A. Sample preparation

Mixtures of MgH₂ (98%, Alfa Aesar) with 5 mol% of additive were ball milled under argon atmosphere using tungsten carbide (WC) hard alloy vial (80 ml) and balls (10 mm diameter) in a Fritsch Pulverisette P4 planetary mill. As additives, commercially available chemicals were used: vanadium pentoxide (V₂O₅, 98 + %, Aldrich), titanium hydride (TiH₂, 98%, Aldrich), zirconium hydride (ZrH₂, 99%, Aldrich), vanadium chloride (VCl₃, 97%, Aldrich), and scandium chloride (ScCl₃, 99.9%, Aldrich). The typical mass of each mixture constituted 3 g with 1 : 41 powder-to-ball weight ratio. The milling was performed in time intervals of 30 min at a rotation speed of 300 rpm (Huot et al., Reference Huot, Ravnsbæk, Zhang, Cuevas, Latroche and Jensen2013).

Experimental details regarding samples composition and preparation can be found in Table I. All handling and manipulation of the chemicals were carried out under argon atmosphere in an MBraun Unilab glove box with a recirculation gas purification system and gas/humidity sensors. Oxygen and water levels were kept well below 1 ppm during all operations.

B. In situ synchrotron radiation powder X-ray diffraction (SR-PXD)

The detailed reaction mechanisms and kinetics of hydrogen release and uptake in MgH₂ with different additives were studied by in situ SR-PXD experiments. Data were collected at beamline I711 at the synchrotron MAX II, in the research laboratory MAX-lab, Lund, Sweden using a Mar165 CCD detector. A LaB₆ external standard was used to determine sample-detector distance, detector orientation, and the wavelength. The samples were mounted in sapphire single crystal tubes (Al₂O₃, o.d. 1.09 mm, i.d. 0.79 mm) in an argon-filled glove box (p(O₂, H₂O) <1 ppm). The sample holder was specially developed for the study of gas/solid reactions and hydrogen pressures were allowed up to 20 MPa and temperatures from room temperature (RT) to 800 °C (Jensen et al., Reference Jensen, Nielsen, Filinchuk, Jorgensen, Cerenius, Gray and Webb2010). The selected wavelength was 0.94608 Å and the X-ray exposure time was 5 s per PXD pattern. Samples for investigation of mechanism and kinetics of hydrogen release and uptake were heated stepwise to fixed temperatures of 270 and 320 °C (15 °C min⁻¹) in a hydrogen pressure of 100–150 bar. The hydrogen pressure was reduced to p(H₂) ~10⁻² bar (dynamic vacuum) when the fixed temperature was reached and then again increased to 100–150 bar when the sample was fully dehydrogenated (after 15 to 30 min) measuring one full hydrogen release and uptake cycle at each temperature. Selected samples were additionally heated to 450 °C and two extra H₂ desorption/absorption cycles were performed. Finally, the samples were cooled to 270 °C and one dehydrogenation was conducted.

All obtained raw images were transformed to two-dimensional powder patterns using the FIT2D program (Hammersley et al., Reference Hammersley, Svensson, Hanfland, Fitch and Hausermann1996), which was also used for calibration using measurements of the standard NIST LaB₆ sample. Crystal structure refinements were performed using the Rietveld method implemented in the Fullprof program (Rodríguez-Carvajal, Reference Rodríguez-Carvajal1993). The backgrounds were described by linear interpolation between selected points, while pseudo-Voigt profile functions were used to fit the diffraction peaks. Sequential refinements provide the weight fractions for the compounds present in the sample as a function of time. The extracted normalized curves represent phase formation or decomposition curves as a function of time denoted α(t). These sigmoidal shaped curves can be used for evaluation of the activation process.

The decomposition curves for MgH₂ can be fitted well with the equation of shrinking core model:

where α is the phase fraction at time t and k is the rate constant (Liang et al., Reference Liang, Huot, Boily and Schulz2000; Varin et al., Reference Varin, Czujko and Wronski2009).

An apparent activation energy, E _A, and a pre-factor, A, for the overall reaction can be calculated using Arrhenius law:

where R is the gas constant and T is the absolute temperature.

A. Reactions during mechano-chemical and heat treatment

Dispersion of transition metals in Mg/MgH₂ system is difficult because of the ductility of the metals. Therefore, binary salt-like oxides, chlorides, and hydrides based on early d-block metals were selected as additives in the present investigation, i.e., V₂O₅, VCl₃, ScCl₃, TiH₂, and ZrH₂.

