Surface tension is governing a lot of phenomena in nature, e.g., formation of water droplet, insects floating on a water surface. In these cases, surface tension results from the large attraction of liquid molecules to each other. For solid state materials, however, we normally could not see these effects. Although atoms in solid materials also have larger attraction to each other, a shape change of the solid materials toward the minimization of surface energy is usually un-seeable. However, in metallic bonded materials and in some extreme cases, this liquid-like behavior will be seen and is very interesting. More interesting, these cases are all related to electron devices.

The first case is in high temperature (but below the melting temperature of the solid material). Prior to solid state electron device, an issue for the vacuum tube electronic devices is their lifetime, approximately hundreds of hours but with large variation. The sub-mm tungsten filamentary in vacuum tube works in high temperature. Liquid-like behavior were observed, e.g., blunting of the filamentary tip, ovulation of spheres from a longer filamentary, all happening below the melting temperature of tungsten. Pioneers of theoretical study in electron device (not only in solid state device) attributed this liquid-like behavior to surface tension. Unlike in the liquid cases, the surface tension driven shape change is through the flux of surface atom migration (i.e., surface diffusion) other than volume flux, since in the solid case, the bulk atoms are hard to move. However, these theory works are soon forgotten with the dim of vacuum tube.

The second case is in nanoscale and in recent. As the semiconductor industry challenging the physical scaling limitation, the size of the electron device structure is approaching the sub-10 nm or even atomic level. This results a large surface-to-ratio structure, where the surface could play a much more important role. A lot of issues are being attributed to surface tension or surface energy, e.g., high surface energy of noble metals favors the formation of isolated islands rather than continuous ultrathin films hindering the fabrication of ultrathin interconnection layer, ultrathin filament or nanowire tends to clusters together hindering the stability of a designed structure. On the other side, the liquid-like behavior of nanoscale might be beneficial, providing the possibility to obtain a self-recovery structure. Unfortunately, no quantitative studies have been made on this surface tension or surface energy effect in nanoscale structure. Researchers are using concepts like Gibbs-Thomson effect, Ostwald ripening, Rayleigh instability, etc. (theories borrowed from the liquid case), to superficially explain the above phenomena in nanoscale structures.

In metallic filamentary resistive switching device, a silver or copper filament with relatively large surface-to-volume ratio can be formed, and its size can be controlled by the compliance current. Focus on the lifetime or stability of the silver or copper filament, in this Nature Communication paper, we confirmed the surface diffusion mechanism for the nanoscale filament shape evolution in room temperature. A quantitative size-dependent filament lifetime is obtained, which accounts for the volatile and non-volatile behavior of the resistive switching device. The driving force is surface tension, furthermore, we also connected it with the Gibbs-Thomson effect and Ostwald ripening, for the spontaneous shape evolution. We would rather viewing it as a case study for a broad range of emerging nanscale solid state electron device, in which large surface-to-volume ratio structure are being fabricated or being electrically formed after the fabrication, and in which atom interactions and their transport (vs. electron transport) should be intensively considered for a comprehensive device physics understanding. Of course, it is also a tentative study, more factors should be considered for a complete theory and more experimental evidences should be provided to consolidate it.