The structure of water: A historical perspective.

Attempts to understand the molecular structure of water were first made well over a century ago. Looking back at the various attempts, it is illuminating to see how these were conditioned by the state of knowledge of chemistry and physics at the time and the experimental and theoretical tools then available. Progress in the intervening years has been facilitated by not only conceptual and theoretical advances in physics and chemistry but also the development of experimental techniques and instrumentation. Exploitation of powerful computational methods in interpreting what at first sight may seem impenetrable experimental data has led us to the consistent and detailed picture we have today of not only the structure of liquid water itself and how it changes with temperature and pressure but also its interactions with other molecules, in particular those relevant to water's role in important chemical and biological processes. Much remains to be done in the latter areas, but the experimental and computational techniques that now enable us to do what might reasonably be termed "liquid state crystallography" have opened the door to make possible further advances. Consequently, we now have the tools to explore further the role of water in those processes that underpin life itself-the very prospect that inspired Bernal to develop his ideas on the structure of liquids in general and of water in particular.

Detailed crystallographic analysis of the ice V to ice XIII hydrogen-ordering phase transition.

Ice V is a structurally highly complex material with 28 water molecules in its monoclinic unit cell. It is classified as a hydrogen-disordered phase of ice. Yet, some of its hydrogen-bonded water molecules display significant orientational order. Upon cooling pure ice V, additional orientational ordering cannot be achieved on the experimental time scale. Doping with hydrochloric acid has been shown to be most effective in enabling the phase transition of ice V to its hydrogen-ordered counterpart ice XIII. Here, we present a detailed crystallographic study of this phase transition investigating the effects of hydrochloric and hydrofluoric acid as well as lithium and potassium hydroxide doping. The magnitudes of the stepwise changes in the lattice constants during the phase transition are found to be more sensitive indicators for the extent of hydrogen order in ice XIII than the appearance of new Bragg peaks. Hydrofluoric acid and lithium hydroxide doping enable similar ordering processes as hydrochloric acid but with slower kinetics. The various possible space groups and ordered configurations of ice XIII are examined systematically, and the previously determined P21/a structure is confirmed. Interestingly, the partial hydrogen order already present in ice V is found to perpetuate into ice XIII, and these ordering processes are found to be independent of pressure. Overall, the hydrogen ordering goes along with a small increase in volume, which appears to be the origin of the slower hydrogen-ordering kinetics under pressure. Heating pressure-quenched samples at ambient pressure revealed low-temperature "transient ordering" features in both diffraction and calorimetry.

Detailed crystallographic analysis of the ice VI to ice XV hydrogen ordering phase transition.

The D2O ice VI to ice XV hydrogen ordering phase transition at ambient pressure is investigated in detail with neutron diffraction. The lattice constants are found to be sensitive indicators for hydrogen ordering. The a and b lattice constants contract whereas a pronounced expansion in c is found upon hydrogen ordering. Overall, the hydrogen ordering transition goes along with a small increase in volume, which explains why the phase transition is more difficult to observe upon cooling under pressure. Slow-cooling ice VI at 1.4 GPa gives essentially fully hydrogen-disordered ice VI. Consistent with earlier studies, the ice XV obtained after slow-cooling at ambient pressure is best described with P-1 space group symmetry. Using a new modelling approach, we achieve the atomistic reconstruction of a supercell structure that is consistent with the average partially ordered structure derived from Rietveld refinements. This shows that C-type networks are most prevalent in ice XV, but other structural motifs outside of the classifications of the fully hydrogen-ordered networks are identified as well. The recently proposed Pmmn structural model for ice XV is found to be incompatible with our diffraction data, and we argue that only structural models that are capable of describing full hydrogen order should be used.

Renaissance of Bernal's random close packing and hypercritical line in the theory of liquids.

