In disordered metals the ideal assumption that independent sources of scattering add up without interfering, known as Matthiessen’s rule, is violated. Even more puzzling is the widespread experimental observation that upon introducing more and more randomness, the temperature coefficient of the resistivity (TCR) decreases and can even change sign, while still supporting metallic conduction. As first noticed by Mooij , this negative evolution of the TCR linearly correlates with a positive increase of the residual zero-temperature resistivity ρ0, a phenomenon that has eluded a proper microscopic explanation up to now . Here we show that the emergence of Mooij correlations and the associated high-temperature breakdown of Matthiessen’s rule originate from the same processes at the microscopic level, as they can be consistently understood from the constructive interplay between disorder and local lattice deformations . By combining extensive numerical evidence together with an analysis of the dominant diagrams in the perturbation series we demonstrate that disorder scattering is efficiently screened by thermal lattice fluctuations, which can overcome the thermal increase in resistivity from the phonons themselves, giving rise to the observed negative TCRs. This identifies a broad non-Fermi liquid region located between the good metal in clean systems and the Anderson transition at strong disorder.
 J.H.Mooij phys.stat. sol. 17, 521 (1973)
 C. C. Tsuei Phys. Rew. Lett. 57, 1943 (1986)
 Domenico Di Sante, Simone Fratini, Vladimir Dobrosavljević, and Sergio Ciuchi Phys. Rev. Lett 118 036602 (2017)