The Influence of Pr and Nd Substitution on Hydrogen Storage Properties of Mechanically Alloyed (La,Mg)₂Ni₇-Type Alloys

Applied commercial materials: La powder—grated from rod (Alfa Aesar, 99.9%), Mg powder (Alfa Aesar, 45 µm, 99.8%), Ni powder (Aldrich, 5 µm, 99.99%), Pr powder—grated from rod (Alfa Aesar, 99.1%), Nd powder—grated from rod (Alfa Aesar, 99.9%).

All electrochemical measurements were taken in deaerated 6 M KOH solution prepared from the KOH flakes and 18 MΩ cm⁻¹ water.

The La_1.5−xPr_xMg_0.5Ni₇ and La_1.5−xNd_xMg_0.5Ni₇ alloy powders (x = 0, 0.25, 0.5, 1) were prepared by the MA and annealing processes. The elemental powders were weighted, blended and poured into vials in a glove box filled with the argon atmosphere. Mechanical alloying lasted 48 h in all cases and was carried out in the argon atmosphere. The ratio of the weight of the hard steel balls (12 mm in diameter) to the weight of powder was 4.2:1.

A heat treatment of the obtained as-milled powder materials was carried out in a furnace at 1123 K for 0.5 h. This process was conducted in high-purity argon.

The structure characterization of the obtained alloys was carried out to find a correlation between the structures of the alloys and their phase compositions with the observed hydrogenation and dehydrogenation characteristics.

A PANalytical Empyrean powder diffractometer (CuKα radiation, λ = 1.54056 Å) was used to investigate the phase compositions of the materials after synthesis. The XRD measurements were taken at a room temperature with a voltage and anode current of 45 kV and 40 mA, respectively. The XRD profiles were refined with the Rietveld method using the Maud software.

The microstructure morphology of the obtained materials was determined by scanning electron microscopy (SEM)—VEGA TS5135 TESCAN digital microscopy imaging.

All synthesized samples with the addition of carbonyl nickel powder (10 wt.%) were used as negative electrodes in the Ni-MH_x systems. It is noteworthy that in this paper, the discharge capacities have been presented per unit gram of the measured material. A detailed description of the electrochemical measurements was included in the authors’ previous work (Ref 8). The cycle stability of the materials was evaluated by the capacity retaining rate after 30th and 50th cycles: R = (C_x/C_max) × 100%, where the C_x and the C_max are discharge capacities at cycle x (30, 50) and maximum discharge capacity, respectively.

The obtained La-Mg-Ni-type alloys were studied by a conventional Sievert’s type PCI apparatus, Particulate Systems, HPVA-200 with a vacuum pump. The mass of the sample for each measurement run was approx. 0.6 g. Before the hydrogen sorption–desorption measurements, no activation method was used. All alloys were degassed under vacuum at 303 K for 1 h before the tests.

Firstly, three cycles of hydrogen absorption kinetics measurements were performed under the hydrogen pressure of 3 MPa at 303 K. These studies also provided information on the activation properties of these materials. Between the cycles, the alloys were degassed at 673 K. After the kinetic measurements, the degassed La-Mg-Ni-type materials were tested to obtain the pressure–composition–isotherm curves (at 303 K and under a hydrogen pressure range of up to approx. 7 MPa). A detailed description of the apparatus and the measurements has been presented in the authors’ previous work (Ref 17).

All of the La-Mg-Ni-type materials after the PCT tests were tested by a TA DSC Q20 scanning calorimeter (DSC). The tests were performed up to 873 K, with the heating rate of 10 K/min in the argon atmosphere.

Figure 1 presents the SEM images of the La_1.0Pr_0.5Mg_0.5Ni₇ alloy powders—one of the investigated (La,Mg)₂Ni₇-type alloys. All alloys are composed of irregular porous particles. Moreover, all materials are characterized by a bimodal particle size distribution. The size of the alloy particles varies approximately from 100 to 800 µm. The rest of the particles have a size lower than 10 µm. (Part of them are attached to the surface of the larger particles.)

The authors did not observe any significant differences between the morphologies of the studied samples. The grains refinement after the substitution of La with Pr reported by Zhang et al. (Ref 16) for the melt-spun A₂B₇-type alloys was not observed.

Figure 2 presents the XRD patterns of the La_1.5−xPr_xMg_0.5Ni₇ and La_1.5−xNd_xMg_0.5Ni₇ alloy powders (x = 0, 0.25, 0.5, 1) after 48 h of MA and an additional heat treatment at 1123 K for 0.5 h in high-purity argon. The most important data from the Rietveld refinement of the XRD patterns are presented in Table 1. Compared to the results reported by Xiangqian et al., the annealing process described in this paper appears to be more preferable. Alloys with the main (La,Mg)₂Ni₇-type phase are obtained after a heat treatment at lower temperatures. Moreover, the time of the heat treatment process is much shorter (Ref 9).

