Crystal Structural Investigations for Understanding the Hydrogen Storage Properties of YMgNi4-Based Alloys.

The hydrogen storage properties and crystal structures of YMgNi4-based alloys, which were synthesized from (2 - x)YNi2 and xMgNi2 (0.6 ≤ x ≤ 1.2), were investigated by pressure-composition-temperature measurements and powder neutron diffraction at a deuterium gas pressure to understand the hydrogen absorption and desorption reactions viewed from atomic arrangements around H atoms. Reducing the amounts of MgNi2, which was utilized as a Mg source in YMgNi4-based alloys, has been observed to lower the hydrogen absorption and desorption pressures and increase the hydrogen storage capacities. However, the reversible hydrogen capacity attained a maximum value of 1.2 mass % at x = 0.8 because of the formation of a thermodynamically stable hydride in which hydrogen was not released at x = 0.6. In the case of x = 0.6, the presence of excessive Y atoms around the H atoms in the hydrogen-absorbed phase would lead to the formation of a hydride with stronger interaction between Y and H because of the affinity between them. Moreover, the presence of small amounts of D atoms with short interatomic D-D distances (1.6 and 1.9 Å) in the deuterium-absorbed phase (Y0.81Mg1.19Ni4.00D3.35 and Y1.06Mg0.94Ni4.00D3.86) at <5 MPa and 323 K was proposed by the crystal structural investigations. The D atoms with short D-D interatomic distances were located in the same local atomic arrangements of D atoms in a deuterium-absorbed phase, which were formed at a higher-pressure range, and had higher hydrogen storage capacities than the deuterium-absorbed phases in this study.

Introduction

Hydrogen is a promising renewable energy carrier. Thus, it is necessary to develop efficient hydrogen storage technologies. Several approaches involving compressed H2 gas or liquid H2 as well as material-based storage, e.g., utilizing metal hydrides, have been proposed. In these methods, metal hydrides are used to store hydrogen with high gravimetric and volumetric hydrogen densities at moderate pressures and temperatures compared with compressed H2 gas and liquid H2.[1,2] To date, numerous metal hydrides have been reported as potential hydrogen storage materials. Among them, Mg, which is an abundant and light element, reacts with H2 to form MgH2 with high gravimetric and volumetric hydrogen densities, 7.7 mass % and 109 kgH2/m3, respectively. Therefore, Mg has been shown as a particularly promising metal for hydrogen storage. Although MgH2 has high hydrogen storage capacities, the thermodynamic stabilities and kinetics of hydrogen absorption and desorption reactions (e.g., enthalpy = −74 kJ/mol H2) are not suitable for the utilization of Mg as a hydrogen storage material.[3] Higher thermodynamic stabilities would relate to the ionic bonding in MgH2 that is composed of Mg2+ and H–. To improve the thermodynamic stability, the development of Mg alloys with transition metals, such as Ni, can be an effective approach. For example, RE(2 – MgNi4 (RE: rare-earth element, 0 < x < 2) has been applied to reversible hydrogen absorption and desorption reactions at ambient temperature.[4−11] In the case of RE(2 – MgNi4, the H atom is in the elemental state, which is qualitatively associated with less interaction between the metal elements and H than MgH2. Then, the subsequent hydrogen absorption reaction steps of LaMgNi4 at 373 K in a hydrogen (deuterium) atmosphere have been reported[6]

Figure shows the crystal structures of LaMgNi4 obtained during the hydrogen absorption reaction.[6,10]