Figures 1–5 show in situ SR-PXD data for the ball-milled samples S1–S5 (see Table I), respectively. Bragg diffraction peaks of β-, γ-MgH_2, and MgO are observed for sample S1 (Figure 1) at RT. There is no observed diffraction from V₂O₅, which suggests that a redox reaction has occurred during mechano-chemical treatment [see reaction scheme (3)]. Decomposition of MgH₂ initiates at 270 °C by decreasing excess hydrogen pressure below equilibrium at this temperature. All γ-MgH₂ disappears within <1 min of heating at 270 °C and most of the present β-MgH₂ (90%) is decomposed after 10 min (during measurement of 40 PXD patterns). The PXD data show the presence of Mg, MgO, and possibly V at 270 °C. The latter two phases are difficult to distinguish owing to overlapping Bragg reflections.

Figure 1. In-situ SR-PXD data measured for sample MgH₂–V₂O₅ (S1) heated under p(H₂) = 100 bar from RT to 270 °C (15 °C min⁻¹) and subsequently dehydrogenated and hydrogenated at this temperature applying alternately p(H₂) = 10⁻² and 100–150 bar (λ = 0.946 08 Å).

Figure 2. In-situ SR-PXD data measured for sample MgH₂–ZrH₂ (S2) heated under p(H₂) = 100 bar from RT to 270 °C (15 °C min⁻¹) and subsequently dehydrogenated at this temperature applying p(H₂) = 10⁻² (λ = 0.946 08 Å).

Figure 3. In-situ SR-PXD data measured for sample MgH₂–TiH₂ (S3) heated under p(H₂) = 100 bar from RT to 270 °C (15 °C min⁻¹) and subsequently cycled at this temperature applying alternately p(H₂) = 10⁻² and 100–150 bar (λ = 0.946 08 Å).

Figure 4. In-situ SR-PXD data measured for sample MgH₂–ScCl₃ (S4) heated under p(H₂) = 100 bar from RT to 270 °C (15 °C min⁻¹) and subsequently cycled at this temperature applying alternately p(H₂) = 10⁻² and 100–150 bar (λ = 0.946 08 Å).

Figure 5. In-situ SR-PXD data measured for sample MgH₂-VCl₃ (S5) heated under p(H₂) = 100 bar from RT to 270 °C (15 °C min⁻¹) and subsequently cycled at this temperature applying alternately p(H₂) = 10⁻² and 100–150 bar (λ = 0.946 08 Å).

PXD patterns of the samples S2 and S3 measured at RT reveal diffracted intensities from the reactants only, i.e., no reaction has occurred between ZrH₂ or TiH₂ and MgH₂. Notice that there is no observed diffraction from MgO in sample S2 and S3. This indicates that the presence of MgO in S1 is because of a reaction between V₂O₅ and MgH₂. Magnesium metal formation in both samples, S2 and S3, initiates at 270 °C when the pressure is reduced.

The as-milled chloride doped sample S4 reveals diffraction from MgH₂ and scandium(II) hydride ScH₂ and no diffraction from the additive, ScCl₃. This indicates that a redox reaction has occurred during ball-milling [see reaction scheme (4)]. Formation of magnesium initiates when the pressure is reduced at 270 °C. Formation of MgCl₂ is observed during the rehydrogenation at p(H₂) = 100 bar.

Sample S5 (MgH₂ − VCl₃) behaves in a similar way, i.e., no observation of VCl₃ after milling, crystallization of MgCl₂ during rehydrogenation (see Figure 5). Furthermore, unidentified reflection reveals a continuous change in composition at 2θ = 24.9–25.4° under vacuum.

B. Evaluation of additives kinetic effect

In situ SR-PXD data are used to investigate the effect of selected early d-block hydrides, oxides, and chlorides additives on the mechanism of hydrogen release and uptake in magnesium hydride system at 270 °C. The diffracted intensities from MgH₂ are integrated, normalized, and converted to phase fraction and used to visualize decreasing amounts of MgH₂ during dehydrogenation reactions (Figure 6). At 270 °C the weakest kinetic effect for hydrogen desorption is provided by ScCl₃ (S4) and TiH₂ (S3) and the strongest influence is provided by VCl₃ (S5) and V₂O₅ (S1). In the early stages of MgH₂ decomposition V₂O₅ demonstrates the fastest kinetics, which is somewhat slowed down in the late stage of the decomposition. In contrast, the effect from VCl₃ additive is also significant in the late stage of the decomposition. Zirconium hydride (S2) has an intermediate kinetic effect. A similar analysis of additives effect was also performed at higher temperature of 320 °C and the results are presented in Figure 7. The time scale for these experiments is clearly shorter as compared with those performed at 270 °C. In addition, the additives ability to accelerate the reaction kinetics follows the sequence: ScCl₃ < TiH₂ < ZrH₂ < VCl₃ < V₂O₅. These experiments reveal that the effect of increasing temperatures on the acceleration of reaction rate is more pronounced than the different additives effect. It should be emphasized that commercial MgH₂ ball-milled without additives in the same conditions does not desorb hydrogen at 270 nor at 320 °C. Slight yield of hydrogen begins to be observed only at temperatures above 350 °C.

