The Importance of Rare-Earth Additions in Zr-Based AB 2 Metal Hydride Alloys

Effects of substitutions of rare earth (RE) elements (Y, La, Ce, and Nd) to the Zr-based AB2 multi-phase metal hydride (MH) alloys on the structure, gaseous phase hydrogen storage (H-storage), and electrochemical properties were studied and compared. Solubilities of the RE atoms in the main Laves phases (C14 and C15) are very low, and therefore the main contributions of the RE additives are through the formation of the RENi phase and change in TiNi phase abundance. Both the RENi and TiNi phases are found to facilitate the bulk diffusion of hydrogen but impede the surface reaction. The former is very effective in improving the activation behaviors.  ́40  ̋C performances of the Ce-doped alloys are slightly better than the Nd-doped alloys but not as good as those of the La-doped alloys, which gained the improvement through a different mechanism. While the improvement in ultra-low-temperature performance of the Ce-containing alloys can be associated with a larger amount of metallic Ni-clusters embedded in the surface oxide, the improvement in the La-containing alloys originates from the clean alloy/oxide interface as shown in an earlier transmission electron microscopy study. Overall, the substitution of 1 at% Ce to partially replace Zr gives the best electrochemical performances (capacity, rate, and activation) and is recommended for all the AB2 MH alloys for electrochemical applications.


Results and Discussion
Results from the Y- [7], La- [17], and Nd-[18] substitutions have been previously published.In this paper, new data from the Ce-substitution are organized and presented.Five alloys with different Ce-contents were made by arc melting.Their design compositions are listed in Table 2.The base Ce-free alloy, Ce0 (Ti 12 Zr 22.8 V 10 Cr 7.5 Mn 8.1 Co 7.0 Ni 32.2 Co 0.4 ), is the same base alloy used in the La- [17] and Nd-[18] substitution studies and a high-rate derivative of the base alloy in the Y-substitution study, Y0 (Ti 12 Zr 21.5 V 10 Cr 7.5 Mn 8.1 Co 8.0 Ni 32.2 Sn 0.3 Co 0.4 ) [7].Compared to alloy Y0, alloy Ce0 contains 1.3 at% higher Zn-content, 1.0 at% lower Co-content, and no Sn.ICP results of the current batch of alloys (alloys Ce0 to Ce5) are also listed in Table 2 for comparison.Most samples show very close compositions to the design values with the exception of a slight deficiency in Cr observed in alloy Ce2.Inconsistent uniformity of Cr in the AB 2 MH alloy has been seen previously [17].Comparing among Zr, V, and Cr, which are the higher-content elements with the highest melting points, Cr has the highest eutectic temperature with Ni (1345 ˝C, 960 ˝C for Zr-Ni, and 1202 ˝C for V-Ni) [38].Therefore, Cr's distribution may not be as consistently uniform compared to the other constituent elements among alloys.In combination with the small sample size for ICP, the sampling from alloy Ce2 may exclude a Cr-rich region that is not dissolved in the main phases and results in the observed lower Cr-content compared to the design composition.Average electron densities (e/a) of these alloys are just slightly under the C14/C15 threshold (6.83) [54], so a C14-predominating microstructure is predicted.The B/A ratios are slightly over the design value (1.87) due to the loss of Zr during slag formation.The slight hypo-stoichiometry design is used to balance between the degree of disorder (DOD, from the secondary phase abundances) and electrochemical properties [55].

X-Ray Diffraction Analysis
XRD analysis was used to study the microstructures of the alloys.The obtained XRD patterns are shown in Figure 2 with four phases identified: C14, C15, CeNi, and TiNi.Full XRD pattern fitting was performed using the Rietveld refinement and Jade 9 Software to obtain the lattice parameters and crystallite size of the main C14 phase and the abundances of all four phases, and the results are summarized in Table 3.With the increase in Ce-content (and the corresponding decrease in Zr-content) in the alloy, both lattice constants a and c in the C14 phase decrease, and the a/c ratio increases slightly (Figure 3).Lattice constant a of the C15 phase follows the same trend as the one in the C14 phase, and both decreasing trends are caused by the reduction in Zr-content (second largest among all constituent elements) and zero or close to zero solubility of Ce (largest among all constituent elements) in the C14 and C15 phases as the Ce-content in the alloy design increases.Un-shifted positions of the major peaks for the TiNi and CeNi phases, shown in Figure 2, indicate that the unit cell sizes of these two phases stay the same as the Ce-content increases in the alloy design, which can be explained by the relatively unchanged TiNi and CeNi phase compositions as revealed by the EDS analysis (Section 4.2).Crystallite size of the main C14 phase first increases with the initial introduction of Ce in the alloy but then decreases with further increases in Ce-content, and this trend is very similar to that observed in the La-substitution study [17].One possible explanation for the C14 crystallize size evolution is by the phase abundance of the non-Laves secondary phase.The higher abundance of non-Laves phase reduces the crystallization time of the matrix phase and consequently reduces the crystallite size of the matrix phase.In the beginning, when the CeNi phase abundance is small, the C14 crystallite size increases due to the decrease in the TiNi abundance.After more Ce is added, the CeNi phase abundance started to increase and caused a decrease in the C14 crystallite size.
Batteries 2016, 2, 25 5 of 19 be explained by the relatively unchanged TiNi and CeNi phase compositions as revealed by the EDS analysis (Section 4.2).Crystallite size of the main C14 phase first increases with the initial introduction of Ce in the alloy but then decreases with further increases in Ce-content, and this trend is very similar to that observed in the La-substitution study [17].One possible explanation for the C14 crystallize size evolution is by the phase abundance of the non-Laves secondary phase.The higher abundance of non-Laves phase reduces the crystallization time of the matrix phase and consequently reduces the crystallite size of the matrix phase.In the beginning, when the CeNi phase abundance is small, the C14 crystallite size increases due to the decrease in the TiNi abundance.After more Ce is added, the CeNi phase abundance started to increase and caused a decrease in the C14 crystallite size.Table 3. Lattice constants a and c, a/c ratio, unit cell volume (VC14), full-width at half-maximum (FWHM) for the (103) peak, and crystallite size for the C14 phase, lattice constant a for the C15 phase, and phase abundances in wt% calculated from the XRD analysis.Evolutions of the TiNi and RENi (where RE = Y, La, Ce, and Nd) phase abundances with increasing RE-content in the alloy design are plotted in Figure 4 and demonstrate that larger amounts of Ce (>1 at%) is very effective in suppressing the TiNi phase (beneficial to low-temperature performance [7,8]) and promoting the RENi phase (detrimental to low-temperature performance in the case of YNi [7,8]).Furthermore, the RENi phase abundances in various RE-substituted alloys do not correlate very well with the heats of formation or melting temperatures of the corresponding RENi phases.Table 3. Lattice constants a and c, a/c ratio, unit cell volume (V C14 ), full-width at half-maximum (FWHM) for the (103) peak, and crystallite size for the C14 phase, lattice constant a for the C15 phase, and phase abundances in wt% calculated from the XRD analysis.Evolutions of the TiNi and RENi (where RE = Y, La, Ce, and Nd) phase abundances with increasing RE-content in the alloy design are plotted in Figure 4 and demonstrate that larger amounts of Ce (>1 at%) is very effective in suppressing the TiNi phase (beneficial to low-temperature performance [7,8]) and promoting the RENi phase (detrimental to low-temperature performance in the case of YNi [7,8]).

