To outline the effect of mechanical milling and the additional effects of MgO, the materials were characterised by DSC measurements. As shown Figure 1, as-received MgH₂ released hydrogen in a single endothermic peak at 414 °C. Upon milling MgH₂, the hydrogen desorption temperature was reduced to 336 °C. However, a significant decrease of the hydrogen desorption temperature was obtained when MgH₂ was milled with MgO as the desorption peak shifted from 336 °C to 262 °C. In our previous report, such a drastic improvement of hydrogen desorption was related to an important reduction in the size of MgH₂ particles as MgO would act as a process control agent facilitating the grinding of MgH₂ particles [32]. Further investigations on the potential effect of MgO have raised the possibility of a defective oxide layer at the surface of MgH₂ particles catalysing the splitting/recombination of hydrogen [37]. However, such a hypothesis challenges common assumptions as an oxide layer would seriously impede hydrogen absorption/desorption. In fact, pre-oxidising MgH₂ before milling in an attempt to enhance the number of potential oxygen vacancies resulted in poor hydrogen desorption characteristics as demonstrated by the DSC measurement carried out after milling (Figure 1). In this case, it is most likely that the oxide layer generated during the pre-oxidation of MgH₂ still inhibits hydrogen desorption despites the high energy involved in the milling process and the possibility of forming fresh surfaces and defective surface sites facilitating the desorption of hydrogen. Other investigations have also found similar results with hydrogen kinetics deteriorating as an oxide layer is allowed to grow at the surface of MgH₂ particles [27,34]. Although hydrogen could strongly absorb at defective oxide sites and dissociate, such a heterolytic dissociation would most likely result in the formation of surface hydroxyls that would be further reduced upon heating [38,39,40]. Accordingly, a catalytic activity of MgO should not be expected.

Figure 2 shows the hydrogen kinetics of MgH₂ milled for 100 h at 300 °C. As previously reported, hydrogen absorption is faster than desorption with ~30 min necessary to desorb MgH₂ whereas only ~8 min are required to complete the absorption process up to 90% of the total capacity. Hydrogen kinetics were drastically improved when MgH₂ was milled with MgO (Figure 3). Hydrogen absorption and desorption could be achieved in less than 100 s at 300 °C. At a lower temperature of 250 °C, full absorption was also completed in less than 200 s, and a fast desorption rate with full hydrogen release occurring in less than 800 s was still achievable. However, below 250 °C the desorption rate of hydrogen drastically decreased and no significant release of hydrogen could be observed in agreement with the DSC measurements. It is noteworthy that the shape of hydrogen absorption curves was not influenced by the temperature or the addition of MgO. Conversely, the hydrogen desorption curves evolved from a sigmoidal shape for milled MgH₂ to a non-sigmoidal curve when MgO was used as a milling additive.

In order to identify possible mechanisms governing the hydrogen kinetics of MgH₂ modified with MgO and better understand the improvements observed, the kinetic curves obtained under isothermal conditions were fitted using the general equation for solid-state reactions [41]:

(1)

where α is the fraction reacted in time t, k = k(T,P,r) is the reaction rate and G(α) is a functiondepending on the mechanism controlling the reaction. The main theoretical functions G(α) are summarized in Table 1 and grouped according to four major reaction mechanisms governed by: (i) the surface; (ii) the nucleation and growth of the new phase; (iii) the reaction at the new phase/shrinking core interface with the interface reaction proceeding at constant velocity; and (iv) the reaction at the new phase/shrinking core interface with the interface reaction controlled by diffusion. In order to facilitate the fitting of the hydrogen curves measured and ease the distinction of reaction mechanisms, the method of Hancok and Sharp was used [42]. Since nucleation and growth processes in condensed systems follow the almost universal Equation (2), the method consists of plotting ln(−ln(1 − α)) vs. ln(t).

(2)

where B is a constant that depends on the nucleation frequency and linear rate of the grain growth, and m is a constant that varies according to the geometry of the system.

(3)

upon plotting Equation (3), the determination of m allows the identification of the reaction mechanism by comparison with tabulated theoretical values (Table 1).

