Behavior of Compacted Magnesium-Based Powders for Energy-Storage Applications

: Energy storage is one of the main challenges to address in the near future—in particular due to the intermittent energy produced by extensive renewable energy production plants. The use of hydrides for this type of energy storage has many positive aspects. Hydride-based systems consist of absorption and desorption reactions that are strongly exothermic and endothermic, respectively. Heat management in the design of hydrogen storage tanks is an important issue, in order to ensure high-level performance in terms of the kinetics for hydrogen release / uptake and reasonable storage capacity. When loose powder is used, material in the form of pellets should be considered in order to avoid detrimental e ﬀ ects including decreased cycling performance. Moreover, sustainable materials in large-scale hydrogen reactors could be recovered and reused to improve any life cycle analysis of such systems. For these reasons, magnesium hydride was used in this study, as it is particularly suitable for hydrogen storage due to its high H 2 storage capacity, reversibility and the low costs. Magnesium hydride was ball-milled in presence of 5 wt % Fe as a catalyst, then compacted with an uniaxial press after the addition of expanded natural graphite (ENG). The materials underwent 45 cycles in a Sievert’s type apparatus at 310 ◦ C and eight bar, in order to study the kinetics and cycling stability. Scanning electron microscopy was used to investigate microstructural properties and failure phenomena. Together with Rietveld analysis, X-ray di ﬀ raction was performed for phase identiﬁcation and structural information. The pellets demonstrated suitable cycling stability in terms of total hydrogen storage capacity and kinetics.


Introduction
In the near future, energy storage will be one of the main issues in the field of renewable energy sources (RES). Hence, suitable energy-storage facilities must be developed [1]. For portable devices and the automotive sector, batteries are currently the best energy-storage solution while for stationary large-scale applications batteries are no longer viable and other technologies need to be explored. Another problem correlated to the use of batteries is related to raw material supply highlighting further the need for alternative energy-storage solutions [2,3]. Hydrogen is an energy vector that could be used for energy-storage applications-in particular for large-scale storage. Hydrogen can be stored as a compressed gas in large stationary tanks or underground cavities. Other options include liquid at low temperatures, or in suitable carriers such as ammonia and liquid organic hydrogen carriers (LOHCs). However, using compressed hydrogen requires large amounts of energy for the compression process. Reinforced composite materials must be used when designing the tanks. In addition, in the case of storing hydrogen in liquid form, vast amounts of energy consumption is required for hydrogen liquefaction and liquid boil off is inevitable, the system requiring the use of open tanks to avoid harmful and absorption are highly endothermic and exothermic reactions (about 75 kJ/molH 2 in the case of MgH 2 [77]), heat must be supplied to and removed from the system in order to allow the reactions to take place and prevent the slowing down of sorption reactions. It has been reported that carbon-based compounds not only increase thermal conductivity, but also enhance the mechanical properties of pellets which are affected by swelling during cycling. Due to cycling, pellets disaggregate with the formation of cracks and increasing porosity resulting in the formation of loose powder [78][79][80][81][82]. Use of carbon-based compounds increases mechanical stability and thermal conductivity of pellets resulting in reduced gravimetric hydrogen capacity [16,71,75,76,83]. Thermal conductivity of MgH 2 powders can be increased from about 0.25 W/mK to more than 4 W/mK when compressed into pellets with 5 wt % ENG [71,83]. Another advantage of pellets is that they can be safely handled with reduced moisture and oxygen contaminations compared to loose powder [75].
In the present work, hydrogen sorption behavior and microstructural characterization of MgH 2 -5wt % Fe-5wt % ENG (MgH 2 -5Fe-5ENG) pellets were studied. The results demonstrate that these pellets have suitable properties for the realization of hydrogen-storage reactors including scale up capabilities.