Structural
Furthermore, the RENi phase abundances in various RE-substituted alloys do not correlate very well with the heats of formation or melting temperatures of the corresponding RENi phases.

Scanning Electron Microscope/Energy Dispersive Spectroscopy Analysis
Both SEM secondary electron images (SEI) and back-scattering electron images (BEI) were taken from each alloy in this study.While the former carry surface topological information, the latter demonstrate both topological and compositional information and are shown in Figure 5  ) is always greater than 1 even with the assignment of V in the A-site [16][17][18].The Ti-Ni binary phase diagram shows a solubility range of TiNi on the Ni-rich side at higher temperatures [38].Therefore, a higher B/A ratio for the TiNi phase is expected due to the quick cooling nature of arc melting technique.The main phase is composed of two regions with very close contrasts.The slightly brighter region (Figure 5a-3,b-3,c-3,d-3,e-3) has a smaller B/A ratio and an e/a value higher than the C14/C15 threshold of 6.83 [54], and it was consequently assigned to the C15 phase.The other phase (Figure 5a-4,b-4,c-4,d-4,e-4) with an e/a value lower than the C14/C15 threshold was assigned to the C14 phase.Since the solubility of Ce, same as those of the other RE elements, in the AB2 phase is almost zero, the lattice constants of the

Scanning Electron Microscope/Energy Dispersive Spectroscopy Analysis
Both SEM secondary electron images (SEI) and back-scattering electron images (BEI) were taken from each alloy in this study.While the former carry surface topological information, the latter demonstrate both topological and compositional information and are shown in Figure 5  ) is always greater than 1 even with the assignment of V in the A-site [16][17][18].The Ti-Ni binary phase diagram shows a solubility range of TiNi on the Ni-rich side at higher temperatures [38].Therefore, a higher B/A ratio for the TiNi phase is expected due to the quick cooling nature of arc melting technique.The main phase is composed of two regions with very close contrasts.The slightly brighter region (Figure 5a-3,b-3,c-3,d-3,e-3) has a smaller B/A ratio and an e/a value higher than the C14/C15 threshold of 6.83 [54], and it was consequently assigned to the C15 phase.The other phase (Figure 5a-4,b-4,c-4,d-4,e-4) with an e/a value lower than the C14/C15 threshold was assigned to the C14 phase.Since the solubility of Ce, same as those of the other RE elements, in the AB2 phase is almost zero, the lattice constants of the