We review the scientific history of random close packing (RCP) of equal spheres, advocated by J D Bernal as a more plausible alternative to the non-ideal gas or imperfect crystal as a structural model of simple liquids. After decades of neglect, computer experiments are revealing a central role for RCP in the theory of liquids. These demonstrate that the RCP amorphous state of hard spheres can be well defined, is reproducible, and has the thermodynamic status of a metastable ground state. Further evidence from simulations of square-well model liquids indicates an extended role of RCP as an amorphous ground state that terminates a supercooled liquid coexistence line, suggesting likewise for real liquids. A phase diagram involving percolation boundaries has been proposed in which there is no merging of liquid and gas phases, and no critical singularity as assumed by van der Waals. Rather, the liquid phase continuously spans all temperatures, but above a critical dividing line on the Gibbs density surface, it is bounded by a percolation transition and separated from the gas phase by a colloidal supercritical mesophase. The colloidal-like inversion in the mesophase as it changes from gas-in-liquid to liquid-in-gas can be identified with the hypercritical line of Bernal. We therefore argue that the statistical theory of simple liquids should start from the RCP reference state rather than the ideal gas. Future experimental priorities are to (i) find evidence for an amorphous ground state in real supercooled liquids, (ii) explore the microscopic structures of the supercritical mesophase, and (iii) determine how these change from gas to liquid, especially across Bernal's hypercritical line. The theoretical priority is a statistical geometrical theory of RCP. Only then might we explain the coincident values of the RCP packing fraction with Buffon's constant, and the RCP residual entropy with Boltzmann's ideal gas constant.

Small-angle neutron scattering study of micropore collapse in amorphous solid water.

Vapor-deposited amorphous solid water (ASW) is the most abundant solid molecular material in space, where it plays a direct role in both the formation of more complex chemical species and the aggregation of icy materials in the earliest stages of planet formation. Nevertheless, some of its low temperature physics such as the collapse of the micropore network upon heating are still far from being understood. Here we characterize the nature of the micropores and their collapse using neutron scattering of gram-quantities of D2O-ASW of internal surface areas up to 230 ± 10 m(2) g(-1) prepared at 77 K. The model-free interpretation of the small-angle scattering data suggests micropores, which remain stable up to 120-140 K and then experience a sudden collapse. The exact onset temperature to pore collapse depends on the type of flow conditions employed in the preparation of ASW and, thus, the specific surface area of the initial deposit, whereas the onset of crystallization to cubic ice is unaffected by the flow conditions. Analysis of the small-angle neutron scattering signal using the Guinier-Porod model suggests that a sudden transition from three-dimensional cylindrical pores with 15 Å radius of gyration to two-dimensional lamellae is the mechanism underlying the pore collapse. The rather high temperature of about 120-140 K of micropore collapse and the 3D-to-2D type of the transition unraveled in this study have implications for our understanding of the processing and evolution of ices in various astrophysical environments.

The polymorphism of ice: five unresolved questions.

Our recent discovery of three new phases of ice has increased the total number of known distinct polymorphs of ice to fifteen. In this Perspective article, we give a brief account of previous work in the field, and discuss some of the particularly interesting open questions that have emerged from recent studies. These include (i) the effectiveness of acid and base dopants to enable hydrogen-ordering processes in the ices, (ii) the comparison of the calorimetric data of some of the crystalline phases of ice and low-density amorphous ice, (iii) the disagreement between the experimental ice XV structure and computational predictions, (iv) the incompleteness of some of the hydrogen order/disorder pairs and (v) the new frontiers at the high and negative pressure ends of the phase diagram.

How many amorphous ices are there?

Many acronyms are used in the literature for describing different kinds of amorphous ice, mainly because many different preparation routes and many different sample histories need to be distinguished. We here introduce these amorphous ices and discuss the question of how many of these forms are of relevance in the context of polyamorphism. We employ the criterion of reversible transitions between amorphous "states" in finite intervals of pressure and temperature to discriminate between independent metastable amorphous "states" and between "substates" of the same amorphous "state". We argue that the experimental evidence suggests we should consider there to be three polyamorphic "states" of ice, namely low-(LDA), high-(HDA) and very high-density amorphous ice (VHDA). In addition to the realization of reversible transitions between them, they differ in terms of their properties, e.g., compressibility, or number of "interstitial" water molecules. Thus they cannot be regarded as structurally relaxed variants of each other and so we suggest considering them as three distinct megabasins in an energy landscape visualization.

