Magnesium transition-metal based lightweight hydrogen storage materials : a NMR study

As the world is running out of fossil fuels, sustainable energy sources and efficient forms of energy storage are studied nowadays. Hydrogen is an energy carrier and has advantages that it is abundant and does not pollute the environment. Metal hydrides are studied for their reversible hydrogen storage properties. MgH2 is a promising candidate for light-weight hydrogen storage material. It can store up to 7.7 wt-% of hydrogen. However, it suffers from high thermodynamic stability and poor sorption kinetics. To improve its performance, Mg is doped with transition metals (TM). The resulting complex alloys are not fully understood on an atomistic level. This doctoral dissertation therefore examines the following central research questions: Is it possible to distinguish between hydrogen atoms with different metal coordination, Qn = H-MgnTM4-n, where n is the number of Mg atoms in the first coordination sphere. Do hydrogen atoms from one metal-coordination environment move to all other coordination environments? How are the different metal coordinations arranged within the lattice? What is a typical length scale separation between different metal coordinations? A single technique alone is not adequate to characterize these complex materials completely. Therefore, we have used a combination of techniques such as fast Magic Angle Spinning (MAS) solid state Nuclear Magnetic Resonance (ssNMR) spectroscopy, powder X-ray diffraction (XRD) and Neutron diffraction (ND) to elucidate the nanostructure of Mg-based hydrogen-storage materials. Diffraction–based techniques are applicable to materials which possess long-range structural ordering. Long-range ordering is not required for NMR. Information about the local environment of the energy carrier i.e hydrogen (deuterium) is necessary to optimize the hydrogen-storage materials. ssNMR is unique and powerful method in this respect. Advanced pulse sequences can be employed to obtain information about the local environment of the energy carrier itself, directly. Moreover, motion of the energy carrier can be monitored which is essential for hydrogen-storage materials. Chapter 2 gives the theoretical understanding of the methodologies that are used in this study to investigate Mg-TM based hydrides. We have studied gas-phase deuterated Mg0.65Sc0.35 and Mg0.65Ti 0.35 materials. In Mg0.65Sc0.35D2.2, with a novel 2 H double-quantum NMR with 45 Sc irradiation and TRAPDOR NMR techniques, we were able to observe that Mg and Sc are not randomly distributed within the XRD determined coherent lattice, instead we found Mg-rich and Sc-rich clusters. With Two-Dimensional Exchange Spectroscopy (2D Exsy), the length separation between these clusters was found to be within few unit cells. Sc is a valuable element and hence, its neighbor in the periodic table, Ti, was investigated. Mg and Ti are immiscible under equilibrium conditions. Therefore, we have investigated Mg0.65Ti0.35 alloy prepared with non-equilibrium methods, namely ball-milling and magnetron sputtering. Chapter 3 describes MgTi alloy prepared by ball-milling and subsequently deuterated at 175 °C and 70 bars. Both NMR and XRD showed that the alloy, after deuteration, phase separates into MgD2 and TiD2. Additionally, NMR indicates the presence of another TiDy phase that is not visible with XRD. Two- Dimensional Exchange Spectroscopy (2D Exsy) reveals deuterium exchange between the XRD invisible TiDy phase and the MgD2. With One-Dimensional Exsy, a weak temperature dependence is found corresponding to an effective activation barrier for deuterium exchange of approximately 12 kJ/mol. This low effective activation energy is probably the result of a broad deuterium-mobility distribution. Comparing Mg0.65Sc0.35D2.2 and Mg0.65Ti0.35D0.65, we propose that the stabilization of the nano-structure of the later maybe a possible outcome from the coherent coupling of individual crystal lattices of MgD2 and TiD2. XRD homogenous MgTi films can be prepared by rf magnetron sputtering. Chapter 4 elucidates the hydrogen siting and dynamics in Mg0.65Ti0.35 prepared by this method. After gas-phase deuterium loading at room temperature, we did not observe a macro-phase separated TiD2 phase unlike in ballmilled material. We do observe, partly resolved signals of deuterium located in non-conductive domains at tetrahedral Mg4 and, possibly, mixed MgnTi4-n sites (4 ppm), and deuterium at Ti4 sites in conducting TiD2 nanodomains (-29 and -68 ppm). No bulk-TiD2 signal at -150 ppm is observed, in contrast to what we find in ball-milled Mg0.65Ti0.35D0.65, which is largely phase separated. 