Vesicle formation in supersaturated hydrous silicate melts is a critical process driving explosive volcanism. The number of H₂O vesicles per unit volume of melt (VND) controls the efficiency of fluid-melt separation during decompression, yet the phase separation mechanisms governing vesicle formation remain unclear. Two theories are proposed to explain this process: (1) nucleation theory and (2) spinodal decomposition.

Nucleation theory applies to metastable melts in the region between the binodal curve, where the first derivative of the free energy curve equals zero, and the spinodal curve, which is defined by its inflection points. Vesicle formation in this regime requires significant work to overcome the activation energy barrier. This energy is proportional to the cube of macroscopic surface tension and depends strongly on supersaturation. During decompression, vesicle nucleation and H2O diffusion reduces supersaturation below the threshold for further vesicle formation, implying a strong dependence of VND on decompression rate as observed in rhyolitic melts.

In contrast, spinodal decomposition occurs spontaneously at the thermodynamic stability limit, where the second derivative of the free energy curve is negative. Here, small concentration fluctuations grow continuously over large distances without sharp phase boundaries, progressing via uphill diffusion until distinct boundaries develop. This process is thought to produce a VND independent of decompression rate as observed in phonolitic and phono-tephritic melts.

Further experiments are needed to clarify phase separation in hydrous silicate melts. Developing a predictive theory is hindered by the lack of an equation of state and mixing model for the silicate melt–H₂O system. The complex structural properties of silicate melts, highly dependent on H₂O concentration, present a significant challenge for future research.