Figure 1

Crystal structures of REMgNi4, the α-phase hydride (LaMgNi4H0.75 in the space group F4̅3m (No. 216)), the β-phase hydride (LaMgNi4H3.70 in the space group Pmn21 (No. 31)), and the γ-phase hydride (LaMgNi4H4.85 in the space group F4̅3m (No. 216)). The β-phase hydride exhibits similar metal atomic arrangements to the alloy, α-phase, or γ-phase hydrides. Two crystal structures are illustrated for the γ-phase hydride: LaMgNi4D4.85, in which the D atoms (D1γ and D2γ) are located at 24g (0.987, 0.25, 0.25) and 4b (0.5, 0.5, 0.5) in the space group F4̅3m (No. 216), and Pr0.6Mg1.4Ni4.0D3.6, in which Pr and some of the Mg atoms coexist at 4a (0, 0, 0), whereas the D atoms (D1γ and D2γ) are located at 48h (0.7363, 0.7363, 0.9979) and 16e (0.8493, 0.8493, 0.8493) in the space group F4̅3m (No. 216).

As it can be seen, LaMgNi4 adopted a face-centered cubic (FCC) AuBe5-type crystal structure, which resembles the structure of the C15-type Laves phase alloy (a MgCu2-type crystal structure). Neutron diffraction experiments at deuterium gas pressure revealed the initial formation of LaMgNi4D0.75 (gravimetric hydrogen density: 0.2 mass %), which adopted an FCC crystal structure while maintaining the Bravais lattice of LaMgNi4, referred to as the α-phase hydride in this study.[6] It was shown that the D atoms (D1α) were located at the tetrahedral sites and coordinated by four Ni atoms. In the second step, LaMgNi4D3.70 (gravimetric hydrogen density: 0.9 mass %) with a primitive orthorhombic crystal structure, which we refer to as the β-phase hydride in this study, was formed. In this crystal structure, the D atoms (D1β and D2β) were located inside of the triangular bipyramid and coordinated by two La, one Mg, and two Ni atoms. The remaining D atoms (D3β) were positioned inside of the tetrahedron and coordinated by one La and three Ni atoms. Furthermore, in the third step, LaMgNi4D4.85 (gravimetric hydrogen density: 1.2 mass %) with an FCC crystal structure, which we refer to as the γ-phase hydride in this paper, was formed. The D atoms (D1γ) were located inside of the triangular bipyramid and coordinated by two La, one Mg, and two Ni atoms. The remaining D atoms (D2γ) in the crystal structure were positioned inside of the tetrahedron and coordinated by four Ni atoms. In the case of Pr0.6Mg1.4Ni4 (Pr0.6Mg1.4Ni4D3.6, gravimetric hydrogen density: 0.9 mass %), it has been reported that the D2γ positions in the γ-phase hydride differed from those in the γ-phase hydride, LaMgNi4D4.85 (Figure ).[10]

Since Y is the lightest RE in REMgNi4, YMgNi4 was expected to have a higher gravimetric hydrogen density than the other REMgNi4 because it has the lightest formula weight. By contrast, the hydrogen storage reaction was halted during the formation of the β-phase hydride, whose gravimetric hydrogen density is comparable to that of REMgNi4 (approximately 1.0 mass %) containing heavier RE elements.[7,11] Although the γ-phase hydride was typically obtained at a higher hydrogen pressure, the reaction pressure and pathways depended on the RE elements or the compositional ratios of RE to Mg. This indicated that if the compositional ratios of Y to Mg could be optimized, YMgNi4 would be able to form hydrides with more hydrogen atoms. Additionally, this implied that the reactions were related to the interactions between the metal elements and H (affinities of the metal elements toward H) in the crystal structures. However, the roles of RE and Mg in REMgNi4-based alloys in hydrogen absorption and desorption reactions have not been understood properly. Hence, in this study, we attempted the synthesis of YMgNi4-based alloys exhibiting systematically adjusted compositional ratios of Y to Mg and studied their crystal structures to understand the local atomic arrangements around the H atoms. On the basis of the investigation of the atomic arrangements, we aimed to improve the gravimetric hydrogen densities and understand the roles of Y and Mg in YMgNi4 as well as in other REMgNi4-based alloys that are used for hydrogen storage.

