Is ambient water inhomogeneous on the nanometer-length scale?

According to a recent article (1), the structure of water should be regarded as inhomogeneous on the 1-nm-length scale. This claim is made on the basis of small-angle x-ray scattering data and x-ray absorption/emission data. For a single component fluid, the Q = 0 value of the structure factor, S(Q), is given by

where N is the number of molecules in any given volume of the fluid and the angle brackets indicate ensemble averages. Diverging density fluctuations near a critical point cause S(Q) to rise substantially at low Q. Does the small low-Q rise in S(Q) observed in noncritical ambient water justify the claim made in (1) that water is heterogeneous on some length scale? We think not.

To understand why, it was shown some time ago that, even in a noncritical fluid, S(Q) can rise at small Q depending on the density and the interplay between attractive and repulsive forces (2), and that the corresponding short-range density fluctuations remain unimodal in character (3). Furthermore, general considerations from small-angle scattering theory for a two-component system (4) suggest the so-called anomalous Q = 0 rise has an amplitude given by

where φ is the volume fraction of one component, Δρ is the density difference between the two components, and v_c is a “correlation volume.” Although we would disagree that either S_A(0) or v_c can be determined with any degree of reliability for a fluid far away from its critical point, even if these values were known, this would still leave the assignment of φ and Δρ from Eq. 2 indeterminate. Moreover, it is not difficult to show that the density differences allowed by Eq. 2, for a given v_c and using the data of figure 3 in ref. 1, have similar magnitude to the stochastic fluctuations in density which arise from applying Eq. 1 to the same volume. Hence, it is not possible to conclude from the rise of S(Q) at small Q that ambient water density fluctuations are qualitatively different from the dynamic number fluctuations which characterize all liquids.

Spectroscopic (Raman, IR, XAS, XES, etc.) data from water also cannot be used to imply two-state behavior. Intensity shifts from one part of a response function to another can wholly be explained as a consequence of a continuous distribution of local environments (5). Further, the relationship between XAS and XES data and the local geometry of water molecules assumed in (1) is strongly contested by many authorities.

With these observations, the currently accepted picture of water, in which each molecule is typically hydrogen-bonded to four others in a roughly tetrahedral arrangement, and with rarely more than one additional nonbonded molecule in the first coordination shell, is retained.