ARTICLES PUBLISHED ONLINE: 9 JANUARY 2011 | DOI: 10.1038/NMAT2934 Disorder-induced localization in crystalline phase-change materials T. Siegrist 1,2 , P. Jost 1 , H. Volker 1 , M. Woda 1 , P. Merkelbach 1 , C. Schlockermann 1 and M. Wuttig 1,3 * Localization of charge carriers in crystalline solids has been the subject of numerous investigations over more than half a century. Materials that show a metal–insulator transition without a structural change are therefore of interest. Mechanisms leading to metal–insulator transition include electron correlation (Mott transition) or disorder (Anderson localization), but a clear distinction is difficult. Here we report on a metal–insulator transition on increasing annealing temperature for a group of crystalline phase-change materials, where the metal–insulator transition is due to strong disorder usually associated only with amorphous solids. With pronounced disorder but weak electron correlation, these phase-change materials form an unparalleled quantum state of matter. Their universal electronic behaviour seems to be at the origin of the remarkable reproducibility of the resistance switching that is crucial to their applications in non-volatile-memory devices. Controlling the degree of disorder in crystalline phase-change materials might enable multilevel resistance states in upcoming storage devices. F ew properties have provided such a wealth of information on solids as that of charge-carrier transport. The electrical resistivity is highly valuable to characterize solids, because the room-temperature resistivity for different materials spans more than 32 orders of magnitude 1 . Two different types of solid can be distinguished on the basis of the temperature dependencies of their resistivity: experimentally the temperature coefficient of the resistivity (TCR) is used to separate metallic (dρ/dT > 0) from insulating behaviour (dρ/dT < 0). Theoretical studies instead focus on the low-temperature limit of the resistivity, or the reciprocal quantity (electrical conductivity). The conductivity of insulators vanishes as T goes to 0 K; metals reveal a finite conductivity. This immediately raises the question of a minimum conductivity for metals. Indeed, Mott 2 argued that no metallic state at 0 K is possible below a minimum conductivity 3 . Experimental investigations of doped semiconductors, however, show a metallic-state conductivity that is lower by at least three orders of magnitude 4 , in line with scaling theory 5 . Although many models of a metal–insulator transition (MIT) invoke a change of the crystal structure on the transition 6 , two well- known concepts exist where the transition occurs without a change in crystallographic phase. According to Mott, an MIT arises if the electron interaction energy (correlation) exceeds the Fermi energy 7 . A second route has been identified by Anderson, who showed that increasing disorder turns the delocalized electronic states at the Fermi energy into localized states 8 , with a quasiclassical version oc- curring in heavily doped and highly compensated semiconductors 9 . A rich playground to study electronically driven MITs was identified in phosphorus-doped silicon (Si:P; refs 4,10,11), with the MIT occurring on reaching a critical concentration of free carriers with doping. An MIT predicted by Mott is expected because the increasing carrier concentration reduces the electron-correlation- to-Fermi-energy ratio. However, doping also introduces disorder because dopant atoms statistically occupy sites of the silicon lattice. Hence this MIT has also been discussed as a manifestation of an Anderson transition 12–14 . As aspects of both a Mott 1 I. Physikalisches Institut (IA), RWTH Aachen University, 52056 Aachen, Germany, 2 Department of Chemical and Biomedical Engineering, Florida State University, Tallahassee, Florida 32310, USA, 3 JARA-FIT, RWTH Aachen University, I. Physikalisches Institut (IA), 52056 Aachen, Germany. *e-mail: wuttig@physik.rwth-aachen.de. (correlation) and an Anderson (disorder) transition are prevalent, the MIT in doped semiconductors is now considered to have both signatures 12–14 . Until now, no three-dimensional crystalline solid is known where the MIT can solely be attributed to a varying degree of disorder (Anderson localization). Here, we present electronic-transport measurements for several phase-change materials (PCMs) in the crystalline state, which reveal unusual transport properties. PCMs that reversibly switch between an amorphous and a crystalline state show a pronounced change of optical and electrical properties. They may contain chalcogens, in particular Te, or pnicogens such as Sb, for example GeTe, GeSb 2 Te 4 , Ge 2 Sb 2 Te 5 and doped Sb or Sb 2 Te. The resistivity can be used for data storage owing to a large drop on crystallization, often exceeding four orders of magnitude 15 . As the amorphous-to-crystalline-state transformation proceeds at moderately elevated temperatures on a nanosecond timescale 16–18 , PCMs have attracted strong interest for optical and electronic data storage 19 . Resistivity measurements of two as-deposited amorphous thin films of GeTe and GeSb 2 Te 4 are depicted in Fig. 1. Starting at room temperature, the films are heated at a constant rate of 5 K min −1 ; the resistivity smoothly decreases because the amorphous phase is semiconducting. With the onset of crystallization, the resistivity suddenly drops by several orders of magnitude. Interestingly, in one material (GeTe), the resistivity in the crystalline state does not depend on the annealing temperature and the resistivity change proceeds within a narrow temperature window in one single step. It is generally accepted that crystalline GeTe is a degenerate semiconductor (see for instance 20 ), that is, the Fermi energy lies within the valence band (details in Supplementary Information), resulting in metal-like behaviour and thus a metallic TCR. In GeSb 2 Te 4 , crystallization is also accompanied by a steep decrease in resistivity. In contrast to GeTe, however, the resistivity in the crystalline state of GeSb 2 Te 4 shows a pronounced annealing dependence. We note that for GeSb 2 Te 4 the TCR changes from non-metallic (TCR < 0) to metallic (TCR > 0) behaviour on annealing and the corresponding transition is drawn out over a 202 NATURE MATERIALS | VOL 10 | MARCH 2011 | www.nature.com/naturematerials © 2011 Macmillan Publishers Limited. All rights reserved