Solid-state reactions and hydrogen storage in magnesium mixed with various elements by high-pressure torsion: experiments and first-principles calculations

Abstract

Magnesium hydride is widely known as an interesting candidate for solid-state hydrogen storage. However it is too stable and does not desorb hydrogen at ambient conditions. Although MgH₂ suffers from slow kinetics, its hydrogenation kinetics can be significantly improved by addition of catalysts and/or decreasing the grain size. Reducing the thermodynamic stability of MgH₂ is now the main challenging task. In this study, 21 different elements were added to magnesium in atomic scale by using the High-Pressure Torsion (HPT) technique and different kinds of nanostructured intermetallics and new metastable or amorphous phases were synthesized after HPT (Mg₁₇Al₁₂, MgZn, MgAg, Mg₂In, Mg₂Sn, etc.) or after post-HPT heat treatment (MgB₂, Mg₂Si, Mg₂Ni, Mg₂Cu, MgCo, Mg₂Ge, Mg₂Pd, etc.). In most of the compounds, the desorption temperature decreases by addition of elements, even though that the ternary hydrides are formed only in limited systems such as Mg–Ni and Mg–Co. Appreciable correlations were achieved between the theoretical binding energies obtained by first-principles calculations and the experimental dehydrogenation temperatures. These correlations can explain the effect of different elements on the hydrogenation properties of the Mg-based binary systems and the formation of ternary hydrides.

---

1. Introduction

Mg is considered as a potential candidate for solid-state hydrogen storage because of its high storage capacity, reversible hydriding features and low cost. However, its applications for hydrogen storage are not efficient due to its slow kinetics for hydrogen absorption and desorption as well as the high thermodynamic stability of its hydride (enthalpy of hydrogen desorption for MgH₂ is −76 kJ mol⁻¹ H₂). Despite significant progress on the improvement of hydrogenation kinetics, the reduction of thermodynamic stability of Mg-based hydrides is still a challenging task.¹

Many studies attempted to improve the hydrogenation kinetics of Mg by employing different approaches. Addition of transition metals to Mg as catalytic agents can improve the hydrogen sorption kinetics. As an example, previous studies demonstrated that the hydrogen absorption and desorption kinetics in Mg highly improved by addition of Nb, V, Pd and Ni.^2–10 The addition of transition metal oxides (such as Nb₂O₅, V₂O₅, Cr₂O₃ and Fe₃O₄) also enhance the hydrogen sorption kinetics.^11,12 The most promising effect on the improvement of hydrogen absorption/desorption kinetics is obtained by addition of Nb₂O₅ to Mg during milling by formation of additional sites for hydrogen dissociation in Mg powder.¹¹ Another route to improve the reaction kinetics is the formation of composites of Mg with other intermetallics which react with hydrogen in milder condition. Previous studies showed that, the desorption temperature reduces significantly by formation of the Mg-based composite with LaNi₅, ZrNi and TiFe.^13–15 More recent studies also demonstrated a new class of Mg-based composite with light hydrides such as LiBH₄, NaBH₄ and Ca(BH₄)₂.¹⁶ The formation of carbon-based composite with Mg also leads to the increase of hydrogen storage capacity and decrease of desorption temperature, owing to their high specific surface area and adsorption properties.^17–19

Previous studies also demonstrated that ball milling (mechanical milling), which can lead to the formation of structural defects, phase changes, grain refinement and/or amorphization, can improve the kinetic properties of Mg.^20,21 Earlier works by Zaluska et al. showed that ball milling can improve the morphology and surface activity of Mg powders for hydrogenation.²² Milling under a hydrogen atmosphere is even more effective for grain refinement and storing high content of hydrogen with fast hydrogen sorption kinetics.²² Microstructural modification by severe plastic deformation (SPD) through equal-channel angular pressing (ECAP) technique also leads to acceleration of the hydrogen desorption kinetics.^23–25

The previous works demonstrated that the thermodynamics of hydrogenation reaction in Mg can be improved by alloying Mg with different elements in the atomic scale and formation of intermetallic compounds. The Mg₂Ni is the most famous Mg-based hydrogen storage material.^26–28 Alloying Mg with Ni lowers the enthalpy of hydride formation for Mg₂NiH₄ (ΔH = −64.5 kJ mol⁻¹) comparing to MgH₂ (ΔH = −76 kJ mol⁻¹). In Mg₂NiH₄, hydrogen atoms interact with both Ni and Mg, but the interaction of Ni–H is stronger than Mg–H in spite of the higher affinity of Mg for hydrogen than Ni in binary metal hydrides.²⁹ The Mg₂Cu intermetallic also absorbs hydrogen, but the Mg₂Cu intermetallic decomposes to MgH₂ and MgCu₂ in an irreversible reaction during the hydrogenation.³⁰ Al is also considered as an elemental additive for destabilizing MgH₂.^31,32 Another interesting Mg-based ternary hydride is Mg₂FeH₆. The high volumetric (150 kgH₂ per m³) and gravimetric (5.5 wt%) capacities of Mg₂FeH₆ make it interesting for hydrogen storage applications,³³ but synthesis of this hydride is quite difficult since Mg and Fe cannot form any intermetallics. Mg₂FeH₆ can be formed by mechanical milling of Mg and Fe in the presence of hydrogen. The Mg₂CoH₅ hydride with appreciable volumetric (100 kgH₂ per m³) and gravimetric (4.5 wt%) capacities can also be formed by the reaction of Mg and Co under the atmosphere of hydrogen,³⁴ however synthesizing this hydride is not easy since the Mg₂Co intermetallic does not exist. Ti and Mg are also immiscible elements, however some previous studies showed that Mg₇TiH_x can be formed by the reaction of MgH₂ and TiH_1.9 in a high-pressure anvil.³⁵ The Mg₇TiH_x hydride decompose to Mg and TiH_1.9 by desorbing 4.7 wt% of hydrogen at 332 °C. In addition, the Mg₃Cd intermetallics can also absorb and desorb 2 wt% of hydrogen reversibly at 250–300 °C without any hysteresis between the absorption and desorption curves.³⁶

