6 Neutron scattering studies for analysing solid-state hydrogen storage D. K. R O S S, University of Salford, UK
6.1
Introduction
In the search for useful hydrogen storage materials, whether solid state or H2 adsorbed on surfaces, the most immediately relevant data are, of course, the macroscopic thermodynamics and kinetics of the ab(ad)sorption/desorption process. However, if we seek to use scientific reasoning to guide our search, it is clearly necessary to understand the behaviour of hydrogen on the atomic scale both within the material and when entering or leaving it. It is in the latter context that thermal neutron scattering can make a crucial contribution, particularly when linked to ab initio modelling. Neutron scattering is a widely used technique for studying the structure and dynamics of solids and liquids. This is because a thermalised neutron emerging from a moderator at a nuclear reactor or pulsed neutron source has a wavelength that is comparable to interatomic distances and an energy that is comparable to the energies of atomic excitations in the solid. Moreover, we have good techniques for measuring diffraction patterns and energy transfers very accurately. There are many cases where neutron diffraction (ND) has significant advantages over X-ray or electron diffraction while inelastic neutron scattering (INS) often has advantages over IR and Raman measurements, for instance, in the measurement of phonon dispersion curves in single crystal samples. Other areas of application include small angle neutron scattering (SANS) for measuring longer range spatial correlations, where neutrons have many advantages over X-rays and quasi-elastic neutron scattering (QENS), which probes the diffusive motions, particularly of hydrogen, in a unique way. One notable property of the neutron, which we will not be particularly concerned with here, is its magnetic moment which allows a whole array of methods for investigating the magnetic structure and dynamics of solids. General textbooks are available on the subject, such as Squires [1]. A full description of quasi-elastic neutron scattering may be found in the book by Bee [2]. 135
136
Solid-state hydrogen storage
6.2
The neutron scattering method
6.2.1
Advantages of the neutron scattering technique for investigating solid-state hydrogen storage systems
The particularly significant contribution that neutron scattering can make to the study of hydrogen in solids arises from a number of factors: •
•
•
• •
• •
The neutron scattering cross-section of the proton (80 × 10–28 m–2 or, colloquially, 80 barns) is between one and two orders of magnitude greater than that of any other isotope so that it is easy to detect hydrogen in quite low concentrations in solids. The hydrogen cross-section is predominantly incoherent (σinc = 79.7 b, σcoh = 1.8 b) for reasons described below and so the measured scattering is dominated by scattering from individual atoms summed over all H atoms present. Incoherent inelastic scattering from polycrystalline solids reproduces the frequency distribution of the hydrogen vibrations in a solid, weighted with the square of the hydrogen vibration amplitudes and inversely with the mass of the hydrogen atom and so normally other elements can be ignored in a solid containing hydrogen. It should be noted that this amplitude-weighted density of vibrational states can also be directly derived from ab initio simulations of the solid’s dynamics and hence provides a direct check on the accuracy of these calculations. Moreover, the enthalpy change on hydriding/dehydriding can be estimated using this density of vibrational states. This technique is particularly useful for electrical conductors because IR and Raman techniques are not available. However, the technique is still very valuable for insulators because the electromagnetic techniques only provide frequencies at the zone centre and the intensities of these features are strongly influenced by the selection rules involved in the electronic response. Incoherent quasi-elastic neutron scattering from diffusing hydrogen atoms allows us to investigate the elementary jumps involved in the tracer diffusion process. On the other hand, the deuteron (2D), has a relatively large cross section which is predominantly coherent (σcoh = 5.6 b and σinc = 1.8 b) and so the diffraction pattern from a crystalline solid contains terms that enable us to locate D in crystal structures through the interference between scattered waves from all pairs of nuclei. H and D have scattering lengths of opposite sign so that there are a whole range of phase contrast techniques that can identify the role of hydrogen, for instance in SANS. A general advantage of neutrons is that they are relatively penetrating so
Neutron scattering studies
•
137
that samples can be contained in cans that are able to withstand considerable gas pressures and temperatures. This feature enables us to measure the actual process of (de)hydrogenation in situ under working hydrogen pressures and so we can directly investigate processes that determine the kinetics of hydrogen cycling into and out of a hydrogen store. Being the lightest isotope, hydrogen has very pronounced quantum mechanical properties. As the neutron interacts with the nucleus, it directly observes these properties, formally observing the transition state probability between two quantum states of the solid. Of particular importance to us here is the nature of the scattering from para- and ortho-molecular hydrogen because selection rules require that these molecules have antisymmetric total wave functions. Hence we find that para-hydrogen, with anti-parallel spins has only even rotational quantum numbers (J) while ortho-hydrogen, with parallel spins can only have odd values of J. Moreover, the scattering lengths combine in such a way that para-hydrogen only scatters significantly via a spin-flip process involving a J = 0 to 1 transition while ortho-hydrogen has large cross-sections for both spinflip and non-spin-flip scattering. Also, the energy difference between the lowest energy para-state (J = 0) and the lowest ortho state (J = 1) for free hydrogen molecules is 14.7 meV, which is easily measured with high accuracy. These features make it instantly possible to identify whether hydrogen is present in its molecular form and to probe the trapping states with progressively less trapping energy as hydrogen is added. The basic theory of neutron scattering from molecular hydrogen has been described by Young and Koppel [3].
The application of neutron scattering to understanding the role of hydrogen in solids has been described in various general reviews [4,5]. Specific applications to intermetallic compounds are described by Richter et al. [6]. Its use in vibrational spectroscopic studies in chemistry, biology, materials science and catalysis are described by Mitchell et al. [7]. For a review of the basic properties of metal–hydrogen systems, we would refer the reader to the book by Fukai [8].