Figure 6. (color online) Integrated normalized diffracted intensities for samples S1–S5 showing changes in the relative amount of MgH₂ at 270 °C as a function of time.

Figure 7. (color online) Integrated normalized diffracted intensities for samples S1–S5 showing changes in the relative amount of MgH₂ at 320 °C as a function of time.

C. Activation of magnesium hydride via a chemical reaction

Analysis of the additives effect (Section B) revealed that the best results were obtained using the addition of V₂O₅. Here we further analyse the mechanism for this effect. Bragg diffraction peaks of β-, γ-MgH₂ and MgO are observed for MgH₂–V₂O₅ (S1) (see Figure 1), at RT. There is no observed diffraction from V₂O₅, which suggest that a redox reaction has occurred during mechano-chemical treatment [see reaction scheme (3)]. To crystallize possible reaction products the same sample was heated to 450 °C and 2 cycles of hydrogen release and uptake were conducted (see Figure 8). Diffraction peaks from V₂H are clearly observed at 2θ = 24.5° and 35.9° in the hydrogenated state when the temperature reaches 450 °C and for V in the dehydrogenated state at 2θ = 25.4° and 36.4°. This leads to the splitting of the peak previously assigned to MgO and V, at 2θ ≈ 26° in Figure 1. Apparently, the heating procedure fully reduced the possibly remaining V₂O₅ and facilitated crystallization of V₂H/V. The experiment was continued using the same sample (see Figure 9), the temperature was decreased to 270 °C and one dehydrogenation was conducted similar to what is shown in Figure 1. Presence of vanadium and vanadium hydride is more clearly observed in Figure 9 as compared with Figure 8 and during dehydrogenation diffraction peaks from V₂H (at 2θ = 24.4°) continuously change position to 2θ = 25.5° because of the release of hydrogen and formation of V.

Figure 8. In-situ SR-PXD data of the activation process for sample MgH₂–V₂O₅ (S1) heated under p(H₂) = 100 bar from 320 to 450 °C (15 °C min⁻¹) and subsequently cycled at this temperature applying alternately p(H₂) = 10⁻² and 100–150 bar (λ = 0.946 08 Å).

Figure 9. In-situ SR-PXD data of post-activation desorption measured for sample MgH₂–V₂O₅ (S1) cooled from 450 to 270 °C under p(H₂) = 100 bar (15 °C min⁻¹) where the pressure was decreased to p(H₂) = 10⁻² bar (λ = 0.946 08 Å).

The activation procedure, i.e., cycling at 450 °C clearly improves the desorption kinetics. This is illustrated in Figure 10 where extracted decomposition curves for MgH₂ are shown using data from Figures 1 and 9. After the activation process (Figure 8) the release of hydrogen from MgH₂ is both faster and more complete. This may be related to the formation of more V₂H/V in the sample and the fact that the hydrogen permeability of vanadium is large and exceeds that of palladium (Dolan, Reference Dolan2010). It is known that Nb and V based alloys are promising membranes materials, which currently are developed instead of Pd based alloys (Yukawa et al., Reference Yukawa, Nambu and Matsumoto2012). Therefore, the V₂H/V system may be the active catalytic component for the MgH₂–V₂O₅ system.

Figure 10. (color online) Comparison of the kinetic curves for MgH₂ decomposition at 270 °C before and after the activation of MgH₂–V₂O₅ (S1).

D. Apparent activation energy for the MgH₂ decomposition

The decomposition curves for MgH₂–V₂O₅ (S1) can be fitted well using the contracting surface model (two-dimensional growth) (Bamford and Tipper, Reference Bamford, Tipper, Brown, Dollimore and Galwey1980), e.g., kinetic shrinking core model, which suggests that hydrogen desorption is interface controlled (Liang et al., Reference Liang, Huot, Boily and Schulz2000; Varin et al., Reference Varin, Czujko and Wronski2009).

