Logo tutelle


On this website

On the whole CNRS Web

Home page > Strongly correlated oxide materials: from models to devices

M. Gabay, M. Rozenberg

In transition metal based oxides (TMO) where electronic interactions are quite strong, metal to insulator or to superconductor transitions can be triggered by slight variations of the carrier concentration. These phase changes may be produced in a well defined region of the sample using field effects and this property opens new venues for engineering next generation devices. Strontium titanate (STO) is a prime candidate for applications and the two studies that are presented below are based on this material; they explain the mechanisms at play in the transitions, using theoretical considerations pertaining to strongly correlated systems.

The discovery of the Resistive Switching (RS) effect in CMR manganites at the Univ. of Houston and in STO at IBM (Zurich) in 2000 set the stage for an intense experimental activity in a new line of research. The physical process involves controlling the resistance of a TMO structure (usually a capacitor like structure with a TMO thin film dielectric layer) by the application of voltage pulses (i.e., strong electric fields). The systems are observed to switch reversibly between well defined and non-volatile states, thus enabling the use of such a device as a memory (the OS of computers could remain in the memory even after the machine is powered off). Despite the clear huge potential for practical applications, at this point, this field remains an area with mostly experimental vigorous activity and relatively few theoretical works. One reason may be the lack of contact that historically the electronics semiconductor community has had with the strongly correlated electron one. In this context, we have proposed a basic model that provides a successful description of several experimental observations. It constitutes a helpful theoretical reference frame for the experimental activity, and we have recently extended it to propose the idea that a doping driven Mott transition may be responsible for the effect in some materials that include STO. This work captured the attention of several experimental groups at (IBM-Zurich), at Julich, at the Univ. of Tokyo, at Thales and at Sharp Electronics (Fukuyama).

Reference : Strong electron correlation effects in nonvolatile electronic memory devices. M. J. Rozenberg, I. H. Inoue, and M. J. Sánchez, Appl. Phys. Lett. 88, 033510 (2006).

A high quality heterostructure was engineered at the University of Geneva, consisting of an STO substrate, a single crystal film of a superconductor oxide (Nb-STO) and a thin layer of ferroelectric oxide material (PZT) playing a role similar to the gate in a field effect transistor. Applying a DC voltage to the tip of an STM allows one to change reversibly and in a controlled fashion the ferroelectric polarization of the gate layer, which causes a local change in the carrier concentration of Nb-STO; the region swept by the STM tip has a sizably higher superconducting temperature (P- state) than the rest of the film (P+ state). The nature of the transition is quite unusual because of the strong electronic correlations. For a film, vortices (V) and antivortices (AV) are large thermal excitations which affect the phase Φ of the order parameter. Their concentration and their mutual distances are set by the temperature. In the ordered phase, a diluted gas of bound V-AV pairs is obtained. At the Kosterlitz-Thouless (KT) transition, the binding energy of the pairs goes to zero and the unbound V and AV destroy phase coherence, i.e., superconductivity. If this scenario were to apply in the case of Nb-STO, it would predict a KT temperature almost equal to the BCS critical Tc where the gap vanishes. But the difference in Tc for the P- and the P+ states is much too small to allow a local on-off switching of superconductivity, and this fact contradicts the experimental observation of a commutation effect. Besides, the core energy of the phase defects is tiny, and hence their concentration would be high enough to kill superconductivity in the very range of temperatures where it is obtained. We have demonstrated that correlations stabilize a dense crystal of V and AV. The lattice spacing is set by a balance between attractive interactions and thermal repulsions of the V and AV. Superconductivity is a stable phase since Φ has a well defined value everywhere. Now, in 2D perfect crystalline order cannot exist; dislocations must be present at finite temperature. These defects of the lattice... of defects (V-AV) cause the loss of quasi long range order at a KT transition where the lattice melts, releasing V and AV. This melting implies the loss of ordering for Φ, hence of superconductivity. The KT temperature is proportional to the carrier concentration of the (P+ or P-) state. Lowering T in the superconducting phase entails a decrease of the V and AV concentrations. We have also shown that if the concentration gets too small, the crystal structure is not stable anymore and it sublimates, yielding a gas of bound V-AV pairs. Nevertheless the system is still in the superconducting phase. This scenario is predicted to apply quite generally to films of correlated materials (for instance to cuprate superconductors).

Reference : Local switching of 2D superconductivity using the ferroelectric field effect K.S. Takahashi, M. Gabay, D. Jaccard, K. Shibuya, T. Onishi, M. Lippmaa, J.M. Tricone, Nature, 441, 195 (2006).