It can be easily seen that all of the samples had a multi-phase structure. The main phase in all alloys is the (La,Mg)₂Ni₇-type one having a hexagonal structure crystallizing in the P63/mmc space group (Ce₂Ni₇-type)—marked in Fig. 2 and Table 1 as La₂Ni₇-type (H). The (La,Mg)₂Ni₇-type phase is obtained by stacking two AB₅ slabs with one A₂B₄ (Ref 1, 20). According to all the XRD patterns, no separate La, Pr, Nd, Mg or Ni elements were found.

In addition to the main (La,Mg)₂Ni₇-type phase, the unmodified La_1.5Mg_0.5Ni₇ alloy is composed of two minor phases. The first one is the LaNi₅-type phase having a hexagonal structure crystallizing in the P6/mmm space group (CaCu₅-type). The second one is the rhombohedral (La,Mg)₂Ni₇-type phase crystallizing in the R-3 m space group—marked in Fig. 2 and Table 1 as La₂Ni₇-type (R). The same phase composition was obtained by Li et al. (Ref 1) for inductive melted La_0.60R_0.20Mg_0.20(NiCoMnAl)_3.5 alloys (R = La, Ce, Pr, Nd). However, the phase abundances are different than those presented in Table 1, which most likely results from different chemical compositions.

A partial substitution of La by the Pr atoms caused changes in the crystal structure of the alloys. The rhombohedral (La,Mg)₂Ni₇-type phase is replaced with the LaNi₃-type phase that crystallizes in the same space group (R-3 m). The phase abundance of the main hexagonal (La,Mg)₂Ni₇-type phase increased, while the LaNi₃-type phase decreased with increasing Pr content in the material. Moreover, the LaNi₃-type phase is replaced by the PrMgNi₄-type phase in La_0.5PrMg_0.5Ni₇. The phase abundance of the LaNi₅-type phase remains at the same level (1-2 wt.%) regardless of the Pr content in the alloy.

The lattice parameters of all the (La,Mg)₂Ni₇-type, LaNi₅-type and LaNi₃-type phases were reduced with increasing Pr content in the alloy. It is due to the smaller atomic radius of Pr (0.1828 nm) compared to La (0.1877 nm).

A partial substitution of La by the Nd atoms caused changes in the minor phase composition of the alloys. The LaNi₅-type and the rhombohedral (La,Mg)₂Ni₇-type phases forming the unmodified La_1.5Mg_0.5Ni₇ alloy are replaced by the La₅Ni₁₉-type phase in the Nd-containing alloys. It is noteworthy that the La₅Ni₁₉ theoretical hydrogen storage capacity is lower than that of La₂Ni₇. On the other hand, this phase is characterized by good catalytic properties (Ref 11). The phase abundance of the hexagonal (La,Mg)₂Ni₇-type phase increased, while the La₅Ni₁₉-type phase decreased with the increasing Nd content in the La_1.5−xNd_xMg_0.5Ni material. Based on the XRD studies, we can conclude that the maximum content of the hexagonal La₂Ni₇-type phase is observed for the La_0.5Nd₁Mg_0.5Ni₇ alloy.

The lattice parameters of the hexagonal (La,Mg)₂Ni₇-type phase were firstly increased (for La_1.25Nd_0.25Mg_0.5Ni₇) and then decreased with increasing Nd content in the alloy. The authors think that the initial increase in the hexagonal (La,Mg)₂Ni₇-type phase lattice parameters was caused by the change of the phase compositions occurring after a partial substitution of La with the Nd atoms, as described above. As was shown, the La_1.5Mg₀₅Ni₇ alloy is composed of the hexagonal La₂Ni₇-type phase and two secondary phases. The authors think that until the material is partially composed of the secondary LaNi₅-type and/or rhombohedral La₂Ni₇-type phase, the chemical composition of the hexagonal La₂Ni₇-type is depleted in lanthanum. After a partial substitution of La with the Nd atoms, a change in the secondary phases was observed. This change could cause a release of some of the La atoms embedded in the structure of minor phases prior to chemical modification. After a partial substitution, these La atoms could be incorporated into the main phase taking up the other atoms’ place in the structure, thus creating the La-richer hexagonal La₂Ni₇-type phase. Because La has the largest atomic radius among the used elements, an increase in the lattice parameter was observed. To prove this claim, however, additional high-quality structure measurements have to be taken.

The lattice parameters of the secondary La₅Ni₁₉-type phase decreased with increasing Nd content in the alloy. As in the case of Pr, it is caused by a smaller atomic radius of Nd (0.1821 nm) compared to La (0.1877 nm). The same phenomenon was observed in the past for Nd- and Pr-substituted La-Ni alloys (Ref 9, 14).

