We turn next to the very close crystallographic relationships in structure between silica and
water as noted by Bernal and Fowler as early as 1933 [23]. They already assumed the
existence of three “nano-regions” with structures analogous to SiO2- quartz and SiO2-
tridymite. Weyl and Marboe and many others have developed these structural affinities
between solutions in H2O and SiO2 [24]. (See Eitel for a general discussion [25].) Unknown
to most readers concerned with biological effects, ordinary water forms (noncrystalline) glass
fairly easily, e.g. by emulsions being poured into liquid N2. Unfortunately, many recent
papers on H2O glass appear to have missed the enormous literature on SiO2 which in
crystalline and glassy forms is so similar to water. In spite of the debates recorded in Mazurin
and Porai-koshits, for pure silica glass, a tetrahedrally coordinated, quenched liquid, with
structures like water (but much more viscous), is implicit (see e.g. the work by Patel et al.,
Konnert and Karle, and Roy: i.e. that SiO2 glass also consists of regions with different
packings or structural units [11, 26, 27, 28]).

By applying pressures of ≈200 kbar to SiO2-glass at room temperature, Bridgman and Simon
first established that SiO2 glass could easily be prepared and retained under laboratory p and t
conditions in two very different structures [29]. Cohen and Roy in a series of papers then
definitively established this phenomenon of unambiguous structural change with pressure, as
a general property of virtually all strongly bonded glasses [30—32]. Thirty years later,
apparently unaware of the early work, confirming the parallel between SiO2 and H2O, Angell
et al. and Kieffer via their data for glassy water: the latter saying that the evidence “provides
strong support for the concept of polyamorphism, i.e. different non-crystalline structures in
structures of glassy water” [3, 33].

While this “nano-scale heterogeneous” perspective on water, and the possible “phase
behaviour of metastable water” itself have recently also started to appear in the literature, the
possible extension to “stable water” via this early very rich and relevant background given
here is hardly known and never referenced [34, 35]. Recently Soper, Tulk et al., and
DeBenedetti and Stanley explicitly accepted nanoheterogeneity in glassy
While this “nano-scale heterogeneous” perspective on water, and the possible “phase
behaviour of metastable water” itself have recently also started to appear in the literature, the
possible extension to “stable water” via this early very rich and relevant background given
here is hardly known and never referenced [34, 35]. Recently Soper, Tulk et al., and
DeBenedetti and Stanley explicitly accepted nanoheterogeneity in glassy H2O [4, 36, 37].
They also infer that there are discontinuous steps and first order transitions among “distinct
metastable forms” in the changes from one to the other, in H2O glass. The paper by
Kawamoto et al. (see Fig. 6) shows the existence of (so far only) two “polymorphs” of stable
liquid water in a P-T diagram exactly parallel to those for S, etc., discussed above [22]. This
occurs not in glassy or metastable water, but in liquid-stable water. Thus they take this line of
argument (via exactly analogous P-T equilibria studies,) to the same conclusion we have
derived from the data cited above on the P-T diagrams for S, Se, Te, etc.: that the presence of
different crystalline structures are excellent hints for potential differences in liquid structures. 

Fig. 6 The paper by Kawamoto et al. shows a projection of at least two water structures into the stable liquid
water region exactly analogous to Fig. 4’s experimental data on several liquid structures in liquid sulfur some
35 years earlier [22].

The significance of these data on the thermodynamics of liquid water, following the earlier
studies of S, Se, Te, etc., can now be summarized, although they may not be obvious to those
unfamiliar with this branch of thermodynamics. It has been an established part of
conventional thermodynamics (as see in any textbook on phase diagrams) that the gas and one
liquid stable regions of a fixed composition can only have one phase, in contrast to solids
where one can, and often does, find even a dozen phases. There are no phase transitions of
liquid A liquid B at a fixed composition. Hence these data—the extensive earlier work
and now the paper on water—require a major re-thinking on the structure(s) of water.
These data also provide some important indications on the kinetics of change of such
structures. The conventional wisdom typically uses the argument that if new clusters (or
nano-“structures”) form they must be very transient because “the lifetime of a bond can be
estimated by the two relations:


where O is the vibration period (≈h/kT • 1.6 x 10-13 s at room temperature). Inserting data for
the bond energy of typical hydrogen bonds one gets a lifetime of an average single hydrogen
bond of about a microsecond. (For a strong covalent bond it jumps to 106+ years.) However,
this is not what is at issue. Consider some of the larger oligomers shown in Martin Chaplin’s
website reference (Figs. 8 & 9) containing say 200 H2O molecules [38]. These presumably do
not completely break up and reform via some cooperative bond breakage every microsecond.
The Kawamoto et al. phase diagram (Fig. 6) proves that at least for the duration of the
experiment (minutes-hours) under the P-T conditions specified there are structurally distinct
phases, with characteristic properties, which make the phase boundary detection possible [22].
Likewise, the analogy of H2O to the other liquids described is not that their strongest covalent
bonds are identical but that the bonds holding such clusters together are likely to be more
similar because they enable one to study closely analogous phase changes in the same P-T
range, with temperature as the principal bond-breaking vector.

The intuitively reasonable concept of continuing the structures of the crystalline phases into
the liquid phase was the basis of Bernal’s connecting H2O and SiO2 [23]. Konnert and Karle
identified explicitly the tridymite structure of SiO2 as being present in SiO2 glass [27].
Robinson’s two state model for water is based on dense and less dense ice, and recently
Beneditti and Stanley suggested that fragments of two different crystalline ice structures must
persist into the liquid water region [4, 39]. The point being made here is that the obviously
relevant kinetics are those of the persistence of structural elements (crystalline form
determined clusters, non-heterogeneous regions, etc.) under near ambient conditions. It is
absolutely certain that at least some of these are reasonably long lived, since they give us the
distinctive properties.

One can therefore summarize that the actual experimental data, ranging over 50 years, on the
structure of many glasses and liquids shows the following:

a. The ubiquity of nanoscale heterogeneity in the structure of many covalently
bonded liquids
b. That such heterogeneity on the nanometer scale is the rule rather than the
exception for the structure of all strongly bonded liquids (i.e.
principally excepting ionic and metallic melts).

Roy summarized the case for this “nano-heterogeneity” as the most generalized model for
glasses in a review paper on the structure of glasses and their nucleation and crystallization
[15]. Figure 7a, taken from his paper (confirmed by the later data such as those of Mazurin
and Porai-koshits), presents a very crude schematic visual image which should replace Fig. 1
in our memories, as a closer approximation to reality for the “structure of (most, covalently
bonded) liquids” [11, 15]. 

Fig. 7.a. The cartoon version of the more generalized structure of glass clearly indicating its heterogeneous
(with respect to structure or structure and/or composition) nature from Roy [15].Note that water is mentioned in
the third column. This is the new minimalist schematic representation of the structure of water.

Strikingly similar is the cartoon image (Fig. 7b) from the major text on Electrochemistry by
Bockris and Reddy [40]. 

Fig. 7.b. A similar representation of the water structure by Bockris and Reddy [40].

Such liquids as the ones we are dealing with, similar to H2O, consist of statistically distributed
molecular aggregates of different sizes, structures and (where relevant) compositions.
Furthermore, it is also thoroughly established, that major changes of such structures (i.e. the
3-D arrangement of such aggregates or clusters in space) readily occur as a function of
temperature and pressure for all common glasses (even of monotonic glasses). Many of these
fine-structure changes in such glasses remain stable (i.e. exhibit a kind of memory) for years.
Following the discovery by Bridgman and Simon with extensive work by Roy and Cohen and
Cohen and Roy showed that the density and refractive index of SiO2 glass and indeed glasses
of all compositions examined and, hence their structures, were a continuous function of
pressure, and that these high-density solid forms could be recovered and retained metastably
under room temperature ambients for years [29—32].