Activity and dynamics of an enzyme, pig liver esterase, in near-anhydrous conditions.

Water is widely assumed to be essential for life, although the exact molecular basis of this requirement is unclear. Water facilitates protein motions, and although enzyme activity has been demonstrated at low hydrations in organic solvents, such nonaqueous solvents may allow the necessary motions for catalysis. To examine enzyme function in the absence of solvation and bypass diffusional constraints we have tested the ability of an enzyme, pig liver esterase, to catalyze alcoholysis as an anhydrous powder, in a reaction system of defined water content and where the substrates and products are gaseous. At hydrations of 3 (±2) molecules of water per molecule of enzyme, activity is several orders-of-magnitude greater than nonenzymatic catalysis. Neutron spectroscopy indicates that the fast (≤nanosecond) global anharmonic dynamics of the anhydrous functional enzyme are suppressed. This indicates that neither hydration water nor fast anharmonic dynamics are required for catalysis by this enzyme, implying that one of the biological requirements of water may lie with its role as a diffusion medium rather than any of its more specific properties.

Molecular self-assembly in a model amphiphile system.

The physical origin of the large and negative excess entropy of mixing of alcohols and water remains controversial. In contrast to standard explanations that evoke concepts of water structuring, recent work has shown that, at ambient conditions, it can be quantitatively explained in terms of molecular scale partial demixing of the two components. Here, we estimate the negative excess entropy (DeltaS(E)) of aqueous methanol at low temperature and high pressure using experimentally-derived structural data and a recently introduced cluster model. On cooling to 190 K the cluster sizes increase, but the change in DeltaS(E), which according to this method of calculation depends on the surface area to volume ratio of the clusters, is not significant, suggesting that the topology of the clusters must change with decreased temperature. On compression the cluster sizes also increase, and DeltaS(E) is now positive, suggesting an even more pronounced change in cluster topology with increased pressure. This work suggests that it is the amphiphilic nature of a molecule that determines aggregation and self-assembly processes in aqueous solution. The results therefore give useful insight into the processes of cold and pressure denaturation of proteins.

Ice XV: a new thermodynamically stable phase of ice.

A new phase of ice, named ice XV, has been identified and its structure determined by neutron diffraction. Ice XV is the hydrogen-ordered counterpart of ice VI and is thermodynamically stable at temperatures below approximately 130 K in the 0.8 to 1.5 GPa pressure range. The regions of stability in the medium pressure range of the phase diagram have thus been finally mapped, with only hydrogen-ordered phases stable at 0 K. The ordered ice XV structure is antiferroelectric (P1), in clear disagreement with recent theoretical calculations predicting ferroelectric ordering (Cc).

A calorimetric study on the low temperature dynamics of doped ice V and its reversible phase transition to hydrogen ordered ice XIII.

Doped ice V samples made from solutions containing 0.01 M HCl (DCl), HF (DF), or KOH (KOD) in H(2)O (D(2)O) were slow-cooled from 250 to 77 K at 0.5 GPa. The effect of the dopant on the hydrogen disorder --> order transition and formation of hydrogen ordered ice XIII was studied by differential scanning calorimetry (DSC) with samples recovered at 77 K. DSC scans of acid-doped samples are consistent with a reversible ice XIII <--> ice V phase transition at ambient pressure, showing an endothermic peak on heating due to the hydrogen ordered ice XIII --> disordered ice V phase transition, and an exothermic peak on subsequent cooling due to the ice V --> ice XIII phase transition. The equilibrium temperature (T(o)) for the ice V <--> ice XIII phase transition is 112 K for both HCl doped H(2)O and DCl doped D(2)O. From the maximal enthalpy change of 250 J mol(-1) on the ice XIII --> ice V phase transition and T(o) of 112 K, the change in configurational entropy for the ice XIII --> ice V transition is calculated as 2.23 J mol(-1) K(-1) which is 66% of the Pauling entropy. For HCl, the most effective dopant, the influence of HCl concentration on the formation of ice XIII was determined: on decreasing the concentration of HCl from 0.01 to 0.001 M, its effectiveness is only slightly lowered. However, further HCl decrease to 0.0001 M drastically lowered its effectiveness. HF (DF) doping is less effective in inducing formation of ice XIII than HCl (DCl) doping. On heating at a rate of 5 K min(-1), kinetic unfreezing starts in pure ice V at approximately 132 K, whereas in acid doped ice XIII it starts at about 105 K due to acceleration of reorientation of water molecules. KOH doping does not lead to formation of hydrogen ordered ice XIII, a result which is consistent with our powder neutron diffraction study (C. G. Salzmann, P. G. Radaelli, A. Hallbrucker, E. Mayer, J. L. Finney, Science, 2006, 311, 1758). We further conjecture whether or not ice XIII has a stable region in the water/ice phase diagram, and on a metastable triple point where ice XIII, ice V and ice II are in equilibrium.

Protein dynamics and stability: the distribution of atomic fluctuations in thermophilic and mesophilic dihydrofolate reductase derived using elastic incoherent neutron scattering.

The temperature dependence of the dynamics of mesophilic and thermophilic dihydrofolate reductase is examined using elastic incoherent neutron scattering. It is demonstrated that the distribution of atomic displacement amplitudes can be derived from the elastic scattering data by assuming a (Weibull) functional form that resembles distributions seen in molecular dynamics simulations. The thermophilic enzyme has a significantly broader distribution than its mesophilic counterpart. Furthermore, although the rate of increase with temperature of the atomic mean-square displacements extracted from the dynamic structure factor is found to be comparable for both enzymes, the amplitudes are found to be slightly larger for the thermophilic enzyme. Therefore, these results imply that the thermophilic enzyme is the more flexible of the two.

Structural characteristics of a 0.23 mole fraction aqueous solution of tetrahydrofuran at 20 degrees C.

Hydrogen/deuterium isotopic substitution neutron diffraction techniques were used to measure the structural correlation functions in a 0.23 mole fraction solution of tetrahydrofuran in water at room temperature. Empirical potential structure refinement (EPSR) was used to build a three-dimensional model of the liquid structure that is consistent with the experimental data. Detailed analysis shows a preference for nonpolar interactions between the cyclic ether molecules plus polar interactions between the ether and solvent water and hydrophobic hydration of the nonpolar regions of the solute. The increase in the number of hydrogen-bond-acceptor sites relative to the number of hydrogen-bond-donor sites in this system, compared to the balanced situation that would be found in pure water, has a marked compressive effect on the structure of the solvent. Despite the small size of the solvent water molecules, the 0.23 mole fraction aqueous solution is still found to contain small voids akin to those in pure liquid tetrahydrofuran. In contrast to the positive surface charge of the voids in the pure system, the average void in this aqueous solution is found to have a net negative charge. This is due to contributions from the water oxygen atoms that are negatively polarized by their intramolecular bonding.

Raman spectroscopic study of hydrogen ordered ice XIII and of its reversible phase transition to disordered ice V.

Raman spectra of recovered ordered H(2)O (D(2)O) ice XIII doped with 0.01 M HCl (DCl) recorded in vacuo at 80 K are reported in the range 3600-200 cm(-1). The bands are assigned to the various types of modes on the basis of isotope ratios. On thermal cycling between 80 and 120 K, the reversible phase transition to disordered ice V is observed. The remarkable effect of HCl (DCl) on orientational ordering in ice V and its phase transition to ordered ice XIII, first reported in a powder neutron diffraction study of DCl doped D(2)O ice V (C. G. Salzmann, P. G. Radaelli, A. Hallbrucker, E. Mayer, J. L. Finney, Science, 2006, 311, 1758), is demonstrated by Raman spectroscopy and discussed. The dopants KOH and HF have only a minor effect on hydrogen ordering in ice V, as shown by the Raman spectra.

The structure of liquid tetrahydrofuran.

