Experimental Evidence for the Breakdown of Uniform-Electron-Gas Models in Warm Dense Aluminium

TL;DR

The study addresses the reliability of uniform-electron-gas models in warm dense matter by performing momentum-resolved X-ray Thomson scattering on shock-compressed aluminium at ~50 GPa and comparing the results to both UEG-based and ab initio simulations. Using angle-resolved XRTS, XRD, and hydrodynamic constraints, the authors test how well RPA, static local-field corrections, and finite-temperature TDDFT reproduce the plasmon dispersion and line shapes across a broad range of k within a density window ρ ≈ 3.75–4.5 g cm⁻³ and electron temperature Te ≈ 0.6 eV. The key finding is that UEG-based models systematically overestimate the plasmon energy by up to ~8 eV at the largest k and fail to capture the high-energy tail, whereas TDDFT with MD-derived ionic snapshots accurately matches both dispersion and damping within experimental uncertainty. This provides direct, momentum-resolved evidence that ab initio approaches are essential for reliable inference of thermodynamic conditions in warm dense aluminium and establishes a benchmark for validating dynamic exchange–correlation kernels in WDM.

Abstract

The robust diagnosis of thermodynamic conditions in warm dense matter experiments remains a central challenge. We report angle-resolved femtosecond x-ray Thomson scattering measurements of shock-compressed aluminum at approximately 50 GPa over a wide range of scattering wavevectors at the European XFEL. The measured plasmon dispersion and line shape demonstrate that the standard analysis of x-ray Thomson scattering spectra based on uniform electron gas models systematically overestimates the plasmon resonance energy by up to 8 eV. In contrast, an analysis based on ab initio calculations reproduces both the dispersion and spectral shape within experimental uncertainty. Our results show that shock-induced disorder plays a critical role in the interpretation of x-ray Thomson scattering from compressed solids and provide direct evidence that ab initio approaches are required for reliable inference of thermodynamic conditions in warm dense aluminum.

Figures (11)

• Figure 1: Experimental geometry and diagnostics. The seeded XFEL probe (purple, $E_X\simeq8.3$ keV, $\Delta E\approx1$ eV) is focused onto the aluminium target by upstream Be compound refractive lenses (CRLs). The target consists of a $50~µm$ Al foil bonded to a $25~µm$ Kapton ablator and is shock-compressed by a nanosecond DiPOLE drive laser (green) incident. Both beams are incident at $22.5^\circ$ relative to the target normal in the horizontal plane, and an angle between the beams of $45^\circ$. Inelastically scattered x-rays are energy-dispersed by a cylindrically bent HAPG(004) crystal in von Hámos geometry and recorded on a Jungfrau detector, defining the scattering angle $\Theta$ for XRTS in the vertical plane. A second Jungfrau detector collects simultaneous x-ray diffraction (XRD), while velocity interferometer system for any reflector (VISAR, red) provides independent shock-timing and velocity diagnostics.
• Figure 2: Momentum-resolved XRTS and concurrent structural diagnostics in shock-compressed aluminium.(a) Representative ambient (blue) and driven (red) spectra at four momentum transfers, normalised to the elastic line and vertically offset for clarity. The thin black curve shows the measured SIF at $k=0.99~\per\angstrom$. The driven plasmon shifts to higher energy loss and broadens with increasing $k$, becoming asymmetric near $k=2.57~\per\angstrom$. (b) Ionic static structure factor $S_{\mathrm{ii}}(k)$ extracted from XRD (orange line; band: $1\sigma$ shot-to-shot variation) compared with DFT-MD predictions at selected densities (coloured lines) and the density-averaged DFT-MD curve (black dashed) corresponding to the experimentally constrained density window. Vertical dashed lines indicate the experimental XRD $k$-range. The agreement constrains the density window used for theory comparisons.
• Figure 3: Low-$k$ XRTS spectra and model comparison.(a)$k=0.99~\per\angstrom$; (b)$k=1.28~\per\angstrom$. Inelastic x-ray scattering spectra from laser-compressed aluminium at two representative momentum transfers, normalised to the elastic peak. Experimental data (red lines, shaded band: $1\sigma$ uncertainty) are compared with theoretical predictions from TDDFT (blue), RPA (orange), and static LFCs (light blue). A Voigt-profile plasmon (black dashed) is shown for reference and is used to extract the experimental plasmon peaks. All theoretical spectra and the Voigt model are convolved with the experimentally determined SIF, modelled as the sum of two Voigt profiles (seed spike plus weak pedestal; Supplemental Material, Sec. \ref{['sec:S2']}). Only two $k$ values are shown here for clarity; spectra for all four measured $k$ are provided in Supplemental Fig. \ref{['fig:fullset']}.
• Figure 4: Plasmon dispersion in warm dense aluminium and comparison to theory. Extracted plasmon peak positions $E_{\mathrm{peak}}(k)$ (red markers) compared with theoretical predictions from TDDFT (blue line), static LFCs (orange), and RPA (green). For reference, a representative data point from Preston et al.preston2019momentum (black marker) illustrates the much larger experimental uncertainty typical of previous measurements at similar conditions. Error bars include both fitting uncertainties and shot-to-shot experimental variation. Lines connecting the data points are guides to the eye only. TDDFT tracks the experimental dispersion within error bars at all measured $k$, whereas UEG RPA and static LFCs systematically overshoot the plasmon energy.
• Figure S1: Hydrodynamic profiles at the XFEL probe time. Density (solid blue, left axis) and temperature (dashed red, right axis) from HELIOS-CR at $t_\mathrm{pp}=6.3$ ns. The simulation predicts a $\sim30~\mu$m-wide plateau of compressed aluminium with a density of $3.75~\mathrm{g\,cm^{-3}}$ to $4.5~\mathrm{g\,cm^{-3}}$ and nearly constant $T\approx0.6$ eV. The density gradient ($\sim20\%$ across the plateau) is explicitly included in the density-averaged theoretical spectra.