Mixing Mg with different elements in the atomic scale is in general challenging because of low boiling point of Mg (1091 °C), its high reactivity in the air and its thermodynamic immiscibility in many other elements such as Ti, Zr, Nb, V, and so on. Non-equilibrium processing techniques are usually employed to overcome these technical and thermodynamic limitations. The possible synthesizing routes are vapor deposition and mechanochemistry techniques. The vapor deposition techniques are mostly used for production of thin films.^37–39 However, the mechanochemistry techniques allow the preparation of higher quantity and wider range of materials.⁴⁰ The mechanical alloying by ball milling is one of the most common synthesizing techniques.^41–45 Previous studies showed that, the mechanical alloying is even effective to synthesize new metastable phases from the elements that are totally immiscible in the equilibrium condition. For example, although Mg and Ti are totally immiscible, their processing by ball milling under an Ar atmosphere results in the formation of metastable Mg–Ti phases.⁴⁶ Reactive ball milling under a hydrogen atmosphere was also applied for in situ formation of Mg–Ti hydride.^47,48 Despite significant benefits of ball milling, the whole process should be performed under the inert or hydrogen atmospheres and the processed materials should be handled in a controlled atmosphere to avoid oxidation, and this limits the applicability of the method to synthesize hydrogen storage materials.

In this study, a Severe Plastic Deformation (SPD) technique under high pressure, so-called High-Pressure Torsion (HPT),^49–51 is applied to mix 21 different elements with Mg at the atomic scale. In the HPT method, a high plastic shear strain is introduced to the disc-shaped or powder samples by torsional deformation under high pressure (>1 GPa). As a result, significant grain refinement occurs in the samples. Earlier studies showed that the HPT method is effective for atomic-scale mixing of elements,^52–55 synthesis of intermetallics,^56–58 nanostructuring and/or amorphization of intermetallics,^59,60 acceleration of polymorphic phase transformations,^61–63 activation of hydrogen storage materials^64,65 and enhancement of hydrogenation kinetics.^66–70 Moreover, in this technique the whole process is performed under air and consequently the processed material is less sensitive to surface oxidation. This advantage makes the HPT technique quite distinctive comparing to other non-equilibrium processing techniques such as ball milling. In this study, in addition to the experimental studies, the first-principles calculations are also performed to investigate the electronic effect of elemental additives on the hydrogen binding energy. In order to improve the thermodynamic properties of Mg-based alloys for hydrogen storage applications, it is essential to understand the effect of added elements on the hydrogen binding energies and electric charge density, as discussed in earlier publications.^71–73

2. Experimental materials and methods

Mg powders were mixed with 33.3 mol% powders of 21 elements to form Mg₂X compositions. The purity levels and particle sizes of powders were 99–99.999% and 2–300 μm, respectively, as given in Table 1. HPT was conducted at room temperature for N = 100 turns under a pressure of P = 3 GPa and the powder mixtures were mechanically mixed and consolidated to discs with 14 mm diameter and 0.8 mm thickness. A bulk form of annealed Mg, which was used in a previous study,⁶⁹ was also used as reference.

Table 1 Purity level and particles sizes of elemental additives and phases formed after HPT, after hydrogenation in DSC and after dehydrogenation in TGA/DTA for “Mg₂X” materials

Additive in Mg2X	Atomic no.	Purity (%)	Particle size (μm)	Phases after HPT	Phases after hydrogenation	Phases after dehydrogenation
B	5	99	<45	Mg, B		Mg, MgB2
C	6	99.99	10	Mg, C		Mg, C
Mg	12	99.9	<180	Mg	MgH2, Mg	Mg
Al	13	99.99	<75	Mg, Al, Mg17Al12, amorphous		Mg, Mg17Al12
Si	14	99.999	<75	Mg, Si	Mg2Si, MgH2, Mg, Si	Mg, Si, Mg2Si
Ti	22	99.9	63–90	bcc, fcc, hcp	Mg, MgH2, TiH2	Mg, Ti, TiH2
V	23	99.9	<300	Mg, V		Mg, V
Cr	24	99.99	<180	Mg, Cr		Mg, Cr
Mn	25	99.9	<75	Mg, Mn		Mg, Mn
Fe	26	99.999	<10	Mg, Fe	MgH2, Mg, Fe	Mg, Fe
Co	27	99	5	Mg, Co	Mg2CoH5, MgH2, Co	Mg, MgCo
Ni	28	99.99	<150	Mg, Ni	Mg2NiH4, MgH2, Ni	Mg, Ni, Mg2Ni
Cu	29	99.9	<75	Mg, Cu		Mg, Cu, Mg2Cu
Zn	30	99.99	<75	MgZn, amorphous	MgZn, Mg	Mg, MgZn
Ge	32	99.99	<45	Mg, Ge	Mg2Ge, MgH2, Mg, Ge	Mg, Ge, Mg2Ge
Zr	40	99.5	—	Mg, α-Zr, ω-Zr		Mg, α-Zr, ZrH2
Nb	41	99.9	<45	Nb, Mg		Nb, Mg, NbH0.9
Mo	42	99.99	2–3	Mg, Mo	MgH2, Mg, Mo	Mg, Mo
Pd	46	99.9	—	Mg, Pd		Mg, Mg2Pd, MgPd
Ag	47	99.99	<45	Mg, Ag, MgAg	MgH2, Mg, Ag	Mg, MgAg
In	49	99.99	<45	Mg, Mg2In, MgIn	MgH2, Mg, In	Mg, Mg2In, MgIn
Sn	50	99.99	<38	Mg, Sn, fcc-Mg2Sn, hcp-Mg2Sn		Mg, Sn, fcc-Mg2Sn

---

Following the HPT processing, Vickers microhardness was measured with an applied load of 100 g for 15 s at 2–6 mm away from the disc center. Hydrogenation properties of the samples were examined using differential scanning calorimetry (DSC) at 25–500 °C with a heating/cooling rate of 5° min⁻¹ under a hydrogen pressure of 2 MPa. The samples after cooling in DSC were quickly examined by thermogravimetry analysis (TGA) and differential thermal analysis (DTA) at 25–500 °C with a heating/cooling rate of 5° min⁻¹ under an Ar pressure of 0.1 MPa. X-ray diffraction (XRD) analysis was performed before HPT, after HPT, after hydrogenation in DSC and after dehydrogenation in TGA/DTA using the Cu Kα radiation. For microstructural examinations, thin foils were prepared from the disc samples at ∼6 mm away from the disc center with a focused ion beam system. Transmission electron microscopy (TEM) was performed at 300 kV for microstructural observation and for recording selected-area electron diffraction (SAED) patterns.

