La2Ni4] + n[LaNi5] subunits with n = 1, 2 and 3, respectively.13,14 Mg does not substitute La in the LaNi5 subunits, but it replaces the rare earth within the La2Ni4 subunits. This led to a real breakthrough in terms of increased gravimetric H densities as magnesium allows significant decrease of the molar mass. On the basis of the described concept, new materials with various rare earths and transition metals have been developed offering significant electrochemical storage capacities up to 400 mAh g−1.2−4,15 Among them, La3−xMgxNi9 are key compounds as they correspond to a simple stacking [La2−xMgxNi4] + [LaNi5] (x = 0−2),14 allowing Mg-rich phases. LaMg2Ni9 (2 [MgNi2] + INTRODUCTION Metallic hydrides are very convenient materials for hydrogen storage.1 They can store a large amount of hydrogen gas at thermodynamic conditions close to ambient pressure and room temperature, making them suitable for energy storage. Metallic hydrides have been also developed for electrochemical storage. Indeed, they can react with water to electrochemically exchange one proton and one electron. This allows the design of efficient negative electrodes for alkaline secondary batteries (NiMH) which offer better properties that the actual NiCad batteries.2−4 For two decades, the most developed materials for these negative electrodes were derived from the LaNi5 intermetallic compound. By proper substitutions on either the La site (by other rare earths) or the Ni sites (by other transition or pmetals) and by playing with the stoichiometry (according to the narrow domain of existence of the LaNi5+x compound; 0 < x < 0.4), a complex but efficient class of materials has been developed.5−10 According to the thermodynamic rules, during charge and discharge, reactions at the negative electrode should occur only between two solid phases, the intermetallic (α) and the hydride (β). However, it was evidenced by several authors that for the best materials in term of cycle life, the appearance of an out-ofequilibrium, metastable, transient γ phase is observed during charge.6,8,11 This unexpected γ phase (with intermediate volume and H-loading) was reported to form at the interface between the α and β phases, helping to accommodate the heavy © 2014 American Chemical Society Received: April 1, 2014 Revised: May 19, 2014 Published: May 19, 2014 12162 dx.doi.org/10.1021/jp503226r | J. Phys. Chem. C 2014, 118, 12162−12169 The Journal of Physical Chemistry C Article porosity of the electrode. The potential was monitored against a solid-state Cd/Cd(OH)2 reference electrode.5,9 NPD data were recorded at room temperature and ambient pressure with the high-resolution powder diffractometer for thermal neutrons (HRPT) in high intensity mode (flux of about 1014 n cm−2 s−1) at SINQ, PSI in Switzerland. Using the large position sensitive (PSD)3He detector, measurements were performed in the scattering angle ranging from 5 to 165° with angular step of 0.05°. Wavelength was set to 1.494 Å. The patterns were recorded by batches of eight, varying the detector angle by step of 0.15° between 3.95° and 5°. Typical time acquisition was 225 s per pattern. The eight patterns were finally combined into one signal file leading to a time resolution of about 1800 s for each diffraction measurement. Because of the high flux of the spallation source and the use of a large position-sensitive detector, the high-rate capability of such electrodes can be followed in situ. The diffraction patterns were sequentially refined using the program FULLPROF.18 Preliminary Work. The composite electrode was first charged ex situ before the NPD experiment at 20 mA g−1 (115 mA or C/16) up to a total capacity of 2300 mAh. Three activation cycles were performed with a cutoff potential of 200 mV versus Cd/Cd(OH)2. The discharge capacities measured at each cycles are given in Table 1. [LaNi5]) can be considered to be the parent compound of the La3−xMgxNi9 family. Earlier synchrotron X-ray diffraction and neutron powder diffraction (NPD) studies revealed that substitution of La by Mg in LaNi3 proceeds in an ordered way and leads to the formation of LaMg2Ni9 at maximum solubility of magnesium. Gradual increase in Mg content is accompanied by a linear decrease of the unit cell volume. Hydrogen interaction with the La3−xMgxNi9 alloys was investigated by in situ neutron powder diffraction and pressure−composition−temperature (PCT) studies.13 Magnesium was found to influence structural features of the hydrogenation process and determine various aspects of the hydrogen interaction with the La3−xMgxNi9 intermetallics causing: (i) more than 1000-fold increase in equilibrium pressures of hydrogen