Magnesium (Mg) is a promising candidate for hydrogen storage with a high gravimetric capacity of 7.6 mass% H₂ and a volumetric density of ~110 kg m⁻³ H₂ (Sun et al., 2017). However, the high stability of MgH₂ (75 kJ mol⁻¹ H₂) induces the need of high temperatures (>300°C) to enable hydrogen release and the slow kinetics for the hydrogen sorption limit practical utilization (Lillo-Ródenas et al., 2008; Aguey-Zinsou and Ares-Fernández, 2010; Jain et al., 2010). Nanosizing is the promising approach to improve the hydrogen storage properties of Mg (Sun et al., 2017). This can be done via the synthesis of isolated Mg nanoparticles (Norberg et al., 2011; Shissler et al., 2014; Zou et al., 2014, 2015; Zhang et al., 2015), or their confinement in a porous host material to facilitate their stabilization (Wagemans et al., 2005; Adelhelm and de Jongh, 2011; Zlotea et al., 2015). Carbon materials have been widely used to nanoconfine Mg and achieve better hydrogen storage properties especially in terms of kinetics (Yao et al., 2006; Lillo-Ródenas et al., 2008; de Jongh and Adelhelm, 2010; Skripnyuk et al., 2010; Adelhelm and de Jongh, 2011). This includes graphene, carbon nanotubes (CNTs), and carbon aerogels (Liu et al., 2013; Cai et al., 2015; Xia et al., 2015; Shinde et al., 2017). It is expected that upon particle size reduction, hydrogen kinetics will be improved owing to the shorted diffusion paths and enhanced hydrogen mobility (Corey et al., 2008; Au et al., 2014; Zlotea et al., 2015). For example, Zlotea et al. synthesized ultra-small Mg particles of 1.3 nm confined in ordered microporous carbon which started to release hydrogen below 100°C with a peak at 200°C (Zlotea et al., 2015). Nanoconfinement of Mg may also result in improved thermodynamic properties for ultrasmall particles (<2 nm) according to theoretical calculations (Cheung et al., 2005; Wagemans et al., 2005). However, experimentally, little improvement has been observed owing to an enthalpy/entropy compensation effect (Zhao-Karger et al., 2010; Sun et al., 2017). In this context, the main improvements achieved so far through nanoconfinement have been in terms of hydrogen kinetics. The carbon scaffold is also expected to limit the growth of the Mg nanoparticles upon hydrogen cycling and, thus, lead to stable hydrogen kinetics (Zlotea et al., 2013).

Currently, Mg nanoconfinement with carbon materials is usually done via melt infiltration or thermal decomposition of an Mg precursor initially impregnated in the porosity of the carbon scaffold (Konarova et al., 2012; Liu et al., 2014). However, via these approaches control over the particle size of the nanoconfined material is often difficult because it relies on the synthesis of ordered mesoporous scaffolds. These scaffolds are also difficult to fully impregnate and the exact location of the Mg nanoparticles is hard to control.

Herein, we investigated an alternative electrochemical deposition method whereby an Mg precursor is directly reduced at the surface of carbon particles in suspension in the electrolyte. Since carbon is conductive, it is expected that every time a carbon particle touches the working electrode, the Mg precursor will be reduced at its surface and supported Mg particles will further grow. Through the functionalization of carbon particles, such an approach should not only enable control over particle size but also the location of the deposited particles. By further adjusting the deposition current, particles of different sizes can also be grown (Reetz et al., 2001). The present study reports on the potential of the approach through the electrochemical reduction of di-n-butylmagnesium at the surface of CNTs suspended in tetrahydrofuran (THF). CNTs were chosen as support for Mg as they can be readily dispersed in THF and have good electrical conductivity (Wang et al., 2001) to enable electron transfer between the reduced MgBu₂ and the electrode surface. The physical and hydrogen sorption properties of Mg nanoparticles confined/supported on these CNTs are reported and the potential of such a novel route to prepare Mg/carbon composite materials for hydrogen storage is discussed.

All operations were carried out under an inert atmosphere in an argon-filled LC-Technology glove box (< 1 ppm O₂ and H₂O).

We previously reported on the electrochemistry of MgBu₂ and at a current of 0.5 mA cm⁻², this Mg precursor is readily reduced into metallic Mg (Shen and Aguey-Zinsou, 2016). The XRD pattern of the pristine CNTs (Figure 1A) shows the (002) peak around 26 corresponding to the inter-planar spacing of 0.34 nm (Kawasaki et al., 2005). However, the as-prepared Mg/CNTs composites showed in addition to the diffraction peaks of the CNTs, peaks related to the typical hexagonal Mg phase (Figure 1A). Furthermore, with increasing amounts of CNTs in the electrolyte, the diffraction peaks related to hexagonal Mg were found to significantly decrease in intensity and become broader; and this was interpreted in first instance as a reduction of Mg crystallite size. For the same amount of Mg electrochemically reduced, elemental analysis confirmed that the amount of Mg deposited on the CNTs is decreasing with increasing amounts of CNTs in the electrolyte (Figure 1B), and this was considered as an indirect proof that Mg was effectively deposited on CNTs. This also showed that even at high amount of CNTs in the electrolyte, the reduction of MgBu₂ still occurred. The evolution of the amount of Mg on CNTs also followed theoretical trends with a gap attributed to the fact that (in addition to the CNTs) some Mg is also directly deposited on the Pt working electrode.

