The Effect of Y Content on Structural and Sorption Properties of A2B7-Type Phase in the La–Y–Ni–Al–Mn System

Metal hydrides are an interesting group of chemical compounds, able to store hydrogen in a reversible, compact and safe manner. Among them, A2B7-type intermetallic alloys based on La-Mg-Ni have attracted particular attention due to their high electrochemical hydrogen storage capacity (∼400 mAh/g) and extended cycle life. However, the presence of Mg makes their synthesis via conventional metallurgical routes challenging. Replacing Mg with Y is a viable approach. Herein, we present a systematic study for a series of compounds with a nominal composition of La2-xYxNi6.50Mn0.33Al0.17, x = 0.33, 0.67, 1.00, 1.33, 1.67, focusing on the relationship between the material structural properties and hydrogen sorption performances. The results show that while the hydrogen-induced phase amorphization occurs in the Y-poor samples (x < 1.00) already during the first hydrogen absorption, a higher Y content helps to maintain the material crystallinity during the hydrogenation cycles and increases its H-storage capacity (1.37 wt.% for x = 1.00 vs. 1.60 wt.% for x = 1.67 at 50 °C). Thermal conductivity experiments on the studied compositions indicate the importance of thermal transfer between powder individual particles and/or a measuring instrument.


Introduction
Integrating renewables with various energy storage technologies into the electrical grid is a compelling approach for further expansion of the green energy production and utilization [1]. Hydrogen as a fuel and an energy carrier is a part of the sustainable energy revolution, while energy can be stored in a form of liquid or gaseous hydrogen, a considerable amount of energy is lost during the gas compression and/or liquefaction processes [2,3]. An alternative storage method utilizes metal hydrides, in particular intermetallic alloys, which can reversibly sorb hydrogen gas at moderate temperatures and pressures [3,4], in a safer way than the mentioned, conventional solutions [5].
Previous studies have shown that by performing multiple substitutions of both the A element, and the B elements, the stability of the structure during cycling and storage capacities are affected [12][13][14][15]. It has been found that by substituting Mg for La in rareearth (RE) superlattice based A 2 B 7 -type metal hydrides, it is possible to achieve a higher storage capacity (compared to commercial AB 5 -type alloys), a lower self-discharge if used in batteries and improved cycle stability by preventing the occurrence of hydrogen-induced amorphization (HIA), when compared to pure La 2 Ni 7 [12,[16][17][18][19][20][21][22].
However, an issue related to using RE-Mg-Ni based materials is due to the high vapour pressure of Mg, making compositional control difficult while also being an explosion hazard [12]. Further, RE-Mg-Ni based materials still suffers from poor cycle stability.
To solve some of these issues, new manufacturing techniques such as smelting protection with helium gas [20,23], sintering techniques [24,25], ball milling [26] or other processing techniques [23,27] have been proposed. These are however either very expensive and/or are highly complicated processes. Evidence suggest that though substitution with Mg is preferable from an economic standpoint, substitution with Y eliminates the issues of Mg synthesis.
Substitution of the La element for the heavier lanthanides is of limited interest for industrial applications due to the cost of these elements, alongside the increase in weight of the final product. However, using Y instead of Mg is a viable solution, since it is lighter and cheaper than the heavier lanthanides. Baddour-Hadjean et al. [28] found that by combining LaNi 5 and YNi 2 in appropriate amounts, a La-Y-Ni based AB 3 -type system was formed. It was found that the LaY 2 Ni 9 alloy was a PuNi 3 -type compound, similar to that of LaMg 2 Ni 9 , and could absorb a higher hydrogen content than that of LaMg 2 Ni 9 under identical conditions. However, using LaY 2 Ni 9 as an electrode, showed an electrochemical capacity of only 265 mAh/g. Further, Belgacem et al. [29] showed that LaY 2 Ni 9 only maintained 54% capacity after 100 cycles. Other studies showed that multi-substituted materials based on the A 2 B 7 -type in the La-Y-Ni system, show great improvements in performance, outperforming already commercialized AB 5 -type metal hydrides [12,[30][31][32].
The A 2 B 7 intermetallic compounds crystallise either with the hexagonal (H-; the Ce 2 Ni 7 structure type; space group: P6 3 /mmc, a high-temperature phase) or rhombohedral (R-; the Gd 2 Co 7 structure type, space group: R-3m, a low-temperature phase) symmetry. Both polymorphs belong to the family of superlattice compounds with a composite structure, built up by AB 5 (the CaCu 5 structure type) and A 2 B 4 (AB 2 ) subunits (the MgZn 2 and MgCu 2 structures types, known as C-14 and C-15 Laves phases, in the hexagonal and rhombohedral system, respectively) alternated along c-axis, with a 2:1 ratio. Due to this, the A 2 B 7 phases combine the advantage of the fast activation of AB 5 phases with the high discharge capacity of AB 2 -type compounds.
