Figure 2. Revealing in-depth spatial inhomogeneity of H₂O ice X at megabar pressures.

(a): Typical time-resolved reflectivity signal in ice compressed in a DAC to 84 GPa. The vertical arrow marks the time of transmission of the laser-generated acoustic pulse across the interface of ice with diamond. Insert: zoom of the signal in the vicinity of the ice/diamond interface. (b) TDBS signal obtained by subtracting in signal (a) the time-varying thermo-reflectance contribution caused by transient heating of the sample. Insert: The Fourier spectra of the signal inside two temporal windows, marked by rectangles, demonstrate the shift of the Brillouin frequency from 60 GHz to approximately 73 GHz, indicating spatial inhomogeneity of the ice sample. For each time window, the obtained frequency value is an average over multiple crystallites in the volume from which the signal is collected. The lateral size of the probed volume is determined by the elliptical intensity correlation function with the axes of 3 μm and 4.5 μm FWHM for the pump and probe beams (see Methods). The cross-sectional area of the tested volume can be estimated as 10 μm². The axial, i.e., in-depth, dimensions of the probed sample volumes are determined by the duration of the time windows and exceed, in this particular case, 3 μm. (c): Temporal profile of the Brillouin frequency of the TDBS signal shown in (b). The Fourier transform was performed in the rectangular moving temporal window of 100 ps duration, corresponding to the in-depth distance of approximately 1.5 μm. (d): Spatial variation of the longitudinal sound velocity obtained from the temporal dependence of the Brillouin frequency shown in (c) using the extrapolated refractive index values of ice³⁹. The coordinate in (d) is evaluated by integrating the sound velocity over acoustic propagation time only in a part of the experimental window, which does not include short delay times (shadowed in Fig. 2), where the determination of the Brillouin frequency is not sufficiently exact because of imperfect filtering of the thermo-reflectance contribution at GHz frequencies. Thus, the distance in (d) should be measured relative to the ice/diamond interface.