Hydrogen/deuterium isotopic substitution neutron diffraction techniques have been used to measure the structural correlation functions of liquid tetrahydrofuran at room temperature. Empirical potential structure refinement (EPSR) has been used to build a three-dimensional model of the liquid structure that is consistent with the experimental data. Analysis to the level of the orientational correlation functions shows that the liquid displays a preference for T-like configurations between the tetrahydrofuran molecules, a local structure that results in void-like regions of approximately 1.25 angstroms radius within the bulk liquid. The surface chemistry of these voids suggests a slightly positive electrostatic character. These findings are consistent with the known propensity of the liquid to solvate free electrons.

The preparation and structures of hydrogen ordered phases of ice.

Two hydrogen ordered phases of ice were prepared by cooling the hydrogen disordered ices V and XII under pressure. Previous attempts to unlock the geometrical frustration in hydrogen-bonded structures have focused on doping with potassium hydroxide and have had success in partially increasing the hydrogen ordering in hexagonal ice I (ice Ih). By doping ices V and XII with hydrochloric acid, we have prepared ice XIII and ice XIV, and we analyzed their structures by powder neutron diffraction. The use of hydrogen chloride to release geometrical frustration opens up the possibility of completing the phase diagram of ice.

Excess entropy in alcohol-water solutions: a simple clustering explanation.

We show that the anomalous negative excess entropy of mixing characteristic of aqueous lower alcohols containing hydrophobic groups is quantitatively consistent with a model exploiting only the experimentally observed molecular-scale segregation of the components across the entire concentration range. The simple model presented here, which uses plausible interatomic distances as its only free parameters, obviates the need to invoke "iceberg" or other water restructuring concepts which, though frequently postulated in explaining the hydrophobic interaction, are unsupported by recent experiments.

Enzyme activity and flexibility at very low hydration.

Recent measurements have demonstrated enzyme activity at hydrations as low as 3%. This raises the question of whether hydration-induced enzyme flexibility is important for activity. Here, to address this, picosecond dynamic neutron scattering experiments are performed on pig liver esterase powders at 0%, 3%, 12%, and 50% hydration by weight and at temperatures ranging from 120 to 300 K. At all temperatures and hydrations, significant quasielastic scattering intensity is found in the protein, indicating the presence of anharmonic, diffusive motion. As the hydration increases, a temperature-dependent dynamical transition appears and strengthens involving additional diffusive motion. The implication of these results is that, although the additional hydration-induced diffusive motion in the protein detected here may be related to increased activity, it is not required for the enzyme to function.

Experimental configurational landscapes in aqueous solutions.

Structures and interactions between molecules in solution are modulated by the solvent. Changes in solvent conditions can lead to structural changes and transitions such as the assembly processes seen in micelle formation and protein folding. In the case of even quite complex liquid systems, we can now explore experimentally the configurational energy landscapes that underlie these processes. Using an aqueous solution of an amphiphile as an example, the structural transitions induced by changes in temperature, concentration and added salt are examined at the molecular level, and some critical regions of the landscape identified. Moreover, the potentials of mean force that quantitatively describe the solvent-modulated interaction between molecules in solution can now be experimentally accessed.

Neutron frequency windows and the protein dynamical transition.

Proteins undergo an apparent dynamical transition on temperature variation that has been correlated with the onset of function. The transition in the mean-square displacement, , that is observed using a spectrometer or computer simulation, depends on the relationship between the timescales of the relaxation processes activated and the timescale accessible to the instrument or simulation. Models are described of two extreme situations---an "equilibrium" model, in which the long-time dynamics changes with temperature and all motions are resolved by the instrument used; and a "frequency window" model, in which there is no change in the long-time dynamics but as the temperature increases, the relaxation frequencies move into the instrumental range. Here we demonstrate that the latter, frequency-window model can describe the temperature and timescale dependences of both the intermediate neutron scattering function and derived from molecular dynamics simulations of a small protein in a cryosolution. The frequency-window model also describes the energy-resolution and temperature-dependences of obtained from experimental neutron scattering on glutamate dehydrogenase in the same solvent. Although equilibrium effects should also contribute to dynamical transitions in proteins, the present results suggests that frequency-window effects can play a role in the simulations and experiments examined. Finally, misquotations of previous findings are discussed in the context of solvent activation of protein dynamics and the possible relationship of this to activity.