absorption and desorption for the Mgrich LaMg 2Ni 9 as compared to that of the Mg-poor La2.3Mg0.7Ni9, which implies a substantial modification of the thermodynamics of the formation and decomposition of the hydrides according to the van’t Hoff relation; (ii) an increase of the reversible hydrogen storage capacities following increase of Mg content in the La1−xMgxNi3 to ∼1.5 wt % H (380 mAh g−1 in equivalent electrochemical units) for La2MgNi9; (iii) improvement of the resistance against hydrogen-induced amorphization and disproportionation; (iv) change of the cell expansion mechanism of the hydrogenation from anisotropic to isotropic. Thus, optimization of the magnesium content provides different possibilities for improving properties of the studied alloys as hydrogen storage and battery electrode materials. The present work deals with the electrochemical formation and decomposition of the hydride of the representative parent La2MgNi9 alloy for battery electrode applications. Studies of the interaction with gaseous hydrogen revealed that the hydride forms at pressures slightly below one bar, which is suitable for electrochemical formation and decomposition in open cells. Indeed, the electrochemical discharge capacity of such alloys is larger than that for LaNi5-based electrodes.16 Importantly, La2MgNi9 alloy can be synthesized directly from the melt by using the rapid solidification technique.17 The outcome of the present study is the optimization of the chemical composition of ternary La−Mg−Ni compounds with respect to their electrochemical hydrogen storage performance. Table 1. Discharge Capacities at Each Cycle during Activation of the Composite Electrode cycle number (C/16) capacity first cycle second cycle thrid cycle 949 mAh (177 mAh g−1) 1257 mAh (234 mAh g−1) 1375 mAh (256 mAh g−1) From the pressure−composition−temperature (PCT) isotherm curve of La2MgNi9 measured at 20 °C (Figure 1; data from ref 13), the electrochemical capacities can be estimated according to the practical pressure windows. This defines about 310 mAh g−1 of reversible capacity (between 0.001 and 0.1 EXPERIMENTAL SECTION The raw metallic material La2MgNi9 was prepared by powder metallurgy at IFE, Norway.13 From this starting material, a composite electrode was made from intermetallic powder sieved below 63 μm and mixed with carbon black and PTFE in the weight ratio 90:5:5. This mixture was spread out in sheets approximately 1.5 mm thick and compressed at 5 tons on both sides of a 5 × 3.14 cm2 nickel grid which plays the role of current collector. The final thickness of the composite electrode was about 0.9 mm. These electrode plates were then rolled up on themselves to form a cylinder of about 50 mm height and 10 mm diameter. The final working electrode contains 5.366 g of active material and was introduced in a specially designed silica cell. The electrode is sandwiched between inner (⌀ 8 mm) and outer (⌀ 12 mm) counter-electrode cylinders made of nickel grid rolled on themselves, with silica sheaths as separators on each side of the working electrode. The electrode is immersed in NaOD 5.5 M electrolyte and pumped under primary vacuum in order to fully impregnate the working electrode with the liquid and to remove any gaseous species trapped in the ■ Figure 1. Pressure−composition−temperature (PCT) isotherm curve for La2MgNi9 during a full absorption−desorption cycle at 20 °C. The dotted lines stand for the electrochemical window (between 0.1 and 0.001 MPa) and define about 310 mAh g−1 of reversible capacity (data from ref 13), whereas the total capacity at 0.1 MPa reaches 380 mAh g−1. 12163 dx.doi.org/10.1021/jp503226r | J. Phys. Chem. C 2014, 118, 12162−12169 The Journal of Physical Chemistry C Article 543.65 Å3), as the low quantity of this phase and the bulky pattern do not allow accurate refinement of these parameters. These values are higher than those of the La 2 MgNi 9 intermetallic (a = 5.0314(2); c = 24.302(1) Å; V = 532.78 Å3; 1.2 Å3/f.u. La2MgNi9)19 and indicate the formation for the active grains of the H/D solid solution containing about 0.5−1 atom H/D/f.u. La2MgNi9. This is consistent with the PCT diagram shown in Figure 1. MPa) and 380 mAh g−1 of total capacity (between 0 and 0.1 MPa). Thus, the reversible capacity obtained for the third activation cycle represents about 82% of the expected reversible electrochemical capacity. The electrode was then fully charged just before the in situ experiment and was measured once in beam under open circuit voltage (OCV). The diffraction data were interpreted on the basis of the structural parameters given for La2MgNi9 (i.e., the phase α)13 and La2MgNi9D13 (i.e., the phase β)19 (see Supporting Information for details). Beside the strong nickel lines coming from the current collector and the counterelectrodes, the diffraction lines belonging to the main hydride phase (i.e., the phase β) are clearly observed (Figure 2). Some EXPERIMENTAL RESULTS The electrode was first discharged at a constant current of 180 mA (33 mA g−1 or D/10) with a cutoff potential of 0.5 V (vs ECd/Cd(OH)2). The total discharged capacity was 1494 mAh (278 mAh g−1). During the process, starting from 0.044 V, an increase of the potential of the working electrode is observed. An unexpected kink appears after 3 h of discharge at 0.158 V (Figure 3) but was not observed during the previous or ■ Figure 2. Refined neutron powder diffraction pattern (measured, dots; calculated, solid line) for the electrode at initial charged state. Crystallographic structures are taken from Denys and Yartys.13,19 Vertical bars correspond to diffraction line positions for each phase: La2MgNi9 (α) and deuterated La2MgNi9D13 (β) phases. Nickel lines (heavily textured) arise from the current collector of the working electrode and from the counter electrodes. Background around 30° comes from the silica cell and the NaOD/D2O electrolyte. One can also note an extra peak at 2θ = 17.5° that is attributed to PTFE. Figure 3. Evolution of the potential Ew versus the Cd/Cd(OD)2 reference electrode during the first in-beam charge−discharge (C/10) of the electrode at a current rate of 33 mA g−1. weak lines coming from the deuterium-free phase (i.e., the phase α) can also be observed, and this phase was also introduced in the refinement. The presence of this phase indicates that a small amount (around 14 wt %) of the intermetallic compound is electrochemically nonactive because of loss of electronic or electrolytic contacts for a few grains. This partly explains the 82% observed for the reversible capacity at the third cycle. The cell parameters of the β phase are 5.3692 and 26.1328 Å for a and c, respectively (space group R3m ̅ ), and the cell volume is 652.42 Å3. This is slightly lower than the volume reported by Denys and Yartys13,19 (675.11 Å3) for the fully charged hydride under gas pressure (13 D/f.u.), but this latter value was obtained under 1 MPa of gas pressure, 1 order of magnitude larger than the present ambient conditions. The volume decrease of 22.7 Å3 or 2.5 Å3/f.u. La2MgNi9 can be translated into the loss of hydrogen storage capacity of approximately 0.8 at H/f.u. from La 2 MgNi 9 D 13 to La2MgNi9D12. This volume shrinking is obviously associated with a depopulation of the D sites occupied in La2MgNi9D13.19 However, for the simplicity of the evaluation of the data, only changes of the unit cell parameters where accounted, whereas the atomic structure was considered as unaltered. The cell parameters of the α phase were kept fixed to 5.0712 and 24.4105 Å for a and c, respectively (space group R3̅m; V = following measurements. It is then assumed that it is probably an artifact (related to bubble formation close to the reference capillary). This is supported by the analysis of the EMH−ENi signal (not shown here), which shows a flat plateau during the whole discharge. After 8.5 h of discharge, the potential rises rapidly to reach 0.500 V and the current was switched off. The potential then decreases rapidly to stabilize around 0.030 V during the rest period (OCV). Following 1 h of rest (OCV), the current was then set to −180 mA (33 mA g−1 or C/10) for 12 h assuming full charge of the electrode over this time scale. Potential decreases rapidly to reach a plateau around −0.163 V. It remains nearly constant though slightly decreasing and becoming bulky after 7 h (i.e., t = 16 h) of charge (Figure 3). This is attributed to hydrogen evolution leading to large production of deuterium gas bubbles close to the reference electrode capillary and thus involving a noisy signal. The 3D view of the NPD pattern evolution as a function of time during the first cycle (D/10 + C/10) is shown in Figure 4. One can observe that the strong diffraction lines belonging to the β phase decrease rapidly during the discharge (time 0−8.5 h) whereas those of the α phase increase. After 9.5 h, the transformation is almost fully completed and the charge was 12164 dx.doi.org/10.1021/jp503226r | J. Phys. Chem. C 2014, 118, 12162−12169 The Journal of Physical Chemistry C Article Figure 5. Comparison between the α and β phase amounts and the capacity Q during the in-beam charge−discharge cycle at D/10 + C/10 for the working electrode. The initial charge state Q at t = 0 is set at 380 mAh g−1, a value determined from the PCT curve (Figure 1) at 0.1 MPa. Figure 4. 