Results and Discussion

Synthesis

Figure shows the powder X-ray diffraction (PXD) patterns of the as-synthesized materials. Although small amounts of Y2O3 and (Y, Mg)Ni3 were identified after the synthesis, the main Bragg peaks were indexed by an FCC unit cell with a = 6.9–7.1 (Table ).

Figure 2

PXD patterns of (2 – x)YNi2 + xMgNi2 after the syntheses. The Si powder was added as an internal standard to correct the Bragg peak positions. The blue, black, red, and green lines correspond to x = 1.2, 1.0, 0.8, and 0.6.

Table 1

Unit Cell Parameters after the Syntheses and Pressure–Composition–Temperature (PCT) Measurements at 323 Ka

	unit cell parameter a (Å)
(2 – x)YNi2 + xMgNi2	as-synthesized	after PCT measurement	refined composition
x = 1.2	6.9888(4)	6.9856(07)	Y0.78Mg1.22Ni4.00
x = 1.0	7.0066(3)	7.0072(22)	Y0.81Mg1.19Ni4.00
x = 0.8	7.0297(2)	7.0308(26)	Y1.06Mg0.94Ni4.00
x = 0.6	7.0572(2)	7.0680(14)	Y1.16Mg0.84Ni4.00

The data include the refined composition of the main phase of (2 – x)YNi2 + xMgNi2. The unit cell parameters were obtained from PXD, whose Bragg peak positions were corrected by Si as the internal standard. The refined compositions were obtained from the Rietveld method on the synchrotron radiation powder X-ray diffraction (SR-PXD) data.

The unit cell parameter decreased with an increasing amount of MgNi2, which was the source of Mg in the YMgNi4-based alloys. YMgNi4 with a Y:Mg ratio of 1:1 has been previously evaluated.[4,11,12] Nevertheless, the reported unit cell parameters (refs (4, 11), and (12)) are inconsistent. The discrepancies probably originated from correction methods of the Bragg peak positions in PXD in refs (4) and (11) or from the partial replacement of the Mg atoms with the Y one and vice versa (ref (12)). In this study, the chemical compositions of the final products were refined from those of the starting materials by the Rietveld method using the SR-PXD data obtained at room temperature. The established chemical compositions are summarized in Table . The Rietveld refinement fits (Figures S1–S4) and crystallographic parameters (Tables S1–S4) are presented in the Supporting Information. The observed SR-PXD data of all of the samples were reasonably reproduced by an FCC structure. The inconsistencies in the quantity of Mg between the ratios of the starting materials and final products were suspected to be due to impurities in the starting materials or the evaporation of Mg during the syntheses. However, the overall trend of the quantities of Mg in the final products is consistent with the ratios of the starting materials.

Hydrogen Storage Properties

Figure shows the PCT curves of YMgNi4-based alloys at 323 K. The PCT curves were obtained after several hydrogen absorption and desorption reactions at 323 K. Table lists the equilibrium pressures of the hydrogen absorption and desorption, i.e., the plateau pressures, as well as the hydrogen storage capacities.

Figure 3

PCT curves of (2 – x)YNi2 + xMgNi2 at 323 K. Using the sample with x = 1.0, the cycle degradation was confirmed up to five cycles after exhibiting stability from the second to fifth cycles. The sample with x = 1.0 featured the fifth hydrogen absorption and desorption reaction cycles. Except for x = 1.0, the third hydrogen absorption and desorption reaction cycles are plotted. The blue, black, red, and green circles correspond to x = 1.2, 1.0, 0.8, and 0.6. The closed and open circles indicate the hydrogen absorption and desorption reactions, respectively.