Yan et al. [12] showed that the hexagonal LaY 2 Ni 9.7 Mn 0.5 Al 0.3 could absorb up to 1.48 wt.% of hydrogen, which is close to the reported hydrogen storage capacities of the H-A 2 B 7 phases in the RE-Mg-Ni system (up to 1.57 wt.% of H 2 ) [18,33].
Liu et al. [34] and Zhao et al. [35] studied the effects of yttrium substitution in A 2 B 7type phases of La-Y-Ni-Mn-Al and found that Y induced the formation of the H-A 2 B 7 compositions. These observations however contradict other studies indicating that the yttrium presence favoured the R-A 2 B 7 phase formation [36,37]. Zhao et al. [35] also reported that the studied La 1 -x Y x Ni 3.25 Mn 0.15 Al 0.10 (x = 0.00-1.00) intermetallic compounds had a single hydrogen ab/desorption plateau, while Liu [34] reported on the formation of two hydrogen ab/desorption plateaus for a similar sample series, La 3 -x Y x Ni 9.7 Mn 0.5 Al 0.3 (x = 1, 1.5, 1.75, 2, 2.25, 2.5), which is in line with observations reported for other A 2 B 7 -type materials [12,37,38].
With the aim to clarify inconsistencies on structural and functional properties of A 2 B 7 -type materials in the La-Y-Ni-Al-Mn system, we have conducted a systematic study of the sample series La 2 -x Y x Ni 6.50 Mn 0.33 Al 0.17 (x = 0.33, 0.67, 1.00, 1.33, 1.67), focusing on the analysis of the material structural/microstructural properties and hydrogen sorption behaviour.

HR SR-PXD and PND
The structural properties of the intermetallic powders were investigated using both HR SR-PXD and PND techniques to obtain sufficient contrast among the system elements, while SR-PXD data helped to differentiate between La and Y, the PND measurements allowed to distinguish the Mn and Ni atoms. The combined simultaneous analysis of both data sets, carried out for each composition, appears essential for the accurate analysis of the material crystal structures (Figures 1-3 and Tables 1-4). The results confirm the formation of H-A 2 B 7 as a main phase in all studied samples. The higher Y content results in the shifting of Bragg peak positions towards higher 2θ angles in the collected diffraction patterns, which indicates the formation of a smaller unit cell, as expected (atomic radius of La = 1.87 Å and Y = 1.79 [39]) and previously reported [34,35]. The substitution of Ni by Al or Mn atoms (atomic radius of Ni = 1.26 Å, Al = 1.41 Å, Mn = 1.32 Å [39]), were expected to influence the average unit cell size, as also reported by other studies [30,38].
The Rietveld refinement results indicate the formation of a single hexagonal Ce 2 Ni 7type phase only in Y 0.67 and Y 1.00 . In all other samples, additional secondary phases are present (Table 1). Y 0.33 consists of H-A 2 B 7 (95 wt.%) and the minor AB 5 -type (5.3 wt.%) phase. The concentration of the major H-A 2 B 7 increases to 100 wt.% in Y 0.67 and Y 1.00 but again decreases to 89 wt.% and 53 wt.% for Y 1.33 and Y 1.67 , respectively. In the samples with the higher Y concentration (x ≥ 1.33), the formation of the R-A 2 B 7 and R-AB 3 phases has also been observed, which confirms but also contradicts some of the previously reported results. Liu et al. [34] (La 3 -x Y x Ni 9.7 Mn 0.5 Al 0.3 (x = 1, 1.5, 1.75, 2, 2.25, 2.5)) claimed that the increased Y concentration stabilised the formation of H-A 2 B 7 , whereas Zhao et al. [35] (La 1 -x Y x Ni 3.25 Mn 0.150 Al 0.10 (x = 0.00-1.00)) observed the presence of the H-A 2 B 7 phase only up to x ≥ 0.85, and the R-AB 3 phase occurrence for higher Y content. The findings presented herein confirm that the Y presence induces the formation H-A 2 B 7 in the entire sample series, but the higher Y concentration (x ≥ 1.33) leads additionally to crystallisation of the R-A 2 B 7 and R-AB 3 phases. Our observations are in line with earlier reports, suggesting that A 2 B 7 compositions with a higher concentration of smaller A elements, e.g., Y, prefer to take rhombohedral symmetry instead of hexagonal one [40,41], which is in line with the higher abundance of the R-A 2 B 7 phase in Y 1.33 and Y 1.67 . The refined lattice parameters of the major H-A 2 B 7 phase decrease with the increasing Y content (a = b = 5.0740 and 5.0004 Å for Y 0.33 and Y 1.67 , respectively; c = 24.643 and 24.322 Å for Y 0.33 and Y 1.67 , respectively), resulting in smaller subunit (∆V A 2 B 4 = 6.4% and ∆V AB 5 = 2.9%) and unit cell volumes (∆V = 4.2%) as seen in Table 1. (The subunit volumes are calculated using | V cell c · ((z 1 − z 2 ) · c| where V cell is the unit cell volume, c is the lattice parameter, and z 1 and z 2 are z value coordinates between 0 and 1 of the atoms at the edge of the subunit).   To help determine the quality of the fitted parameters, the R wp and R Bragg are shown in Tables 2 and 3, respectively. As seen for R wp , the values might seem high. Still, one must remember that when dealing with higher quality datasets, these may, in turn, provide higher R wp values due to imperfections not considered in the data refining step. On the other hand, the R Bragg values are generally small, indicating that the chosen model is satisfactory.
