Neutron diffraction studies of dense molecular solids

Reproduced from the ISIS 96 annual report

J S Loveday, R J Nelmes (University of Edinburgh), S Klotz , G Hamel, J M Besson (Université P et M Curie, Paris),  W G Marshall (ISIS),

Studies of simple-molecular systems under pressure allow investigation of a wide range of phenomena, such as progressive orientational ordering and eventual molecular dissociation. In addition, the outer planets are thought to be composed of dense, simple-molecular solids whose high-pressure behaviour is thus important for planetary modelling. Neutron diffraction is the technique of choice for structural studies of the many hydrogen-containing systems. The recent development at ISIS of the Paris-Edinburgh pressure cell has opened up this field, extending the available pressure range to 10 GPa (100 kbar) and even to above 20 GPa in some cases. The most extensive work so far has been done on ammonia, D2S and ice.

Figure F1.1 The orthorhombic structure of ammonia phase IV. The H-bonds are shown as green dashed lines. An example of bifurcated bonding is shown in yellow.

The P-T phase diagram of ammonia has three different phases in the range up to 3 GPa, and phase IV is stable from 3 to 13 GPa. This latter phase is thus the important first step from the low-pressure structures towards the eventual molecular dissociation expected to occur at ~60 GPa. On the basis of X-ray studies, phase IV has long been thought to have a simple hexagonal close packed (hcp) structure. However, neutron diffraction data collected on POLARIS reveal several strong non-hcp peaks, with the pattern fitting better to an orthorhombic unit cell. The full structure has now been solved (Phys. Rev. Lett., 76 (1996) 74). Although the nitrogen atoms form the pseudo-hcp arrangement indicated by the X-ray data, the deuterium atoms do not. The molecules, which were expected to be freely rotating on the basis of the hcp solution, in fact form a fully ordered, H-bonded arrangement. The bonding is also unexpectedly complicated. One of the deuterium atoms appears to be bonded to two nitrogen atoms; such bifurcated H-bonds are occasionally found in larger molecules containing NH3 and NH2 groups, but it is a surprising feature in ammonia. Previously, ammonia was thought to progress towards dissociation through a series of simple hexagonal and cubic structures. Our results show that the behaviour is considerably more complex.

Figure F1.2 Part of the D2S phase 1 structure showing a central sulphur atom, the six associated deuterium sites (ellipsoids) and the six nearest neighbour sulphur atoms which lie in a plan normal to <111>. The arrows denote the direction of displacement of the sulphur atoms from fcc sites and the dashed lines denote possible H-bond contacts.

Hydrogen sulphide is unusual amongst molecular solids in that the transition sequence on cooling, from spherical disordering of the molecules to full orientational order, involves an intermediate phase (phase II) in which the molecular rotation is thought to be axial, with the hydrogen atoms confined to two rings, one each side of the S atom. A similar phase (phase I´) has been found at high pressure, but with the interesting difference that there appear to be accompanying distortions of the sulphur arrangement. This system is ideal for complementary X-ray and neutron studies because X-rays scatter mostly from the S atoms and neutrons from the H (or D) atoms. X-ray data collected from phase I´ on station 9.1 at SRS Daresbury allowed us to determine the S positions to be displaced ~0.1 Å along <111> from an fcc arrangement (arrows, figure F1.2). We could then proceed to a full solution of the neutron data collected on POLARIS. The model adopted was based on the known geometry of the D2S molecule, using libration to reproduce the scattering from a rotating molecule. A model with the D atoms localised on six sites around each S atom (three above and three below the plane of S atoms see figure F1.2) gives a better fit to the data than models with the deuterium more uniformly distributed around complete rings. The 50% probability envelopes obtained give very interesting evidence that the deuterium distribution is non-uniform, with maxima directed towards the three neighbouring S atoms that are displaced in a direction which reduces the D...S distance. This is not seen in the low-pressure phase II, and suggests that phase I´ marks the onset of H-bonding under pressure.

Figure F1.3 Tetragonal Ice VIII (left) and cubic ice VII (right). D sites are 50% occupied in ice VII.

A long-standing problem involving disorder is the precise structural relationship between phases VII and VIII of ice. These are the only known phases above 2 GPa, one occurring above and the other below 273 K in the range up to 10 GPa. The structures are closely related, as shown in figure F1.3. Both consist of two interpenetrating networks, with molecular orientations that are disordered in ice VII but ordered in ice VIII. It has long been known that neutron diffraction studies give an anomalously short OD distance in ice VII, and it has been suggested by Kuhs et al that this arises because the O atoms are disordered

Figure F1.4 Temperature dependence of the thermal motion of dueterium (upper plot) and oxygen (lower plot) atoms in ice VII (red symbols) and ice VIII (blue symbols).

over six sites displaced ~0.1 Å along <100> from the mean oxygen positions. To look for direct evidence for such disorder, we have made careful measurements on POLARIS of the atomic thermal motion on either side of the transition. As shown in figure F1.4, hysteresis allowed measurements to be made at overlapping temperatures in both phases, and there are clear steps at the transition in both the overall oxygen thermal motion, Uiso(O), and the component of the deuterium thermal motion perpendicular to the

OD bond, Uperp(D). However, the along-bond motion of the deuterium, Ull(D), remains nearly constant. This, taken with other evidence that the lattice stiffness does not change at the transition, strongly suggests that true thermal motion is the same in both phases. The steps in Uiso(O) and Uperp(D) can then be attributed to the onset of site disorder in ice VII, and the magnitude of the steps allows us to obtain the size of the O and D atom displacements. One clear result is that the <100> displacement model for the O atoms can be shown to give an OD distance that is no longer smaller but significantly larger for ice VII than for ice VIII suggesting that this model may not be correct. This is interesting and important because a different model would open up a quite new understanding of the structural relationship and transition between the two phases. Various other possibilities are now being tested, such as displacements along <111> or <110>.

These three examples illustrate the important new information that can be obtained from high-pressure neutron diffraction studies of molecular systems, for which ISIS now has unique facilities. Similar studies are in progress on a range of other simple systems of fundamental interest, including methane (which is also found in planetary interiors), DBr (which has similarities to D2S), and nitrogen (in which there are important unsolved problems of partial orientational ordering).