Facing the energy crisis as we are in the 21st century, many contributions have been made to develop new energy resources. Hydrogen is a promising alternative energy carrier that has been highly regarded because of its prominent advantages such as high energy density (142 MJ/kg [1]), great variety of potential sources, light weight, and environmental friendliness. Besides the application of hydrogen energy, many kinds of hydrogen storage materials have been studied in the past decades, such as lithium compounds [2,3], ammonia borane [4,5], as well as the hydrides of light metals [6,7]. Magnesium hydride is one of promising candidates for hydrogen storage materials due to its high hydrogen capacity (7.6 wt%), which fulfills the ultimate target of 7.5 wt% [8] set by the Department of Energy, USA. However, the absorption and desorption reaction of Mg/MgH₂ itself is too slow. Recently, scientists found out that the composite of MgH₂ and transition metal compounds prepared by ball-milling showed improved kinetics for both hydrogen absorption and desorption [9,10,11,12,13], which can be seen as a good indicator for practical application. Barkhordarian et al. reported that Nb₂O₅ had a superior catalytic effect for absorption and desorption, compared to other metal oxide catalysts. The magnesium hydride catalyzed with 0.5 mol% Nb₂O₅ and ball-milled for 100 h finished full desorption within 90 s under vacuum conditions at 300 °C. The desorbed products could also absorb roughly 7 wt% of hydrogen within 60 s under 8.4 bar of hydrogen at 300 °C [12]. Hanada et al. reported that the composite, MgH₂ with 1 mol% Nb₂O₅ milled for 20 h, was able to absorb ∼4.5 wt% of hydrogen after full desorption even at room temperature, under a pressure lower than 1.0 MPa within 15 s [13]. The exciting achievements reflect a glorious prospect for future application.

On the other hand, the catalytic mechanism of Nb₂O₅ has also been well investigated. According to the progress so far, the potential mechanism may rest in some aspects. The first possibility should be the refinement of the material size. In the recent Transmission Electron Microscope (TEM) observations on MgH₂-Nb₂O₅ ball-milled composites, the size of both the hydride and the additive was found to be within nanoscale. The refinement of size and the consequently homogeneous distribution may lead to the fast kinetics [14,15]. The second possibility comes to the phase changes of Nb₂O₅ brought by high energy ball-milling. Hanada et al. performed X-ray absorption spectroscopy for investigating the valences of metal oxide additives. The results showed that Nb₂O₅ had been reduced by ball-milling with MgH₂ for 20 h. Therefore, they suggested that the improvement of desorption properties was caused by the catalytic effect of reduced niobium oxides [16]. Meanwhile, some complex compounds, such as MgNb₂O_3.67, have been discovered by another group [15]. Finally, the increased defects in the ball-milled composites may act as the hydrogen pathway facilitating the transportation of hydrogen. Friedrichs et al. proposed their “pathway” model, in which Mg-Nb-O ternary phases diffuse through MgH₂ and form a pathway for hydrogen desorption [17]. However, the completion of the exact mechanism in MgH₂-Nb₂O₅ needs more effort and investigation.

In the present work, we did a series of ball-milling using MgH₂ and 1 mol% Nb₂O₅ composites. The desorption properties were evaluated by the Thermal Desorption Spectroscopy (TDS). Moreover, the crystallography and chemical state were investigated in order to discuss the catalytic mechanism.

Thermal desorption properties of the ball-milled compositesare shown in Figure 1. The peak temperatures of desorption for the composites ball-milled for 0 min (hand mixed), 12 min, 120 min, as well as the pure MgH₂ ball-milled for 120 min turned out to be 378 °C, 268 °C, 226 °C and 315 °C, respectively. It can be seen that the desorption properties have been improved with the increase of ball-milling time. This milling time dependence reveals the fact that the catalytic effect can be activated gradually by ball-milling. In the sample milled for 12 min, a very wide peak with a shoulder around 310 °C was obtained. It suggests that Nb₂O₅ in the sample is not well-distributed yet, so that only parts of the sample have been affected by the catalyst during thermal desorption. On the other hand, bycomparing the catalyzed and pure MgH₂ ball-milled for 120 min, we can see a decrease of 89 °C in the peak temperature. This significant improvement of the desorption properties should be ascribed to the addition of Nb₂O₅.

The XRD patterns of the samples were obtained in order to evaluate the crystalline information, as presented in Figure 2. The peaks widened and the intensity drastically decreased after the samples were milled for a long time. It indicates that the crystallites become smaller during ball-milling. This refinement of size may suggest a better distribution of catalyst in the samples, leading to an improvement on the catalytic effect. The size of crystallites can be estimated using Sherrer Equation:

(1)

where K is the shape factor, typically 0.89, λ is the X-ray wavelength, 1.54 Å for Cu radiation, β is the full width at half maximum (FWHM), θ is the Bragg angle, and τ is the mean size of the crystallites [18]. We selected the three strongest peaks of the catalyzed and pure MgH₂ milled for 120 min, shown in Figure 3, to apply the equation and listed the results in Table 1. It can be seen that the crystalline size of the catalyzed composites is a little smaller than the pure MgH₂ after ball-milling. It seems Nb₂O₅ with ceramic feature acts as abrasive facilitating the refinement of MgH₂. However, the difference of several ten angstroms is not so sufficient to bring the obvious improvement on the desorption properties. There should be other more essential factors.