As shown in Table 4, the Y occupancy factor at the 4f 1 and 4f 2 sites in H-A 2 B 7 increases from 0.14 to 0.70, and from 0.09 to 0.26 for Y 0.33 and Y 1.00 , respectively. With the higher Y content, the element concentration at the 4f 1 and 4f 2 sites rise further to reach eventually 93 and 84%, respectively, in Y 1.67 . This suggests that while initially, Y occupy the A 2 B 4 subunits, once filled up (x ≥ 1), it prefers entering the AB 5 subunits. Such behaviour has been previously reported and was related to a lower coordination number of 16 at the 4f 1 site when compared with coordination number 20 at the 4f 2 site [36]. The higher Y concentration in the samples also shortens the 4f 1 -4f 1 distance in the neighbouring unit cells from 3.282(3) to 3.170(7) Å for Y 0.33 and Y 1.67 , respectively. Table 4. Refined sample compositions, fractional occupancies (n), displacement factors (B iso ), atomic positions, atomic distance 4f 1 -4f 1 in the A 2 B 4 subunit and goodness-of-fit (χ 2 ) for the H-A 2 B 7 phase in the studied intermetallic compounds, as obtained from Rietveld refinement. All crystallographic sites are assumed to be fully occupied; the remaining fractional occupancies are completed to 100% by either La (A atoms) or Ni (B atoms).
3.282 (3) 3.260 (2) 3.243 (3) 3.229 (4) 3.170(7) The Rietveld refinement results indicate that Mn atoms are present at the 4e and 6h sites in all studied compositions. The obtained fractional occupancies are low but larger than the estimated standard deviation values. In Y 0.33 , Y 0.67 and Y 1.00 , Mn appears additionally at the 4f 3 site. This site depletion observed in the Y-rich compositions, correlates with a subtle increase of the Mn atom occupancies at 4e and 6h, similar to the result reported by Deng et al. [31] for Y 0.75 La 0.25 Ni 3.2 Mn 0.3 with H-A 2 B 7 crystal structure.
In Y 0.33 , Y 0.67 and Y 1.00 , the Al atoms are distributed over the 6h and 12k sites, with the latter being less occupied. This is in line with the earlier report by Wang et al. [38]. They found that for H-A 2 B 7 in the LaY 1.9 Ni 10.2 -x Al x Mn 0.5 (x = 0, 0.2, 0.4, 0.6) system the presence of Al was only detected at the 6h site, between two AB 5 subunits. Similar results were also observed for H-A 2 B 7 formed in La 0.77 Mg 0.23 Ni 3.41 Al 0.09 and Nd 0.9 Mg 0.1 Ni 3.3 Al 0.2 [21,42]. In Y 1.33 and Y 1.67 , the 6h site is no longer populated by Al, but at the same time, a slight increase of aluminium concentration at the 12k site is observed. These findings, compared with the changes in population of the 4f 1 and 4f 2 sites, may indicate that the higher amount of Y atoms in H-A 2 B 7 , and its increasing concentration at the 4f 2 site, have an effect on (correlates with) the B atom distribution over the available crystallographic sites in this crystal structure, however the nature of interatomic relationships must be further investigated.

