Materials science. Now you see them.

PERSPECTIVES C rystallization lies at the heart of many natural and technological processes, from the production of pharmaceuticals and nanomaterials to the formation of bones and teeth, frost heave, and scale deposi-tion. Crucial features of these crystals, such as lattice orientation, particle size, and size distribution , are defined by conditions during the earliest stages of precipita-tion—at nucleation. Yet, nucle-ation from solution is poorly understood, because experimental studies of nucleation are highly challenging (1). Recent studies have highlighted the possible role of clusters in nucleus formation (2, 3). On page 1819 of this issue, Gebauer et al. provide support for this thesis (4) by demonstrating the presence of large, well-defined clusters before nucleation of one of the phases of calcium carbonate. Crystallization appears to proceed through aggregation of these clusters. The results challenge the conventional picture of crystal nucleation. Classical nucleation theory provides a simple understanding of how crystals nucleate. Nucleation is often slow because of a free-energy barrier originating from the interface between the nucleus and its surroundings. The theory assumes that nuclei grow one molecule at a time (see the figure, top). As the nuclei grow, their Gibbs free energy increases, until a free-energy maximum is reached at the critical size. At least in simple systems such as argon, critical nuclei are expected to persist for microseconds or less, making them virtually impossible to observe. Beyond the critical size, the nuclei are stable and release energy during growth. In their investigation of calcium carbonate nucleation, Gebauer et al. observe long-lived precritical clusters, about 2 nm in diameter, and suggest that they grow by colliding and coalescing. These results are clearly in contrast to the picture of nucleation presented by classical nucleation theory. The theory assumes that the structure of the nucleus is like a piece of the bulk phase and that its surface has the same interfacial tension as a bulk phase. However, if stable, precritical clusters are to exist, they must lie in a free-energy minimum. Such a minimum would only occur if the classical theory's assumptions are wrong, perhaps because the structure of the clusters is different from that of the bulk. The possible structure of the precritical calcium carbonate clusters is open to speculation. If the ions are present in a bulk crystal lattice, then it is surprising that the clusters of about 70 ions neither shrink nor grow. Alternatively, the structures may be ordered …