Although surface ice does insulate a body of liquid water below from loss of heat, it also has a higher albedo than water liquid. When water ice floats, it reflects more light from the Sun, which leads to more cooling, more ice on the surface, a resultant higher albedo, and still more cooling. Floating water ice amplifies a glaciation event, leading to more glaciation. Indeed, it appears to have done just this many times on Earth, including within the past few million years. Thus, the fact that water ice floats causes water to amplify, not damp, perturbations in energy flux coming to a planet. If a useful property of a bulk solvent is that it supports a stable environment conducive to life, then the fact that water ice floats might be viewed as a disadvantage.

Another example is that a focus on terran sea-level atmospheric pressure can influence researchers’ views of the range of temperatures at which water is a liquid. On the surface of Mars liquid water cannot exist, except possibly transiently, because the pressure is too low, and water ice sublimes directly to water vapor without going through an intermediate liquid phase. Further, over most ranges of pressure, formamide has a larger liquid temperature range (255 to 480 K) than water and is also an excellent solvent for polar materials.

The view of water as an ideal biosolvent is geocentric in other ways. Although the temperature range over which water is a liquid is large on the Kelvin scale, it is important not to be tied to a linear view of the Celsius scale. The temperature range from 1 to 2 K is much more significant than from 273 to 274 K, both in fractional terms and in terms of how physical processes occur over that range.Although ammonia is liquid at lower temperatures than water, the temperature range over which ammonia is liquid for relevant planetary surface pressures is greater than for water.

Considered in a cosmological context, water is a liquid over a temperature range present in only a very small number of objects in the universe. Assuming that a fluid is necessary for life, and accepting the premise that water has properties that make it well suited to support life in the cosmos, what do water’s freezing and boiling points imply about the likelihood of its fluidity over a significant range of the space in the galaxy, or even around a star? This question has been examined recently by Ward and Brownlee,1 who concluded that complex life (i.e., animal life) is distributed sparsely, perhaps extremely sparsely,in the cosmos precisely because it requires water as a liquid, and that liquid water is sparsely distributed in the cosmos. However, Kasting has raised questions about this conclusion.2

Despite the apparent match between terran metabolism and water, water does not seem to be unquestionably preferable as a biosolvent. For example, although much of terran metabolism exploits C=O molecules, the electrophilicity of carbon doubly bonded to nitrogen (C=N) is equally satisfactory for supporting carbon-carbon bond-forming reactions. Yet, with the exception of aromatic heterocycles such as the purines, C=N is hardly used in terran biochemistry, because most other compounds containing C=N units spontaneously hydrolyze in water to generate the corresponding C=O compound and the corresponding N-containing by-product. But if a solvent other than water were the matrix for life, and if that solvent tolerated C=N units, would life not have been able to evolve in that solvent to exploit C=N units just as effectively as terran life today exploits C=O units?

The ability of water to form strong hydrogen bonds disrupts the hydrogen bonding useful for supramolecular structures. Water does not support protein folding, because it disrupts hydrogen bonds that stabilize the fold. Indeed, an examination of the chemical literature for examples of work on self-organizing molecules indicates that chemists consciously trying to achieve this outcome rarely use water, precisely because it disrupts noncovalent directional bonding such as hydrogen bonding. The situation is similar for genetic molecules. The DNA double helix is joined by hydrogen bonds that in water are only barely stable because water offers opportunities for a single strand of DNA to form hydrogen bonds, not with its complementary strand, but with water itself. But DNA, because it has a repeating charge in its backbone, is not expected to work well in nonwater solvents. This implies that a genetic molecule in nonwater solvents would not be DNA, or not be a polyelectrolyte. Conversely, if one accepts the notion of repeating charges as universal in genetic molecules, then one might have to conclude that a somewhat polar solvent (like water) is necessary for life.

Nevertheless, certain features of terran metabolism might benefit from a biosolvent whose properties differ from water’s. For example, the instability of C=N in water constrains the structure of metabolites in water. The compound HN=C=NH, an analog of O=C=O (carbon dioxide), immediately hydrolyzes in water to give urea (H2NCO-NH2), whose thermodynamic instability with respect to hydrolysis in water yields carbon dioxide and ammonia. Thus, water as a solvent requires that the dominant form of carbon at the +4 oxidation state be carbon dioxide.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement