1. Introduction
The use of hydrogen as a clean energy carrier has become a viable idea to meet the stringently required decarbonization of the energy sector [
1,
2]. However, the effective implementation of a full-scale economy of hydrogen requires more efforts to overcome the main technical issue related to hydrogen, i.e., the extreme low density of the dihydrogen gas that limits both the efficient storage and transportation. Pressurized H
2 gas, cryogenic liquid, and physically/chemically bonded hydrogen to solid-state materials are the current existing ways for hydrogen storage [
3,
4]. Nevertheless, storing hydrogen in the form of either gaseous or liquid states presents many disadvantages, such as, high cost, large volume tanks, and safety issues. Hydrogen storage in metals and metallic alloys is a clean, safe, and efficient alternative only if the following characteristics are fulfilled: high absorption capacities, reaction close to ambient pressure and temperature, fast kinetics, and good cycling stability.
Many metals and metallic alloys have been studied for their hydrogen sorption properties and depending on the nature of elements and their concentrations in the material, the properties may change remarkably [
5]. The search for the right combination of elements and their optimal stoichiometry to fulfill all the above-mentioned features is a great challenge. Many metallic materials are currently being studied for hydrogen storage. For example, LaNi5 and TiFe intermetallic compounds present good hydrogen storage reversibility, but they are disadvantaged either by the low storage capacities, activation difficulties, or the high price of raw materials [
6,
7,
8]. Along with intermetallics, alloys are promising candidates to find the most efficient material for hydrogen storage applications. Traditionally, the alloying strategy consists of the addition of several elements in small amounts within one or two main elements to obtain compositions with enhanced properties. Along with conventional alloys, another metallurgic paradigm was proposed in 2004 based on the mixing of at least five elements near equimolar composition (concentration of each element is between 5 to 35 at.%) [
9,
10]. These materials are defined as high-entropy alloys (HEA) and have been initially proposed for their interesting mechanical properties [
11,
12]. Interestingly, HEAs are known to adopt simple crystalline structures, such as body-centered cubic (
bcc), face-centered cubic (
fcc), and hexagonal close compact (
hcp). In such simple crystalline structures, the mixing of elements with different atomic radii might result in an important lattice distortion with possibly large interstitial sites that might facilitate the hydrogen insertion and increase the amount of stored hydrogen. The lattice distortion is described by the empirical parameter δ, defined elsewhere [
13]. Additionally, another empirical parameter is the valence electron concentration, VEC, which is described by the number of valence electrons per formula unit [
14]. Presently,
bcc HEAs containing refractory elements are studied for solid-state hydrogen storage applications due to their interesting properties and large chemical versatility (the nature, the number, and the concentration of elements can be tuned) [
15,
16,
17,
18].
In this framework, the series of
bcc HEAs Ti
0.
30V
0.
25Zr
0.
10Nb
0.
25M0.
10 (
M = Mg, Al, Cr, Mo, and Ta) [
19,
20,
21,
22,
23] has been thoroughly studied in our group to explore the effect of the addition of 10 at.% of a fifth
M element into the quaternary alloy Ti
0.
325V
0.
275Zr
0.
120Nb
0.
275 [
24]. The hydrogen capacity is affected by the nature of the
M element: the quaternary alloy can absorb 2.5 wt.% hydrogen, whereas the capacities are 2.5, 2.6, 2.7, 2.8, and 3.0 wt.% after adding 10 at.% of Ta, Al, Mg, Mo, and Cr, respectively [
19,
20,
21,
22,
23]. Furthermore, the hydrogen desorption properties are also optimized after the addition of the fifth element, mainly for compositions containing Al, Ta, and Cr. Following these studies, a relationship between the VEC and the storage capacity has been proposed in this series of HEAs [
25]. For a VEC < 4.9, the capacity of absorption is in the range of 1.5–2.0 H/M, whereas for HEAs with a VEC ≥ 4.9, the capacity of hydrogen absorption drastically decreases. In light of these findings, the Ti
0.
30V
0.
25Mn
0.
10Zr
0.
10Nb
0.