Scanning Electron Microscope/Energy Dispersive Spectroscopy Analysis
Both SEM secondary electron images (SEI) and back-scattering electron images (BEI) were taken from each alloy in this study.While the former carry surface topological information, the latter demonstrate both topological and compositional information and are shown in Figure 5  ) is always greater than 1 even with the assignment of V in the A-site [16][17][18].The Ti-Ni binary phase diagram shows a solubility range of TiNi on the Ni-rich side at higher temperatures [38].Therefore, a higher B/A ratio for the TiNi phase is expected due to the quick cooling nature of arc melting technique.The main phase is composed of two regions with very close contrasts.The slightly brighter region (Figure 5a-3,b-3,c-3,d-3,e-3) has a smaller B/A ratio and an e/a value higher than the C14/C15 threshold of 6.83 [54], and it was consequently assigned to the C15 phase.The other phase (Figure 5a-4,b-4,c-4,d-4,e-4) with an e/a value lower than the C14/C15 threshold was assigned to the C14 phase.Since the solubility of Ce, same as those of the other RE elements, in the AB 2 phase is almost zero, the lattice constants of the C14 and C15 phases do not increase with increasing overall Ce-content; on the contrary, they decrease due to the decrease in Zr/Ti ratio (Zr is larger than Ti) in both the C14 and C15 phases as seen in Table 4. Upon a closer examination of the C14 and C15 phase compositions, we found that the main differences occur in the V-and Ni-contents.More specifically, the Ni/V ratio in the C14 phase is lower compared to that in the C15 phase (a lower Ni-content and a higher V-content in the C14 phase compared to those in the C15 phase), which results in a lower e/a in the C14 phase.In addition, the C14 phase is predicted to have a larger H-storage capacity due to its higher V-content but also exhibit a lower electrochemical high-rate dischargeability (HRD) due to its lower Ni-content according to our previous findings from a phase contribution study on the AB 2 MH alloys [56].Finally, phase distribution analysis reveals the following solidification sequence: AB 2 phase first (M.P. « 1450 ˝C), followed by TiNi (1310 ˝C), Zr 7 Ni 10 (1158 ˝C), and finally CeNi (680 ˝C) as the last solid formed from the liquid.C14 and C15 phases do not increase with increasing overall Ce-content; on the contrary, they decrease due to the decrease in Zr/Ti ratio (Zr is larger than Ti) in both the C14 and C15 phases as seen in Table 4. Upon a closer examination of the C14 and C15 phase compositions, we found that the main differences occur in the V-and Ni-contents.More specifically, the Ni/V ratio in the C14 phase is lower compared to that in the C15 phase (a lower Ni-content and a higher V-content in the C14 phase compared to those in the C15 phase), which results in a lower e/a in the C14 phase.In addition, the C14 phase is predicted to have a larger H-storage capacity due to its higher V-content but also exhibit a lower electrochemical high-rate dischargeability (HRD) due to its lower Ni-content according to our previous findings from a phase contribution study on the AB2 MH alloys [56].Finally, phase distribution analysis reveals the following solidification sequence: AB2 phase first (M.P. ≈ 1450 °C), followed by TiNi (1310 °C), Zr7Ni10 (1158 °C), and finally CeNi (680 °C) as the last solid formed from the liquid.PCT isotherms were measured at 30, 60, and 90 ˝C, and results from the first two temperatures are shown in Figure 6.These isotherms are typical examples of multi-phase alloys with coherent synergetic effects among constituent phases [18].Gaseous phase H-storage characteristics obtained from the PCT isotherms are summarized in Table 5.Both the maximum and reversible H-storage capacities first increase (maximized at alloy Ce1) and then decrease as the Ce-content in the alloy increases.Compared to the other RE additives, as shown in Figure 7a, alloy Ce1 has the second highest maximum H-storage capacity (alloy Y4 with 4 at% Y has the highest maximum H-storage capacity capacity).It is interesting to observe that alloys with 5 at% La, Ce, and Nd show very similar maximum H-storage capacities.Since no obvious pressure plateau can be identified, desorption pressures at 0.75 wt% H-storage capacity were compared and used in the calculations for hysteresis, heat of hydride formation (∆H h ), and change in entropy (∆S h ).For the Ce-containing alloys, both the 30 ˝C and 60 ˝C desorption pressure at 0.75 wt% H-storage first decrease (minimized at alloy Ce1) and then increase as the Ce-content in the alloy increases, and this trend is opposite to the trend observed in H-storage capacity.Desorption pressure is plotted against maximum H-storage capacity for all four RE-substituted series of alloys in Figure 8a, and a trend emerges where alloy with a higher metal-hydrogen (M-H) bond strength demonstrates both a lower plateau pressure and a higher H-storage capability, which was previously proposed [19].This trend is also related to the evolution of the main phase unit cell volume as seen in Figure 8b, where a larger unit cell makes for a more stable hydride with a stronger M-H bond and increases the maximum H-storage capacity.Slope factor (SF) is defined as the ratio of the storage capacity between 0.01 MPa and 0.5 MPa to the total capacity in the desorption isotherm.This ratio can be used as an indicator for DOD in a multi-phase MH alloy system [6,20,57].A large SF corresponds to a flatter plateau and lower DOD (fewer components or variations) in the system.The 30 °C SFs of this series of alloys are similar with the exception of alloy Ce1, which has the lowest SF and thus the highest DOD.Hysteresis of the PCT isotherm is defined as ln(Pa/Pd), where Pa and Pd are the absorption and desorption equilibrium pressures, respectively, at 0.75 wt% H-storage.Hysteresis is believed to be associated with the irreversible energy loss during plastic deformation of the hydride phase (β) in the alloy matrix (α) [58][59][60] and was found to be related to both the a/c ratio and pulverization rate of the alloy [61].PCT hystereses at 30 °C of the Ce-containing alloys are smaller than that of the Ce-free alloy Ce0, indicating that the addition of the CeNi phase can reduce the stress built from the α to β transformation; however, such effect was not observed with the other RE-substitutions (Figure 7b).60 °C H-storage characteristics are similar to those measured at 30 °C with the only exception of SF.SF measured at 60 °C decreases with increasing Ce-content in the alloy.One possible explanation is that the CeNi phase contributes more to DOD at a higher temperature with a higher participation rate (CeNi has a stronger M-H bond than AB2 and will not release hydrogen at room temperature [52]).
Desorption equilibrium pressures at 0.75 wt% H-storage at 30, 60, and 90 °C were used to estimate ΔHh and ΔSh with the equation: Slope factor (SF) is defined as the ratio of the storage capacity between 0.01 MPa and 0.5 MPa to the total capacity in the desorption isotherm.This ratio can be used as an indicator for DOD in a multi-phase MH alloy system [6,20,57].A large SF corresponds to a flatter plateau and lower DOD (fewer components or variations) in the system.The 30 °C SFs of this series of alloys are similar with the exception of alloy Ce1, which has the lowest SF and thus the highest DOD.Hysteresis of the PCT isotherm is defined as ln(Pa/Pd), where Pa and Pd are the absorption and desorption equilibrium pressures, respectively, at 0.75 wt% H-storage.Hysteresis is believed to be associated with the irreversible energy loss during plastic deformation of the hydride phase (β) in the alloy matrix (α) [58][59][60] and was found to be related to both the a/c ratio and pulverization rate of the alloy [61].PCT hystereses at 30 °C of the Ce-containing alloys are smaller than that of the Ce-free alloy Ce0, indicating that the addition of the CeNi phase can reduce the stress built from the α to β transformation; however, such effect was not observed with the other RE-substitutions (Figure 7b).60 °C H-storage characteristics are similar to those measured at 30 °C with the only exception of SF.SF measured at 60 °C decreases with increasing Ce-content in the alloy.One possible explanation is that the CeNi phase contributes more to DOD at a higher temperature with a higher participation rate (CeNi has a stronger M-H bond than AB2 and will not release hydrogen at room temperature [52]).
Desorption equilibrium pressures at 0.75 wt% H-storage at 30, 60, and 90 °C were used to estimate ΔHh and ΔSh with the equation: Slope factor (SF) is defined as the ratio of the storage capacity between 0.01 MPa and 0.5 MPa to the total capacity in the desorption isotherm.This ratio can be used as an indicator for DOD in a multi-phase MH alloy system [6,20,57].A large SF corresponds to a flatter plateau and lower DOD (fewer components or variations) in the system.The 30 ˝C SFs of this series of alloys are similar with the exception of alloy Ce1, which has the lowest SF and thus the highest DOD.Hysteresis of the PCT isotherm is defined as ln(P a /P d ), where P a and P d are the absorption and desorption equilibrium pressures, respectively, at 0.75 wt% H-storage.Hysteresis is believed to be associated with the irreversible energy loss during plastic deformation of the hydride phase (β) in the alloy matrix (α) [58][59][60] and was found to be related to both the a/c ratio and pulverization rate of the alloy [61].PCT hystereses at 30 ˝C of the Ce-containing alloys are smaller than that of the Ce-free alloy Ce0, indicating that the addition of the CeNi phase can reduce the stress built from the α to β transformation; however, such effect was not observed with the other RE-substitutions (Figure 7b).60 ˝C H-storage characteristics are similar to those measured at 30 ˝C with the only exception of SF.SF measured at 60 ˝C decreases with increasing Ce-content in the alloy.One possible explanation is that the CeNi phase contributes more to DOD at a higher temperature with a higher participation rate (CeNi has a stronger M-H bond than AB 2 and will not release hydrogen at room temperature [52]).
Desorption equilibrium pressures at 0.75 wt% H-storage at 30, 60, and 90 ˝C were used to estimate ∆H h and ∆S h with the equation: where R is the ideal gas constant, and T is the absolute temperature at which the measurement was performed.∆H h is negative due to the exothermic nature of the hydrogenation reaction.∆H h increases (becomes less negative) with increasing Ce-content in the alloy and can be correlated to the shrinking C14 unit cell volume, resulting in a less stable hydride (lower capacity and higher equilibrium pressure).∆S h is related to DOD in a hydride from a completely ordered solid (e.g., solid hydrogen) (´130.7 J¨mol ´1¨K ´1 for H 2 (g) at 300 K and 0.1 MPa [27]).The Ce-containing alloys have slightly higher ∆S h (less negative), suggesting that the addition of the CeNi phase increases DOD of the hydride, which is another indication of the CeNi phase's participation in H-storage.

Electrochemical Analysis
Activation characteristics were studied by half-cell capacity measurements in a flooded configuration.Electrode was made from ground and sieved powder without any annealing or surface acid/alkaline treatments.Low-rate capacities (sum of discharge capacity of 50 mA¨g ´1 and discharge capacities from two additional pulls at 12 mA¨g ´1 and 4 mA¨g ´1) and HRDs (ratio of discharge capacity at 50 mA¨g ´1 to that at 4 mA¨g ´1) for the first 13 cycles for each alloy in this study are plotted in Figure 9a,b, respectively.Activations in both capacity and HRD were facilitated by the addition of Ce, and therefore alloys with higher Ce-content activated faster.The improvement in the activation by the addition of RE element is commonly seen and was attributed to the formation of RENi secondary phase with a higher solubility in KOH solution [8,[16][17][18].Both the low-rate discharge capacities and HRDs of the two base alloys (alloys Y0 and Ce0) and alloys with 1 at% RE are compared in Figure 10a,b, respectively.For the low-rate capacity performance, alloy Y1 exhibits the easiest activation due to the ease of oxidation (Y has the lowest oxidation potential in Table 1), and alloy La1 shows a degradation in capacity as a result of the thick and passive surface oxide formation [62].For the HRD activation, which is more related to the new surface formation from cracking, alloys La1 and Ce1 show the fastest formations due to the relative high solubilities of lanthanum and cerium hydroxides in 30 wt% KOH electrolyte (pH « 14.7 in Equation ( 1)).Discharge capacities measured at 50 mA¨g ´1 and 4 mA¨g ´1 and HRD at the third cycle are listed in Table 6.As the Ce-content in the alloy increases, capacities measured at both rates first increase and then decreases while HRD continues to increase.The highest discharge capacity obtained from this series of alloys is from alloy Ce1 (400 mAh¨g ´1), which also has the highest gaseous phase storage capacity.Low-rate capacities and HRDs measured at the third cycle for alloys with different RE substitutions are compared in Figure 11a,b, respectively.Alloys Y5 and Ce1 show the highest and second highest low-rate capacities, respectively, and the La-and Ce-substitutions demonstrate the best HRD performances among all the RE-substitutions.Except for YNi, the RENi phases are beneficial for HRD with the trade-off occurring in capacity.By examining both the capacity and HRD results, it is concluded that alloy Ce1 has the best electrochemical properties among all the RE-substituted alloys.where R is the ideal gas constant, and T is the absolute temperature at which the measurement was performed.ΔHh is negative due to the exothermic nature of the hydrogenation reaction.ΔHh increases (becomes less negative) with increasing Ce-content in the alloy and can be correlated to the shrinking C14 unit cell volume, resulting in a less stable hydride (lower capacity and higher equilibrium pressure).ΔSh is related to DOD in a hydride from a completely ordered solid (e.g., solid hydrogen) (−130.7 J•mol −1 •K −1 for H2 (g) at 300 K and 0.1 MPa [27]).The Ce-containing alloys have slightly higher ΔSh (less negative), suggesting that the addition of the CeNi phase increases DOD of the hydride, which is another indication of the CeNi phase's participation in H-storage.

Electrochemical Analysis
Activation characteristics were studied by half-cell capacity measurements in a flooded configuration.Electrode was made from ground and sieved powder without any annealing or surface acid/alkaline treatments.Low-rate capacities (sum of discharge capacity of 50 mA•g −1 and discharge capacities from two additional pulls at 12 mA•g −1 and 4 mA•g −1 ) and HRDs (ratio of discharge capacity at 50 mA•g −1 to that at 4 mA•g −1 ) for the first 13 cycles for each alloy in this study are plotted in Figure 9a,b, respectively.Activations in both capacity and HRD were facilitated by the addition of Ce, and therefore alloys with higher Ce-content activated faster.The improvement in the activation by the addition of RE element is commonly seen and was attributed to the formation of RENi secondary phase with a higher solubility in KOH solution [8,[16][17][18].Both the low-rate discharge capacities and HRDs of the two base alloys (alloys Y0 and Ce0) and alloys with 1 at% RE are compared in Figure 10a,b, respectively.For the low-rate capacity performance, alloy Y1 exhibits the easiest activation due to the ease of oxidation (Y has the lowest oxidation potential in Table 1), and alloy La1 shows a degradation in capacity as a result of the thick and passive surface oxide formation [62].For the HRD activation, which is more related to the new surface formation from cracking, alloys La1 and Ce1 show the fastest formations due to the relative high solubilities of lanthanum and cerium hydroxides in 30 wt% KOH electrolyte (pH ≈ 14.7 in Equation ( 1)).Discharge capacities measured at 50 mA•g −1 and 4 mA•g −1 and HRD at the third cycle are listed in Table 6.As the Ce-content in the alloy increases, capacities measured at both rates first increase and then decreases while HRD continues to increase.The highest discharge capacity obtained from this series of alloys is from alloy Ce1 (400 mAh•g −1 ), which also has the highest gaseous phase storage capacity.Low-rate capacities and HRDs measured at the third cycle for alloys with different RE substitutions are compared in Figure 11a,b, respectively.Alloys Y5 and Ce1 show the highest and second highest low-rate capacities, respectively, and the Laand Ce-substitutions demonstrate the best HRD performances among all the RE-substitutions.Except for YNi, the RENi phases are beneficial for HRD with the trade-off occurring in capacity.By examining both the capacity and HRD results, it is concluded that alloy Ce1 has the best electrochemical properties among all the RE-substituted alloys.Table 6.Summary of electrochemical half-cell and magnetic measurements, including the capacities at the third cycle, HRD at the third cycle, number of cycles needed to achieve 92% HRD, bulk diffusion coefficient, surface exchange current, results from AC impedance measurements, saturated magnetic susceptibility, and applied field at half of the saturated magnetic susceptibility.Since the MH alloy can be charged either in the gaseous phase by reacting with hydrogen gas or in the electrochemical environment with protons from water splitting that occurs with voltage, it is always interesting to compare the capacities obtained by these two different paths.After converting the unit of gaseous phase H-storage to that of electrochemical discharge capacity, 1 wt% hydrogen content was found to be equivalent to a discharge capacity of 268 mAh•g −1 .Our earlier studies indicated that both the high-rate (50 mA•g −1 for AB2) and low-rate (4 mA•g −1 for AB2) electrochemical capacities are between the boundaries set by the gaseous phase maximum and reversible H-storage  Table 6.Summary of electrochemical half-cell and magnetic measurements, including the capacities at the third cycle, HRD at the third cycle, number of cycles needed to achieve 92% HRD, bulk diffusion coefficient, surface exchange current, results from AC impedance measurements, saturated magnetic susceptibility, and applied field at half of the saturated magnetic susceptibility.Since the MH alloy can be charged either in the gaseous phase by reacting with hydrogen gas or in the electrochemical environment with protons from water splitting that occurs with voltage, it is always interesting to compare the capacities obtained by these two different paths.After converting the unit of gaseous phase H-storage to that of electrochemical discharge capacity, 1 wt% hydrogen content was found to be equivalent to a discharge capacity of 268 mAh•g −1 .Our earlier studies indicated that both the high-rate (50 mA•g −1 for AB2) and low-rate (4 mA•g −1 for AB2) electrochemical capacities are between the boundaries set by the gaseous phase maximum and reversible H-storage Table 6.Summary of electrochemical half-cell and magnetic measurements, including the capacities at the third cycle, HRD at the third cycle, number of cycles needed to achieve 92% HRD, bulk diffusion coefficient, surface exchange current, results from AC impedance measurements, saturated magnetic susceptibility, and applied field at half of the saturated magnetic susceptibility.Since the MH alloy can be charged either in the gaseous phase by reacting with hydrogen gas or in the electrochemical environment with protons from water splitting that occurs with voltage, it is always interesting to compare the capacities obtained by these two different paths.After converting the unit of gaseous phase H-storage to that of electrochemical discharge capacity, 1 wt% hydrogen content was found to be equivalent to a discharge capacity of 268 mAh¨g ´1.Our earlier studies indicated that both the high-rate (50 mA¨g ´1 for AB 2 ) and low-rate (4 mA¨g ´1 for AB 2 ) electrochemical capacities are between the boundaries set by the gaseous phase maximum and reversible H-storage capacities but do not necessarily exhibit strong correlations with the gaseous phase capacities [19].In order to investigate the connection between the gaseous phase and electrochemical measurements, low-rate discharge capacity vs. maximum H-storage capacity and high-rate discharge capacity vs. reversible H-storage capacity for the RE-substituted AB 2 MH alloys are plotted in Figure 12a,b, respectively.From these two subfigures, linear correlations can be established in both cases, and the electrochemical discharge capacities occur between the boundaries set by the maximum and reversible H-storage capacities.It is worthwhile to point out that the Ce-containing alloys have the highest reversible H-storage and high-rate discharge capacities as shown in Figure 12b.

