Figure 2

(a) The experimental compressibility factor Z (rhombi), obtained under the assumption of hard rotationally-isotropic interaction potentials, is completely missed by the Volmer isotherm (solid red line). The agreement is poor for even an unphysical A₀ value (blue dashes). A much better match, with no free parameters, is achieved for virial expansions (dash-dotted curve) and HDs simulations (circles). Note, since the deduction of Z from our experiments assumes hard interactions, we do not compare these Z to WCA or LJ fluid simulations, with the later⁴⁸ strongly deviating from the rhombi. As the experimental error bars on η are of about the symbol size, the η of the two rightmost rhombi coincide within error. (b) The u(r) used in WCA and LJ simulations are displayed on top of the u(r) obtained by inversion of the experimental g(r), for η = 0.19; note the presence of an attraction well in the last (olive dash-dot-dot curve). Here we use for WCA and LJ potentials, as in the corresponding g(r) fits, where σ is the experimental particle size measured by light scattering, as in previous work⁴. For comparison, no attractions are observed when the inversion algorithm is applied to EDBD-simulated g(r) of ideal HDs, at η = 0.44 (short blue dashes). Inset: The g(r) of a simulated zero-dipole purely-repulsive WCA fluid (dashes; ρ = 0.31) is compared to that with point dipoles (solid curve; the dipole moment is |p| = 1.5 in the LJ units³⁷ and ρ = 0.2). Note the much higher first peak in the dipolar fluid, indicative of an enhanced pair formation. The higher-order peaks have roughly the same amplitude. A similar increase in the first peak of the g(r), with unchanged higher-order peaks, was observed for the experimental fluids, when compared to the purely repulsive (HD or WCA) simulations. Thus, while a more advanced model, beyond the point dipole approximation, is necessary to fully reproduce the experimental g(r), our simple simulations are successfully imitating the qualitative trends observed in the experiment.