25 with VEC = 4.8 shows one of the best capacities (2.0 H/M or 2.98 wt.%). Therefore, the aim of the present work is to thoroughly investigate the composition Ti
0.
30V
0.
25Mn
0.
10Zr
0.
10Nb
0.
25, with a focus on the structure, the microstructure, the chemical homogeneity, the hydrogen sorption properties, and the phase transition during the reversible hydrogenation reaction. Regarding the latter aspect, both ex situ and in situ characterizations using neutrons and synchrotron radiation have been carried out to analyze both the average and local crystalline structures of different phases.
2. Results and Discussion
The calculated empirical parameters for the composition Ti
0.
30V
0.
25Mn
0.
10Zr
0.
10Nb
0.
25 are δ = 5.7% and VEC = 4.8, based on the definitions proposed elsewhere [
13]. Compared to the quaternary alloy Ti
0.
325V
0.
275Zr
0.
120Nb
0.
275, the δ decreases from 6% to 5.7% after the addition of Mn. This could be explained by the fact that the Mn radius is close to the values of the other constituent elements (r
V = 1.31 Å < r
Mn = 1.37 Å < r
Nb = 1.43 Å < r
Ti = 1.45 Å < r
Zr = 1.59 Å as taken from [
26]) and it results in the reduction of the atomic size disparity. The VEC increases from 4.55 to 4.8 after the addition of Mn to the quaternary alloy.
The Ti
0.
30V
0.
25Mn
0.
10Zr
0.
10Nb
0.
25 alloy crystallizes in a single
bcc (Im
m) structure (see laboratory XRD in
Figure 1A), with a lattice parameter a
bcc = 3.239(5) Å, which is close to a
bcc= 3.261(1) Å, previously obtained for the quaternary alloy Ti
0.
325V
0.
275Zr
0.
120Nb
0.
275 [
24].
Figure 1B displays the SEM-EDS chemical mapping results of the as-cast Ti
0.
30V
0.
25Mn
0.
10Zr
0.
10Nb
0.
25. A dendritic morphology can be noticed, similarly to previously reported refractory HEAs synthesized by arc melting [
21,
22,
27]. The results of the EDS chemical analysis are listed in
Table 1, proving a good agreement between the overall and the nominal compositions.
The dendritic regions are enriched mainly by Nb and slightly by Mn, whereas the interdendritic regions are rich in Zr, Ti, and V, and are close to the nominal compositions, irrespective of the areas. The Nb enrichment of dendritic zones can be explained by the difference of the melting temperatures between the constituent elements. During the solidification process, a Nb-rich solid phase (dendrite) is formed in the beginning because of the higher melting temperature of Nb as compared to the other elements, whereas a liquid phase poor in Nb will later form the interdendritic areas. This microstructure is commonly encountered in refractory HEAs synthesized by high-temperature arc melting and usually has no influence on hydrogen sorption properties [
21,
22,
27,
28].
The structural and microstructural characterizations demonstrate that this composition is a single-phase bcc alloy with typical dendritic microstructure. Thus, the next section is dedicated to the study of the hydrogen sorption properties, such as kinetics of absorption, thermodynamics of the reaction with hydrogen, structure investigation of hydride phases, and the phase transformation pathway.
The kinetics of hydrogen absorption was acquired at 25 °C under 44 bars of hydrogen pressure (
Figure 2A), after an activation step described in the “Materials and Methods” section. This curve shows extremely fast kinetics of absorption, with 95% of the maximum absorbed capacity reached within 30 s. The maximum capacity of absorption is 2.0 H/M (2.98 wt.%), which is higher than 1.7 H/M (2.5 wt.%) obtained for the quaternary alloy [
24] and comparable to the maximum uptakes already reported for Ti, V, Zr, and Nb elemental hydrides [
29]. Moreover, the capacity obtained for Ti
0.
30V
0.
25Mn
0.
10Zr
0.
10Nb
0.
25 perfectly matches the values from similar compositions, such as Ti
0.
30V
0.
25Zr
0.
10Nb
0.
25M0.
10, with
M = Cr [
21], Mo [
22], and Ta [
25]. The dihydride and the corresponding deuterated phases were characterized by SR-XRD (
Figure 2B) and neutron diffraction (
Figure 2C), respectively. The related Rietveld analyses of the diffraction patterns confirm that the dihydride adopts a
fcc structure (
) with a lattice parameter listed in
Table 2. However, a small amount of secondary phase, possibly the
bcc monohydride phase, is present in the neutron diffraction pattern (star symbols in
Figure 2C). Rietveld refinement on neutron diffraction data confirms that the deuterium (hydrogen) atoms occupy the tetrahedral sites of the
fcc structure (¼, ¼, ¼), in good agreement with previous findings for similar HEAs [
19,
21,
22,
30].
To determine the thermodynamic properties of a reaction with hydrogen, i.e., the enthalpy and entropy of dihydride formation, several pressure-composition isotherms were measured at 25, 227, 255, and 280 °C, and the Van’t Hoff method was applied on the three data obtained at high temperature only (
Figure 3). A two-step reaction with hydrogen is observed in the PCI curves recorded at high temperature, whereas at 25 °C, this behavior is not noticeable due to the low equilibrium pressure values (within the error bar of the pressure transducer). A first transition occurs at a very low pressure (below the minimum measurable value of the pressure transducer), resulting in the formation of a monohydride phase with a capacity of around 1 H/M. A second reaction arises at a higher pressure, forming a dihydride phase (2 H/M) with a very sloped plateau. This two-step behavior is similar to the hydrogen absorption reaction observed for Ti
0.
30V
0.
25Cr
0.
10Zr
0.
10Nb
0.
25 [
21], and contrasts the one-step transition noticed for Ti
0.
30V
0.
25Zr
0.
10Nb
0.
25Mo
0.
10 [
22]. This highlights the importance of the nature of the fifth element added in the pristine quaternary alloy Ti
0.
325V
0.
275Zr
0.
120Nb
0.
275 [
24].
With the increase in temperature from 227 to 280 °C, the equilibrium pressure of the second transition increases, which allows for the calculation of the variation of the enthalpy (
) and the entropy (
) (
Table 3). The modulus value of the enthalpy is quite large, suggesting the formation of a very stable dihydride. However, the entropy value is larger than the expected value of H
2 gas entropy, −130 J/K∙molH
2. Two reasons might be invoked to explain this behavior: the slopping nature of the PCI measurements and the short range of the PCI temperatures (from 227 to 280 °C). The first reason makes the determination of mid-plateau values difficult, whereas the second one may introduce errors in the linear regression of the Van’t Hoff plot.
For the sake of comparison, the enthalpy and entropy of hydrogen absorption in other refractory HEAs are also listed in
Table 3. It is obvious that the Mn addition in this series of alloys is responsible for the formation of a very stable dihydride phase, which is more stable than the related Cr-containing composition (Ti
0.
30V
0.
25Cr
0.
10Zr
0.
10Nb
0.
25) or other HEAs [
21].
An investigation of the local structure of Ti
0.
30V
0.
25Mn
0.
10Zr
0.
10Nb
0.
25H
2 at short- and medium-range order was carried out using X-ray total scattering and related pair distribution function (PDF) analysis, using the G(R) formalism. The local structure of the hydride in terms of the real-space distribution of interatomic distances is shown from the PDF profile in
Figure 4 together with the corresponding refinement, using an average
fcc lattice with a random distribution of metallic atoms. The refinement results for the
fcc lattice parameter and the isotropic atomic displacement parameter (U
iso) are given in
Table 2. A good agreement is observed between the values of the lattice parameters obtained by SR-XRD and PDF.
The difference curve between the experimental and the calculated profiles suggests that the medium structure (high R region) and the local structure (low R range) can be well described by an undistorted random
fcc lattice. All observed features in the PDF data are well explained by an average random
fcc structure without any sign of strong lattice distortion at a short range order or elemental segregation. Recently, it has been demonstrated by combined X-ray and neutron PDF analysis that the lattice distortion initially present in as-cast refractory
bcc alloys can be relieved by the formation of dihydrides with an
fcc lattice [
32]. Previously, we have reported the PDF profiles for two related dihydrides: Ti
0.