Electrochemical and Magnetics
Batteries 2016, 2, 25 13 of 19 capacities but do not necessarily exhibit strong correlations with the gaseous phase capacities [19].In order to investigate the connection between the gaseous phase and electrochemical measurements, low-rate discharge capacity vs. maximum H-storage capacity and high-rate discharge capacity vs. reversible H-storage capacity for the RE-substituted AB2 MH alloys are plotted in Figure 12a,b, respectively.From these two subfigures, linear correlations can be established in both cases, and the electrochemical discharge capacities occur between the boundaries set by the maximum and reversible H-storage capacities.It is worthwhile to point out that the Ce-containing alloys have the highest reversible H-storage and high-rate discharge capacities as shown in Figure 12b.Origins of the variations observed in HRD can be studied by the changes in bulk diffusion coefficient (D) and surface reaction current (Io) measured electrochemically.Details of the measurements and data analysis can be found in our earlier publication [17].Both D and Io of the Cecontaining AB2 MH alloys were measured, and the results are listed in Table 6.In general, both quantities increase and then decrease with increasing Ce-content in the alloy with the exception of Io from alloy Ce4.Both trends are different from the monotonic increase seen in HRD.Therefore, other Origins of the variations observed in HRD can be studied by the changes in bulk diffusion coefficient (D) and surface reaction current (I o ) measured electrochemically.Details of the measurements and data analysis can be found in our earlier publication [17].Both D and I o of the Ce-containing AB 2 MH alloys were measured, and the results are listed in Table 6.In general, both quantities increase and then decrease with increasing Ce-content in the alloy with the exception of I o from alloy Ce4.Both trends are different from the monotonic increase seen in HRD.Therefore, other factors may also play important roles in affecting HRD for this series of alloys.Comparing among various RE-substitutions, D increases with increasing La-, Y-, and Ce-contents but remains low in the case of Nd-substitution (Figure 13a), and I o increases with a small amount of Ce, La, and Y but decreases with a small amount of Nd (Figure 13b).Alloy Ce1 shows the highest I o , which can be attributed to its unique microstructure that contains Ce metal instead of the CeNi phase as in the alloys with higher Ce-contents (Table 4).It appears that Ce metal is more reactive compared to CeNi with KOH electrolyte and thus contributes more to the surface reaction.
In order to study the roles of the TiNi and RENi phases in D and I o , four sets of correlations are shown in Figures 14 and 15.In general, both the TiNi and RENi phase abundances show positive correlations to D with the exception of the Nd-containing alloys, which have the highest RENi-and TiNi-contents but also the lowest Ds (Figure 4).It is possible that the TiNi and RENi phases are beneficial to D only up to certain percentages (approximately 10 wt% and 5 wt% for TiNi and RENi, respectively).Correlations with the TiNi and RENi phase abundances for I o are less obvious, but the general decreasing trends with increasing RENi and TiNi phase abundances are observed (Figure 15).In an earlier effort to differentiate the increasing rates of the YNi and TiNi phases, an off-stoichiometric composition design was adopted [8].In that study, the increase in YNi phase abundance was suppressed, and I o increased with the increase in TiNi phase abundance (data sets with symbol * in Figures 14 and 15).Origins of the variations observed in HRD can be studied by the changes in bulk diffusion coefficient (D) and surface reaction current (Io) measured electrochemically.Details of the measurements and data analysis can be found in our earlier publication [17].Both D and Io of the Cecontaining AB2 MH alloys were measured, and the results are listed in Table 6.In general, both quantities increase and then decrease with increasing Ce-content in the alloy with the exception of Io from alloy Ce4.Both trends are different from the monotonic increase seen in HRD.Therefore, other factors may also play important roles in affecting HRD for this series of alloys.Comparing among various RE-substitutions, D increases with increasing La-, Y-, and Ce-contents but remains low in the case of Nd-substitution (Figure 13a), and Io increases with a small amount of Ce, La, and Y but decreases with a small amount of Nd (Figure 13b).Alloy Ce1 shows the highest Io, which can be In order to study the roles of the TiNi and RENi phases in D and Io, four sets of correlations are shown in Figures 14 and 15.In general, both the TiNi and RENi phase abundances show positive correlations to D with the exception of the Nd-containing alloys, which have the highest RENi-and TiNi-contents but also the lowest Ds (Figure 4).It is possible that the TiNi and RENi phases are beneficial to D only up to certain percentages (approximately 10 wt% and 5 wt% for TiNi and RENi, respectively).Correlations with the TiNi and RENi phase abundances for Io are less obvious, but the general decreasing trends with increasing RENi and TiNi phase abundances are observed (Figure 15).In an earlier effort to differentiate the increasing rates of the YNi and TiNi phases, an offstoichiometric composition design was adopted [8].In that study, the increase in YNi phase abundance was suppressed, and Io increased with the increase in TiNi phase abundance (data sets with symbol * in Figures 14 and 15).Low-temperature performance of the alloys was studied by AC impedance measured at ´40 ˝C.Surface charge-transfer resistance (R) and surface double-layer capacitance (C) of each alloy were calculated from curve fitting of the Cole-Cole plot and are listed in Table 6.Very similar to the case of Nd-substitution, the Ce-containing alloys have much lower R compared to the RE-free base alloy (Figure 16a).The main reason for the reduction in R in the Ce-containing alloys is the increase in surface area represented by the increase in C as the Ce-content increases in the alloy (Figure 16b).RC product, a quantity inversely proportional to the surface catalytic ability, from the Ce-containing alloys are higher (less catalytic) than those from the La-and Nd-containing alloys.Low-temperature performance of the alloys was studied by AC impedance measured at −40 °C.Surface charge-transfer resistance (R) and surface double-layer capacitance (C) of each alloy were calculated from curve fitting of the Cole-Cole plot and are listed in Table 6.Very similar to the case of Nd-substitution, the Ce-containing alloys have much lower R compared to the RE-free base alloy (Figure 16a).The main reason for the reduction in R in the Ce-containing alloys is the increase in surface area represented by the increase in C as the Ce-content increases in the alloy (Figure 16b).RC product, a quantity inversely proportional to the surface catalytic ability, from the Ce-containing alloys are higher (less catalytic) than those from the La-and Nd-containing alloys.