Table 2

Equilibrium Pressures for Hydrogen Absorption and Desorption Reactions and Hydrogen Storage Capacities of (2 – x)YNi2 + xMgNi2 at 323 K

	equilibrium pressure (MPa)
(2 – x)YNi2 + xMgNi2	absorption	desorption	gravimetric hydrogen density (mass %)
x = 1.2	3.74	1.48	0.73
x = 1.0	2.71	1.00	1.14
x = 0.8	0.99	0.33	1.23
x = 0.6	0.34	0.10	0.86

The regions of the plateau pressure on the curves correspond to the hydrogen absorption that is due to the formation of the β-phase hydride, which was confirmed by powder neutron diffraction (PND). A decrease in the amounts of Mg resulted in lowered hydrogen absorption and desorption pressures and increased hydrogen storage capacities. By contrast, the maximum hydrogen storage capacity reached 1.2 mass % at x = 0.8. Moreover, the unit cell parameters and PXD patterns after the PCT measurements were listed (Table ). It was observed that the unit cell parameter of the YMgNi4-based alloy with x = 0.6 did not return to the initial value after the PCT measurement. Furthermore, in the hydrogen absorption involving LaMgNi4 (eqs –3), the α-phase hydride displayed an FCC lattice, which was maintained from the initial alloy. Hence, the unit cell parameter at x = 0.6 suggested the formation of the α-phase hydride. In the first hydrogen absorption reaction at x = 0.6, the hydrogen capacity was comparable to that at x = 0.8. However, after the second cycle, the hydrogen capacity at x = 0.6 was lower than that at x = 0.8. This indicated that the α-phase hydride was thermodynamically too stable to release hydrogen at 323 K. Furthermore, a decreased crystallinity with decreasing amounts of Mg was observed after the PCT measurement (Figure S5 in the Supporting Information). It was attributed to the radius ratios of the Y, Mg, and Ni atoms, similarly to the Laves phase hydrides reported by K. Aoki et al.[13] Although the ratios of Y and Mg affected the control of hydrogen storage capacities and plateau pressures, lower amounts of Mg and an excessive amount of Y caused the formation of the thermodynamically stable α-phase hydride. The α-phase hydride resulted in decreased hydrogen storage capacities after the second cycle and reduced crystallinity. The hydrogen storage properties are discussed in the next section based on the atomic arrangements. At x = 0.8, hydrogen absorption was observed at a gravimetric hydrogen density of ca. 1.0–1.2 mass % and above 1.0 MPa at 323 K (Figure ). This reaction might be a transformation from the β- to the γ-phase hydride.

Since the material with x = 0.8 exhibited the highest hydrogen storage capacity in this study, the PCT measurements were performed at 303–343 K (Figure S6 in the Supporting Information) to obtain the thermodynamic parameters, specifically enthalpy (ΔH) and entropy (ΔS), of the hydrogen absorption and desorption. Table summarizes the ΔH and ΔS values. Notably, ΔH of the analyzed material was comparable to that of a hydrogen storage alloy (LaNi5) (ΔH ≈ −30 kJ/mol H2).[14]

Table 3

Thermodynamic Parameters (ΔH and ΔS) of Hydrogen Absorption and Desorption on 1.2YNi2 + 0.8MgNi2

	absorption	desorption
enthalpy ΔH (kJ/mol H2)	–29.2	–35.1
entropy ΔS (J/mol H2 K)	110.5	119.8

Atomic Arrangements during Hydrogen Absorption

The crystal structures of Y0.81Mg1.19Ni4.00 (1.0YNi2 + 1.0MgNi2) and Y1.06Mg0.94Ni4.00 (1.2YNi2 + 0.8MgNi2) during hydrogen absorption were investigated by PND in vacuum and at a deuterium gas pressure of 3–5 MPa at 323 K. Figure shows the crystal structures and local atomic arrangements around the D atoms in Y1.06Mg0.94Ni4.00D3.86 only since the crystal structure of Y0.81Mg1.19Ni4.00D3.35 fundamentally exhibited the same atomic arrangements with Y1.06Mg0.94Ni4.00D3.86 with different site occupancies. The Rietveld refinement fit of the PND pattern for the 1.2YNi2 + 0.8MgNi2 under a D2 pressure of 3 MPa is shown in Figure ; the Rietveld refinement fit of the PND patterns for 1.0YNi2 + 1.0MgNi2 under vacuum and D2 pressure of 5 MPa and that for 1.2YNi2 + 0.8MgNi2 under vacuum are shown in Figures S7–S9 in the Supporting Information; and the crystallographic parameters of the main phases are listed in Tables S5–S10 in the Supporting information. Based on the crystal structural investigations, the β-phase hydrides (deuterides) with orthorhombic unit cells with a = 5.0274(2) Å, b = 5.3823(2) Å, c = 7.2853(3) Å, V = 197.13(1) Å3, and Z = 2 in Pmn21 (No. 31) in Y0.81Mg1.19Ni4.00D3.35 and a = 5.0407(1) Å, b = 5.4162(2) Å, c = 7.3035(2) Å, V = 199.39(1) Å3, and Z = 2 in Pmn21 (No. 31) in Y1.06Mg0.94Ni4.00D3.86 were observed. The chemical compositions of the deuterides were established as Y0.81Mg1.19Ni4.00D3.35 and Y1.06Mg0.94Ni4.00D3.86, which are consistent with the results obtained from the PCT measurements.

Figure 4

Crystal structures of (a) Y1.06Mg0.94Ni4.00 and (b) Y1.06Mg0.94Ni4.00D3.86, and the (c–f) local atomic arrangements around the D atoms. The Y, Mg, Ni, and D atoms are represented by the gray, orange, green, and blue circles, respectively. In the crystal structures, the corner-shared tetrahedra formed by four Ni atoms are represented as the green polyhedra. The gray lines connect the Y atoms in Y1.06Mg0.94Ni4.00D3.86 to compare it with the unit cell of Y1.06Mg0.94Ni4.00. In the local atomic arrangements around the D atoms, the hydrogen sites in the triangular bipyramid and tetrahedra are represented as pink (D1), yellow (D2), cyan (D3), and green (D4) polyhedra.

Figure 5

Rietveld refinement fits of the PND for 1.2YNi2 + 0.8MgNi2 at (left) 0 < d ≤ 6.0 Å, (middle) 5 ≤ d ≤ 5.8 Å under D2 pressure of 3 MPa at 323 K, and (right) simulated PND patterns from hypothetical crystal structure models based on Y1.06Mg0.94Ni4.00D3.86 with different occupancies of the D4 atoms. The top and bottom profiles are a crystal structure model with and without D4, respectively. In the Rietveld refinement fits, the observed, calculated background and difference between observed and calculated patterns are indicated by circles, red, green, and blue lines, respectively. The positions of Bragg reflection are shown for (top) Y1.06Mg0.94Ni4.00D3.86 (91 wt %), (middle) Y1.06Mg0.94Ni4.00 (5 wt %), and (bottom) Y2O3 (4 wt %). In the simulated PND pattern, occupancies of the D4 atoms with 0.0, 0.1, 0.18, 0.3, 0.4, and 0.5 are represented by black, yellow, red, green, cyan, and blue lines, respectively. The occupancies of the D4 atoms with 0.0, and 0.18 indicate Y1.06Mg0.94Ni4.00D3.86 without and with the D4 atoms, respectively.

The crystal structural investigations revealed that the D atoms were located in four crystallographically different sites, D1, D2, D3, and D4. The D1 and D2 atoms were located inside of the triangular bipyramids and coordinated by two Y, one Mg, and two Ni atoms. The D3 atoms were positioned inside of the tetrahedron and coordinated by one Y and three Ni atoms. Although the positions of D1, D2, and D3 were analogous to those previously reported for the β-phase hydrides of REMgNi4 (D1β, D2β, and D3β), small amounts of D4 atoms, which were located inside of the tetrahedron and coordinated by one Y and three Ni atoms, were identified in our crystal structural investigations (Figure ).