Results and Discussions
X-ray diffraction patterns (XRD) were obtained for MgH 2 -5Fe-5ENG pellets after 20 and 45 cycles ( Figure 1). Initially, the XRD shows the presence of the γ-MgH 2 phase in the as-milled powders. However, this phase was not detected in the cycled samples. In fact, γ-MgH 2 is a metastable phase formed only due to high energy ball milling and the results show that it disappears after cycling. The formation of Mg 2 FeH 6 after ball milling and repeated cycles under hydrogen has not been observed. This compound could likely be formed in trace amounts during cycling under hydrogen pressure, but its presence was not detected by XRD. In respect to the process conditions used in these experiments, higher energy and stoichiometric Mg/Fe ratio is required to obtain bulk formation of the phase Mg 2 FeH 6 [65,66]. XRD patterns and phases present in the MgH 2 -5Fe-5ENG pellets after 20 and 45 cycles are shown in Figure 1. Rietveld analysis performs the full profile fitting of the pattern and it refines the crystal structure of the crystalline phases present in the sample. In Figure 1, calculated and measured XRD patterns are shown as solid red line and as black hollow dots, respectively.   In Figure 2, the images of the pellets before and after 20 and 45 cycles are shown. It can be seen that pellets are not heavily damaged by cycling process. The volume variation after 20 and 45 cycles is about 1.9% and 18.7%, respectively.
The pellets were cycled at 310 °C at eight-bar hydrogen pressure during absorption and 1.2 bar for desorption. Figure 3 shows the kinetics for sorption reactions indicating a slight increase in time required for desorption after 45 cycles than 20 cycles. This could have been due to MgH2 crystallite coarsening which tends to stabilize with cycling.  Table 1 gives structural refinement parameters of different samples obtained by Rietveld analysis performed using MAUD software. Figure of merits for the refinements with sig < 2% and Rw < 15% was considered acceptable [84]. It was observed that MgH 2 crystallite size with respect to pristine MgH 2 , increases by an order of magnitude in the first 20 cycles and this trend slowing down with further cycling. A similar trend could be observed for Mg, but the opposite is true in the case of MgO and Fe. Table 1. Structural refinement parameters of phases present in pristine MgH 2 (as-prepared), ball milled MgH 2 , MgH 2 -5Fe-5ENG after 20 and 45 cycles, obtained by Rietveld refinement.    (5) Values in parentheses are estimated standard deviations on the last significant digit.
In Figure 2, the images of the pellets before and after 20 and 45 cycles are shown. It can be seen that pellets are not heavily damaged by cycling process. The volume variation after 20 and 45 cycles is about 1.9% and 18.7%, respectively.
The pellets were cycled at 310 • C at eight-bar hydrogen pressure during absorption and 1.2 bar for desorption. Figure 3 shows the kinetics for sorption reactions indicating a slight increase in time required for desorption after 45 cycles than 20 cycles. This could have been due to MgH 2 crystallite coarsening which tends to stabilize with cycling.
respectively was observed with cycling. In the case of desorption this may be due to the coarsening of MgH2 crystallites with cycling and for absorption due to the increased porosity of the pellet, as a result of cycling, as also reported by S. Nachev et al. [78]. A similar trend has been observed for compacted powders of MgH2 ball milled with Nb2O5 and ENG [79].  respectively was observed with cycling. In the case of desorption this may be due to the coarsening of MgH2 crystallites with cycling and for absorption due to the increased porosity of the pellet, as a result of cycling, as also reported by S. Nachev et al. [78]. A similar trend has been observed for compacted powders of MgH2 ball milled with Nb2O5 and ENG [79].    In order to evaluate the stability due to the cycling of the pellets, wt % vs. t 90 is shown in Figure 4. t 90 is the time in minutes to reach the 90 wt % of H 2 total capacity. The pellets demonstrated high stability in terms of hydrogen capacity and cycling kinetics, as the maximum quantity of hydrogen stored remains constant. A slight increase and decrease of t 90 during desorption and absorption respectively was observed with cycling. In the case of desorption this may be due to the coarsening of MgH 2 crystallites with cycling and for absorption due to the increased porosity of the pellet, as a result of cycling, as also reported by S. Nachev et al. [78]. A similar trend has been observed for compacted powders of MgH 2 ball milled with Nb 2 O 5 and ENG [79].
(c) (d)     In Figure 6a,b, radial and axial cross sections are shown. ENG could be identified in the images as it appears as dark strips. In radial direction ENG seems to be randomly distributed in the matrix of hydride while in the axial direction it is aligned along the direction perpendicular to the compaction axis. This microstructural configuration, which has been observed previously [75,81], is particularly suitable in the case of cylindrical reactors with heat flowing in the radial direction.   In Figure 6a,b, radial and axial cross sections are shown. ENG could be identified in the images as it appears as dark strips. In radial direction ENG seems to be randomly distributed in the matrix of hydride while in the axial direction it is aligned along the direction perpendicular to the compaction axis. This microstructural configuration, which has been observed previously [75,81], is particularly suitable in the case of cylindrical reactors with heat flowing in the radial direction.