Transmission electron microscopy analysis revealed that Mg nanoparticles (~1 nm) were effectively deposited at the surface of the CNTs. As compared to the pristine CNTs (Figures 2A,B), darker particles were observed after Mg deposition and this was confirmed by dark field analysis (Figures 2C–G). SAED analysis also corroborated the crystalline nature of the nanoparticles observed at the CNTs surface with diffraction planes corresponding to that of hexagonal Mg (Figure 2H). With higher amounts of CNTs in the electrolyte, the overall coverage of individual carbon tubes with Mg was also found to decrease (Figures 2I–L).

Considering the small particle size of the Mg observed at the CNTs surface, no diffraction peaks related to hexagonal Mg should be expected by XRD owing to the lack of long-range order of such small particles. However, the presence of Mg diffractions peaks with a crystallite size of ~ 26 ± 2 nm (as determined from the Scherrer equation) indicates that larger Mg particles are also contained within the material. Indeed, in addition to small Mg particles deposited at the surface of the CNTs, larger Mg particles with a size ranging from 20 to 100 nm were also observed by TEM (Figure 2E), in particular at low CNTs concentrations. Accordingly, the diffraction peaks observed by XRD were assigned to large Mg particles contained within the CNTs decorated with ultra-small Mg. This, thus, indicates that amounts of CNTs >5 mg ml–¹ should be used to achieve a better homogeneity of the material with the majority of Mg deposited at the CNTs surface.

The hydrogen sorption properties of the Mg/CNTs composite materials were also investigated and, as an example, only the behavior of the composite material synthesized with 2 mg ml^–1 of CNTs, i.e., containing both populations of Mg at the CNTs surface and “free” larger Mg particles is reported here. As shown by XRD analysis of the materials hydrogenated at different temperatures, 300°C was not sufficient to achieve the full hydrogenation of the large Mg particles (Figure 3). At 100°C, no MgH₂ phase appeared by XRD. However, verification of any hydrogen uptake by MS revealed that the material effectively absorbed hydrogen. Hence, as shown in Figure 4, hydrogen was released by the hydrogenated material from 150°C with a broad peak followed by significant hydrogen release peaking at 334°C. The low temperature hydrogen release—akin to the observations of Zlotea et al. on ultra-small Mg particles confined within an ordered mesoporous carbon (Zlotea et al., 2015)—was assigned to the ultra-small Mg particles at the CNTs surface and related to the shorted hydrogen diffusion paths and higher hydrogen mobility observed in nanosized Mg [9]. By contrast, the desorption of hydrogen observed above 250°C was attributed to the larger Mg particles contained within the composite material. It is noteworthy that as the temperature for hydrogenation was increased to 200 and then 300°C, the low-temperature hydrogen desorption profile disappeared in favor of a single hydrogen desorption peak shifted at 405°C; and this indicated that the CNTs supported Mg nanoparticles were extremely sensitive to temperature and their hydrogen properties degraded with increasing temperatures. One possible hypothesis is the reaction of the ultra-small Mg particles with the carbon support. However, additional SAED analysis of the material after hydrogen cycling did not reveal any Mg carbide phase or oxide phase other than the Mg hexagonal phase (Figure 5). Previous investigations have suggested some migration/agglomeration and recrystallization of nanosized Mg confined with carbon scaffold (Zlotea et al., 2013). However, this was not clearly visible from TEM analysis (Figure 5). Mg has a melting point of 650°C and it can be expected that this melting point will significantly decrease upon particle size reduction although this has never been reported for Mg. Nanosized Al (~ 2 nm) has been reported to have a reduced melting temperature of 200°C instead of the 660°C of the bulk material (Puri and Yang, 2007; Sun and Simon, 2007). Bulk Mg having a similar melting temperature, a reduction in melting temperature of the same order could be expected for the ultra-small Mg particles observed here. Hence, at 200°C and above significant melting of the Mg nanoparticles and potential diffusion within the carbon layers may lead to a deterioration of the hydrogen properties. Accordingly, our attempt to cycle this composite material and determine the kinetics for hydrogen uptake and release at low temperatures (<200°C) was unsuccessful. Only, a clear hydrogen release was observed at 300°C but this was attributed to the large Mg nanoparticles contained within the material.

Magnesium–CNTs composites were successfully synthesized through electrochemical deposition with the CNTs suspended in the electrolyte, and this demonstrated for the first time an alternative method to effectively generate ultra-small Mg particles at the surface of carbon materials. Upon hydrogen absorption at 100°C hydrogenation, the composite material was found to release hydrogen from 150°C with a broad peak followed by additional hydrogen release peaking at 334°C. The hydrogen release at low temperatures was assigned to the ultra-small Mg particles on CNTs’ surface, while the high temperature hydrogen desorption most likely corresponded to larger Mg particles contained within the composite material. In addition, the hydrogen storage properties of the Mg nanoparticles supported on the CNTs was found to be extremely sensitive to temperature, with a shift of hydrogen desorption to higher temperatures as soon as the material was taken to 200°C. Such a drastic loss of low-temperature hydrogen properties was assigned to the melting of the Mg nanoparticles and the potential diffusion of Mg within the carbon layers. It is, thus, apparent that the strategy of nanosizing Mg to enable low hydrogen cycling properties not only depends upon the ability to stabilize Mg nanoparticles but also to impede excessive reaction with the support in particular during the exothermic process of hydrogen absorption. However, assuming further functionalization of the carbon support to grow nanoparticles at specific equidistant sites, and some porosity to effectively “lock” the nanoparticle in place and restrict atomic surface diffusion of Mg, the current approach could lead to an effective way to prepare low temperature Mg-based hydrogen storage materials.

Both authors contributed equally.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.