In Situ XRD
In situ PXD data were collected to determine: (1) the experimental conditions required for the hydrogenation of the studied compounds, and (2) the phase compositions of the hydrogenated materials. After initial activation of sample Y 0.33 and Y 0.67 , various hydrogen pressures and temperatures were applied, within the ranges reported previously in literature [12]. Regardless of the tested experimental conditions, both materials became amorphous upon hydrogenation. Figure 4a shows the results obtained for Y 0.67 at 30 • C and under 8 bar of H 2 (red line). Due to the amorphous nature of the H 2 exposed Y 0.33 and Y 0.67 , the samples were excluded from further studies. The Y-rich samples (x ≥ 1) were successfully hydrogenated at 20 bar of H 2 and 30 • C, with well-preserved crystallinity of the formed metal hydride phases (red plots in Figures 4b and S1 in the Supplementary Information).  The previously reported investigation also indicated the occurrence of HIA for similar intermetallics (e.g., La 1 -x Y x Ni 3.25 Mn 0.15 Al 0.10 ) at lower Y concentration (x < 0.5) [35]. Aoki and Masumoto [43] related the probability of the HIA occurrence with the ratio of the Goldschmidt radii between the A and B atoms (R A /R B ). Based on the empirical data, if the ratio was below 1.37 (R A /R B < 1.37) for the A 2 B 4 subunit, then HIA was effectively prevented, whereas if above amorphization was expected to happen. The calculated Goldschmidt radii ratio in our samples decreased from 1.48 for Y 0.33 to 1.43 for Y 1.67 , being higher than the limit found by Aoki and Masumoto. A similar result was reported for La 1 -x Y x Ni 3.25 Mn 0.15 Al 0.1 in [35] where a ratio of 1.453 was reported, with the 4f 1 site being completely filled with Y. The difference in the calculated ratios, was expected to be due to different atomic sizes being used, with the source not being provided in [35]. As such, all samples were expected to get amorphous.
It has also been found that the hydrogenated samples reported in this paper spontaneously release hydrogen, when stored in a sealed stainless-steel vial stored in a glovebox under ambient argon atmosphere ( Figure S2 in Supplementary Information).

SEM and EDX
Investigations with SEM-EDX were carried out for Y 1.00 , Y 1.33 and Y 1.67 to: (1) map out the chemical composition of the intermetallics, and (2) to study the effect of the Y concentration on the particle size distribution in the pristine intermetallics and hydrogenated samples. The difference in holes in the carbon tape has no influence to the analysis. These are caused by the quality and type of the carbon tape used. The differences in coverage can be due to three factors. (1) The samples were produced three months apart manually (no reproducible results). (2) The finer material obtained after hydrogenation formed a blanket that was stuck to the carbon tape. (3) Since the micrographs were taken three months apart, the stickiness of the carbon tape could have changed (different roll). Before hydrogen exposure, the Y 1.00 material (Figure 5a) revealed the most homogeneous particle size distribution as compared to other samples. In contrary, the Y 1.67 powders (Figure 5e) were characterized by the largest spread of the particle sizes. A zoom onto the particle surface at 25,000× before and after exposure to hydrogen can be seen in Figure S3 in the Supplementary Information. After hydrogenation, smaller particles with a comparable size distribution were observed in all studied compositions (Figure 5b,d,f). Although not directly comparable, Liu et al. [34] found that when the Y concentration increased in La 3 -x Y x Ni 9.7 Mn 0.5 Al 0.3 (x = 1, 1.5, 1.75, 2, 2.25, 2.5) from x = 1 to x = 1.5, the average particle size increased. Increasing the Y amount further, lowered the average particle size.
EDX investigations ( Table 5, Supplementary Information Figure S4) consisted of two sets of data acquired for Y 1.00 , Y 1.33 and Y 1.67 ; one focusing on a small region of a seemingly flat particle, with the second covering a larger area (referred to as "overview" in Table 5). A general trend presented in the EDX data confirmed the lower Ni content than in the sample nominal compositions when looking at single particles. Still, when considering at the overview, Ni concentrations were closer to the nominal values. When comparing the nominal composition, the obtained (average) EDX compositions and the refined compositions, it was still observed that the amount of Ni detected in the EDX data was smaller than for the pristine and refined data ( Table 6). Hao et al. [44] found that when annealing samples of La 0.33 Y 0.67 Ni 3.25 Mn 0.15 Al 0.1 , that the composition in the surface change. It is thus believed this is the reason for the lower Ni contant on the surface, and that it instead is located in the bulk. Further, since the samples are not perfectly flat, this can also influence the final results, altering the ratio of different elements detected. Figure 5. SEM micrographs collected at 250× for: Y 1.00 , Y 1.33 and Y 1.67 before (a, c and e, respectively) and after (b, d and f, respectively) hydrogen exposure. Additional micrographs obtained at higher magnification (25,000×) can be found in Supplementary Information (Figure S3).