Materials and Methods
High purity materials were used for powder and pellet preparation: MgH2 (98%, Alfa Aesar, Kandel, Germany), iron (Sigma-Aldrich, Darmstadt, Germany), 325 mesh and expanded natural graphite (ENG, Carbon Lorraine, La Défense, France). Sample preparation was performed following a procedure similar to that reported previously [79,85]. In order to increase matrix defects and enhance hydrogen mobility in MgH2, it was ball milled in a SPEX 8000 (SPEX, Metuchen, NJ, USA) with ball to powder ratio 10:1 for 10 h. Before milling, MgH2 was mixed with 5 wt % of Fe. After milling-in order to increase thermal conductivity and mechanical stability of the pellets-5 wt % of ENG, previously dried at 120 °C under vacuum, was mixed into the MgH2-5Fe compound by milling for 2 min. A manual uniaxial press (Specac, Orpington, UK) was used to prepare the pellets. The powder was pressed in a 8 mm diameter die at 600 MPa to prepare pellets of about 3 mm height. A quantity of about 250 mg of powder was compacted. Sieverts-type apparatus (from Advanced Materials Corporation, Pittsburgh, PA, USA) was used to study the kinetics and cycling behavior. Briefly, the compound was inserted in a stainless-steel cylindrical chamber which could be  In Figure 6a,b, radial and axial cross sections are shown. ENG could be identified in the images as it appears as dark strips. In radial direction ENG seems to be randomly distributed in the matrix of hydride while in the axial direction it is aligned along the direction perpendicular to the compaction axis. This microstructural configuration, which has been observed previously [75,81], is particularly suitable in the case of cylindrical reactors with heat flowing in the radial direction.

Materials and Methods
High purity materials were used for powder and pellet preparation: MgH2 (98%, Alfa Aesar, Kandel, Germany), iron (Sigma-Aldrich, Darmstadt, Germany), 325 mesh and expanded natural graphite (ENG, Carbon Lorraine, La Défense, France). Sample preparation was performed following a procedure similar to that reported previously [79,85]. In order to increase matrix defects and enhance hydrogen mobility in MgH2, it was ball milled in a SPEX 8000 (SPEX, Metuchen, NJ, USA) with ball to powder ratio 10:1 for 10 h. Before milling, MgH2 was mixed with 5 wt % of Fe. After milling-in order to increase thermal conductivity and mechanical stability of the pellets-5 wt % of ENG, previously dried at 120 °C under vacuum, was mixed into the MgH2-5Fe compound by milling for 2 min. A manual uniaxial press (Specac, Orpington, UK) was used to prepare the pellets. The powder was pressed in a 8 mm diameter die at 600 MPa to prepare pellets of about 3 mm height. A quantity of about 250 mg of powder was compacted. Sieverts-type apparatus (from Advanced Materials Corporation, Pittsburgh, PA, USA) was used to study the kinetics and cycling behavior. Briefly, the compound was inserted in a stainless-steel cylindrical chamber which could be