325V
0.
275Zr
0.
125Nb
0.
275H
1.
7 and Ti
0.
30V
0.
25Zr
0.
10Nb
0.
25Mo
0.
10H
2 [
22]. These PDF profiles are very similar to the present one, hinting to the same local order. This might be understood by the comparable crystalline structure (
fcc) and close chemical compositions (90% of atoms are identical).
Based on PCIs, SR-XRD, neutron diffraction and PDF results, the hydrogen absorption follows a two-step reaction: from the as-cast alloy to an intermediate monohydride at a low pressure and subsequently, to a dihydride phase at higher pressures. The same sequence has been already reported for other refractory HEAs, such as Ti
0.
30V
0.
25Cr
0.
10Zr
0.
10Nb
0.
25 [
21], TiVNb, and Al
0.
10Ti
0.
30V
0.
30Nb
0.
30 [
31]. Therefore, it is important to investigate the crystalline structure as a function of hydrogen content. Several samples with different hydrogen absorption capacities, such as 2.0 H/M, 0.8 H/M, 0.5 H/M, and a fully desorbed sample (0 H/M), were studied by X-ray diffraction (laboratory XRD and SR-XRD), as displayed in
Figure 5. The refined lattice parameters and phase fractions are listed in
Table 4.
The dihydride phase crystallizes in an fcc lattice, as previously demonstrated by several experimental methods, whereas the intermediate monohydride with 0.8 H/M is a bcc single phase with abcc = 3.340 (6) Å, which is larger than abcc = 3.239 (5) Å of the as-cast alloy. The material with 0.5 H/M consists of a mixture of two bcc phases: one is the monohydride phase with a slightly smaller lattice parameter (abcc = 3.261 (1) Å and phase abundance: 54%) and the second one is a bcc solid solution with hydrogen (abcc = 3.243 (1) Å and phase abundance: 46%).
The completely desorbed sample has the same bcc lattice as the as-cast sample, without any sign of phase segregation, but with a slightly larger lattice parameter abcc = 3.251 (2) Å than the pristine as-cast alloy abcc = 3.239 (5) Å. This discrepancy might be accounted for by an uncomplete hydrogen desorption under the present experimental conditions.
Further investigation of the reversibility of hydrogen/deuterium desorption in Ti
0.
30V
0.
25Mn
0.
10Zr
0.
10Nb
0.
25 was carried out using in situ neutron diffraction. The diffraction patterns have been recorded for the deuteride Ti
0.
30V
0.
25Mn
0.
10Zr
0.
10Nb
0.
25D
2 phase during heating from 40 to 450 °C, with a temperature rate of 1 °C/min under dynamic vacuum (
Figure 6).
Starting from the bottom to the top of
Figure 6 (left), the initial
fcc deuteride phase (marked with #) is stable until a sharp transition occurs in the temperature range 240–320 °C, with a maximum desorption rate at 280 °C. The deuteride
fcc phase abruptly disappears at 280 °C due to deuterium desorption and only weak diffraction peaks (marked with *) from the
bcc-desorbed phase are barely visible. This is due to the low thermal neutron cross section of the desorbed alloy, as also reported for the quaternary Ti
0.
325V
0.
275Zr
0.
125Nb
0.
275 alloy. The latter composition undergoes a similar sharp transition due to deuterium desorption in a single event, at around 270 °C [
23]. The very close desorption temperatures noticed by in situ neutron thermo-diffraction hints to a negligible effect of Mn addition on the deuterium desorption properties. This finding is in contrast with the results of the additions of 10 at.% of Al, Cr, and Ta in the quaternary alloy that have proven to significantly lower the desorption temperature [
21,
23].
A structural analysis of the
bcc desorbed phase cannot be exploited from in situ neutron diffraction due to very weak diffraction peaks in the absence of deuterium. Thus, for an in-depth investigation of the phase transition during the desorption, in situ SR-XRD patterns were analyzed starting from the dihydride Ti
0.