Because the chemical modification of La_1.5Mg_0.5Ni₇ by Pr and Nd affected the lattice parameters of all phases, it can be stated that the Pr and Nd atoms co-form all of these phases. According to the work of Liu et al. (Ref 11), it can be concluded that the Pr and Nd atoms co-form an [AB₅] subunit instead of [A₂B₄]. However, when the [AB₅] subunit of the superlattice phase is saturated (especially for Pr-rich and Nd-rich samples), the extra Pr and Nd can co-form other minor phases.

Phase compositions of the described La_1.5−xNd_xMg_0.5Ni₇ alloys differ from the compositions obtained in the past for the melted La_0.8Nd_xMg_0.2Ni_3.1Co_0.25Al_0.15 samples (Ref 9). For these materials, instead of the La₅Ni₁₉-type minor phase, LaNi₅-type and LaMgNi₄-type phases were observed. It means that even a small chemical modification (in this case by the Co and Al atoms) can significantly affect the structure of the A₂B₇-type alloys (Ref 9).

Based on the above-described XRD studies, it can be concluded that the substitution of La by Pr and Nd did not affect the change of the major phase. The addition of Pr and Nd stabilizes and increases the hexagonal (La,Mg)₂Ni₇-type phase abundance (Ref 15).

The results of the electrochemical measurements taken on the La_1.5−xPr_xMg_0.5Ni₇ and La_1.5−xNd_xMg_0.5Ni₇ alloys are presented in Fig. 3. Moreover, the most important data from the electrochemical measurements are summarized in Table 2.

All of the electrodes displayed the maximum capacity at the first five cycles. The activation properties result from the phase composition and lattice parameters of the material. Such good activation properties were also observed for other La-Ni-type alloys, which means that this could be a common property of the La₂Ni₇-type materials (Ref 1, 9, 10).

The maximum discharge capacity of the La_1.5Mg_0.5Ni₇ alloy chemically modified by Pr is firstly reduced (for La_1.25Pr_0.25Mg_0.5Ni₇) and then increased to reach 308 mAh/g for the La_0.5PrMg_0.5Ni₇ alloy. An opposite trend was observed for alloys modified by Nd. The more the Nd in the alloy, the lower the discharge capacity. The best maximum discharge capacity (311 mAh/g) was obtained for the La_1.25Nd_0.25Mg_0.5Ni₇ alloy characterized by the highest La₅Ni₁₉-type phase content. As was stated above, the La₅Ni₁₉ phase is characterized by very good catalytic properties. Moreover, studies on the electrochemical performance of the A₂B₇-type single-phase La₃MgNi₁₄ alloy revealed that both the (La,Mg)₅Ni₁₉ and the LaNi₅ phases have a catalytic effect on the fast discharge of the alloy (Ref 11).

The discharge capacities result from several factors such as phase composition and phase abundances. Moreover, the hydrogen uptake during charging is also strongly related to the cell volume of the main phase. A reduction of the cell volume may cause a decrease in the interstitial sites for hydrogen and thus reduce the discharge capacity. In this work, the authors showed that the chemical modification of the La_1.5Mg_0.5Ni₇ alloy affects all the said factors: phase composition, phase abundances and cell volumes. The measured discharge capacities result from all of the factors, but the exact impact of each factor is unknown.

A substitution of the La by Pr or Nd elements resulted in increased cycle stability of the electrodes. The best cycle stability after 50 cycles was obtained for the alloys with the highest Pr and Nd content. After 50 cycles, the cycle stability of MA and annealed La_0.5PrMg_0.5Ni₇ and La_0.5NdMg_0.5Ni₇ reached 64 and 71%, respectively. As was shown in the past, a partial substitution of La with rare earth elements is an effective way to improve the cycle life and high rate dischargeability.

The discharge capacities of all the studied alloys were degraded with the charge/discharge cycles. Most often, it was due to the following reasons:

Because the Nd and Pr addition suppresses the oxidation to the hydrogen-absorbing elements such as Mg and La, it can be concluded that the pulverization should be the dominant factor of the reduction of the discharge capacity (Ref 14). It is noteworthy that the cell volume expansion rate of the LaNi₅ phase is different from those of the superstacking phases—the discrete expansion leads to a serious pulverization and capacity degradation (Ref 11).

Manometric, Sievert’s type device was used to measure PCI and the kinetic properties of all the studied alloys. The most important data obtained from these studies are summarized in Table 3.

