Abstract
The development of electric potentials in ice crystals under the influence of temperature gradients is investigated both theoretically and experimentally. The maintenance of a steady temperature gradient across a piece of ice is accompanied by concentration gradients of H$^+$ and OH$^-$ ions; because of the much greater mobility of H$^+$ ions, these diffuse more rapidly into the colder part of the ice and, in the steady state, a potential difference is set up across the ice crystal, the colder end being positive. A theory of this effect predicts a surface density of charge on the ends of the ice of $\sigma$ = 5 x 10$^{-5}$ (dT/dx) e.s.u. cm$^{-2}$ and a potential difference across a uniform specimen of about 2$\Delta T$ mV, where $\Delta T$ is the temperature difference across the ends. These values are quite well confirmed by a series of experiments on specimens of highly purified ice. When two pieces of ice of initially different temperatures are brought into temporary contact and separated, the warmer acquires a negative charge and the colder an equal positive charge. The theory indicates that a maximum charge transfer of 3 x 10$^{-3}$ $\Delta T$, e.s.u. cm$^{-2}$ should occur with a contact time of about 0.01 s and that it should thereafter decline as the two pieces of ice become more nearly equal in temperature. The theoretical value for the charge developed for a contact time of $\sim$ 0.01 s is well confirmed by experiments which also show that very little charge separation occurs if the contact period exceeds $\frac{1}{2}$ s. Experiments in which the ice was contaminated with carbon dioxide, hydrofluoric acid, and sodium chloride in concentrations of up to 50 times that normally present in rain water, showed that the electrification was not greatly influenced by these impurities. These phenomena are thought to be of basic importance in the generation of electric charge in thunderstorms, this aspect being developed in the following paper.