30V
0.
25Mn
0.
10Zr
0.
10Nb
0.
25H
2 (
Figure 7).
Figure 7A shows in situ SR-XRD thermo-diffractograms during the hydrogen desorption from the dihydride Ti
0.
30V
0.
25Mn
0.
10Zr
0.
10Nb
0.
25H
2, while heating from 25 to 450 °C under dynamic vacuum (5 °C/min). The
fcc dihydride phase (marked with white circles) is stable up to around 275 °C. Above this temperature, the dihydride undergoes a sharp transition to a
bcc-desorbed phase (marked with diamond symbols), in very good agreement with in situ neutron diffraction data. Rietveld analysis was applied to gain in-depth information on the phase transition: the thermal variation of lattice parameters and phase fractions are illustrated in
Figure 7B,C, respectively. A thermal expansion of the
fcc lattice parameter from 4.478 (1) Å to 4.482 (1) Å is noticed upon a temperature increase from 25 to around 211 °C (
Figure 7B). Above this temperature, the lattice parameter slightly decreases due to the beginning of the hydrogen desorption. In the 282–312 °C range, a progressively vanishing
fcc dihydride coexists with a newly formed
bcc phase. The lattice parameter of the dihydride
fcc phase strongly decreases from 4.482 (1) Å to 4.443 (1) Å, and the phase fraction diminishes from 100 to 6% (
Figure 7C). Above 312 °C, only the desorbed
bcc phase is visible with the lattice parameter continuously decreasing from 3.318 (1) Å to 3.250 (1) Å at the maximum temperature of 450 °C. This value is slightly larger than 3.239 (5) Å for the as-cast alloy that might be described by thermal expansion.
Intriguingly, both in situ SR-XRD and neutron diffraction experiments hint to a single step desorption (
fcc dihydride →
bcc phase), whereas the PCIs and ex situ XRD at different hydrogen concentrations have demonstrated a two-step reaction with hydrogen:
fcc dihydride ↔
bcc monohydride ↔
bcc desorbed phase. The second transition from
bcc monohydride to a completely desorbed
bcc phase could not be detected by in situ SR-XRD. One possible explanation might be based on the very small displacive rearrangement of atoms involved into the transition from the
bcc monohydride to the
bcc solid solution, as already observed for the closely related alloy Ti-V-Cr-Zr-Nb [
21]. However, the in situ experiments clearly demonstrated full reversibility, suggesting a structurally stable alloy during the absorption/desorption cycling of hydrogen.
3. Materials and Methods
The Ti
0.
30V
0.
25Mn
0.
10Zr
0.
10Nb
0.
25 alloy was prepared by high-temperature arc melting under Ar atmosphere to avoid any contaminations, as described previously [
25]. This is the most widespread method to prepare refractory HEAs [
27]. The samples as both coarse and fine powders can be manipulated in air without any pyrophoric risk. However, they are stored into an Ar glove box to minimize the air exposure and surface oxidation.
The microstructure of the as-cast alloy was analyzed with a scanning electron microscopy (SEM), using a Zeiss Merlin microscope. Furthermore, energy dispersive X-ray spectroscopy (EDS) was used for the chemical mapping. Prior to the analysis, the sample was exposed to air for at least 24 h, in order to be immobilized in an epoxy resin, polished and finally coated with 1.9 nm of Pd, as already reported for the SEM-EDS analysis of HEAs [
33,
34,
35].
The crystalline structure of the materials in different forms (as-cast, hydride, desorbed, etc.) was studied via X-ray powder diffraction (XRD), using both a laboratory X-ray diffractometer D8 advance Bruker (Cu Kα radiation, Bragg-Brentano geometry) and synchrotron radiation measurements (SR-XRD) at the CRISTAL beamline in the SOLEIL facility. The powder samples were handled and measured in air for all ex situ experiments. Ex situ (λ = 0.7289 Å) and in situ (λ = 0.67156 Å) SR-XRD diagrams were recorded using the Mythen2 detector system (Dectris Ltd., Switzerland installed on the powder diffractometer. To minimize X-ray absorption during the in situ experiment, the sample was mixed with a fumed silica powder and enclosed in 0.2 mm diameter quartz capillary tube. The powder mixture was manipulated and loaded into the quartz capillary under air, with a limited time exposure. For ex situ measurements, the powder samples were loaded and measured in glass capillaries in air.