An Arrhenius plot of the natural logarithm to the rate constant as a function of inverse temperature is provided in Figure 11. The calculated apparent activation energy E _A = 43 kJ mol⁻¹ is obtained for the MgH₂–V₂O₅ sample S1 after milling. This value compares well to the activation energy for long range hydrogen diffusion in bulk Mg determined from experimental data to be E _A = 40.2 kJ mol⁻¹ (Renner and Grabke, Reference Renner and Grabke1978). For the fully activated sample S1 the compounds V/V₂H are identified to be the active catalytic compounds, which facilitate hydrogen release. The in situ SR-PXD data also allow to extract an apparent activation energy for the activated sample, E _A = 34 kJ mol⁻¹ (see Figure 11). These measured values are low, in particular when compared with that for bulk (unmilled) and air-exposed MgH₂, E _A = 300 kJ mol⁻¹ (Jensen et al., Reference Jensen, Andreasen, Vegge, Andreasen, Stahl, Pedersen, Nielsen, Molenbroek and Besenbacher2006). Apparent activation energies of E _A = 52 and 120 kJ mol⁻¹ are measured for mechano-chemically treated MgH₂ with and without vanadium (5 mol%) as additive (Liang et al., Reference Liang, Huot, Boily, Van Neste and Schulz1999b, Reference Liang, Huot, Boily and Schulz2000).

Figure 11. (color online) Arrhenius plot of the kinetic data for dehydrogenation of MgH₂–V₂O₅ (S1).

Thus, this kinetic analysis reveals that in situ synthesis of small particles of V/V₂H in the sample during cycling appears to have a stronger catalytic effect on hydrogen release and uptake in magnesium/magnesium hydride system. This may be because of a more uniform distribution of catalytic active particles compared with mechano-chemically prepared samples of MgH₂–V. There are inherent difficulties with obtaining uniform distribution of ductile metals in MgH₂ in contrast to brittle oxides that tend to form more homogenous distributed samples. The drawback is that considerable amounts of inert MgO are formed during reduction of V₂O₅ in the MgH₂–V₂O₅ system.

E. On the catalytic properties of vanadium pentoxide and vanadium metal

There are several studies that report high catalytic activity of V₂O₅ for release and uptake in magnesium hydride (Oelerich et al., Reference Oelerich, Klassen and Bormann2001a, Reference Oelerich, Klassen and Bormann2001b; Barkhordarian et al., Reference Barkhordarian, Klassen and Bormann2006; Jung et al., Reference Jung, Lee and Lee2006). Vanadium(V) oxide, similar to niobium(V) oxide, is readily reduced in contact with hydrogen or mechano-chemically treated with MgH₂ (Jung et al., Reference Jung, Lee and Lee2006; Nielsen and Jensen, Reference Nielsen and Jensen2012). Several lower oxides may act as intermediates during the formation of MgO. The final product is shown here to be vanadium, which also participate in hydrogen release and uptake via V/V₂H system.

As mentioned above vanadium does not form alloy or intermetallic compounds with Mg (Schimmel et al., Reference Schimmel, Huot, Chapon, Tichelaar and Mulder2005) thus it can act as a pure catalyst. MgH₂-5 mol%V composite desorbs hydrogen at 473 K (under vacuum) and re-absorbs hydrogen rapidly even at RT (Liang et al., Reference Liang, Huot, Boily, Van Neste and Schulz1999b). The desorption rate was the highest for this mixture and decreased in the following order of additives: Ti, Mn, Fe, and Ni (Liang et al., Reference Liang, Huot, Boily, Van Neste and Schulz1999a).

Moreover, MgH₂-5 mol%V system has the fastest kinetics for hydrogen release and uptake in magnesium hydride in comparison with Ti and is similar to Nb additive (Charbonnier et al., Reference Charbonnier, de Rango, Fruchart, Miraglia, Pontonnier, Rivoirard, Skryabina and Vulliet2004). Mg-10 wt.%V system exhibits the highest rate of hydrogen desorption and shows the lowest activation energy among Y and Zr additives (Czujko et al., Reference Czujko, Varin, Chiu and Wronski2006). In addition, vanadium itself reversibly absorbs and desorbs 2.3 wt% of hydrogen at RT, which is also larger than that of the known alloys, LaNi₅, FeTi (Varin et al., Reference Varin, Czujko and Wronski2009; Nakamura et al., Reference Nakamura, Kato, Takamatsu, Fuura, Oosawa and Tsunokake2012).