3D view of the NPD pattern evolution as a function of time during the first cycle (D/10 + C/10) of the working electrode at a current rate of 33 mA g−1. • from 9.5 to 12 h, a region attributed to the α phase solid solution domain for which phase amounts do not vary significantly; • from 12 to 16 h, a two-phase domain where the α and β phases are in equilibrium and transform into each other rapidly; • beyond 16 h, a region where the β phase solid solution domain starts but in competition with the hydrogen evolution. Then a slow increase of the β phase amount is observed despite the linear augmentation of Q. From the analysis of the diffraction patterns, the evolution of the cell volumes of both phases can be plotted as a function of time (Figure 6). During the discharge, the volume of the β phase decreases continuously, which indicates that the phase is progressively depleted in deuterium. Assuming that ΔV/atom D = 3.26 Å3 (as shown later in the paper), we can estimate that the lower content of deuterium in the β-phase reached during started. The α phase then transforms reversibly into the β phase, but contrary to the discharge, strong overlap of the diffraction peaks of both phases is observed. After about 20 h, the charge is completed and a full recovery of the diffraction peaks of the β phase takes place. For the whole cycle, the diffraction patterns have been refined sequentially assuming three phases as observed in the first diffraction pattern (Figure 2): α deuterium-free and β deuteride La2MgNi9 phases plus the nickel one from the counter electrodes. Only scale factors, cell parameters, and the U parameters of the Caglioti function were refined (8 parameters). In addition, as the level of the background can follow slight variations depending on the amount of produced deuterium gas (leading to various amounts of NaOD/D2O electrolyte in the beam), the 25 points of the interpolated background were also refined leading to a total of 33 refined parameters. From analysis of these data, the relative amount of each phase (α and β) can be extracted. The amount of Ni was considered to be a constant. The results are shown in Figure 5 and are compared to the electrochemical capacity Q passed through the electrode (right-hand scale). However, one should keep in mind that during the charge step, because of hydrogen (deuterium) evolution, the capacity Q does not reflect the exact state of charge of the working material. To overcome this difficulty, the fully charged state was set at 380 mAh g−1, a value determined from the PCT curve (Figure 1) at 0.1 MPa, i.e., at the top of the electrochemical window. The results of the galvanostatic cycle can be split into different domains. During the discharge: • from 0 to 3 h, a first step attributed to the β phase solid solution domain (i.e., the formation of the α phase is hardly detected); • from 3 to 8.5 h, a two-phase domain where the β and α phases are in equilibrium and transform into each other; • from 8.5 h to 9.5 h, a relaxation period for which phase amounts remain nearly constant. During the charge: Figure 6. Evolution of the cell volumes for the α and β phases during the in-beam charge−discharge cycle at C/10 for the working electrode. For sake of comparison, the evolution of the potential Ew versus the Cd/Cd(OD)2 reference electrode is also shown on the right-hand scale. 12165 dx.doi.org/10.1021/jp503226r | J. Phys. Chem. C 2014, 118, 12162−12169 The Journal of Physical Chemistry C Article its discharge is close to La2MgNi9D11.2. This is much less the case for the α phase for which the cell volume is nearly constant though slightly decreasing. During the charge, the determination of the cell volumes is less accurate, especially during the α-to-β transformation (t = 12−16 h) when diffraction peaks from both phases strongly overlap (see 3D patterns in Figure 4). Nevertheless, one can observe that both volumes increase rapidly during the first 6 h of charge and then stabilize progressively though the β phase volume continues to increase until the end of the charge. This indicates that the charge still takes place in the β solid solution domain. At the end of the charge, the β phase volume recovers its initial value (t = 0). Finally, the evolution of the diffraction peak broadening, as determined by the fitting of a common U parameter for both phases, is given in Figure 7. For the discharge and the charge, Figure 8. Evolution of the phase amounts for the α and the β phases during the in-beam charge−discharge cycle at C/5 for the working electrode. For sake of comparison, the evolution of the potential Ew versus the Cd/Cd(OD)2 reference electrode is also shown on the right-hand scale. Figure 7. Evolution of the half-width of the diffraction peaks for the β and α phases during the in-beam charge−discharge cycle at D/10+C/ 10 for the working electrode. For sake of comparison, the evolution of the capacity Q is also shown on the right-hand scale. The initial charge state Q at t = 0 is set at 380 mAh g−1, a value determined from the PCT curve (Figure 1) at 0.1 MPa. Figure 9. 