Figure 4 presents the time–capacity curves of the La_1.5−xPr_xMg_0.5Ni₇ and La_1.5−xNd_xMg_0.5Ni₇ alloys in the activation/kinetic measurements obtained at 303 K. These studies were conducted without any activation procedure. It can be seen that all of the alloys need to be activated before they obtain the best kinetic properties. Moreover, for most of the samples the hydrogenation process in the first cycle was initiated after some incubation time. (The only exception was the LaPr_0.5Mg_0.5Ni alloy.) In the following cycles, the hydrogen sorption started immediately after filling the reaction chamber with hydrogen. The best hydrogenation properties were obtained in the second or third cycle. This means that all of the materials need to be activated. However, the process of activation is much easier compared to what described by Zhang et al. (Ref 15) described for inductive melted Pr-Mg-Ni alloys—those samples were four times hydrided (2 MPa, 1 h) and dehydrided (0.0001 MPa, 1 h) at 318 K.

The chemical substitution of La with Pr and Nd affected the kinetics of the hydrogen sorption. A partial replacement of La with the Nd atoms is effective in the reduction of time needed for the hydrogen absorption. Time needed to reach 95% of the maximum hydrogen capacity is reduced from 13 min for unmodified La_1.5Mg_0.5Ni₇ to 4-5 min for the Nd-containing alloys. Such good properties should be connected to the presence of the La₅Ni₁₉-type minor phase characterized by good catalytic properties.

The maximum hydrogen storage capacity strongly resulting from the structure and phase composition of the alloy was reduced after a chemical modification by Pr and Nd. This reduction is caused by the shrinkage of the cell volume caused by a partial substitution of La with the Pr and Nd atoms (Table 1). In smaller cells, the interstitial sites for hydrogen are smaller and the insertion of hydrogen is harder, thus leading to the reduction of the hydrogen storage capacity. The highest value was obtained for the unmodified La_1.5Mg_0.5Ni₇ alloy—1.6 wt.%.

Figure 5 presents a comparison of the PCI curves obtained for all the studied alloys. These curves show that all materials actively absorb hydrogen at 303 K. For all samples, an absorption plateau region (which is very flat) was observed. This region corresponds to the transformation from the α-solid solution of hydrogen in the intermetallic alloy to the β-hydride phase. Despite the fact that the alloys are multi-phase, only one plateau region is visible for each sample. It is caused by a low secondary phase abundance. Moreover, the equilibrium pressures of the A₂B₇ and A₅B₁₉ phases are very close due to their similar superstacking structures (Ref 11).

The plateau pressure heavily depends on the stability of the metal hydride, which results from the unit cell volume. It was previously shown that increasing the Mg content in the La_2−xMg_xNi₇ alloy leads to an increase in the plateau pressure until it reaches the level of 10 MPa for x = 1.0 (Ref 14).

The plateau pressure (sorption process) observed on the PCT curves of the La_1.5−xPr_xMg_0.5Ni₇ alloys increases with increasing Pr content in the sample, to reach the highest value for La_0.5PrMg_0.5Ni₇. This increase is related to the reduction of the cell volume—in smaller cells the interstitial sites for hydrogen are smaller and the insertion of hydrogen is harder. Therefore, the plateau pressure increases. The result is a decreased stability of the hydride phase. For La_0.5PrMg_0.5Ni₇, a decrease in the plateau pressure was observed—as caused by the change in the phase composition.

The chemical modification of the (La,Mg)₂Ni₇-type alloy by the Nd atoms also caused a change in the plateau pressure firstly decreased for La_1.25Nd_0.25Mg_0.5Ni₇ (caused by the change in the phase composition—replacement of the LaNi₅-type and rhombohedral La₂Ni₇-type minor phases by the La₅Ni₁₉-type minor phase) and then increased with increasing content of the Nd atoms in the La_1.5−xNd_xMg_0.5Ni₇ alloys. This increase was caused by the shrinkage of the cell volume (as was in the case of the modification by Pr described above).

The sorption plateau pressure for all of the studied samples has a level close to the atmospheric pressure, which is best in terms of their application. Due to the very low pressure of the desorption plateau for most of the samples, it is not possible to determine the reversibility of the hydriding–dehydriding process. (The low-pressure end of the plateau region is not visible.) Between the absorption and the desorption curves, a hysteresis appeared. It is most likely due to the transformation strains resulting from the cell volume expansion and the contraction (that cannot be relaxed) during the formation of hydrides formation and the decomposition.

The observed hysteresis can be used to predict the pulverization rate of the alloy during cycling. Alloys with a larger hysteresis have higher pulverization rates during the hydride/dehydride cycling (Ref 15).

All the materials after the PCI measurements were also studied by the DSC method—the obtained profiles are presented in Fig. 6. Additionally, the hydrogen desorption peak temperatures are listed in Table 3. All of the curves were characterized by one endothermic peak corresponding to the desorption of hydrogen from the alloys (hydrogen that was undesorbed during the PCI measurements). It can be seen that the hydrogen desorption peak temperature heavily depends on the phase composition. A partial substitution of La with Pr caused an increase in the hydrogen desorption peak temperature. The hydride stability of the Nd-containing alloys increased with the higher content of the La₅Ni₁₉-type minor phase.