I must explain the extension to my title. I wrote a Parting Shot with the title ‘Nooks and Crannies’ for Elements, 8 (2), 2012, in an issue on ‘Minerals, Microbes and Remediation’, devoted to nooks and crannies on weathered alkali feldspar surfaces. For international readers, I explained that (according to the Concise Oxford English Dictionary) a ‘nook’ is ‘a corner or recess; a secluded place’ and a ‘cranny’ is ‘a chink, a crevice, a crack’. The reactivity of complex mineral surfaces is a complicating factor in ‘reactive transport’, particularly near Earth's surface. The present issue provides me with an opportunity to dust-off a few old micrographs, images of considerable beauty in themselves, to remind readers that minerals are not just chemical compounds. It also allows me to introduce an intriguing discovery of the last five years. Some, but not all, feldspars are extremely effective at nucleating ice in clouds, and this may be related to nooks and crannies, not just to chemistry.

In my 2012 Parting Shot, I explained how the surface complexities of alkali feldspars arise. What follows applies to the alkali feldspars, the third most abundant mineral in Earth's crust and the main repository of potassium, and specifically alkali feldspars in granite. At their growth temperatures, the alkali feldspars form a continuous solid solution, as indicated by their general formula (Na,K)AlSi3O8, but, as the crystals cool, the difference in ionic radius of the Na and K ions produces strain in the enclosing Al–Si–O framework.

Strain energy can be minimized if the alkali ions form clusters, a process usually called exsolution by mineralogists and ‘phase separation’ by physicists. The original single-phase crystal becomes a two-phase assemblage, one phase rich in Na (the albite-rich phase), the other in K (the ‘orthoclase’-rich phase). The intergrowth is called ‘perthite’. As the feldspar cools, the intergrowths coarsen but, in the absence of circulating fluids, the framework remains continuous and is said to be ‘coherent’. To maintain coherency while minimizing elastic strain energy, the Na-rich phase (which is typically 20–30 volume % of the crystal) forms planar lamellae in a plane close to (601).

Up to this point, the crystal structure, although gently bent in places, is continuous, with no imperfections that might lead to nook formation. However, as cooling continues, the exsolution texture coarsens and the phase compositions become increasingly different. Eventually, the strains become too great to sustain and periodic dislocations nucleate along the interface between the two phases. If a granite is in a damp weathering regime, like the hill-side near Shap in northern England from which Figures 1A–1D come, dissolution of the strained structure around the dislocations turns exposed surfaces into a warren of tiny tubes (Fig. 1A). They form loops extending deep into the crystal in two directions, one roughly normal to (001), the other parallel to the b-axis, forming an intersecting mesh that I illustrated in my 2012 article. It is enlargement and coalescence of these features that leads to the extreme complexity of weathered feldspar surfaces.

As weathering continues, albite dissolves away leaving a texture that, to a British eye, resembles a toast-rack (Fig. 1B). (This analogy may not work for Americans. I know they do not share the British liking for cold toast). When you collect them, crystal surfaces like that in Figure 1B look like normal crystal faces or cleavage surfaces, perhaps slightly dulled. The slices are on the order of only a thousandth of a millimetre thick, after all. But the increase in reactive surface area is enormous, and the complexity of the surfaces only increase with time (Fig. 1C). Tubes following dislocations penetrate 100s of micrometres into crystals (Fig. 1D). Freeze–thaw and local surface disturbances cause the slices to flake off, and balls of clay minerals form. Into this world of crystal-lographic cosmos turning to chaos come filaments of the biosphere (Fig. 1E), levering off slices of the toast rack. It is fascinating to realize that tiny defects, born during cooling from the granite solidus, help define mountain range longevity and control rates of soil formation.

The possibility that feldspar nooks and crannies have a role in the behaviour of clouds is very new. If you enter ‘feldspar’, ‘ice’ and ‘nucleation’ as simultaneous search terms into Web of Science before 2013 you will find only one paper, published in 1992, about the rheology of ice, averaging 4 citations per year. Everything changed when Atkinson et al. (2013) reported laboratory experiments that suggested that feldspars were particularly good at nucleating ice in clouds. By February 2019, we have 52 papers and a total of 970 citations.

In the atmosphere, water droplets in clouds can supercool to −35 °C without freezing. ‘Mixed phase’ clouds contain a mixture of super-cooled water droplets and ice. The nucleation of ice changes the radiative properties of the cloud, leads to rain, and shortens the life of the cloud. Clay minerals have long been regarded as the most effective ice-nucleating particles. Atkinson et al. (2013) used a well-established technique in which 15 mm droplets containing suspensions of mineral particles are cooled at a fixed rate and observed with an optical microscope. Droplets containing K-feldspar froze completely 15 °C above those containing montmorillonite. This cheap, common member of the Earth's dust was as effective at nucleation as expensive silver iodide, which has been used in the past for ‘seeding’ clouds to make rain.

On the Richter scale of interdisciplinarity, mineralogy and atmospheric physics are rather widely separated and, understandably, this shows in many of the subsequent papers. If a physicist buys a jar of white powder labelled ‘potassium feldspar’ he or she is likely to assume that it contains the same substance as a similarly labelled jar from another source. Elements readers know differently. Another common practice has been to use freshly crushed powders. We know that Zeus does not roam the Sahara crushing rocks with a giant pestle and mortar. As my micrographs show, most rock degradation is passive.

Of the numerous papers that have appeared since Atkinson et al. (2013), two stand out. Harrison et al. (2016) showed that not all feldspars are equally good at nucleating ice. Potassium feldspars are better than all the plagioclase feldspars they tried, with one inexplicable exception. Amelia albite from Amelia Courthouse (Virginia, USA), world famous as a ‘gem quality’ standard, comes second in the nucleation race. A second paper, Whale et al. (2017), is particularly interesting because the authors used a set of samples supplied by myself and co-workers which we had studied more than twenty years ago in the context of feldspar weathering. The worst ice nucleators were the most perfect crystals, including the sanidine from Eifel that is one of the most defect-free crystalline materials known. And the best nucleator? The perthitic microcline from Shap (UK) of my present Figures 1A–1D.

Problem solved, you may think. Potassium feldspars with very complex surfaces are very efficient at nucleating ice. You would be wrong. The nucleation experiments on all three studies quoted here were done using freshly crushed particles, without the complexities shown in my figures. And the particles in natural atmospheric dust are very small, just a few micrometres in diameter. They would be a mixture of small fragments of those surfaces, not complex individuals with nooks and crannies. The solution to the atmospheric ice nucleation problem has to involve the use of natural dust particles with surfaces that have been characterised with great care, perhaps using transmission electron microscopy.

The readers has free access to the “free” material but MSA holds the rights