Fig. 4: reconstruction of water Brillouin dependency with temperature.

The same measurement has been used as a benchmark for evaluating the reliability of standard BMs from all over the world²⁰. Importantly, this acquisition was completely automatic and lasted ≈12 h, during which the spectrometer was never realigned. After the temperature setpoint was reached, the system automatically performed the EOM-control loop and calibration and finally acquired 500 Brillouin spectra, then waited for the next setpoint in a recursive manner. A example of 2 Brillouin raw spectra (dots) extracted from the data at 27 °C (blue data) and 42 °C (red data) together with corresponding fits (continuous lines). The EOM frequency was set to ν_EOM = 13 GHz. B Water Brillouin shift (blue) and deconvolved FWHM (orange) dependency with temperature. C EOM Anti-Stokes peak position (blue) and FSR (orange) in function of temperature, both remaining stable throughout 12 h of experiment. D water velocity of sound (V_S) extrapolated from Brillouin shifts (blue datapoints and curve); in red, theoretical model³⁸; black squares show acoustic spectroscopy (AS) data^20,46. Our data match the model curve and AS data within 99% confidence intervals (shaded blue areas). E Water longitudinal kinematic viscosity (ν_L) extrapolated from deconvolved Brillouin FWHMs (blue datapoints and curve). We compared our results with already published data from ultrasounds (US)^47,48 (purple triangle) and Inelastic X-ray Scattering (IXS)^47,49 (red star). We also show the curve of the approximation ν_L ~ 4 ν_S (where ν_s are water kinematic shear viscosity data^47,50). Our data match both US and IXS data within 99% confidence intervals (shaded blue areas) and are slightly lower than ν_L ~ 4 ν_S approximation curve. Data of panels B-E are shown as mean ± SD performed over 500 repeated measurements of a single acquisition; shaded areas refer to 99% confidence intervals of the fits. Source data are provided as Source Data file.