3. Calculation methods

Density Functional Theory (DFT) calculations were performed with all-electron Projector Augmented Wave (PAW) method⁷⁴ as implemented in Vienna Ab initio Simulation Package (VASP).^75,76 The exchange correlation potential was of the Generalized Gradient Approximation (GGA) type as formulated by Perdew, Burke and Ernzerhof (PBE).⁷⁷ For all calculations, the structures were fully relaxed by considering the force and stress minimization. In order to have comparable results for the calculated free energies, the same cutoff energy and similar k-grid density were used for convergence. An energy cut-off of 400 eV was used for plane-wave basis set for all geometry optimizations. In all calculations, 8 × 8 × 8 k-points in the whole Brillouin zone were used for MgH₂-type structure and a similar density of k-points was used for all structural arrangement. The calculations were performed for the stable structures according to the experimental results or simplified structural models.

The binding energy of hydrogen for all hydride structures, E_H, was calculated as below:

E_H = E₀ − (E₁ + 0.5E_H–H) (1)

where, E₀ is the energy of hydride, E₁ is the energy of hydride after removing one hydrogen atom (without allowing the structure to be relaxed) and E_H–H is the energy of one hydrogen molecule in the hydride unit cell. The charge density difference, C_H, was calculated as:

C_H = C₀ − (C₁ + C⁰_H) (2)

where C₀ is the total charge density distribution in the hydride, C₁ is the charge density after removing one hydrogen atom and C⁰_H is the charge density of hydrogen atom located in the same position as the removed hydrogen atom in the hydride unit cell. Bader population analysis⁷⁸ was performed to allocate electron density and a VESTA program was used for three-dimensional visualization of the density of electrons.^79,80

4. Results and discussion

4.1. Experimental results

Following the HPT process, the Vickers microhardness was measured for all samples. Fig. 1 shows the hardness variation for different materials after HPT processing. The hardness increases with HPT processing and/or with addition of elemental additives in accordance with many earlier publications.^58–63 The unusual increase in the hardness in several systems such as Mg–Zn, Mg–Ti, Mg–Al, Mg–Sn, Mg–In and Mg–Ag is the first indication of the formation of intermetallics, metastable alloys or amorphous phases. Earlier studies also showed that the hardness increase becomes more significant in Al-based materials by in situ formation of intermetallics during the HPT processing.^55–57

Fig. 1 Vickers microhardness against atomic number of elemental additives in “Mg₂X” samples processed by HPT.

Table 1 summarizes the XRD analysis and gives the crystal structures of “Mg₂X” materials after processing with HPT, after hydrogenation in DSC and after dehydrogenation in TGA/DTA. XRD spectra are shown in Fig. 2 for selected materials. Table 1 and Fig. 1 and 2 indicate several important aspects: (i) solid-sate reactions occur during HPT and Mg₁₇Al₁₂, MgZn, MgAg, Mg₂In, MgIn and Mg₂Sn intermetallics are formed. (ii) Metastable intermetallics are also formed after the HPT processing such as a high-temperature form of Mg₂Sn with the hexagonal crystal structure and metastable phases of Mg–Ti with the fcc, bcc and hcp structures. Note that no binary phases form in the equilibrium phase diagram of Mg–Ti, but the formation of these phases was previously reported by means of ball milling.^81,82 Details of phase transformation in the Mg–Ti system during HPT processing are given in an earlier publication.⁸³ (iii) The significant peak broadening in the Mg–Al and Mg–Zn systems indicates that amorphization should have occurred. (iv) Numerous intermetallics form after heat treatment in DSC and TGA such as MgB₂, Mg₂Si, Mg₂Ni, Mg₂Cu, MgCo, Mg₂Ge, Mg₂Pd and MgPd. (v) Hydrides of ZrH₂ and NbH_0.9 remain stable after heating to 500 °C in an Ar atmosphere. (vi) Ternary Mg-based hydrides could be detected only in the Mg–Co and Mg–Ni systems. The current results indicate that the HPT processing and post-HPT heating are quite effective for the formation of binary Mg-based intermetallics.

Fig. 2 XRD profiles for 5 selected “Mg₂X” samples before and after processing with HPT.

In order to have better understanding about the grain size and microstructure of samples which exhibited the formation of intermetallic phases after the HPT processing, the TEM observations were performed on these samples. TEM micrographs and corresponding SAED patterns are shown in Fig. 3 for several materials after HPT processing. The microstructures of “Mg₂X” samples consist of nanograins with average grain sizes smaller than 100 nm, while the grain size of pure Mg is ∼1 μm after HPT processing.⁶⁹ In “Mg₂Ti” sample, a texture develops and a nano-lamellar structure forms, as shown in Fig. 3(b). In “Mg₂Al” and “Mg₂Zn” samples, an amorphization occurs as evident from the hallo SAED pattern in Fig. 3(a) and (c) as well as from the XRD patterns in Fig. 2(a) and (c). Formation of amorphous phases was also reported in the TiNi⁶⁰ and TiFe⁶⁴ intermetallics after processing by HPT. The formation of nanograins in the “Mg₂X” samples after HPT processing should be due to the presence of several phases^52,53 and/or intermetallics.^56–60

Fig. 3 TEM bright-field images (right), SAED patterns (center) and dark-field images (left) for 5 selected “Mg₂X” samples. Dark-field images were taken with diffracted beams indicated by arrows in SAED patterns.

TGA and DTA curves are shown in Fig. 4 for several representative samples and the whole TGA/DTA results are summarized in Fig. 5(a) and (b). The samples before TGA/DTA were hydrogenated by heating and cooling during DSC at 25–500 °C under a hydrogen pressure of 2 MPa. Decomposition of hydrides results in a weight loss in TGA curves and an exothermic peak in the DTA curves. Fig. 4, 5(a) and (b) indicate three important aspects.

Fig. 4 TGA/DTA profiles for 5 selected “Mg₂X” samples including pure Mg processed by HPT. Hydrogenation was conducted during DSC at 25–500 °C under a hydrogen pressure of 2 MPa.

Fig. 5 (a) Hydrogen desorption temperature and (b) amount of hydrogen measured by TGA/DTA under Ar atmosphere compared with nominal hydrogen storage in MgH₂ against atomic number of elemental additive in “Mg₂X” samples. Hydrogenation was conducted during DSC at 25–500 °C under a hydrogen pressure of 2 MPa.