During the hydrogen desorption from the hydride phase, the SR-XRD patterns were recorded while applying a constant temperature ramp (5 °C/min) from 25 to 450 °C under dynamic secondary vacuum. These in situ measurements were acquired with a 3 min/pattern acquisition time in the 0.2 to 65° 2θ range. The time of acquisition was 10 min, with a scanning range from 1° to 85° (2θ) for these ex situ experiments.
High-energy synchrotron XRD measurements on the powder hydride phase were carried out at beamline ID15A at ESRF, Grenoble (France) [
36]. The incident flux was normalized using a diode placed in front of the sample. Diffraction patterns were collected in transmission geometry by using a Pilatus3 X CdTe 2M hybrid photon-counting detector. The powder sample was handled and loaded into a glass capillary under air. A number of diffraction patterns were collected at slightly different detector positions, in order to minimize inhomogeneity effects and cover dead zones. The detector was off-centered with respect to the 99 keV incident beam and positioned close to the sample to access Q values of up to ~30 Å
−1. The images were corrected for detector geometry, response, distortion, transparency, and for background contributions and X-ray polarization. Data were azimuthally integrated using a modified version [
37] of the PyFAI library [
38] to give one-dimensional scattering patterns. The G(r) were calculated from these patterns using routines from the Diffpy-CMI library [
39] with local modifications for outlier rejection and treatment of background. The structural analysis was carried out using the PDFgui free software [
40].
Powder neutron diffraction experiments on deuterated samples were performed at the D1B beamline at the Institute Laue-Langevin (λ = 1.2858 Å) (DOI: 10.5291/ILL.DATA.CRG-2768), in the scanning range from 1 to 128° (2θ). At room temperature, the ex situ measurement was carried out on a vanadium container, whereas for the in situ measurements, the sample was placed in a silica tube. The powder samples have been manipulated and loaded into the respective containers under air with a limited time exposure. The recording of the diffraction patterns was carried out during the desorption of the deuterated powder sample by applying constant heating (1 °C/min) under secondary vacuum. The evolution of the pressure of evolved gas was continuously recorded by a vacuum gauge during the desorption experiment.
The refinement of the collected neutron and SR powder diffraction data was carried out using the Rietveld method with the Fullprof software [
41], whereas the fundamental parameters approach [
42], as implemented in the TOPAS program (Bruker AXS version 7.0.0.7) [
43], was used to refine the laboratory X-ray diffraction patterns.
Hydrogen sorption properties were studied using a homemade volumetric device, water thermalized at 25 °C. Small pieces of the as-cast alloy (300–400 mg) were cut and loaded under air in a stainless-steel container, with an air exposure of less than 10 min. The stainless-steel container with a minimized dead volume was tightly connected to the volumetric device by a single-use metallic gasket, and a leak test was performed before any activation of the sample. Prior to each experiment, the samples were activated by a heat treatment at 350 °C under a dynamic vacuum for 3 h. The hydrogen uptake was calculated using the real equation of states for hydrogen from the program GASPAK V3.32 (Cryodata. Inc., USA). The absorption kinetic curve was recorded during a first absorption step, under 44 bars of H
2 at room temperature (25 °C). Pressure-composition isotherms (PCI) at 25, 227, 255, and 280 °C have been acquired during absorption and further used to calculate the enthalpy and entropy of the reaction using the Van’t Hoff method. During the PCI measurements, each point was in equilibrium and the minimum time per point was around 6 h. Furthermore, these measurements were conducted following the recommended best practices from Gross et al. [
44], and with careful consideration of the potential limiting factors that could affect the repeatability of the results previously outlined by Broom et al. [
45].