Synchrotron Infrared Measurements of Dense Hydrogen to 360 GPa

Diamond-anvil-cell techniques have been developed to confine and measure hydrogen samples under static conditions to pressures above 300 GPa from 12 to 300 K using synchrotron infrared and optical absorption techniques. A decreasing absorption threshold in the visible spectrum is observed, but the material remains transparent at photon energies down to 0.1 eV at pressures to 360 GPa over a broad temperature range. The persistence of the strong infrared absorption of the vibron characteristic of phase III indicates the stability of the paired state of hydrogen. There is no evidence for the predicted metallic state over these conditions, in contrast to recent reports, but electronic properties consistent with semimetallic behavior are observed.

Figure 1

Representative synchrotron IR spectra of hydrogen in the region of the vibron (${\mathrm{H}}_2$ stretching mode) at selected pressures and 80 K. (a) Spectra from an experiment reaching 300 GPa. (b) Spectra from a higher pressure experiment. Below 150 GPa (in phase I), and at room temperature and at higher pressures (in phase I), the vibron intensity is approximately 3 orders of magnitude lower than at higher pressures and $<100\mathrm{K}$ (phase III). Thus, single-channel transmission spectra measured under these conditions can be used as a reference (or background) spectrum to obtain clear signatures of the vibron. For (a) the single-channel spectrum measured at 140 GPa and 300 K was used as the reference. For the absorption spectra shown in green in (b), a single-channel transmission spectrum measured at 142 GPa and 300 K was used as the reference (see Ref. 28). In order to ratio out the diamond phonon absorption band at these higher pressures, a series of single-channel transmission spectra measured at corresponding pressures and $\ge 250\mathrm{K}$ were also used as reference spectra to determine the absorption, as shown by the black curves in (b). The absorbance bar refers to the amplitude of these spectra; the absorbances for spectra in green are scaled by $\frac{1}{2}$ in the plot. As a cross-check, peak positions were also determined using direct deconvolution of spectra having overlapping bands; the peak positions of the vibrons determined by both approaches were found to be within $\pm 10{\mathrm{cm}}^{-1}$. Further details are provided in Ref. 28.

Figure 2

Low-temperature vibron frequencies as a function of pressure, including results from the present study as well as IR data from Refs. 18, 27, and Raman data from Refs. 18, 19. Different symbols correspond to different experiments. The red lines are our fits of these combined IR and Raman vibron shift datasets. The blue dashed line is the Raman shift reported in Ref. 20, which follows a trend not found in the other studies.

Figure 3

Absorption spectra at selected pressures and temperatures. Synchrotron IR spectra (red lines, offset for clarity) show the persistence of the vibron over the range of conditions studied. The blocked areas at low energy are regions of intrinsic and impurity absorption of the diamond anvils. The initial (160 GPa) sample thickness $d$ for the top three spectra was $3.7\mu \mathrm{m}$ [corresponding to Fig. 1a], and for the lower two spectra $d=2.6\mu \mathrm{m}$ [Fig. 1b]. The blue lines are representative optical conductivities calculated assuming simple Drude model parameters (see Ref. 8) to provide upper bounds on the possible carrier density: (A) $\hslash {\omega }_p=2\mathrm{eV}$, $\hslash /\tau =0.1\mathrm{eV}$; (B) $\hslash {\omega }_p=1\mathrm{eV}$, $\hslash /\tau =0.1\mathrm{eV}$; (C) $\hslash {\omega }_p=0.5\mathrm{eV}$, $\hslash /\tau =0.1\mathrm{eV}$. (D) $\hslash {\omega }_p=0.2\mathrm{eV}$, $\hslash /\tau =0.1\mathrm{eV}$; (E) $\hslash {\omega }_p=0.2\mathrm{eV}$, $\hslash /\tau =0.01\mathrm{eV}$ [28]. A sample thickness $d=2\mu \mathrm{m}$ was assumed. Curves D and E are consistent with the measured data. Sharp features at low energies arise from phonons and hydrogen-diamond interactions and are discussed elsewhere [8, 9, 28]. Left inset: Phase diagram showing the nominal $P\mathrm{\text{−}}T$ region explored (shaded area). The solid lines are I-II-III phase boundaries [6, 11]; the dashed line is the extrapolation of the previously determined I-III phase line [11]. The blue and red dotted lines at higher pressures are predicted melting and plasma phase transition lines, respectively (see Refs. [17, 25]). Right inset: Selected visible absorption spectra. The dashed red lines are from Ref. 18 and are rescaled to fit the plot. The absorption edge appears to have a weak temperature shift under these conditions, but only a limited $P\mathrm{\text{−}}T$ range has been explored [28].