First, we discuss about the D4 atoms. It is noteworthy that the presence of the D4 atoms in the β-phase hydrides has not been previously reported for REMgNi4-based alloys (RE = lanthanides). The crystal structure model of Y1.06Mg0.94Ni4.00D3.86 containing the D4 atoms exhibited better reproducibility of the experimentally observed PND pattern than the crystal structure model without the D4 atoms (Figure ). The crystal structure models of Y0.81Mg1.19Ni4.00D3.35 with and without the D4 atoms demonstrated comparable fitting to the experimentally observed PND pattern (Figure S8 in the Supporting Information) due to less occupancies of the D4 atoms (0.09). In the Rietveld refinement fit of the PND pattern for 1.2YNi2 + 0.8MgNi2 under a D2 pressure of 3 MPa (Y1.06Mg0.94Ni4.00D3.86), the fitting of the (010) reflection observed at d ≈ 5.4 Å clearly revealed the improvement. In simulated PND patterns obtained from hypothetical crystal structure models based on Y1.06Mg0.94Ni4.00D3.86 with different occupancies of the D4 atoms (Figure ), the Bragg peak intensities of the (010) reflection became stronger with increasing D4 atomic occupancies. This is consistent with the improvement fitting of the (010) reflection on the crystal structure model of Y1.06Mg0.94Ni4.00D3.86 with D4 atoms. Though the better Rietveld refinement fits of the PND for Y1.06Mg0.94Ni4.00D3.86 were obtained by the D4 atoms, the interatomic distances between the D4 and the D2 (1.62 Å) or the D3 (1.87 Å) atoms were shorter than a criterion reported by Switendick,[15] in which the minimum interatomic D–D distance should be 2.1 Å. The interatomic distances of the D atoms are listed in Table S11 in the Supporting Information. Shorter interatomic D–D distances than the criterion have been often reported for intermetallic compounds.[10,16−18] Recently, the shortest H–H interatomic distances with 1.5–1.6 Å in C15-type Laves-phase alloy, ZrV2 (ZrV2H3.7), were reported by Borgschulte et al. who investigated the H atomic positions in ZrV2H3.7 by inelastic neutron scattering combined with theoretical calculations.[18] In this study, Y1.06Mg0.94Ni4.00 adopted similar arrangements to ZrV2, as mentioned in the description of the crystal structure of REMgNi4 in the Introduction section. Notably, the local atomic arrangements of the D4 atoms in Y1.06Mg0.94Ni4.00D3.86 corresponded to some atomic positions of D2γ with short D–D interatomic distances (1.40–1.60 Å) in the γ-phase hydride Pr0.6Mg1.4Ni4D3.6 (Figure ),[10] although Y1.06Mg0.94Ni4.00D3.86 were the β-phase hydrides. Since powder neutron (or X-ray) diffraction provides averaged atomic arrangements, the D4–D2 or D4–D3 atomic pairs with the short interatomic distances might not simultaneously exist at each atomic site in Y1.06Mg0.94Ni4.00D3.86 (the β-phase hydride). Considering similarities to ZrV2H3.7 and Pr0.6Mg1.4Ni4D3.6 (γ-phase hydride) and better Rietveld fits with D4 atoms, we propose the presence of the D4 atoms in the averaged crystal structure of the β-phase hydride, Y0.81Mg1.19Ni4.00D3.35 and Y1.06Mg0.94Ni4.00D3.86.

