Why does acoustic cavitation in water fail far below quartz-inclusion strength?

Dynamic methods (focused ultrasound, shock reflection, centrifugation) give reproducible room-temperature cavitation pressures $P_{\mathrm{cav}} \sim -2.6\mathrm{e}8 \text{ to } -3.3\mathrm{e}8\,\mathrm{dyn}/\mathrm{cm}^{2}$ (about $-0.3\,\mathrm{kbar}$), while isochoric cooling of micrometer quartz inclusions reaches about $-1.4\mathrm{e}9\,\mathrm{dyn}/\mathrm{cm}^{2}$ ($-1.4\,\mathrm{kbar}$), close to classical nucleation theory with the bulk surface tension. Identify the mechanism that sets the acoustic value (an unidentified impurity or heterogeneous nucleation, a curvature-dependent surface tension, an equation-of-state anomaly tied to a liquid-liquid critical point, or a flaw in the inclusion experiments) and show that it reproduces the measured $P_{\mathrm{cav}}(T)$ of each method over the temperature range in which that method has data.

An acoustic or inclusion experiment on water of controlled purity that moves $P_{cav}$ continuously between the two values by varying a single identified parameter, or a simulation that reproduces both values from one model.