ANDERSON-MOTT TRANSITION AS A QUANTUM-GLASS PROBLEM

We combine a recent mapping of the Anderson-Mott metal-insulator transition on a random-field problem with scaling concepts for random-field magnets to argue that disordered electrons near an Anderson-Mott transition show glasslike behavior. We first discuss attempts to interpret experimental results in terms of a conventional scaling picture, and argue that some of the difficulties encountered point towards a glassy nature of the electrons. We then develop a general scaling theory for a quantum glass, and discuss critical properties of both thermodynamic and transport variables in terms of it. Our most important conclusions are that for a correct interpretation of experiments one must distinguish between self-averaging and non-self-averaging observables, and that dynamical or temperature scaling is not of power-law type but rather activated, i.e., given by a generalized Vogel-Fulcher law. Recent mutually contradicting experimental results on Si:P are discussed in light of this, and new experiments are proposed to test the predictions of our quantum-glass scaling theory.