Next, we discuss the hydrogen storage properties of the YMgNi4-based alloys from the viewpoint of their atomic arrangements and the affinities of metal elements toward the D atoms. In Y0.81Mg1.19Ni4.00D3.35, excessive Mg atoms were in the Y atomic sites. Moreover, in Y1.06Mg0.94Ni4.00D3.86, excessive Y atoms were located in the Mg atomic sites. As shown in Figure , more Y atoms were found around the D atoms in Y1.06Mg0.94Ni4.00D3.86. The more Y atoms around the D atoms resulted in the formation of larger D atomic sites. The polyhedral volumes of Y0.81Mg1.19Ni4.00D3.35 and Y1.06Mg0.94Ni4.00D3.86 were compared, and the results are listed in Table S12 in the Supporting Information. This would be one of the reasons for higher hydrogen storage capacities of Y1.06Mg0.94Ni4.00D3.86 than Y0.81Mg1.19Ni4.00D3.35. The affinities of Y and Mg toward D are also suspected to affect the hydrogen storage capacity and plateau pressure for hydrogen absorption. Since more Y atoms were located around the D atoms in the crystal structure, the affinity of Y toward D contributed to the interactions with the D atoms more than that of Mg toward D. Y was guessed to exhibit a qualitatively higher affinity toward hydrogen than Mg in YMgNi4-based alloys because of the higher thermodynamic stability of the Y hydrides,[19] such as YH2 and YH3, compared with that of the Mg hydride,[3] e.g., MgH2. Then, the presence of excessive Y atoms in the YMgNi4-based alloys, where excessive Y would facilitate a qualitatively stronger interaction of Y–H than that of Mg–H, might result in a lower plateau pressure to form a hydrogen-absorbed phase and cause increased hydrogen storage capacities in the first hydrogen absorption reaction (e.g., YMgNi4-based alloy with x = 0.6). However, the absorbed hydrogen was not released because of the stronger interaction of Y–H than that of Mg–H. Those results suggested that the plateau pressure was tuned by the substitution of the Mg atoms into the Y atomic sites in the hydrogen-absorbed phase.

Conclusions

YMgNi4-based alloys were synthesized from (2 – x)YNi2 and xMgNi2 (0.6 ≤ x ≤ 1.2) to control the compositional ratios between Y and Mg. The obtained materials exhibited reversible hydrogen absorption and desorption reactions in which the equilibrium pressure decreased with decreasing amounts of Mg. The hydrogen storage capacities also depended on the quantities of Mg, i.e., lower amounts of Mg resulted in higher hydrogen storage capacities. However, a YMgNi4-based alloy with x = 0.8 exhibited the maximum hydrogen capacity (1.2 mass %) because of the formation of a thermodynamically stable hydride phase at x = 0.6, which exhibited irreversible hydrogen absorption and desorption reactions.

PND experiments in vacuum and at deuterium gas pressure were used to investigate the crystal structures that were formed during hydrogen absorptions in Y0.81Mg1.19Ni4.00 (x = 1.0) and Y1.06Mg0.94Ni4.00 (x = 0.8). Formations of β-phase hydrides (deuterides) with orthorhombic structures, a = 5.0274(2) Å, b = 5.3823(2) Å, c = 7.2853(3) Å, V = 197.13(1) Å3, and Z = 2 in Pmn21 (No. 31) in Y0.81Mg1.19Ni4.00D3.35 and a = 5.0407(1) Å, b = 5.4162(2) Å, c = 7.3035(2) Å, V = 199.39(1) Å3, and Z = 2 in Pmn21 (No. 31) in Y1.06Mg0.94Ni4.00D3.86, were observed at 323 K.

In this study, the presence of small amounts of the D atoms with short interatomic D–D distances (1.6 and 1.9 Å) in deuterium-absorbed phases, Y0.81Mg1.19Ni4.00D3.35 and Y1.06Mg0.94Ni4.00D3.86, which were shorter than the criterion for the minimum interatomic D–D distance (2.1 Å), was proposed by neutron diffraction experiments at D2 pressure. Notably, the D atoms with such short interatomic D–D distances in the β-phase hydride of REMgNi4 (RE = lanthanides) have not been previously reported. Although such D–D atomic pairs with short interatomic distances might not simultaneously exist at each atomic site, the D atoms with short interatomic D–D distances were interestingly found in the same local atomic arrangements as those of the previously described γ-phase hydride, which exhibited higher hydrogen capacities than the β phase hydride.