First, for pure Mg, the temperature for hydrogen desorption decreases from 412 °C to 357 °C by HPT processing and the amount of stored hydrogen significantly increases. An examination confirmed that a commercial MgH₂ powder also desorbed hydrogen at 410 °C. The increase in hydrogen storage capacity by HPT should be due to the improvement of hydrogenation kinetics by HPT processing, in consistency with earlier publications,^5,66–70,84 and the decrease in desorption temperature should be also due to the presence of lattice defects and their effect on hydrogenation kinetics.^49–51

Second, in most compositions, one peak appears by hydrogen desorption and the desorption temperature decreases comparing to the one for pure Mg, especially by addition of Fe, Nb and V. The decrease of desorption temperature in these samples should be related to the improvement in the kinetics of hydrogen desorption. Addition of Si and Ge results in no significant change in the desorption temperature. Surprisingly, Mg₂Pd and Mg₂Sn intermetallics do not absorb hydrogen in spite of good catalytic property of Pd and Sn.^4,85 As will be shown later, this behavior of Mg₂Sn and Mg₂Pd is due to the instability of their ternary hydrides, while hydrogen is stored in the form of MgH₂ in Mg with addition of Pd and Sn as catalytic agents.

Third, in some compositions such as “Mg₂Ni” and “Mg₂Co”, ternary hydrides are formed and several exothermic peaks appear in DTA and the slop of TGA curves differs for different hydrides. For these compounds the decrease of desorption temperature appears to be much more significant. This effect could be due to the fact that both the kinetics and thermodynamics of hydrogen desorption improves by addition of these elements. As an example, three desorption temperatures appears in Fig. 4(c) for “Mg₂Co” sample. Ivanov et al.⁴³ suggested that the three peaks correspond to MgH₂, Mg₂CoH₅ and Mg₃CoH₅. Our XRD analyses conducted after hydrogenation in DSC facility also shows the presence of MgH₂ and Mg₂CoH₅ after DSC (see Table 1).

Fourth, hydrogen storage in all selected materials is less than the nominal amounts because of rather large size of samples (∼0.8 mm thickness) and slow kinetics. Among the selected samples, “Mg₂Cu”, “Mg₂Co” and “Mg₂Ni” exhibit excellent kinetics and store more than 70% of their nominal hydrogen capacity. It should be noted that, since mechanical treatment may affect the structural features of the HPT-processed materials, any mechanical crashing or filling of samples was avoided in current experiments.

The destabilization of MgH₂ with addition of X being from group 14th (according to the IUPAC notation) elements (X = C, Si, Ge, Sn) was of great interest for many researchers.^86–90 The Mg₂Si compound especially has received much attention because it can have a high storage capacity due to its lightweight.^86–88 Guinet et al. showed that Mg₂Si can absorb 2 wt% of hydrogen after 10 h at 300 °C.⁸⁶ Some studies indicated that Mg₂Si cannot be hydrogenated by conventional solid–gas reaction at 10 MPa (at 300 °C) or 185 MPa (at 350 °C).^87,88 However, the reversible hydrogenation of Mg₂Si through the conversion reaction 2MgH₂ + Si ↔ Mg₂Si + 2H₂ is reported previously.⁸⁷ The direct reaction could be fully obtained by milling Mg₂Si at the low rotation speed. The reverse reaction also can be achieved by milling at high rotation speeds but in the presence of some amount of Mg₂Si as a catalyst. Previous attempt on hydrogenating the Mg₂Ge and Mg₂Sn, showed that in contrary to Mg₂Si, they do not absorb hydrogen and they do not decompose.⁸⁷ However, some studies reported that the addition of these elements in lower quantity can actually have catalytic property and leads to the decrease of desorption temperature. For example, the effect of 5 at% Ge on the desorption of MgH₂ was examined by Gennari et al. and they reported that the desorption temperature decreases by addition of Ge.⁹⁰

In order to have a better understanding about the effect of group 14th elements on the hydrogenation properties of Mg, three cycles of hydrogen absorption and desorption were performed on “Mg₂X” (X = Si, Ge, Sn) samples by heating and cooling under 2 MPa of hydrogen pressure. Since these samples behaved similarly, as a representation for these measurements only the result corresponding to Mg₂Si compound is shown in Fig. 6(a). Fig. 6(a) shows that the hydrogen absorption and desorption occur in the sample by forming an exothermic and endothermic peak. However, the intensity of these peaks decreases sequentially by increasing the number of cycles. Since the XRD analysis showed only the existence of the Mg and Si phases after the HPT processing, the hydrogen absorption in the first cycle occurs in the Mg phase. At the end of first cycle, the XRD analysis shows the existence of MgH₂ phase beside the intermetallic Mg₂Si and residual Si. In the following cycles, the hydrogen is desorbed from MgH₂ and at the end of second heating cycle larger fraction of Mg₂Si is formed. During the cooling the sample in the second cycle, hydrogen is absorbed in the sample but less Mg is available for hydrogen absorption and thus the intensity of peak referring to the hydrogen absorption decreases. At the end of third heating cycle, the formation of Mg₂Si becomes almost complete and consequently the intensity of peak referring to the hydrogen absorption becomes almost negligible. In agreement with the previous studies, these results show that the Mg₂Si intermetallic cannot absorb hydrogen to form ternary Mg₂SiH_x hydride.⁸⁷

Fig. 6 Cycling behavior of (a) “Mg₂Si” and (b) “Mg₂Co” samples conducted in DSC at 25–500 °C under a hydrogen pressure of 2 MPa.