In order to clarify the mechanism of catalytic effect, XPS measurements were carried out to evaluate the chemical state of Nb on the surface of the catalyst. Niobium 3d spectra were drawn in Figure 4. An obvious decrease of the signal was presented in the composites milled for 12 min, while the signal completely disappeared after the composites were milled for 120 min. The phenomena could be explained as the milled MgH₂ particles gradually covered the surface of Nb₂O₅, preventing the detector from getting enough signal from inside. Yet it still can be seen that the peaks in the composites that was milled for 12 min broadened evidently, and the apex slightly shifted to the lower energy side. It provides the evidence that the chemical state of Nb in this sample has changed. Perhaps new phases, at least on the surface, have been generated during ball-milling, in which the valence of Nb is less than +5, because of the decrease in binding energy. In another word, Nb₂O₅ has been partially reduced during ball-milling. The detailed analysis on peak separation has been reported in another article, which concluded theexistence of a certain or a complication of Nb₂O_5−x phase(s) that may act as a more effective catalystto improve the kinetics of desorption [19]. In our previous work, we have confirmed that hydrogen desorption started on the interface between MgH₂ and Nb₂O₅, where hydrogen was recombined and released with the assistance of the essential catalyst on the surface of Nb₂O₅ [20]. Confirmation of the reduction of Nb₂O₅ reasserts and supports the above mechanism of desorption. The reduced phases on the catalyst surface are likely to be the essential catalyst facilitating the completion of desorption.

Taking advantage of high resolution electron microscopy, we were able to find out one possible state of the reduced niobium phases. The lattice image of Nb₂O₅ particle in the composites milled for 120 min is shown in Figure 5. When we performed the fast Fourier transformation (FFT) and the inverse fast Fourier transformation (IFFT) on the selected area near the edge of Nb₂O₅ particle, we obtained the lattice image with the d-spacing of 2.13Å, which matches the value of MgNb₂O_3.67 (100). The existence of this ternary phase has also been reported elsewhere [15,21]. Nevertheless, the results so far still can not draw a clear conclusion on the catalytic effect of this ternary phase. Further investigation on the phase change of Nb₂O₅ is still on the way in order to make a clear understanding on the mechanism of the catalytic effect in Nb₂O₅ catalyzed MgH₂ composites.

MgH₂ (98%, from Alfa Aesar) and Nb₂O₅ (99.99%, ∼1µm, from Kojundo chemical Laboratory) powder were used as purchased. Ball-milling was performed using a Fritsch P7 device. Mixtures of 300 mg MgH₂ and 1 mol% Nb₂O₅ were ball-milled with 20 steel balls (7 mm in diameter) for 12 min and 120 min. The ball-to-powder ratio was set to 100:1. The hydrogen pressure inside the milling spotwas 1 MPa and the rotating speed was 400 rpm. Pure MgH₂ powder was also ball-milled for 120 min under the same condition. Hand-mixed (HM) sample with the same composition was prepared using agate mortar. The desorption properties were examined by thermal desorption spectroscopy (TDS) with a heating rate of 1 °C/min under a highly pure helium flow. X-ray diffraction (XRD) was performed using JEOL JDX-3500 powder diffractometer with Cu K_α radiation. X-ray Photoelectron Spectroscopy (XPS) was carried out using a JPS-90MX photoelectron spectrometer with Mg K_α radiation. The energy scale of the spectrometer was calibrated by setting oxygen (O1s) peak to 532 eV. TEM observationwas performed under a 1250 kV High Voltage Electron Microscope (JEM-ARM1300). Plastic bag method [22] was used to prevent oxidation of the sample during transport into the TEM instruments.

In summary, we performed a study on the Nb₂O₅ catalyzed MgH₂ ball-milled composites, as well as the pure MgH₂ which serves for comparison. Improved desorption properties were reached by ball-milling with the presence of Nb₂O₅. Especially the peak temperature of the catalyzed composites milled for 120 min was decreased to 226 °C, even 89 °C lower than the pure MgH₂ milled for the same duration. XRD spectra indicated a refinement of crystallites during ball-milling, suggesting a better distribution of Nb₂O₅ in the composites. The estimation of crystalline size using Sherrer equation showed Nb₂O₅ as ceramics acts as abrasive facilitating the size decreasing of MgH₂. Reduction of Nb₂O₅ was confirmed by tracing the chemical state of Nb in the samples, arising the possibility that the reduced phases act as more effective catalysts to improve the desorption properties. We were able to find MgNb₂O_3.67 existing near the edge of Nb₂O₅ particle via TEM, providing further evidence for the reduction of Nb₂O₅ and raising the further investigation on the phase transition of Nb₂O₅ during ball-milling.