Fig. 2: Higgs modes as a function of fluence.

a A_1g Pump-probe Stokes and anti-Stokes Raman spectra at 8 K base temperature and a time delay of 3 ps for selected fluences between 0 μJ cm⁻² (not pumped) and 113 μJ cm⁻² (see Fig. S6 for complete data set). Anti-Stokes intensities are scaled by a factor of 6 for visibility. Data (gray) and fits (black, see SI S.6) are shown, with dashed lines marking zero intensity. The parameterized PB response on the Stokes side is highlighted in blue. With increasing fluence, gap filling occurs, and PB amplitude decreases. Anti-Stokes spectra are shown with Stokes-side fits converted via the Bose function (black lines). Quasi-equilibrium temperatures used for the fits are listed in Table S2. At fluences larger than 50 μJ cm⁻² the in-gap electronic response is dominated by a new NEARS feature (A_1g Higgs mode), resulting in a difference signal compared to the Stokes signal (red). b B_1g Pump-probe Stokes and anti-Stokes Raman spectra at 8 K analogous to (a). c A_1g difference signal between anti-Stokes data and Stokes-fit (corresponding to red area in a). Solid black lines represent fits following eq. (1) (see text and SI S.4). d B_1g difference intensities analogous to (c). e Integrated Raman intensities of the Higgs modes (red) and of the PB response (blue) as a function of fluence. The Higgs mode intensity integrals are the integrated anti-Stokes difference spectra shown in (c) and (d). Error bars are determined based on the noise of the integrated data. Solid black lines are exponential guides to the eye. The red lines represents the equation of inversion population (see SI S.3) with a critical fluence of 21.7 ± 5.3 μJ cm⁻² for A_1g and 31.6 ± 2.3 μJ cm⁻² for B_1g symmetry. f Excitation energy of the Higgs modes vs. fluence. Error bars show the standard uncertainty of the fitted parameter \({\omega }_{H}=\sqrt{2\alpha }\) (see eq. 1). The equilibrium anti-Stokes intensity is limited at higher energies due to thermal Bose factors of the Raman intensity (gray-shaded area, see also Fig. S4) leading to asymmetric error bars.