Structure and Metallicity of Phase V of Hydrogen

A new phase V of hydrogen was recently claimed in experiments above 325 GPa and 300 K. Because of the extremely small sample size at such record pressures the measurements were limited to Raman spectroscopy. The experimental data on increase of pressure show decreasing Raman activity and darkening of the sample, which suggests band gap closure and impending molecular dissociation, but no definite conclusions could be reached. Furthermore, the available data are insufficient to determine the structure of phase V, which remains unknown. Introducing saddle-point ab initio random structure searching, we find several new structural candidates of hydrogen which could describe the observed properties of phase V. We investigate hydrogen metallization in the proposed candidate structures, and demonstrate that smaller band gaps are associated with longer bond lengths. We conclude that phase V is a stepping stone towards metallization.

Figure 1

Relative enthalpies using the (a) BLYP and (b) PBE DFT functionals, and using (c) DMC. The DFT results are at the static lattice level, at $T=0\mathrm{K}$ (including zero-point motion), and at $T=300\mathrm{K}$, and the dashed lines in the static lattice diagrams indicate enthalpies corresponding to structures at saddle points of the energy landscape. The DMC results are at the static lattice level, and the dashed lines between the DMC points are a guide to the eye only.

Figure 2

(a) Raman spectra of $Pc$, $Pca{2}_1$, and phase V at 374 GPa. The absence of data in the range $1500-1900{\mathrm{cm}}^{-1}$ arises from the strong signal from the diamonds at these frequencies. (b) Pressure dependence of the frequencies of the most intense Raman peaks of $Pc$, $Pca{2}_1$, and phase V.

Figure 3

(a) Electronic densities of states of hydrogen candidate mixed layered structures $Pc$, $Pna{2}_1$, $Pca{2}_1$, and $Ibam$ at a pressure of 350 GPa. (b) Static bond lengths of the $Pc$, $Pna{2}_1$, $Pca{2}_1$, and $Ibam$ structures for the layers with longer bonds. The bond lengths of $Pc$ and $Pna{2}_1$ are indistinguishable.