Magnetic Properties
In most MH alloys, the HRD characteristics are heavily affected by the density of metallic Niinclusion embedded in the surface oxide [63][64][65][66].Thus far, the Si-containing AB2 MH alloys are the only type of alloy where the HRD performance is dominated by the clean oxide/metal interface but not the occurrence of metallic Ni [67].Both the total saturated magnetic susceptibility (Ms, related to the total volume of metallic Ni-clusters) and applied field corresponding to half of Ms (H1/2, inversely proportional to the average number of Ni atoms in a cluster) were calculated from the plot of measured M.S. versus applied magnetic field (as in [65]) and are listed in Table 6.As the Ce-content in the alloy increases, Ms increases, and H1/2 drops precipitously and then remains approximately the same.The Ce/CeNi phase in the Ce-containing alloys promotes the formation of metallic Ni-clusters embedded in the surface oxide with a size around two and half times larger than those in the Ce-free base alloy.When these two properties (Ms and H1/2) are compared among all the RE-substitutions, as shown in Figure 17, the Ce-doped alloys have the highest Ms values, which is in agreement of their high HRD (Figure 11b) and Io (Figure 13b) values.The La-substitution shows the lowest Ms but still exibits comparable HRD to that of the Ce-substitution due to the unique clean interface between the alloy and surface catalytic Ni(OH)2 layer [62].All the Ce-and Nd-doped alloys have approximately the same size of metallic Ni-clusters that are larger than those in the undoped base alloy(Figure 17b), which is very different from the unchanged size of Ni-clusters observed in the La-doped alloys.The relatively high solubility of the La ion promotes the formation of a catalytic Ni(OH)2 layer directly onto the metal surface instead of having the metallic Ni embedded in the surface oxide/hydroixde of Nd or Ce.

Magnetic Properties
In most MH alloys, the HRD characteristics are heavily affected by the density of metallic Ni-inclusion embedded in the surface oxide [63][64][65][66].Thus far, the Si-containing AB 2 MH alloys are the only type of alloy where the HRD performance is dominated by the clean oxide/metal interface but not the occurrence of metallic Ni [67].Both the total saturated magnetic susceptibility (M s , related to the total volume of metallic Ni-clusters) and applied field corresponding to half of M s (H 1/2 , inversely proportional to the average number of Ni atoms in a cluster) were calculated from the plot of measured M.S. versus applied magnetic field (as in [65]) and are listed in Table 6.As the Ce-content in the alloy increases, M s increases, and H 1/2 drops precipitously and then remains approximately the same.The Ce/CeNi phase in the Ce-containing alloys promotes the formation of metallic Ni-clusters embedded in the surface oxide with a size around two and half times larger than those in the Ce-free base alloy.When these two properties (M s and H 1/2 ) are compared among all the RE-substitutions, as shown in Figure 17, the Ce-doped alloys have the highest M s values, which is in agreement of their high HRD (Figure 11b) and I o (Figure 13b) values.The La-substitution shows the lowest M s but still exibits comparable HRD to that of the Ce-substitution due to the unique clean interface between the alloy and surface catalytic Ni(OH) 2 layer [62].All the Ce-and Nd-doped alloys have approximately the same size of metallic Ni-clusters that are larger than those in the undoped base alloy (Figure 17b), which is very different from the unchanged size of Ni-clusters observed in the La-doped alloys.The relatively high solubility of the La ion promotes the formation of a catalytic Ni(OH) 2 layer directly onto the metal surface instead of having the metallic Ni embedded in the surface oxide/hydroixde of Nd or Ce.

Conclusions
We have analyzed and compared the effects of various RE-substitutions to the Zr-based AB 2 MH alloys on the structure, gaseous phase H-storage, and electrochemical properties.The additions of RE elements were found to facilitate activation and improve the HRD and low-temperature electrochemical performance.Among all of the substitutions, the Ce-substitution was found to be the most effective in reducing the PCT isotherm hysteresis (improving the cycle stability), increasing the electrochemical capacity, and increasing the total volume of metallic Ni-clusters embedded in the surface oxide, reactive surface area, and consequently surface reaction current.Overall, our observations indicate that a 1 at% Ce-substitution in the AB 2 MH alloy can improve the electrochemical properties and is a viable option for Ni/MH rechargeable batteries.Applied magnetic field strength corresponding to half of saturated magnetic susceptibility

Figure 1 .
Figure 1.(a) CrB-and (b) FeB-type structures.Green and black circles represent the RE and Ni atoms, respectively.Both structures are composed of trigonal prisms containing four RE and four Ni atoms.While the RE atoms form a HCP stacking, the Ni atoms form a zigzag chain in both cases.Reprinted from [48].

Figure 1 .
Figure 1.(a) CrB-and (b) FeB-type structures.Green and black circles represent the RE and Ni atoms, respectively.Both structures are composed of trigonal prisms containing four RE and four Ni atoms.While the RE atoms form a HCP stacking, the Ni atoms form a zigzag chain in both cases.Reprinted from [48].

Batteries 2016, 2 , 25 6 of 19 Figure 3 .Figure 4 .
Figure 3. Evolutions of the C14 lattice constants a and c with increasing Ce-content in the design.
for comparison.Compositions of several representative areas (identified by Roman numerals) in Figure 5 were studied by EDS, and the results are summarized in Table 4.The brightest region in each micrograph belongs to Ce metal (Figure 5a-1 in alloy Ce1) or CeNi (Figure 5b-1,c-1,d-1,e-1 in Alloys Ce2-Ce5, respectively).These observations are in agreement with the close-to-zero CeNi-content in alloy Ce1 revealed by the XRD analysis.Areas with the second brightest contrast (Figure 5a-2,b-2,c-2,d-2,e-2) are identified as a Zr-rich phase with a B/A ratio very close to the theoretical stoichiometry of Zr7Ni10 (1.43) by taking into consideration that V resides in the A-site of the crystal [18].Solubilities of V, Co, Mn, and Cr in the Zr7Ni10 phase are much lower than those in the Laves phases.Moreover, the B/A ratio of the TiNi phase (Figure 5a-5,b-5,c-5,d-5,e-5

Figure 3 .of 19 Figure 3 .Figure 4 .
Figure 3. Evolutions of the C14 lattice constants a and c with increasing Ce-content in the design.
for comparison.Compositions of several representative areas (identified by Roman numerals) in Figure 5 were studied by EDS, and the results are summarized in Table 4.The brightest region in each micrograph belongs to Ce metal (Figure 5a-1 in alloy Ce1) or CeNi (Figure 5b-1,c-1,d-1,e-1 in Alloys Ce2-Ce5, respectively).These observations are in agreement with the close-to-zero CeNi-content in alloy Ce1 revealed by the XRD analysis.Areas with the second brightest contrast (Figure 5a-2,b-2,c-2,d-2,e-2) are identified as a Zr-rich phase with a B/A ratio very close to the theoretical stoichiometry of Zr7Ni10 (1.43) by taking into consideration that V resides in the A-site of the crystal [18].Solubilities of V, Co, Mn, and Cr in the Zr7Ni10 phase are much lower than those in the Laves phases.Moreover, the B/A ratio of the TiNi phase (Figure 5a-5,b-5,c-5,d-5,e-5

