Recent theoretical studies [K. Joshi, D. G. Kanhere, and S. A. Blundell, Phys. Rev. B 67, 235413 (2003)] predict that ${\mathrm{Sn}}_{20}$ melts at around $1200\mathrm{K}$. We have performed calorimetry measurements on unsupported ${\mathrm{Sn}}_{18}^{+}$, ${\mathrm{Sn}}_{19}^{+}$, ${\mathrm{Sn}}_{20}^{+}$, and ${\mathrm{Sn}}_{21}^{+}$ in an effort to test this prediction. We find that these tin clusters disappear well below their predicted melting temperature due to dissociation. Calorimetry measurements performed up to around $650\mathrm{K}$ show some small features (which may be due to localized structural changes) but no clear melting transitions. Hence, tin clusters in this size regime do not melt—they sublime.

• Received 25 September 2004

DOI:https://doi.org/10.1103/PhysRevB.71.073410