Figure 4 shows the fits for the hydrogen absorption and desorption kinetic curves obtained for Mg and Mg milled with MgO. In both cases, a slope close to 0.62 for the absorption kinetic indicates that the absorption process was controlled by the diffusion of hydrogen into MgH₂ in a one dimensional process. Despite the low diffusion coefficient of hydrogen in the MgH₂ phase, such behaviour is possibly due to the high temperatures and pressures used to perform the absorptions. However, upon hydrogen desorption different slopes of 1.04 and 2.86 were obtained for MgH₂ and MgH₂ milled with MgO, respectively. This would indicate a shift from a mechanism where hydrogen desorption is controlled by the nucleation and growth of the Mg phase to a desorption process controlled by the Mg/MgH₂ interface as the MgH₂ core is shrinking. It is widely accepted that the hydrogen desorption of milled MgH₂ follows a nucleation and growth process [5,6]. Accordingly, the results obtained for milled MgH₂ are in agreement with the general trends observed.

As the temperature for hydrogen absorption/desorption in MgH₂ milled with MgO is decreased to 250 °C, the hydrogen absorption kinetic was found to be once again controlled by the diffusion of hydrogen, however hydrogen desorption shifted back to a reaction mechanism controlled by a nucleation and growth process (Figure 5). At such a low temperature, the energy provided may not be sufficient to overcome the activation energy required for the nucleation of the magnesium phase. Therefore, the mechanism would be controlled again by nucleation and growth instead of interface control. Liang et al. reported similar findings with hydrogen desorption from MgH₂ modified with 5 at% of V shifting from a nucleation and growth process to a interface process for temperatures above 250 °C [19]. In this case, the shift observed was explained by the catalytic activity of V taking place above 250 °C by reducing the barrier for nucleation and thus making hydrogen desorption occurring at low driving forces. However, with MgH₂ modified with MgO, it is unlikely that MgO would catalyse the recombination of hydrogen and thus the nucleation of a magnesium phase as previously discussed. The shift in controlling mechanisms observed would therefore correspond to the energy required to effectively recombine hydrogen at the surface of milled MgH₂. After milling, magnesium showed smaller particle sizes and increased specific surface area resulting in a higher density of nucleation sites. As a significant reduction of activation energy for hydrogen desorption has been observed for decreasing particles sizes [43], the smaller particles generated upon milling MgH₂ with MgO would result in faster hydrogen kinetics and depending on the driving forces different reaction mechanisms.

It is also striking that the rate limiting steps identified for MgH₂ milled with MgO are similar to those observed with MgH₂ modified with the best known catalyst, i.e., Nb₂O₅ [24]. At the macroscopic level, hydrogen desorption at 300 °C would proceed through an Mg/MgH₂ interface-controlled growth process regardless the additive (Table 2). The hypothesis can be made that the use of MgO resulted in a better grinding process leading to smaller particle/crystal sizes and thus faster kinetics. However, comparison of the physical properties of the materials obtained in both cases does not revealed any significant differences in terms of particle and crystallite sizes (Table 2). Accordingly, such a finding contradicts current assumptions whereby Nb₂O₅ is considered to as a catalyst reducing the barriers for nucleation and thus facilitating hydrogen desorption at low driving forces [25,27]. The formation of mixed phases of Mg_xNb_yO acting as pathways for hydrogen diffusion with the formation of metastable niobium hydride phases reducing the activation energy of the Mg/MgH₂ transition is another possibility [27]. Alike Nb₂O₅, MgO would also be reduced by MgH₂ upon heating to form new oxide phases. However MgO would not generate nucleation points of partial oxides that could induce the interface controlled mechanism observed with MgO and/or Nb₂O₅. It is noteworthy that the addition of Nb₂O₅ to MgH₂ with fully optimised hydrogen kinetics does not lead to further kinetic improvements. Once fast hydrogen kinetics have been reached at 250 °C by effectively reducing particle size during the milling process, any catalyst addition could not further enhance the hydrogen kinetics [44]. Accordingly, if the modification of MgH₂ with MgO or Nb₂O₅ has led to a final material with equivalent properties, a possible catalytic effect of Nb₂O₅ would be mirror in comparison to other potential effects such as particle size.

