Why thermal Casimir data favor the dissipationless plasma model

Lifshitz theory for Au or Ni sphere-plate geometries at $T = 300\,\mathrm{K}$ and separations $2 \times 10^{-5}\text{ to }10^{-4}\,\mathrm{cm}$: the Drude permittivity $\epsilon(\omega) = 1 - \omega_p^2/(\omega(\omega + i \gamma))$ removes the zero-frequency transverse-electric Matsubara term, while the plasma model ($\gamma = 0$) keeps it. Micromechanical, dynamic AFM and isoelectronic differential measurements agree with the plasma model and exclude Drude, whereas a torsion-pendulum measurement at larger gaps (Sushkov et al., 2011) favored Drude after subtracting electrostatic patch forces. An answer is a physical mechanism (spatial dispersion, surface states, low-frequency response) that reproduces the data while staying consistent with measured optical and dc conductivity, confirmed at separations above $3 \times 10^{-4}\,\mathrm{cm}$ where the thermal term is large.

A measurement of the thermal Casimir force at gaps above $3 \times 10^{-4}$ cm with patch potentials characterized independently, plus a dielectric model that fits it and the metal's measured transport and optics.