Measured second and fourth azimuthal anisotropy coefficients $v_{2,4}(N_{\mathrm{part}},p_T)$ are scaled with the initial eccentricity ${\varepsilon }_{2,4}(N_{\mathrm{part}})$ of the collision zone and studied as a function of the number of participants $N_{\mathrm{part}}$ and the transverse momenta $p_T$. Scaling violations are observed for $p_T\lesssim 3$ GeV/c, consistent with a $p_T^2$ dependence of viscous corrections and a linear increase of the relaxation time with $p_T$. These empirical viscous corrections to flow and the thermal distribution function at freeze-out constrain estimates of the specific viscosity and the freeze-out temperature for two different models for the initial collision geometry. The apparent viscous corrections exhibit a sharp maximum for $p_T\gtrsim 3$ GeV/c, suggesting a breakdown of the hydrodynamic ansatz and the onset of a change from flow-driven to suppression-driven anisotropy.

• Received 26 May 2010

DOI:https://doi.org/10.1103/PhysRevC.82.034910