PCT
The PCT measurements ( Figure 6) were performed for Y 1.00 , Y 1.33 and Y 1.67 to investigate in detail, the material hydrogen absorption behaviours. The corresponding Van't Hoff plots are shown in Figure 6d. The values of the enthalpy and entropy of the hydride formation in the studied samples, along with the expected plateau pressures at 30 • C, are listed in Table 7.
No leaks were detected during the pre-hydrogen exposure setup. For each point of the PCT a step time of 1.5 h has been chosen in order to obtain a PCT measurement with satisfactory accuracy. As an example, Figure S5 of the supplementary information shows a representative kinetics curve obtained for one point during the PCT measurements.
As highlighted in the paper by Rudman [45], great care needs to be taken when measuring the PCT characteristics of a material. A very high ∆P (40 bars) was chosen during activation, and hydrogen was completely absorbed within minutes. During the PCT measurements, a ∆P of 2 bars was set for each point, with the sample experiencing a ∆P of approximately 1 bar due to the reservoir and sample holder volumes described in the experimental section. Here, most hydrogen had been absorbed within minutes, with the absorption being greatly slowed down after 30 min of hydrogen exposure.
While the PCT plot for Y 1.00 (Figure 6a) revealed only one clearly visible plateau pressure, at all studied temperatures, the data for Y 1.33 and Y 1.67 (Figure 6b,c) displayed two hydrogen absorption regions. Furthermore, the single plateau of Y 1.00 occurs at a higher hydrogen pressure value (2.9 bar of H 2 at 50 • C) than any of the highest plateaus observed for Y 1.33 and Y 1.67 (0.6 and 1.6 bar of H 2 at 50 • C for Y 1.33 and Y 1.67 , respectively). Previous studies showed that for La 2 Ni 7 -based materials, either single [46] or multiple [37] plateau pressures can be expected for both single phase (H-A 2 B 7 or R-A 2 B 7 ) or combined phase materials. It has been discussed that if more than one plateaus were present, it was likely due to: (a) the existence of multiple hydrogen-active (absorbing) phases in the sample [47], and/or (b) the hydrogen absorption at different pressures by various structural subunits [37]. The PCT data presented in Figure 6 differ from those reported in literature for similar compositions, both in the number of plateaus and, in the case of Y 1.00 , the equilibrium pressures [34,35,37]. Single plateaus in A 2 B 7 -type compounds can be obtained by adjusting the subunit volumes [37].  As mentioned earlier, there can be two potential reasons for the appearance of the two plateaus as observed in Y 1.33 : The appearance of multiple phases [47], or subunits absorbing at different pressures [37]. Since both samples contained multiple phases, it could be assumed that this was the reason for the multiple plateaus. However, for La 2 -x Y x Ni 7 Zhang et al. [37] argued that the appearance of the multiple plateaus were not due to the H-A 2 B 7 or R-A 2 B 7 phases appearing at the same time, but rather the different subunits absorbing at different pressures due to a difference in volume size.
For Y 1.67 (Table 1), a R-AB 3 phase was present in addition to the H-A 2 B 7 and R-A 2 B 7type phases. Earlier studies have found that the equilibrium pressures of R-AB 3 were expected to be similar to that of H-A 2 B 7 due to the similarities in their superstacking structures [48,49]. As such, it was considered that the appearance of two plateaus in Y 1.33 and Y 1.67 was not due to the appearance of different phases, but due to differences in the subunit volumes in these phases.
By increasing the ratio Y/La our measurements showed a higher gas storage capacity (1.37 wt.% for Y 1.00 vs. 1.60 wt.% for Y 1.67 , both at 50 • C (Table 8)). This was partly attributed to the lower weight of Y compared to La, though the large increase in capacity from Y 1.00 (1.37 wt.% at 50 • C) to Y 1.33 (1.57 wt.% at 50 • C) cannot be explained by the sole relative weight difference between La and Y.