Materials and Methods
High purity materials were used for powder and pellet preparation: MgH 2 (98%, Alfa Aesar, Kandel, Germany), iron (Sigma-Aldrich, Darmstadt, Germany), 325 mesh and expanded natural graphite (ENG, Carbon Lorraine, La Défense, France). Sample preparation was performed following a procedure similar to that reported previously [79,85]. In order to increase matrix defects and enhance hydrogen mobility in MgH 2 , it was ball milled in a SPEX 8000 (SPEX, Metuchen, NJ, USA) with ball to powder ratio 10:1 for 10 h. Before milling, MgH 2 was mixed with 5 wt % of Fe. After milling-in order to increase thermal conductivity and mechanical stability of the pellets-5 wt % of ENG, previously dried at 120 • C under vacuum, was mixed into the MgH 2 -5Fe compound by milling for 2 min. A manual uniaxial press (Specac, Orpington, UK) was used to prepare the pellets. The powder was pressed in a 8 mm diameter die at 600 MPa to prepare pellets of about 3 mm height. A quantity of about 250 mg of powder was compacted. Sieverts-type apparatus (from Advanced Materials Corporation, Pittsburgh, PA, USA) was used to study the kinetics and cycling behavior. Briefly, the compound was inserted in a stainless-steel cylindrical chamber which could be positioned in a cylindrical oven and heated. The chamber was evacuated in order to remove adsorbed gases and residual moisture present on the particles surface. Ar was used as the purge gas. The chamber was heated to 120 • C under vacuum to completely remove residual moisture present in the chamber. Successively, the sample was cycled at 310 • C and eight bar hydrogen during absorption and 1.2 bar during desorption. The equipment monitors pressure variations and records temperature and pressure data. During absorption, every time the pressure decreases to a set value, the valve from a calibrated volume opens allowing hydrogen pressure to remain constant in the sample chamber. In the case of desorption, if the pressure increases automatically, the evacuation valve opens, and the hydrogen gas is removed maintaining a constant pressure. The pellets were cycled for 20 and 45 cycles in order to monitor the effects of cycling on kinetics, and in particular, the samples microstructure. After cycling, the pellets were observed by a Scanning Electron Microscope, SEM EVO MA15 (Zeiss, Oberkochen, Germany) operated at 20 kV and equipped with an X-Act silicon drift detector with AZtec analysis software (Oxford Instruments, Abingdon, UK) for energy dispersive spectroscopy (EDS) analysis. X-ray diffraction (XRD) analysis was performed on X-ray patterns obtained in a SmartLab diffractometer (Rigaku, Tokyo, Japan) equipped with a Cu Kα source radiation and a diffracted beam monochromator operated at 40 kV and 30 mA in Bragg-Brentano geometry. The automatic optics and sample height alignment routines were used to obtain reliable patterns. Rietveld analysis, which performs a full profile fitting, was done with the help of MAUD software [84]. Crystallite sizes were obtained for the different phases present in the samples after cycling. Images of the pellets were acquired with a stereo-microscope SZX12 (Olympus, Tokyo, Japan) equipped with calibrated digital image acquisition system.

Conclusions
The use of cheap, abundant, non-critical and environmentally friendly materials for hydrogen storage must be taken into consideration for the realization of large-scale facilities. In this work, the preparation of MgH 2 with 5 wt % of Fe and 5 wt % of ENG compacted powders and their hydrogen sorption behavior are reported. MgH 2 was ball milled in presence of Fe which acts as a catalyst for hydrogen sorption reactions. ENG was added to the mixture in order to enhance thermal conductivity and mechanical properties. The measurements, performed by means of a Sieverts-type apparatus, demonstrated how Fe is a suitable catalyst for Mg-based compacted powders. XRD patterns showed that MgH 2 crystallite coarsening rates decreases with cycling, while Fe and MgO crystallite size decreases. From a microstructural point of view, no important changes were observed and the pellets demonstrated promising mechanical and hydrogen sorption properties. The concept at the base of this work is that only after the material has been optimized can the most suitable reactor design be evaluated.

Conflicts of Interest:
The authors declare no conflict of interest.