In order to confirm the formation of intermediate hydride phases in “Mg₂Co” sample, the hydrogen absorption and desorption cycling were also performed on this sample, as shown in Fig. 6(b). In agreement with the TGA results of “Mg₂Co” sample, three steps for hydrogen absorption and desorption are also observed during heating and cooling cycles under a hydrogen atmosphere. The previous studies also indicated that hydrogen absorption and desorption in “Mg₂Co” occurs through the formation of intermediate hydride phase.⁴³ The peaks with the higher intensities refer to the hydrogen absorption and desorption in ternary hydrides and the peak with lowest intensity refers to the hydrogen absorption and desorption in residual Mg in the form of MgH₂. As explained earlier, the two steps hydrogen absorption/desorption refers to the formation of two ternary hydride phases of Mg₃CoH₅ and Mg₂CoH₅.⁹¹ At the end of desorption in TGA, MgCo phase is formed, as indicated in Table 1. The previous study performed by Gennari et al. also confirmed the formation of MgCo phase at temperatures below 450 °C. Formation of this phase competes with the formation of MgCo₂ intermetallic phase which usually form at ∼500 °C. However, the XRD analysis in our study only shows the formation of MgCo phase even though the temperature reached 500 °C during TGA. This effect can be due to the kinetics limitation as well as due to the differences in the synthesizing routes used in this study and in previous works.⁹²

4.2. Calculation results

Following the experimental studies, the DFT calculations were performed in order to investigate the electronic effect of substituted elements on the binding energy of hydrogen in Mg-based hydrides. It was attempted to find whether there is any correlation between the hydrogen binding energy and the desorption temperatures in the hydrides with substituted elements. The calculations were first performed for MgH₂ with the tetragonal structure (S.G P4₂/mnm). It was assumed that hydrogen atoms occupy trigonal sites with Wyckoff position 4f (0.3048, 0.3048, 0). These sites are coordinated by three atoms of Mg. In order to investigate the stability of hydrides formed in the “Mg₂X” samples, the phases formed after the HPT processing were used as structural models. For elements which do not form intermetallic phases after HPT processing (X = V, Nb, Fe, etc.), the structure of parent MgH₂ was used as the structural model. However, in order to consider the effect of these elements on the hydrogen binding energy, they were substituted in the Mg site at (0.5, 0.5, 0.5) position. In some compositions the intermetallics are also formed after the HPT processing, however the formation of ternary hydrides was not found for these compounds (X = Zn, Pd, Ag, In, Sn, etc.) neither in this study or previous works. The structural models for these compositions were constructed by inserting the hydrogen atoms in the interstitial sites by considering the Switendick (the length between H–H atoms ≥ 2.1 Å) and Westlake criteria (the radius of hydrogen site ≥ 0.4 Å).^93,94 For elements which produce ternary hydrides with Mg such as Ni and Co, the structures of ternary hydrides were modeled for the DFT calculations.

The calculated hydrogen binding energies of some selected elements, which influence the dehydrogenation temperature more significantly are given in Table 2. The addition of these elements (V, Nb, Fe, Co and Ni) decreases the desorption temperature mostly by improving the kinetics of hydrogen desorption. It has been shown that the addition of Pd as a catalytic agent can reduce the dehydrogenation temperature of MgH₂,¹⁵ thus the calculation was also performed for Pd even though that the addition of this element did not lead to the decrease of desorption temperature in this study. The structure of MgH₂ was used as the structural model for calculation of binding energies. In order to investigate the influence of these elements on the binding energy of hydrogen they have been substituted in Mg site at (0.5, 0.5, 0.5) position. Table 2 shows that the hydrogen binding energy decreases by addition of these elements comparing to the one for MgH₂.

Table 2 Experimentally measured hydrogen desorption temperature, calculated hydrogen binding energy and calculated charge density on hydrogen, Mg and X (X = V, Nb, Fe, Ni and Pd) together with electronegativity of additives

X (in MgXH4)	Electronegativity of X	Desorption temperature (°C)	H-binding energy (eV)	Charge density
				H	Mg	X
Mg	1.31	357	−1.133	−1	+2	—
V	1.83	231	−0.616	−0.88	+2	+1.50
Nb	1.54	218	−0.525	−0.82	+2	+1.30
Fe	1.83	204	−0.405	−0.72	+2	+0.88
Co	1.88	250	−0.473	−0.65	+2	+0.60
Ni	1.60	191	−0.250	−0.63	+2	+0.50
Pd	1.96	No absorption	−0.054	−0.58	+2	+0.30

---

A comparison between the desorption temperatures and the binding energies suggest that there is a correlation between the evolutions of desorption temperatures and binding energies of hydrogen. By addition of elements such as V, Nb, Fe, Co and Ni, the hydrogen binding energy and the desorption temperature decrease comparing to those for MgH₂. The minimum desorption temperature was obtained by addition of Ni, while the calculations indicate that the binding energy also decreases to −0.250 eV by addition of Ni. By substitution of Pd this value further decreases to −0.054 eV, which may indicate that this binding energy is quite low to accommodate any hydrogen in this structure. In agreement to this result, no hydrogen absorption was observed for the “Mg₂Pd” sample. Even though that the addition of these elements mostly influence the kinetics of hydrogen desorption, they may also influence the thermodynamics of hydrogen desorption at the interfaces since an obvious correlation exist between the desorption temperatures and the hydrogen binding energies. The calculated distribution of charge densities are also given in Table 2. It is clear that the elemental additives that reduce the charge density of hydrogen significantly are more effective to reduce the binding energy of hydrogen and reduce the desorption temperature.

In order to investigate the stability of ternary Mg-based hydrides, the calculations were also performed for the intermetallic phases which were formed by HPT processing or by post-HPT heat treatment. The results of these calculations are given in Table 3 for some selected compositions. The hydrogen binding energies for the compositions which form thermodynamically stable ternary hydrides (Mg₂CoH₅ and Mg₂NiH₄) decrease comparing with that for MgH₂. For the intermetallics which do not have any reported hydride phases, the calculations also performed by considering the hydrogen atoms in the interstitial sites of intermetallic phase. In all these compositions, the hydrogen binding energies show positive values indicating that these intermetallics cannot absorb hydrogen in the form of ternary hydrides. The current results explain well why only two ternary hydrides (Mg₂CoH₅ and Mg₂NiH₄) were detected in this study and why Mg₂Sn and Mg₂Pd did not absorb the hydrogen.