When performing PCT at various temperatures, for each sample a loss of storage capacity was observed as the temperature increase ( Figure 6 and Table 8), with a more marked change from 70 • C to 90 • C (between 0.05 and 0.10 wt.% difference) than from 50 • C to 70 • C (between 0.02 and 0.05 wt.% difference). Furthermore, the major changes were observed in Y 1.00 , compared to the changes for Y 1.33 and Y 1.67 . According to various findings, the plateau pressure values are affected by either a local chemical (binding) environment between the hydrogen atoms and the metallic atoms in the crystal structure (both A and B elements) [50], and/or differences in unit cell size [51]. According to Khatamian and Manchester [52], the formation enthalpy of YH 2 is −219.6 kJ/mol(H 2 ) close to other findings [53], whereas that of LaH 2 is −208 kJ/mol(H 2 ) [50,54]. This difference indicates that a higher La-content in the sample should result in higher plateau pressure values. In our case, as indicated by the values obtained for Y 1.33 and Y 1.67 (Figure 6b,c) it was observed that with increasing Y content the plateaus increased. It is important to note that the plateau pressure were higher for Y 1.00 (Figure 6a), when compared to Y 1.33 and Y 1.67 . This goes against the expected trend, seeing as others have shown for A 2 B 7 -type materials based on La-Y-Ni the plateau increased the more Y was present, due to a decrease in the average unit cell size [34,35,37]. Further, Zhang et al. [37] showed that in case the two plateaus were due to absorption in the different subunits, the lower plateau pressure can be related to the volume of the A 2 B 4 subunit, in that if the subunit volume decrease, the lower plateau pressure increase. The higher plateau was more complex and requires a more thorough inspection, since some substitutions lead to plateau pressure stabilization, even with changes to the subunit volumes [37].
The measured hydrogenation pressures for Y 1.00 absorbed from 2.9 bar at 50 • C to 9.3 bar at 90 • C, for Y 1.33 the lower plateau pressure absorb from 0.2 bar at 50 • C to 0.8 bar at 90 • C, the higher pressure go from 0.6 bar at 50 • C to 2.2 bar at 90 • C. Y 1.67 displayed a lower plateau absorbing from 0.4 bar at 50 • C to 1.6 bar at 90 • C and a higher plateau of 1.6 bar at 50 • C to 5.7 bar at 90 • C. From these measurements the enthalpy and entropy of formation, using a Van't Hoff plot as shown in Figure 6d. From these, the pressure plateaus at 30 • C were calculated (Table 7).
Liu et al. [34] found that for La 3  When comparing all of these literature plateau pressure values with the plateau pressures calculated at 30 • C, it was observed that for Y 1.00 the calculated plateau was higher than reported values in the literature for both single and multiple plateaus. When comparing the calculated plateau pressures of Y 1.33 and Y 1.67 to the plateau pressures found by Liu et al. [34] it was observed that all calculated plateau pressures reported here were higher, than the electrochemically measured plateaus reported by Liu et al. By comparing the calculated plateau pressures to those obtained by Zhao et al. [35], it was observed that the single plateaus reported by Zhao et al. have higher plateau pressures, than both of the calculated plateau pressures.
As described, there are no clear trends in the literature regarding this compound, both regarding plateau pressures, but also the number of plateaus. The reported data correspond to this in that no clear trends were found, showing that the picture is more complex, and great care needs to be applied to all known aspects that can affect the hydrogenation properties. Figure 7 shows the PXD data collected for the three samples before and after PCT experiments, with sample completely desorbed (here called after exposure). The sample with the lowest yttrium concentration (Y 1.00 ) become less crystalline after exposure to hydrogen gas as compared to Y-rich compositions. This suggests that higher Y content helps to retain material crystallinity during hydrogen ab/desorption.
Fang et al. [55] found that if the volumes of the A 2 B 4 and AB 5 subunits in a variety of AB 3 -type intermetallics (PrNi 3 , NdNi 3 , SmNi 2.67 Mn 0.33 , SmNi 3 , Sm 0.9 Mg 0.1 Ni 3 and Nd 0.33 Er 0.67 Ni 3 ) were smaller than 89.2 Å 3 and 88.3 Å 3 , respectively, then the material could release all of the absorbed hydrogen. When above these limits, the hydrogen could not be completely desorbed from their respective subunits. The critical volume limits were further confirmed by Zhang et al. [37], who studied a variety of La x A 2 -x Ni 7 , (A = Gd, Sm, Y, Mg). Samples of La x Y 2 -x Ni 7 with x between 0 and 0.8, had subunit volumes below these critical values mentioned above, and showed the best cycling behaviour.