In the light of these findings, alternative explanations should be advanced to describe the role of the oxides used to improve the hydrogen kinetics of Mg/MgH₂. In the initial model of Klassen et al., it is proposed that only transition metal oxides comprising defective oxide sites and high valance state metals capable of high electrons exchange rate during catalysis are suitable for improving Mg/MgH₂ kinetics [15,25]. If such model can in part explain a potential contribution of transition metal oxides in improving the kinetics of the hydrogen absorption/desorption process, it fails to describe the positive effect of TiO₂, MgO and other non-transition metals oxides.

In our early investigations, we have reported that non transition metal oxides such as Al₂O₃ were also good at improving the hydrogen kinetics of Mg/MgH₂ [45]. We also found that upon Nb₂O₅ addition to MgH₂ during the milling process a significant improvement of the grinding process was achieved [44]. Taking into account the poor ability of metal oxides such as SiO₂ to enhance the hydrogen kinetics of Mg [45], and considering the mechanical effect of metal oxides in facilitating the grinding and stabilisation of small MgH₂ particles a correlation was found between the potency of a metal oxide in improving hydrogen kinetics in Mg and the nature of the metal-oxygen bond (Figure 6). Depending on the electronegativity of solid oxides envisaged lower hydrogen desorption temperatures were obtained for oxides of higher electronegativity such as MgO. Such a correlation is comprehensible when considering the processes occurring during high energy ball milling. As MgH₂ particles are initially fractured, the generation of fresh surfaces would result in an excess of surface energy leading to particles agglomeration and cold welding upon balls impact. Hence depending on the experimental conditions, the milling process would become quickly inefficient in reducing particle sizes when the equilibrium between cold-welding and fracturing is established [32,44]. To delay this equilibrium and reach the smallest possible particle size, process control agents such as polar, non-polar organic solvents or polymers maintaining a low friction are usually employed to reduce agglomeration and cold-welding [46]. However, such additives are known to be relatively inefficient with MgH₂ as they quickly degrade and lead to severe carbon contamination due to the high temperatures, i.e., >500 °C, locally generated during milling processes. At such high temperatures, oxides such as TiO₂ and MgO are known to be effective lubricants providing low frictions coefficients to sliding tribological interfaces [47]. In a recently proposed model, Erdemir reported a correlation between friction coefficients and electronegativity [48]. The higher the electronegativity of a solid oxide is, the lower the friction coefficient would be. As metal oxides of higher electronegativity hold their bonding electrons more tightly and have a lower tendency to chemically bind with their neighbouring oxides, Erdemir found that higher electronegativity would lead to lower shearing at the interface of sliding oxide particles and thus lower friction coefficients. The favourable tribological properties of highly electronegative oxides would explain the correlation obtained with respect to the hydrogen kinetics of MgH₂ milled with oxides.

At high temperatures, as the oxide layer is depleted from the particle surface by wear, the surface of MgH₂ particles would be oxidised to some extend at sliding interfaces. Accordingly, the mechanical process of MgH₂ with solid oxides could involve the sliding of two oxidised surfaces: that of the partially oxidised MgH₂ particles and the metal oxide additive. The formation of MgO observed during the milling process of MgH₂ with Nb₂O₅, for example [26], would corroborate the existence of such an interface. As Nb₂O₅ is replaced by more electronegative oxides showing higher friction coefficients, efficient reduction of MgH₂ particle size upon milling would thus become possible. Such a correlation clearly corroborates our observations as the use of MgO having the highest electronegativity of all metal oxides investigated so far led to the most advanced hydrogen kinetics in Mg. It also explains why a simple pre-oxidation of MgH₂ prior to mechanical milling did not help in reducing particle sizes and improving hydrogen kinetics. Furthermore, the correlation found is a clear indicator of the mechanical effect of solid oxides in facilitating the grinding process of MgH₂ and fast hydrogen kinetics through particle size reduction and stabilisation. Although further work is needed to establish the levels of “catalytic/mechanical” contribution of solid oxides to the hydrogen kinetics of magnesium, the correlation found evidently highlights the potential of particle size reduction as a mean for finally controlling the properties of the Mg/MgH₂ system and we are currently exploring the potential of such a route [49].