Figure 4 .
Figure 4. Evolutions of (a) TiNi and (b) RENi phase abundances with increasing RE-content in the design.
for comparison.Compositions of several representative areas (identified by Roman numerals) in Figure 5 were studied by EDS, and the results are summarized in Table 4.The brightest region in each micrograph belongs to Ce metal (Figure 5a-1 in alloy Ce1) or CeNi (Figure 5b-1,c-1,d-1,e-1 in Alloys Ce2-Ce5, respectively).These observations are in agreement with the close-to-zero CeNi-content in alloy Ce1 revealed by the XRD analysis.Areas with the second brightest contrast (Figure 5a-2,b-2,c-2,d-2,e-2) are identified as a Zr-rich phase with a B/A ratio very close to the theoretical stoichiometry of Zr 7 Ni 10 (1.43) by taking into consideration that V resides in the A-site of the crystal [18].Solubilities of V, Co, Mn, and Cr in the Zr 7 Ni 10 phase are much lower than those in the Laves phases.Moreover, the B/A ratio of the TiNi phase (Figure 5a-5,b-5,c-5,d-5,e-5

Figure 7 .Figure 8 .
Figure 7. Evolutions of (a) maximum H-storage capacity and (b) PCT absorption-desorption hysteresis with increasing RE-content in the design.

Figure 7 .Figure 7 .Figure 8 .
Figure 7. Evolutions of (a) maximum H-storage capacity and (b) PCT absorption-desorption hysteresis with increasing RE-content in the design.

Figure 8 .
Figure 8. Correlations between the maximum H-storage and (a) equilibrium pressure at 0.75 wt% H-storage; and (b) C14 phase unit cell volume.

Figure 9 .
Figure 9. Activation characteristics shown by (a) discharge capacities with the lowest discharge current (4 mA•g −1 ); and (b) high-rate dischargeabilities for the first 13 cycles for the alloys in this study.

Figure 9 .
Figure 9. Activation characteristics shown by (a) discharge capacities with the lowest discharge current (4 mA¨g ´1); and (b) high-rate dischargeabilities for the first 13 cycles for the alloys in this study.

Figure 11 .
Figure 11.Evolutions of (a) low-rate capacity and (b) HRD at the third cycle with increasing REcontent in the design.

Figure 11 .
Figure 11.Evolutions of (a) low-rate capacity and (b) HRD at the third cycle with increasing REcontent in the design.

Figure 11 .
Figure 11.Evolutions of (a) low-rate capacity and (b) HRD at the third cycle with increasing RE-content in the design.

Figure 12 .
Figure 12.Comparisons of the electrochemical discharge capacities and gaseous phase H-storages of the RE-containing MH alloys by plotting: (a) low-rate discharge capacity vs. maximum H-storage capacity; and (b) high-rate discharge capacity vs. reversible H-storage capacity.Data set YY is from a previous study on the combination of Y-addition and stoichiometry in the AB2 MH alloys [8].

Figure 12 .
Figure 12.Comparisons of the electrochemical discharge capacities and gaseous phase H-storages of the RE-containing MH alloys by plotting: (a) low-rate discharge capacity vs. maximum H-storage capacity; and (b) high-rate discharge capacity vs. reversible H-storage capacity.Data set YY is from a previous study on the combination of Y-addition and stoichiometry in the AB 2 MH alloys [8].

Figure 12 .
Figure 12.Comparisons of the electrochemical discharge capacities and gaseous phase H-storages of the RE-containing MH alloys by plotting: (a) low-rate discharge capacity vs. maximum H-storage capacity; and (b) high-rate discharge capacity vs. reversible H-storage capacity.Data set YY is from a previous study on the combination of Y-addition and stoichiometry in the AB2 MH alloys [8].

Figure 13 .
Figure 13.Evolutions of (a) bulk diffusion coefficient and (b) surface exchange current with increasing RE-content in the design.

Figure 16 .
Figure 16.Evolutions of (a) surface charge-transfer resistance; (b) double-layer capacitance calculated from the AC impedance results measured at ´40 ˝C; and (c) their product with increasing RE-content in the design.

Figure 17 .
Figure 17.Evolutions of (a) saturated magnetic susceptibility and (b) strength of the applied field corresponding to half of the saturated magnetic susceptibility with increasing RE-content in the design.

Figure 17 .
Figure 17.Evolutions of (a) saturated magnetic susceptibility and (b) strength of the applied field corresponding to half of the saturated magnetic susceptibility with increasing RE-content in the design.

Table 1 .
[27]erties of the rare earth (RE = Y, La, Ce, Pr, and Nd) elements, RENi and RENi5 intermetallic compounds.Data are from[27]unless otherwise cited.HCP and DHCP denote hexagonal close packed, and double-c hexagonal close packed, respectively.

Table 2 .
Design compositions (in bold) and inductively coupled plasma (ICP) results in at%.e/a is the average electron density, and B/A is the ratio of the B-atom-content (V, Cr, Mn, Co, Ni, and Al) to the A-atom-content (Ti, Zr, and Ce).

Table 4 .
Summary of the energy dispersive spectroscopy (EDS) results from several selective spots in the SEM-BEI micrographs shown in Figure5.All compositions are in at%.The main C14 phase is identified in bold.

Table 5 .
Summary of gaseous phase properties, including the maximum and reversible capacities, desorption pressure at 0.75 wt% H-storage capacity, slope factor (SF), hysteresis, and changes in enthalpy and entropy.

Table 5 .
Summary of gaseous phase properties, including the maximum and reversible capacities, desorption pressure at 0.75 wt% H-storage capacity, slope factor (SF), hysteresis, and changes in enthalpy and entropy.