Among the three hydrogenated samples presented here, only the A 2 B 4 and AB 5 subunits of the Y 1.67 intermetallic were below the critical values of 89.2 Å 3 and 88.3 Å 3 (Table 1). When analysing the XRD data before and after hydrogenation, Y 1.67 was also the sample with the best-preserved crystallinity after sample activation and three PCT hydrogenation cycles (Figure 7). Thus, the loss in crystallinity and lower capacity for samples with lower Y content may be due to partial hydrogen release. For Y 1.00 and Y 1.33 , while the volumes of the A 2 B 4 subunits exceeded the critical values, sizes of the AB 5 subunits were rather borderline values, with Y 1.33 being just above the limit. This may explain the appearance of the secondary plateau in the case of Y 1.33 , as well as the better crystallinity and hydrogen release compared to Y 1.00 . In the case of Y 1.67 the volume values were below the critical limit for both subunits and confirmed the link with better crystallinity compared to the other samples, and the most well-defined plateau regions. Further investigations are needed to confirm this hypothesis. Red plots correspond to the intermetallic compound after four absorption/desorption cycles (one activation cycle, and three PCT cycles). The black lines correspond to the pristine intermetallic compound. The data are given a vertical offset, allowing for easier reading.

Thermal Conductance
Thermal conductivity measurements were performed for the intermetallic powders of Y 1.00 , Y 1.33 and Y 1.67 (Figure 8). The measurements were conducted over three rounds, with each round having five measurements. Between rounds, the material was moved around, and re-compacted.
The thermal conductivity of hydrogen storage materials is often an overlooked parameter. For intermetallic compounds, the typical value is low and varies between 0.1 and 1 W/mK. While at lab-scale, the limiting factor of hydrogen absorption/desorption often is related to the intrinsic kinetics of hydrogen absorption of the samples [56], for large batches, the hydrogen uptake and release are dependent on the thermal management of the hydrogen storage systems and thermal conductivity of metal hydrides [56]. As can be observed in Figure 8, the data obtained in this study do not suggest any correlation with the Y content or the sample phase compositions. However, while stirring and re-compacting the powders between measurements, higher values of thermal conductivity were achieved, which may indicate that the contact between particles was essential for enhancing the thermal properties of the samples. It was very likely that the higher values of thermal conductivity for Y 1.67 were due to the greater particle size distribution in this sample, as observed by SEM ( Figure 5).

Experimental Tools and Methods
The series of samples with the following nominal compositions: La 2 -x Y x Ni 6.50 Mn 0.33 Al 0.17 , x = 0.33, 0.67, 1.00, 1.33, 1.67, hereafter referred to as Y 0.33 , Y 0.67 , Y 1.00 , Y 1.33 and Y 1.67 , respectively, were prepared according to the synthesis method reported in [12].
The hydrogen sorption behaviour of the studied compositions was investigated by in situ powder X-ray diffraction carried out with a laboratory Bruker D8 Advance diffractometer, equipped with an in-house built set-up for hydrogen gas pressurisation and thermal sample heating [58]. The data were collected with Cu Kα 1 (λ = 1.5406 Å; 2θ range: 20-95 • ; step size: 0.0176 • 2θ). The powders, sealed in a beryllium sample holder, were first heated up to 70 • C for 1 h, under dynamic vacuum, and subsequently exposed to static hydrogen pressure in the range of 8-20 bar. The hydrogenated materials were subsequently cooled down to temperatures in the range of 28-30 • C.
The HR SR-PXD and PND data collected for the intermetallic compounds were jointly analysed by Rietveld refinements using the Fullprof Suite program [59]. The diffraction profiles were modelled using the pseudo-Voigt peak shape function with the background being defined by interpolation between manually chosen points. For the final refinement cycles the following parameters were allowed to vary: the scale factors of indexed phases, lattice parameters of indexed phases, up to six profile parameters (U, V, W, mixing factor and two asymmetry parameters) and overall or individual displacement parameters.
To model the hexagonal phase, the La 2 Ni 7 crystal structure was used as a prototype. Based on the previously reported data, it was assumed that Y could substitute only La atoms, while Mn and/or Al could replace exclusively Ni. Based on the structure prototype the 4f 1 , 4f 2 crystallographic sites were assumed to be filled with A elements while 4e, 4f 3 , 6h and 12k were assumed to be completely filled with B elements. The 2a site was assumed to be completely filled with Ni and not refined [21,31,38]. During the refinement cycles, the occupancy of the 4f 1 and 4f 2 sites were shared between La and Y, and constrained to 100% for each site. Initially, Ni, Mn and Al atoms were evenly distributed over the 4e, 4f 3 , 6h and 12k sites according to the nominal compositions. However, to correctly determine the multiple (three atoms) atom occupancy at these sites, their distribution in the crystal structure model was constrained as described in [60]. In the final refinement cycle, Ni was distributed over all five sites, with Mn atoms partly present at the 4e, 4f 3 and 6h sites, and Al atoms partially occupying 6h and 12k.