Table 3 Calculated hydrogen binding energies for some selected compositions of ternary hydrides

X (in Mg2X)	Formation of ternary hydride	Composition of hydride	H-binding (eV/H)
Mg	—	MgH2	−1.134
Co	Yes	Mg2CoH5	−0.874
Ni	Yes	Mg2NiH4	−1.007
Si	No	Mg2Si–H	+1.358
Ge	No	Mg2Ge–H	+1.342
Pd	No	Mg2Pd–H	+0.030
	No	MgPd–H	+0.455
Ag	No	MgAg–H	+0.333
In	No	Mg2In–H	+0.047
	No	MgIn–H	+0.478
Sn	No	Mg2Sn–H	+1.405

---

The distribution of electric charges in the structures can give a better understanding about the nature of bindings. The results of electron charge density calculation by using the Bader population analysis⁶⁹ are shown in Fig. 7 for binary MgH₂ and some selected ternary hydrides. In binary MgH₂, Mg atoms lose fully their electrons in valence band (charge density of Mg: C_Mg = +2) and hydrogen atoms receive fully these electrons (C_H = −1). In the case of ternary Mg₂NiH₄ and Mg₂CoH₅, each Mg atom loses its valence electrons. However, in addition to the H atoms the Ni and Co also receive the electron charge from Mg atoms and consequently, the electric charge on hydrogen atoms decreases. The reduction of charge density on hydrogen atoms makes the hydrogen binding weaker comparing to that in MgH₂. In Mg₂Si–H, Mg₂Sn–H and Mg₂Pd–H, similar to Mg₂NiH₄ and Mg₂CoH₅, Mg atoms lose the electrons of the valence band (C_Mg = +2) and Si, Sn, Pd and hydrogen atoms receive the electrons. However due to the high electronegativity of Si, Sn and Pd, the charge density on these atoms is significantly higher than the charge density on H atoms (C_Si = −3.46, C_Sn = −3.47, C_Pd = −3.51, C_H = −0.54 in Mg₂Si–H, C_H = −0.53 in Mg₂Sn–H and C_H = −0.36 in Mg₂Pd–H). As a result, less electron charge is available for hydrogen atoms to make binding in the structure and the hydride phases cannot be formed. These calculations can explain why Mg₂Si, Mg₂Sn and Mg₂Pd could not absorb hydrogen in this study.

Fig. 7 The charge density distribution in (a) MgH₂, (b) Mg₂CoH₅, (c) Mg₂NiH₄, (d) Mg₂Si–H, (e) Mg₂Sn–H and (f) Mg₂Pd–H. Brown atoms refer to Mg, white atoms refer to hydrogen and red atoms refer to substituted elements (X = Co, Ni, Si, Sn and Pd). On the selected positions for hydrogen, blue region corresponds to accumulation of electric charge and yellow regions correspond to depletion of electric charge.

5. Conclusions

In summary, this study shows that the HPT method is promising for atomic-scale mixing of Mg with various elements and for synthesizing intermetallic phases for hydrogen storage application. For most of the synthesized Mg-based compound with nominal composition of Mg₂X (X stands for 21 added elements), the hydrogen desorption temperature decreases when compared to conventional Mg sample, while some compounds such as Mg₂Sn and Mg₂Pd do not absorb hydrogen. In most compositions, the decrease of desorption temperature is due to the improvement in the kinetics of hydrogen desorption (especially in X = V, Nb and Fe). However, in some compositions which form ternary hydrides (X = Ni, Co) or produce very stable intermetallics (Mg₂Si, Mg₂Sn and Mg₂Pd), the thermodynamics also influence the hydrogenation properties but not always positively for reducing the dehydrogenation temperature.

The first-principles calculations show that there is a correlation between the desorption temperature and calculated hydrogen binding energy values. Elements which were effective in decreasing the desorption temperature of MgH₂ are also effective in decreasing the binding energy of hydrogen. Elements such as Ni and Co which forms ternary hydrides (Mg₂CoH₅ and Mg₂NiH₄) reduce the hydrogen binding energy when compared to MgH₂. For many other elements which form stable intermetallics such as Mg₂Pd and Mg₂Sn, the hydrogen binding energy becomes positive indicating that they cannot absorb hydrogen in the form of ternary hydrides.

Acknowledgements

One of the authors (KE) acknowledges a grant from Kyushu University Interdisciplinary Programs in Education and Projects in Research Development (P&P) (No. 27513) and a grant from WPI-I2CNER for Interdisciplinary Researches. This work was supported in part by the Light Metals Educational Foundation of Japan, in part by the Grant-in-Aids from the MEXT, Japan (No. 22102004, No. 26220909 and No. 15K14183). The HPT process was carried out in the International Research Center on Giant Straining for Advanced Materials (IRC-GSAM) at Kyushu University.