2D Exsy indicates deuterium exchange between deuterium states resonating at position 4 and –29 ppm, but not with those giving rise to the resonance at –68 ppm. The signal at –68 ppm probably represents deuterium atoms which are stably bound to Ti. The observed deuterium exchange and the reduced Knight shift compared to bulk TiD2 are explained using a model with TiD2 nano-slabs. At temperatures T ?? 300 K, the intensity of the signal from the Mg-rich sites decreases and a new signal appears at –10 ppm. This is a reversible phase transition and the upfield shift indicates that deuterium is in contact with Ti, probably at Mg-Ti interface. Since we study the energy carriers, i.e. hydrogen or deuterium atoms directly with NMR, it is of most importance to know the visibility of all the energy carriers. Chapter 5 investigates the visibility of deuterium atoms in the melt-cast Mg0.65Sc0.35D2.2.The loss of overall signal intensity in 1D Exsy indicate the presence of NMR-invisible or "dark" deuterium states. We explain the invisibility on the basis of second-order quadrupolar line broadening arising from the unequal charge distribution in the mixed co-ordination states, Q n = H-MgnSc4-n, (1 £ n £ 3) where n is the number of Mg atoms in the first coordination sphere. Approximately 30% of deuterium atoms are invisible. With this correction for deuterium visibility, the distribution of Mg and Sc over the metal within the lattice tends to be closer to statistical distribution rather than clustering of Mg and Sc rich domains. Nano-sizing and confinement is another approach employed to overcome the disadvantages of MgH2. Carbon supported MgH2 nano-composites prepared by Mg-melt infiltration is studied in Chapter 6. The susceptibility of the nano-porous carbon support results in broad resonances in 1 H MAS NMR spectra. For highly packed samples at 11.7 T (500 MHz) the NMR visibility is severely affected by the conductivity of the carbon. We show that the problems associated with conductivity and susceptibility of nano porous carbon can be overcome by working at a lower magnetic field, 4.7 T (200 MHz). The NMR visibility at a lower field is not affected by the packing density of the materials, which implies quantitative NMR is still feasible. MgH2 within the nano-pores of carbon are not detected with XRD. Static and MAS 1 H NMR of a series of MgH2 carbon nanocomposites with different MgH2 content, indicate the presence of two hydride phases with different spin-lattice relaxation time and chemical shift. The component with the broad static 1 H NMR lineshape and long relaxation time (~10 2 s) is assigned to bulk MgH2. The second component has a narrower static 1 H NMR lineshape and a shorter relaxation time (~10 -1 s) and is tentatively assigned to a nanophase consisting of MgH2 and Mg(OH)2. The 1 H NMR lineshape of MgH2/Mg(OH)2 nano-phase is narrower than that of bulk-MgH2, which indicates a higher mobility of the hydrogen atoms in the MgH2/Mg(OH)2 nano-phase. We have tried to separately identify 1 H NMR signals from MgH2 and Mg(OH)2 in the nanophase by using 20-kHz MAS. However, the susceptibility broadening by the nanoporous carbon is too strong. Assuming that the length scale of the susceptibility variation might be longer than the typical distance between MgH2 and Mg(OH)2 we have further tried to enhance the chemical resolution in the inhomegeneous local field caused by susceptibility by use of 2D 1 H MAS NMR Exsy. However, 2D Exsy shows only non specific broadening as a function of mixing time. At the timescale of 10 -1 s, there is a complete non-specific spin exchange or hydrogen exchange over the susceptibility broadened resonance. This indicates that the length scale of susceptibility variation is smaller or equal to that of the average distance between MgH2 and Mg (OH)2 spin- or the susceptibility determined NMR chemical shifts. Finally, chapter 7 summarizes the main findings of the study described in previous chapters