3D view of the NPD pattern evolution as a function of time during the GITT discharge cycle (D/7) of the working electrode at 46 mA g−1. an enlargement of the peaks is seen for the β and α phases when they are in equilibrium and transform into each other (i.e., crossing the plateau region). However, this effect is much larger (1 order of magnitude) during the charge, leading to peak overlapping as observed in Figure 4. For the next 10 h, the electrode was cycled again with a higher current density (75 mA g−1 or ∼C/5). The discharge capacity obtained at this rate is 1332 mAh (248 mAh g−1). Except for this slightly lower capacity, the electrode behaves very similarly to the previous cycle at C/10 involving the same step: α solid solution, α-to-β transformation, β solid solution, and hydrogen evolution at the end of the charge. In the same way, cell volume variations and half-width evolutions are very comparable to those observed at C/10. Phase amounts and potential evolutions are given in Figure 8. Finally, the electrode was discharged by the galvanostatic intermittent titration technique (GITT) using the following protocol: discharge at 46 mA g−1 (D/7) for 1.5 h or Ew > 0.5 V followed by a relaxation period of 1 h. Evolution of the diffraction patterns is shown in Figure 9. The procedure lasts for six steps that are summarized in Table 2 giving discharge time and cumulated capacities. Table 2. Cumulated Discharge Capacities Obtained from GITT Experimenta step discharge time (h) capacity (mAh) capacity (mAh g−1) 1 2 3 4 5 6 1:30 1:30 1:30 1:02 0:10 0:08 370 740 1100 1350 1392 1427 69 138 207 252 259 266 a The current was set at 46 mA g−1 (D/7) for 1.5 h or Ew > 0.5 V followed by relaxation periods of one hour. One can observe in the 3D view of Figure 9 that the diffracted intensities of both phases follow fairly well the different current steps of the GITT. In addition, one can note that the background level is very dependent on the current density. This can be understood if one considers that switching on and off the current involves important changes in the amount of deuterium gas produced in the cell. When the current is off, no gas is produced and the quantity of liquid 12166 dx.doi.org/10.1021/jp503226r | J. Phys. Chem. C 2014, 118, 12162−12169 The Journal of Physical Chemistry C Article (D2O, NaOD) is larger in the beam, causing increased background. Thus, each maximum in the background can be attributed to a relaxation period. The total cumulated discharge capacity obtained at this rate is 1427 mAh (266 mAh g−1), a value very close to that measured during the first in situ cycle at C/10. The evolution of the relative amount of each phase (α and β) is shown in Figure 10. Again, the electrode behaves very closely to the previous cycles involving the same transformations. Figure 11. Evolution of the phase amounts for the α and the β phases during the in-beam charge−discharge cycles at C/10, C/5, and GITT for the working electrode. For sake of comparison, the evolution of the potential Ew versus the Cd/Cd(OD)2 reference electrode is also shown on the right-hand scale. Figure 10. Evolution of the phase amounts for the α and the β phases during the in-beam discharge obtained by GITT experiments. The current was set to 46 mA g−1 (D/7) for 1.5 h or Ew > 0.5 V followed by relaxation periods of 1 h. For sake of comparison, the evolution of the capacity Q is also shown on the right-hand scale. The initial charge state Q is set at 380 mAh g−1, a value determined from the PCT curve (Figure 1) at 0.1 MPa. In addition, small potential plateaus appear during the OCV periods which correspond to zero current flow for which no phase transformation takes place. However, very small changes in the α and β amounts can be observed during the resting periods (Figure 10). This is consistent with deuterium diffusion from the β toward the depleted α phase on the particle surface, transforming some β phase into (saturated) α phase during the electrode relaxation. The charge and discharge capacities for each in-beam cycle described so far are given in Table 3. Figure 12. Evolution of the cell volumes for the α and the β phases during the in-beam charge−discharge cycles at C/10, C/5, and GITT