References

1. I. P. Jain, C. Lal and A. Jain, Int. J. Hydrogen Energy, 2010, 35, 5133–5144.
2. J. F. Pelletier, J. Huot, M. Sutton, R. Schulz, A. R. Sandy, L. B. Lurio and S. G. J. Mochrie, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 63, 052103.
3. H. G. Schimmel, J. Huot, L. C. Chapon, F. D. Tichelaar and F. M. Mulder, J. Am. Chem. Soc., 2005, 127, 14348–14354.
4. G. Liang, J. Huot, S. Boily, A. Van Neste and R. Schulz, J. Alloys Compd., 1999, 291, 295–299.
5. S. Rivoirard, P. de Rango, D. Fruchart, J. Charbonnier and D. Vempaire, J. Alloys Compd., 2003, 356–357, 622–625.
6. J. Charbonnier, P. de Rango, D. Fruchart, S. Miraglia, L. Pontonnier, S. Rivoirard, N. Skryabina and P. Vulliet, J. Alloys Compd., 2004, 383, 205–208.
7. J. Charbonnier, P. de Rango, D. Fruchart, S. Miraglia, N. Skryabina, J. Huot, B. Hauback, M. Pitt and S. Rovoirard, J. Alloys Compd., 2005, 404–406, 541–544.
8. A. Remhof and A. Borgschulte, ChemPhysChem, 2008, 9, 2440–2455.
9. M. Williams, M. V. Lototsky, V. M. Linkov, A. N. Nechaev, J. K. Solberg and V. A. Yartys, Int. J. Hydrogen Energy, 2009, 33, 1171–1179.
10. R. L. Holtz and M. A. Imam, J. Mater. Sci., 1999, 34, 2655–2663.
11. K. F. Aguey Zinsou, J. R. Ares Fernandez, T. Klassen and R. Bormann, Int. J. Hydrogen Energy, 2007, 32, 2400–2407.
12. W. Oelerich, T. Klassen and R. Bormann, J. Alloys Compd., 2001, 315, 237–242.
13. Y. Fu, M. Groll, R. Mertz and R. Kulenovic, J. Alloys Compd., 2008, 460, 607–613.
14. B. Molinas, A. A. Ghilarducci, M. Melnichuk, H. L. Corso, H. A. Peretti, F. Agresti, A. Bianchin, S. Lo Russo, A. Maddalena and G. Pricipi, Int. J. Hydrogen Energy, 2009, 34, 4597–4601.
15. R. Vijay, R. Sundaresan, M. P. Maiya, S. S. Murthy, Y. Fu, H. P. Klein and M. Groll, J. Alloys Compd., 2004, 384, 283–295.
16. G. Barkhordarian, T. Klassen, M. Dornheim and R. Bormann, J. Alloys Compd., 2007, 440, L18–L21.
17. S. H. Joo, S. J. Choi, I. Oh, J. Kwak, Z. Liu, O. Terasaki and R. Ryoo, Nature, 2001, 412, 169–172.
18. C. Z. Wu, P. Wang, X. Yao, C. Liu, D. M. Chen, G. Q. Lu and H. M. Cheng, J. Alloys Compd., 2006, 420, 278–282.
19. C. Z. Wu, P. Wang, X. Yao, C. Liu, D. M. Chen, G. Q. Lu and H. M. Cheng, J. Alloys Compd., 2006, 414, 259–264.
20. V. Berube, G. Radtke, M. Dresselhaus and G. Chen, Int. J. Energy Res., 2007, 31, 637–663.
21. C. Shang and Z. X. Guo, Int. J. Hydrogen Energy, 2007, 32(4), 2920–2925.
22. A. Zaluska, L. Zaluski and J. O. Strom-Olsen, J. Alloys Compd., 1999, 288, 217–225.
23. V. M. Skripnyuk, E. Rabkin, Y. Estrin and R. Lapovok, Acta Mater., 2004, 52, 405–414.
24. M. Krystian, M. J. Zehetbauer, H. Kropik, B. Mingler and G. Krexner, J. Alloys Compd., 2011, 509, 449–455.
25. V. M. Segal, V. I. Reznikov, A. E. Drobyshevskiy and V. I. Kopylov, Russ. Metall., 1981, 1, 99–105.
26. J. J. Reilly and R. H. Wiswall, Inorg. Chem., 1968, 7, 2254–2256.
27. R. Janot, L. Aymard, A. Rougier, G. A. Nazri and J. M. Tarascon, J. Mater. Res., 2003, 18, 1749–1752.
28. Y. Kodera, N. Yamasaki, T. Yamamoto, T. Kawasaki, M. Ohyanagi and Z. A. Munir, J. Alloys Compd., 2007, 446, 138–141.
29. M. Morinaga and H. Yukawa, Mater. Sci. Eng., 2002, 329–331, 268–275.
30. J. J. Reilly and R. H. Wiswall, Inorg. Chem., 1967, 6, 2220–2223.
31. F. E. Pinkerton, M. S. Meyer, G. P. Meisner and C. Udeh, Chem. Commun., 2011, 47, 8001–8003.
32. S. Bouaricha, J. P. Dodelet, D. Guay, J. Hout, S. Boily and R. Schulz, J. Alloys Compd., 2000, 297, 282–293.
33. P. Zoliker, K. Yvon, P. Fischer and J. Schefer, Inorg. Chem., 1985, 24, 4177–4180.
34. S. G. Porutsky, E. A. Zhurakovsky, S. A. Mogilevsky, V. N. Verbetsky, O. S. Bakuma and K. N. Semenenko, Solid State Commun., 1990, 74, 551–553.
35. D. Kyoi, T. Sato, E. Ronnebro, N. Kitamura, A. Ueda, M. Ito, S. Katsuyama, S. Hara, D. Noreus and T. Sakai, J. Alloys Compd., 2004, 372, 213–217.
36. V. M. Skripnyuk and E. Rabkin, Int. J. Hydrogen Energy, 2012, 37, 10724–10732.
37. P. Vermeulen, P. C. J. Graat, H. J. Wondergem and P. H. L. Notten, Int. J. Hydrogen Energy, 2008, 33, 5646–5650.
38. D. M. Borsa, R. Gremaud, A. Baldi, H. Schreuders, J. H. Rector, B. Kooi, P. Vermeulen, P. H. L. Notten, B. Dam and R. Griessen, Phys. Rev. B: Condens. Matter Mater. Phys., 2007, 75, 205408.
39. P. Vermeulen, R. A. H. Niessen and P. H. L. Notten, Electrochem. Commun., 2006, 8, 27–32.
40. E. Ivanov, I. G. Konstanchuk, A. Stepanov and V. Boldyrev, J. Less-Common Met., 1987, 131, 25–29.
41. A. Stepanov, E. Ivanov, I. G. Konstanchuk and V. Boldyrev, J. Less-Common Met., 1987, 131, 89–97.
42. M. Y. Song, E. Ivanov, B. Darriet, M. Pezat and P. Hagenmuller, J. Less-Common Met., 1987, 131, 71–79.
43. E. J. Ivanov, I. G. Konstanchuk, A. Stepanov, Y. Jie, M. Pezat and B. Darriet, Inorg. Chem., 1989, 28, 613–615.
44. G. Liang and R. Schulz, J. Mater. Sci., 2003, 38, 1179–1184.
45. J. A. Puszkiel, P. Arneodo Larochette and F. C. Gennari, J. Alloys Compd., 2008, 463, 134–142.
46. Y. Tsushio, S. J. Choi, H. Enoki and E. Akiba, Proceedings of the Spring Meeting of the Japan Institute of Metals, 2002, p. 499.