Pressure-Composition-Temperature (PCT) measurements were performed with a Hy-Energy PCTPro-2000 Sieverts apparatus. In each measurement, 3 g of the intermetallic samples were placed into a stainless-steel sample holder with a calibrated volume between 12.5 and 13.0 mL. The sample holder was heated using an external thermal couple inserted into the sample holder. The material was first activated under dynamic vacuum at 70 • C for 1 h, and then exposed to hydrogen gas at 40 bar. Subsequently, the hydrogen was removed from the sample, keeping it under dynamic vacuum at 390 • C overnight. Following the activation, the samples were cooled down to 50, 70 and 90 • C, and the PCT hydrogen absorption data collected (∆P = 2 bar; no steady state criteria was used, instead a step time of 1.5 h for each point was decided on, ensuring each measured point was as close to equilibrium as possible; maximum applied P = 50 bar; V reservoir = 11.530 cm 3 ; V sample holder = 12.430-13.298 cm 3 depending on temperature and sample loading). Before the samples were exposed to hydrogen, the system used He gas to perform an automatic leak test. Prior to and after the three PCT experiments reported, sample phase compositions were investigated with a Bruker D2 PHASER (λ = 1.5406 Å; 2θ range: 20-80 • , step size: 0.02 • 2θ).
The intermetallic and desorbed hydride powders were also investigated by scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDX) with a Zeiss EVO MA10 microscope.
The thermal conductivity of the intermetallic powders was determined using the modified transient plane source (MTPS) with a C-Therm Trident system.

Conclusions
The structural and hydrogen storage properties of La 2 -x Y x Ni 6.5 Mn 0.33 Al 0.17 sample series have been investigated. It was found that Y initially induced the formation of a H-A 2 B 7 -type structure, and that higher Y amounts induced the R-A 2 B 7 -type and R-AB 3type structures.
The Rietveld refinement showed that of the two sites (4f 1 and 4f 1 ) where La and Y are located, Y initially prefered to enter into the site of the A 2 B 4 subunit (4f 1 site), with a small amount entering the AB 5 subunit. Instead, when x < 1.00, it was seen that Y primarily entered the AB 5 subunit, similar to the behaviour of single substituted samples. Further, it was found that Al resided at the 6h and 12k sites, while Mn resided at the 4f 3 , 4e and 6h sites at lower Y amounts. When x < 1.00, and Y started entering the AB 5 subunit, Al was only located on the 12k, while Mn was only located on the 4e and 6h sites. This indicates that the presence of the smaller Y atom on the 4f 2 site, can influence which sites the substituted B atoms are located at.
It was found that if the Y amount was low, the material became amorphous immediately during hydrogen exposure, and at higher Y-contents the crystal structure of the material was preserved. This has been coupled to the subunit volumes, since it was also found that the subunit volumes of the main H-A 2 B 7 -type phase decreased with increasing Y content, leading to a better retention of the crystal structure and more defined plateau pressures. Further, by increasing the amount of Y present in the samples, higher hydrogen storage capacities were achieved.
No definitive conclusions could be drawn about the effects of Y on the thermal conductivity nor anti-pulverization ability, and will need further investigation.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/molecules28093749/s1, Figure S1: In-situ data for all samples: (a) Y 0.67 , (b) Y 1.00 , (c) Y 1.33 , (d) Y 1.67 The black lines pristine powder, the red lines after exposure to hydrogen; Figure S2: HR SR-PXD data collected for hydrogenated materials at MCX at Elettra Sinchrotron (Trieste). The samples were stored at ambient conditions in a sealed steel container with Ar gas and in a glovebox over a couple of months. The data were obtained under the same conditions as the SR-HR PXD data presented in Figure 1; Figure S3: SEM micrographs obtained for Y 1.00 before hydrogen exposure (a), after hydrogen exposure (b) Y 1.33 before hydrogen exposure (c), after hydrogen exposure (d) and Y 1.67 before hydrogen exposure (e), after hydrogen exposure (f) at 25,000×; Figure   Acknowledgments: We acknowledge Elettra Sincrotrone Trieste (Italy) and the National Institute of Standards and Technology (USA), for providing access to the synchrotron radiation and to the neutron research facilities, respectively, and the technical staff in obtaining data used in this work. Further E.H.J acknowledges to have recieved funding from UiO:Energy.

Conflicts of Interest:
The authors declare no conflict of interest.

Abbreviations
The following abbreviations are used in this manuscript:

RE
Rare Earh HIA Hydrogen-Induced Amorphisation HR SR-PXD High-Resolution Synchrotron Radiation Powder X-ray Diffraction PXD Powder X-ray Diffraction PND Neutron Powder Diffraction SEM Scanning Electron Microscopy EDX Energy Dispersive X-ray Analysis PCT Pressure-Composition-Temperature