for the working electrode. For sake of comparison, the evolution of the potential Ew versus the Cd/Cd(OD)2 reference electrode is also shown on the right-hand scale. electrochemical capacity does not correspond to that of the electrode materials, mainly because of the deuterium gas evolution at the end of each charge. To overcome this difficulty, data can be analyzed by separately considering capacity of charge Qc and capacity of discharge Qd. This has been done in Figure 13 for the cell volume evolution of both phases. Interestingly, a linear behavior is observed for the two phases though the volumes measured during charge are a bit scattered. The β phase volume increases smoothly as a function of Q following the equation Table 3. Discharge Capacity at Each Cycle during in Situ Measurements of the Working Electrode 4th cycle (D/10) 5th cycle (D/5) 6th cycle (GITT; D/7) capacity (mAh) capacity (mAh g−1) 1494 mAh 1332 mAh 1427 mAh (278 mAh g−1) (248 mAh g−1) (266 mAh g−1) To get an overview of the electrode behavior, the evolutions of phase amounts and cell volumes for the α and the β phases during the in-beam charge−discharge cycles at C/10, C/5, and GITT are given and compared to the potential Ew in Figures 11 and 12. From the precedent data analysis, all measured parameters (phase amounts and cell volumes) have been processed regarding time scale evolution. At this stage, it is worth looking also to the data as a function of the electrochemical charge Q. Once again, one has to be cautious with the values of Q as the Vβ = 0.101(4)Q d + 608.5 (1) whereas that of the α phase is nearly constant in the whole range of Qd. The same figure can also be drawn for the phase amounts, though it is significant only for the discharge. This evolution is shown in Figure 14 for which two domains can be clearly identified. From 100 to 310 mAh g−1, a clear two-phase transition domain can be seen. Then, for Qd larger than 310 12167 dx.doi.org/10.1021/jp503226r | J. Phys. Chem. C 2014, 118, 12162−12169 The Journal of Physical Chemistry C Article Using neutron diffraction data analysis, we observed that the structural properties of the charged electrode material are comparable to that published by Denys et al.19 for the hydride loaded under gas pressure. Starting from this charged state, the electrochemical cycling behavior has been investigated in beam at different rates to follow the mechanism of the reversible charge−discharge process. At first glance, one can consider that the electrode material behaves like a classic one following an intermetallic-to-hydride transformation through the sequences α, α-to-β, and β formation, a typical behavior for a two-phase reaction. This is indeed the case during the discharge according to the variations of the phase amounts, the cell volumes, and the line diffraction half-widths. All these parameters agree well with a two-phase behavior. This is, however, not so obvious for the charge process. Despite the fact that a two-phase transformation seems to take place in both processes, important increase of the half width during charge involves strong overlapping between the α and β diffraction peaks. This unexpected behavior can be attributed to a large concentration gradient in the H concentration or heavy constraints during the charge process. However, this is not supported by the small hysteresis between the absorption and desorption branches or by the relatively flat plateaus of the PCT curves. A better hypothesis might be the formation of an intermediate γ phase as previously observed in the LaNi5-type system.6,8,11 Similarly, the γ phase was observed only during the charge process. This out-of-equilibrium phase was related to constraint relaxation at the α-to-β interface and plays a key role in the cycle life by decreasing the decrepitation process responsible for the high corrosion rate in alkaline medium. As the PuNi3-type structure is built from the stacking of LaNi5 and MgNi2 subunits, such a mechanism may also take place in this system. A better characterization of this intermediate phase could be obtained in the future by performing in-beam GITT during charge with longer relaxation times assuming that the γ phase amount will persist at the phase interfaces because of kinetic barriers. To our knowledge, the formation of this intermediate γ phase was never reported before in these stacking structures. From Figure 13, it is observed that the cell volume for the β phase is strongly dependent on the state of charge and follows a linear dependence. From eq 1, one can derive that this corresponds to 3.26 Å3/D atom, a value in agreement with previously reported values ranging