47. Y. J. Choi, J. Lu, H. Y. Sohn and Z. Z. Fang, J. Power Sources, 2008, 180, 491–497.
48. K. Asano, H. Enoki and E. Akiba, J. Alloys Compd., 2009, 478, 117–120.
49. R. Z. Valiev, R. K. Islamgaliev and I. V. Alexandrov, Prog. Mater. Sci., 2000, 45, 103–189.
50. R. Z. Valiev, Y. Estrin, Z. Horita, T. G. Langdon, M. J. Zehetbauer and Y. T. Zhu, JOM, 2006, 58, 33–39.
51. A. P. Zhilyaev and T. G. Langdon, Prog. Mater. Sci., 2008, 53, 893–979.
52. X. Sauvage, F. Wetscher and P. Pareige, Acta Mater., 2005, 53, 2127–2135.
53. X. Sauvage, P. Jessner, F. Vurpillot and R. Pippan, Scr. Mater., 2008, 58, 1125–1128.
54. K. V. Rajulapati, R. O. Scattergood, K. L. Murty, Z. Horita, T. G. Langdon and C. C. Koch, Metall. Mater. Trans. A, 2008, 39, 2528–2534.
55. Y. Z. Tian, J. J. Zhang, S. D. Wu, Z. F. Zhanng, M. Kawasaki and T. G. Langdon, Acta Mater., 2012, 60, 269–281.
56. K. Edalati, S. Toh, M. Watanabe and Z. Horita, Scr. Mater., 2012, 66, 386–389.
57. K. Oh-ishi, K. Edalati, H. S. Kim, K. Hono and Z. Horita, Acta Mater., 2013, 61, 3482–3489.
58. H. Bahmanpour, Y. Sun, T. Hu, D. Zhang, J. Wongsa-Ngam, T. G. Langdon and E. J. Lavernia, J. Mater. Res., 2014, 29, 578–585.
59. O. Ciuca, K. Tsuchiya, Y. Yokoyama, Y. Todaka and M. Umemoto, Mater. Trans., 2010, 51, 14–22.
60. J. Y. Huang, Y. T. Zhu, X. Z. Liao and R. Z. Valiev, Philos. Mag. Lett., 2004, 84, 183–190.
61. M. T. Perez-Prado, A. A. Gimazov, O. A. Ruano, M. E. Kassner and A. P. Zhilyaev, Scr. Mater., 2008, 58, 219–222.
62. B. B. Straumal, A. A. Mazilkin, B. Baretzky, G. Schutz, E. Rabkin and R. Z. Valiev, Mater. Trans., 2012, 53, 63–71.
63. M. Shirooyeh, J. Xu and T. G. Langdon, Mater. Sci. Eng., 2014, 614, 223–231.
64. K. Edalati, J. Matsuda, H. Iwaoka, S. Toh, E. Akiba and Z. Horita, Int. J. Hydrogen Energy, 2013, 38, 4622–4627.
65. K. Edalati, J. Matsuda, M. Arita, T. Daio, E. Akiba and Z. Horita, Appl. Phys. Lett., 2013, 103, 143902.
66. D. R. Leiva, A. M. Jorge, T. T. Ishikawa, J. Huot, D. Fruchart, S. Miraglia, C. S. Kiminami and W. J. Botta, Adv. Eng. Mater., 2010, 12, 786–792.
67. A. Revesz, Z. Kanya, T. Verebelyi, P. J. Szabo, A. P. Zhilyaev and T. Spassov, J. Alloys Compd., 2010, 504, 83–88.
68. M. Zehetbauer, R. Grossinger, H. Krenn, M. Krystian, R. Pippan, P. Rogl, T. Waitz and R. Wurschum, Adv. Eng. Mater., 2010, 12, 692–700.
69. K. Edalati, A. Yamamoto, Z. Horita and T. Ishihara, Scr. Mater., 2011, 64, 880–883.
70. J. X. Zou, C. F. Perez-Brokate, R. Arruffat, B. Bolle, J. J. Fundenberger, X. Q. Zeng, T. Grosdidier and W. J. Ding, Mater. Sci. Eng., B, 2014, 183, 1–5.
71. P. Vajeeston, P. Ravindran, B. C. Hauback, H. Fjellvag, A. Kjekshus, S. Furuseth and M. Hanfland, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 73, 224102.
72. D. Chen, Y. M. Wang, L. Chen, S. Liu, C. X. Ma and L. B. Wang, Acta Mater., 2004, 52, 521–528.
73. B. Pfrommer, C. Elsasser and M. Fahnle, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 5089–5093.
74. P. E. Blochl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953–17979.
75. G. Kresse and J. Furthmuller, Comput. Mater. Sci., 1996, 6, 15–50.
76. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775.
77. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
78. R. F. W. Bader, Chem. Rev., 1991, 91, 893–928.
79. W. Tang, E. Sanville and G. Henkelman, J. Phys.: Condens. Matter, 2009, 21, 084204.
80. K. Momma and F. Izumi, J. Appl. Crystallogr., 2011, 44, 1272–1276.
81. K. Asano, H. Enoki and E. Akiba, Mater. Trans., 2007, 48, 121–126.
82. S. Rousselot, M. P. Bichat, D. Guay and L. Roue, J. Electrochem. Soc., 2009, 156, A967–A973.
83. K. Edalati, H. Emami, A. Staykov, D. J. Smith, E. Akiba and Z. Horita, Acta Mater., 2015, 99, 150–156.
84. A. M. Jorge, E. Prokofiev, G. F. de Lima, E. Rauch, M. Veron, W. J. Botta, M. Kawasaki and T. G. Langdon, Int. J. Hydrogen Energy, 2013, 38, 8306–8312.
85. H. Imamura, K. Yoshihara, M. Yoo, I. Kitazawa, Y. Sakata and S. Ooshima, Int. J. Hydrogen Energy, 2007, 32, 965–968.
86. P. Guinet, P. Perroud and J. Rebiere, Hydrogen Energy System, Proceedings of the Second World Hydrogen Energy Conference, Zurich, Switzerland, Aug 21-24, 1978, International Association for Hydrogen Energy, Miami, FL, 1978, vol. 3, pp. 1657–1675.
87. R. Janot, F. Cuevas, M. Latroche and A. Percheron-Guegan, Intermetallics, 2006, 14, 163–169.
88. M. Paskevicius, D. A. Sheppard, A. L. Chaudhary, C. J. Webb, E. M. Gray, H. Y. Tian, V. K. Peterson and C. E. Buckley, Int. J. Hydrogen Energy, 2011, 36, 10779–10786.
89. J. Zhang, Z. Li, F. Cuevas and M. Latroche, J. Phys. Chem. C, 2014, 118, 21889–21895.
90. F. C. Gennari, F. J. Castro, G. Urretavizcaya and G. Meyer, J. Alloys Compd., 2002, 334, 277–284.
91. R. Cerny, F. Bonhomme, K. Yvon, P. Fischer, P. Zolliker, D. E. Cox and A. Hewat, J. Alloys Compd., 1992, 187, 233–241.
92. F. C. Gennari and F. J. Castro, J. Alloys Compd., 2005, 396, 182–192.
93. A. C. Switendick, Z. Phys. Chem., 1979, 117, 89–112.
94. D. G. Westlake, H. Shaked, P. R. Mason, B. R. McCart, M. H. Mueller, T. Matsumoto and M. Amano, J. Less-Common Met., 1982, 88, 17–23.

---