6.3
Studies of light metal hydrides
Because of the need to store hydrogen in a system with a low mass so that it can approach the US Department of Energy (DoE) gravimetric target of 6– 9% of the storage system mass, we will be interested in applying the technique to complex metal hydrides such as the alanates, borohydrides and amides and to intermetallic hydrides based on magnesium. In these novel systems, various interesting features can be investigated such as the existence of continuous phase transitions as have been observed in in situ diffraction
138
Solid-state hydrogen storage
studies of the lithium amide system [9,10] and in similar investigations into the hydrogen absorption/desorption process in alanates. Borohydrides are also of interest but from our point of view suffer from the disadvantage that samples would have to be made from the 11B isotope because of the strong neutron capture cross-section of 10B.
6.4
Studies of molecular hydrogen trapping in porous materials
The alternative proposed method of storing hydrogen in a solid matrix involves the physisorption of H2 on a high surface area material. The original interest in this approach was based on some very promising reports of high mass uptakes in single-walled carbon nanotubes [11] and carbon nanofibres [12]. Although these results seem to have been affected by residual water in the vacuum system, there have since been a wide range of experiments on a variety of carbon frameworks, zeolites, ice clathrates, metal oxide frameworks (MOFs), etc. While the normal van der Waals adsorption energy is rather weak (~ 40 meV/atom) and would require that the adsorber was cooled to 80 K to get a reasonable uptake, recent work suggests that it may be possible to enhance this adsorption energy with suitable catalysts, possibly involving spillover effects. The work to be described here (Section 6.7) reported below shows how neutron inelastic scattering can be used to characterise the potential energy surface at different trapping sites in a solid.
6.5
The basic theory of neutron scattering
The neutron is an uncharged nucleon with a mass essentially the same as a proton, a spin of 1/2 and a magnetic moment of –1.913 nuclear bohr magnetons. As we can neglect consideration of the magnetic interaction, being only interested here in the behaviour of hydrogen, we can confine our attention to the neutron–nucleus interaction. Full discussion of the theory can be found elsewhere [1,2] so the present treatment is considerably simplified with emphasis on physical principles rather than mathematical rigour. The probability of a neutron interacting with a nucleus is given in terms of the microscopic cross-section, σ, defined as the rate of interaction/nucleon in unit incident neutron flux (one neutron/unit area/second). Of the possible types of interaction, we are only interested in scattering processes, for which the cross-section is σs. In general, scattering events can result in changes to both the direction and the energy of the neutron so we define a double differential cross-section, d2σ(E0 – E′, θ)/dE′dΩ to be the rate of scattering from an initial energy E0 to a unit range about the final energy E′ and through a scattering angle, θ, into unit solid angle, Ω. Assuming that the nucleus does not change its internal energy level, the scattering process would be a simple
Neutron scattering studies
139
two-body collision until the neutron energy is lowered to near thermal energies where the binding of the target atom in the solid becomes significant and hence influences the scattering process. It is, of course, this near thermal energy range (0–1 eV) that is of interest to us here. From the definitions above, we can write: ∞
∫ ∫ ∞
4π
d 2 σ( E0 – E ′ , θ ) d Ω dE ′ = dE ′ d Ω
∫
4π
d σ( E0 , θ ) d Ω = σ s ( E0 ) dΩ [6.1]
For a fixed nucleus at the origin, this quantity is easily derived in quantum mechanical terms. If we represent the incident flux as a plane wave, we can write ψ(r) = exp (ik0 · r), where k0 is the incident wave vector and hk = mv is the momentum of neutron (with m = neutron mass and v = neutron velocity). On this basis, the neutron probability density in the incident beam, ψψ*, is unity so that the incident flux is v neutrons/unit area/s. We can now represent the scattered wave as an expanding spherical wave centred on the origin which can be written ψ′(r) = (–a/r) exp(ik′ · r). Here, k′ is the scattered wave vector and a is known as the neutron scattering length. The – sign is conventionally inserted here because, for most isotopes, the ‘compound nucleus’ formed from the neutron and the target nucleus is far from a stable nuclear state so we get ‘potential scattering’ with a 180° phase change in the scattering process. The scattering length a obviously depends on the nuclear properties (and is different for every isotope). In particular, if the nucleus has a finite spin I, the neutron–nucleus interaction is spin-dependent and so we need to define a+ and a– for parallel and anti-parallel orientations of the neutron spin relative to the nuclear spin. Here the total spin of the compound nucleus, J+/– = (I +/– 1/2), needs to be defined because it determines the statistical weight of the a+ and a– states. It should be noted that if the compound nucleus has an energy anywhere near the energy of a stable state, the values of a can vary quite widely and indeed can be either positive or negative as here there is strong coupling between the incident neutron and the target nucleus. The proton is a good example of this because in the anti-parallel state, the compound nucleus formed is very close to the ground state of the deuteron (spin zero!). This is why a– is negative and so much larger than for any other nucleus. It also follows that the scattering cross-section for an isolated fixed nucleus will be dσ/dΩ = a2 = σ/4π. We can now consider the scattering from a set of N nuclei rigidly fixed at positions ri (so that the neutron does not exchange energy with the target nucleus). If we observe the scattered waves a long way from the scattering nuclei, we can sum them and then define the scattered flux by multiplying the summed wave function by its complex conjugate and hence we obtain [1] an expression for the angular cross-section defined/nucleon: