Films of ice less than 1 nanometre thick, sandwiched between sheets of graphene, have been observed to adopt a square lattice structure quite different from the widely occurring hexagonal structure of bulk ice. See Letter p.443
Crystals of ice consist of hydrogen-bonded water molecules in a tetrahedral network that has hexagonal symmetry. It is this symmetry that gives rise to the characteristic shapes of snowflakes and the intricate patterns of ice that form on the surfaces of puddles when they freeze. At high pressures, such as 10,000 atmospheres (1 gigapascal; GPa), the underlying tetrahedral geometry of ambient ice is mostly preserved, although the detailed structure undergoes several transformations1. But on page 443 of this issue, Algara-Siller et al.2 describe a high-resolution electron microscopy study of water molecules at high pressure between sheets of the material graphene, and report that the molecules surprisingly adopt a simple, cubic-like symmetry — a bit like scaffolding on the outside of a building, or, more accurately, between two buildings.
The authors' work brings together themes of fluid behaviour near surfaces that date back to at least the 1890s, the time of Johannes van der Waals' studies in this area. From his and related work, we know that the pressure inside a liquid surface is higher than the pressure outside if the liquid surface is convex, and lower than that outside if the surface is concave3 (Fig. 1). If a small droplet of liquid is confined between two sheets, and does not wet the surfaces — that is, it does not bond readily to the surface, as is the case for mercury on glass — then the liquid–vapour interface is convex and the pressure inside the liquid must be greater than that outside.
In fact, the pressure within a sheet-confined droplet is determined by the ratio of the surface tension of the liquid to the radius of curvature of the liquid's surface3. Although the precise details of this effect at the molecular level become complicated by atomic interactions4,5, if the radius of curvature becomes very small (of the order of 1 nanometre or less, as in Algara-Siller and colleagues' work), then large pressures are required to hold the liquid in place. This is, apparently, exactly what happens when water is sandwiched between graphene sheets, because there are no points in the sheets to which water can form hydrogen bonds.
How can that pressure be maintained? To explain this, Algara-Siller et al. draw on another theme from van der Waals' theory, namely, that all atoms must be attracted to each other, irrespective of whether they can form hydrogen bonds or not. This attractive force — called the van der Waals force or London dispersion force6 — increases as atoms approach each other. The authors calculate that when sheets of atoms, such as the carbon atoms in graphene, are separated by distances of less than 1 nm (as used in the experiment), then the van der Waals forces can easily generate pressures as high as 1 GPa.
Water trapped between graphene sheets under these conditions is likely to crystallize, even at room temperature. But the fact that it forms a square structure is unexpected. The researchers' computational molecular-dynamics simulations do suggest that a square lattice can form, as observed, but the detailed origins of this strange arrangement remain a mystery.
Are there any precedents for observing square-like structures formed from water molecules? Yes, there are some. Extensive spectroscopic data and simulations of small clusters of water molecules7,8,9 provide evidence that groups of water molecules can have a near-cubic structure, with 'dangling' hydrogen bonds available in principle to form a more extensive network. And a study10 that combined molecular-dynamics simulations with neutron-scattering experiments concluded that the dynamics of water molecules trapped in carbon nanotubes could be explained if the molecules form a stationary, nearly square array wrapped in a cylinder at the inner surface of the nanotube, through which more water molecules are transported in a chain-like configuration. But Algara-Siller and colleagues are the first to directly observe an extended, two-dimensional, square-like structure in water experiments.
It remains to be seen whether the authors' observations are relevant to water transport through naturally occurring nanometre-scale channels. For example, aquaporin is a widely occurring channel protein that regulates the flow of water across the cell membrane. The flow-control mechanism entails a combination of hydrophobic and hydrophilic interactions between the water and the inside surface of the channel11, which is circular in cross-section, not planar, as in the case of the graphene 'pore'. Nonetheless, the graphene results are highly intriguing, and will probably stimulate much debate about the nature of water in biological channels and at surfaces. No doubt van der Waals would have been delighted to know that the fundamental forces that he identified so long ago have led to the discovery of an unexpected phase of water today.Footnote 1
Petrenko, V. F. & Whitworth, R. W. Physics of Ice (Oxford Univ. Press, 1999).
Algara-Siller, G. et al. Nature 519, 443–445 (2015).
Rowlinson, J. S. & Widom, B. Molecular Theory of Capillarity Int. Ser. Monogr. Chem. (Clarendon, 1982).
Lei, Y. A., Bykov, T., Yoo, S. & Zeng, X. C. J. Am. Chem. Soc. 127, 15346–15347 (2005).
Xue, Y.-Q., Yang, X.-C., Cui, Z.-X. & Lai, W.-P. J. Phys. Chem. B 115, 109–112 (2011).
London, F. Trans. Faraday Soc. 33, 8b–26 (1937).
Gruenloh, C. J. et al. Science 276, 1678–1681 (1997).
Keutsch, F. N. & Saykally, R. J. Proc. Natl Acad. Sci. USA 98, 10533–10540 (2001).
Perera, P. N. et al. Proc. Natl Acad. Sci. USA 106, 12230–12234 (2009).
Kolesnikov, A. et al. Phys. Rev. Lett. 93, 035503 (2004).
Agre, P. & Kozono, D. FEBS Lett. 555, 72–78 (2003).
About this article