between 2.5 and 3.5 Å3/H atom.20,21 On the contrary, the volume of the α phase remains almost constant at all states of charge. Evolution of the phase amounts as a function of the state of discharge allows to clearly discriminate between β solid solution, β to α transformation and α solid solution domains (Figure 14). This latter domain is very small in the range of the electrochemical state of charge studied in the present work, which explains the little variation of the cell volume of the α phase. In other words, the reversible capacity is mainly obtained from hydrogen absorption−desorption in the β solid solution and at the α-to-β transformation but not in the α solid solution. This statement agrees with the shape of the PCT curve (Figure 1) for which the hydrogen trapped in the α solid solution is difficult to extract at low pressure, i.e., out of the practical electrochemical window. Finally, it is interesting to note that the electrode material readily responds to the electrochemical solicitations. A typical example is shown in Figure 10 where the diffraction peak intensities follow very closely the current variation imposed by Figure 13. Evolution of the cell volumes for the α (black squares) and the β (red circles) as a function of the state of charge Qc (solid symbols) and Qd (empty symbols). For Qd, the fully charged state was set at 380 mAh g−1, a value determined from the PCT curve (Figure 1) at 0.1 MPa. Figure 14. Evolution of the phase amounts for the α (black squares) and the β (red circles) phases as a function of the state of discharge Qd. The fully charged state was set at 380 mAh g−1, a value determined from the PCT curve (Figure 1) at 0.1 MPa. mAh g−1, the β solid solution domain starts and extends up to the upper charge state (380 mAh g−1). Interestingly, for the α phase, an instant formation of the electrochemically stable solid solution is observed. As it is not possible to electrochemically decompose it, the domain of the α solid solution with a variable content of H/D is effectively nonexistent. It is also worth noting that a fully charged state (i.e., 100% β) is never reached, whereas the fullly discharged state (i.e., 100% α) is almost complete. DISCUSSION The intermetallic La2MgNi9 readily and reversibly absorbs hydrogen by solid−gas route, and thanks to its suitable equilibrium pressures, the same compound can be used as negative electrode in alkaline medium. Indeed, after a few activation cycles, an excellent correlation between the capacities obtained by solid gas and electrochemical measurements is observed. ■ 12168 dx.doi.org/10.1021/jp503226r | J. Phys. Chem. C 2014, 118, 12162−12169 The Journal of Physical Chemistry C Article (4) Liu, Y.; Cao, Y.; Huang, L.; Gao, M.; Pan, H. Rare Earth−Mg− Ni-based Hydrogen Storage Alloys as Negative Electrode Materials for Ni/MH Batteries. J. Alloys Compd. 2011, 509, 675−686. (5) Latroche, M.; Percheron-Guégan, A.; Chabre, Y.; Bouet, J.; Pannetier, J.; Ressouche, E. Intrinsic Behavior Analysis of Substituted LaNi5-type Electrodes by Means of in Situ Neutron Diffraction. J. Alloys Compd. 1995, 231, 537−545. (6) Latroche, M.; Percheron-Guégan, A.; Chabre, Y. Influence of Cobalt Content in MmNi4.3−xMn0.3Al0.4Cox Alloy (x=0.36 and 0.69) on its Electrochemical Behaviour Studied by In Situ Neutron Diffraction. J. Alloys Compd. 1999, 293−295, 637−642. (7) Cuevas, F.; Joubert, J.-M.; Latroche, M.; Percheron-Guégan, A. Intermetallic Compounds as Negative Electrodes of Ni/MH Batteries. Appl. Phys. A: Mater. Sci. Process. 2001, 72, 225−238. (8) Latroche, M.; Chabre, Y.; Percheron-Guégan, A.; Isnard, O.; Knosp, B. Influence of Stoichiometry and Composition on the Structural and Electrochemical Properties of AB5+y-based Alloys Used as Negative Electrode Materials in Ni-MH Batteries. J. Alloys Compd. 2002, 330−332, 787−791. (9) Latroche, M.; Chabre, Y.; Decamps, B.; Percheron-Guégan, A.; Noréus, D. In Situ Neutron Diffraction Study of the Kinetics of Metallic Hydride Electrodes. J. Alloys Compd. 2002, 334 (1−2), 267− 276. (10) Latroche, M.; Joubert, J.-M.; Guegan, A. P.; Isnard, O. In Situ Neutron Diffraction Study of Deuterium Gas Absorption by AB5+y Alloys Used as Negative Electrode Materials for Ni-MH Batteries. Phys. B: Condensed Matter 2004, 350 (1−3, Suppl. 1), e427−e430. (11) Vivet, S.; Latroche, M.; Chabre, Y.; Joubert, J.-M.; Knosp, B.; Percheron-Guegan, A. Influence of Composition on Phase Occurrence during Charge Process of AB5+x Ni-MH Negative Electrode Materials. Phys. B: Condensed Matter 2005, 362 (1−4), 199−207. (12) Kohno, T.; Yoshida, H.; Kawashima, F.; Inaba, T.; Sakai, I.; Yamamoto, M.; Kanda, M. Hydrogen Storage Properties of New Ternary System Alloys: La2MgNi9, La5Mg2Ni23, La3MgNi14. J. Alloys Compd. 2000, 311 (2), L5−L7. (13) Denys, R. V.; Yartys, V. A. Effect of Magnesium on the Crystal Structure and Thermodynamics of the La3−xMgxNi9 Hydrides. J. Alloys Compd. 2011, 509, S540−S548. (14) Crivello, J.-C.; Zhang, J.; Latroche, M. Structural Stability of ABy Phases in the (La,Mg)−Ni System Obtained by Density Functional Theory Calculations. J. Phys. Chem. C 2011, 115, 25470−25478. (15) Petit-Ferey, A.; Cuevas, F.; Latroche, M.; Knosp, B.; Bernard, P. Elaboration and Characterization of Magnesium-Substituted La5Ni19 Hydride Forming Alloys as Active Materials for Negative Electrode in Ni-MH Battery. Electrochim. Acta 2009, 54, 1710−1714. (16) Hu, W. K.; Denys, R. V.; Nwakwuo, C. C.; Holm, T.; Maehlen, J. P.; Solberg, J. K.; Yartys, V. A. Annealing Effect on Phase Composition and Electrochemical Properties of the Co-Free La2MgNi9 Anode for Ni-Metal Hydride Batteries. Electrochim. Acta 2013, 96, 27−33. (17) Nwakwuo, C. C.; Holm, T.; Denys, R. V.; Hu, W. K.; Maehlen, J. P.; Solberg, J. K.; Yartys, V. A. Effect of Magnesium Content and Quenching Rate on the Phase Structure and Composition of Rapidly Solidified La2MgNi9 Metal Hydride Battery Electrode Alloy. J. Alloys Compd. 2013, 555, 201−208. (18) Rodríguez-Carvajal, J. FULLPROF: A Program for Rietveld Refinement and Pattern Matching Analysis. Phys. B (Amsterdam, Neth.) 1993, 192, 55−69. (19) Denys, R. V.; Yartys, V. A.; Webb, C. J. Hydrogen in La2MgNi9D13: The Role of Magnesium. Inorg. Chem. 2012, 51 (7), 4231−4238. (20) Dorogova, M.; Hirata, T.; Filipek, S. M. Hydrogen-Induced Volume Changes in ZrCr2 and Pseudo-Binary Compounds of ZrCr2, ZrMn2 and ZrV2. Phys. Status Solidi A 2003, 198, 38−42. (21) Yartys, V. A.; Burnasheva, V. V.; Semenenko, K. N. Structural Chemistry of Hydrides of Intermetallic Compounds. Russ. Chem. Rev. 1983, 52, 529−562. the GITT technique. Indeed, good kinetics are obtained at all rates for the electrode though the maximum rate was limited to only C/5 in this experiment. CONCLUSIONS The electrochemical behavior of a composite electrode made of La2MgNi9 has been thoroughly investigated using deuterated samples by in situ neutron powder diffraction at different charge−discharge rates. From the data analysis, combining diffraction and PCT measurements, the mechanisms and the kinetics of the electrode have been determined. A fairly good agreement is observed between the solid−gas and electrochemical capacities. The electrode material works by following a hydride to intermetallic transformation through a β solid solution domain and a β-to-α transformation with low capacity attributed to the α solid solution domain. However, during charge, heavy line broadening is observed, which might be related to the formation of an intermediate γ phase as previously observed in the LaNi5 systems. The electrochemical reaction easily follows the current variations at all rates, indicating no kinetic limitation of hydrogen exchange in the materials between C/10 and C/5. However, for the studied bulky electrode, part of the reversible capacity is lost because of (a) formation of electrochemically stable α H solid solution which contains up to 1 H atom/f.u. and (b) incompleteness of the conversion of the metal hydride anode electrode alloy into the β hydride phase during the charging process. Thus, care should be taken to achieve efficient performance during scaling up of the metal hydride electrodes. ■ ■ ASSOCIATED CONTENT S Supporting Information * Crystallographic data taken from ref 13 for La2MgNi9 (R3̅m; N°166) and from ref 19 for La2MgNi9D13 (R3̅m; N°166). This material is available free of charge via the Internet at http:// pubs.acs.org. ■ AUTHOR INFORMATION Corresponding Author *E-mail: latroche@icmpe.cnrs.fr. Tel.: +33 1 49 78 12 10. Notes The authors declare no competing financial interest. ACKNOWLEDGMENTS This work was financially supported by the project NOVEL MAGnesium based nanomaterials for advanced rechargeable batteries (NOVELMAG) in the frame of the ERA.Net RUS FP7 Programme 225. ■ ■ REFERENCES (1) Latroche, M. Structural and Thermodynamic Properties of Metallic Hydrides Used for Energy Storage. J. Phys. Chem. Solids 2004, 65, 517−522. (2) Notten, P. H. L.; Latroche, M. Secondary Batteries: Nickel Batteries Metal Hydride Alloys. In Encyclopedia of Electrochemical Power Sources; Garche, J., Ed.; ; Elsevier: Amsterdam, 2009; pp 502− 521. (3) Liu, Y.; Pan, H.; Gao, M.; Wang, Q. Advanced Hydrogen Storage Alloys for Ni/MH Rechargeable Batteries. J. Mater. Chem. 2011, 21, 4743−4755. 12169 dx.doi.org/10.1021/jp503226r | J. Phys. Chem. C 2014, 118, 12162−12169