In the crystal structures, more Y atoms were located around the D atoms and caused the formation of larger D atomic spaces. Moreover, it was guessed that the affinity of the Y atoms toward D could contribute to the interactions with the D atoms more than the affinity of Mg toward D. Similarly, in the case of the YMgNi4-based alloys at x = 0.6, excessive Y atoms with qualitatively higher affinities toward hydrogen formed thermodynamically stable hydride phases. However, the equilibrium hydrogen pressure (plateau pressure) could be tuned by the substitution of the Mg atoms into the Y atomic sites. Hence, the size of the hydrogen atomic sites as well as the affinity toward hydrogen optimized by the Y and Mg ratios around the hydrogen atoms could be keys to controlling their hydrogen storage properties.

Methods

The YMgNi4-based alloys were prepared from powder mixtures of YNi2 and MgNi2 at molar ratios of 2 – x:x (x = 0.6, 0.8, 1.0, and 1.2). YNi2 was prepared by arc-melting of a stoichiometric mixture of Y chips (Sigma-Aldrich, 99.9%) and Ni shots (Rare Metallic Co., Ltd., 99.999%). MgNi2 was obtained by the induction melting of Mg (Rare Metallic Co., LTD., 99.9%) and Ni powders (Rare Metallic Co., LTD., 99.9%) at a molar ratio of 1.45:2. The PXD patterns of YNi2 and MgNi2 are shown in Figure S10 in the Supporting Information. YNi2 and MgNi2 were pulverized and mixed at a molar ratio of 2 – x:x. The mixture was pressed into a tablet with a diameter of 6 mm, which was heat-treated at 1073 K for 24 h in an Ar atmosphere. Following the first heat treatment, the sample was collected, pulverized, and pressed into a tablet with a diameter of 6 mm. The tablet was further heat-treated at 1123 K for 24 h in an Ar atmosphere. All of the products were identified by PXD, which was conducted on a PANalytical X’Pert diffractometer with Cu Kα radiation (wavelength λ = 1.5406 and 1.5444 Å for Kα1 and Kα2, respectively).

The hydrogen storage properties of the products were studied by PCT measurements (PCT-3SPWIN, Suzuki Shokan Co., Ltd.) at 323 K up to a hydrogen gas pressure of 9 MPa. Before the PCT measurements, the samples were heat-treated at 473 K in a vacuum for 3 h.

SR-PXD and PND were performed on the BL-8A at the Photon Factory (PF), High Energy Accelerator Research Organization (KEK) in Japan and the NOVA (BL21) at the Material and Life Science Experimental Facility (MLF) of the Japan Proton Accelerator Research Complex (J–PARC) in Japan, respectively. For SR-PXD, the samples were placed in a Lindemann glass capillary with an outside diameter of 0.3 mm and thickness of 0.01 mm. The measurement was conducted in an Ar atmosphere, and the data were collected at room temperature. The wavelength of 0.775777(3) Å was obtained based on the CeO2 powder data. In the PND experiments, the samples were focused on x = 1.0 and 1.2 since x = 1.0 was the basis of this study, whereas x = 1.2 exhibited the highest hydrogen content. The samples were placed in a V sample container with an outside diameter of 10.80 mm and thickness of 1.5 mm. The samples were heat-treated at 423 K in vacuum for 3 h to initiate the hydrogen absorption reaction. Following heat treatment, the PND data were collected at 323 K under vacuum and deuterium gas pressure.

The indexing programs, TREOR97[20] and PIRUM,[21] and the Rietveld refinement program, GSAS with the graphical interface, EXPGUI (version 1.80)[22] were used for the investigations of the crystal structure obtained from the SR-PXD and PND data. In the Rietveld method, the pseudo-Voigt peak shape function with the Finger–Cox–Jephcoat asymmetry correction[23] and TOF profile function extension of a profile function developed by Von Dreele et al.[24] were used for SR-PXD and PND, respectively. The background was modeled using the Chebyschev polynomial function model in GSAS. The isotropic temperature–displacement parameters were used for all elements.

All of the samples were handled in Ar- or He-gas-filled gloveboxes with dew points of <183 K and <1 ppm of O2 to prevent (hydro-)oxidation.