Optical conductivity of GaTa4Se8 under high pressure supports theoretical predictions

The band theory of solids provides the framework for the understanding of the Insulator to Metal Transition (IMT) in semiconductors. This is one of the key achievements of the past century, laying the foundations for an epochal technological revolution. Band theory fails, however, in describing the electronic properties of many transition metal oxides with half-filled d or f orbitals.
When the electronic Coulomb repulsion energy U, dominates over the tendency towards metallization driven by Kinetic Energy, the conduction band splits into so-called lower (LHB) and upper (UHB) Hubbar d-bands, and the material is turned into a Mott insulator. In order to recover the metallic state one can either push the chemical potential away from half-filling or increase the bandwidth (D), so that the Kinetic Energy gain can now catch up with the Coulomb driven localization tendency. The value of the ratio U/D determines the ground state of the system. Experimental control over the U/D parameter can be achieved through the application of High-Pressure in a Diamond Anvil Cell. Pressure reduces the interatomic distances thereby increasing the wave-functions overlap, and thus D.


Figure 1: (a) Schematic view of the evolution ofthe spectral function and of σ1(ω) in a Mott insulator and in a correlated metal on each side of a bandwidth-controlled IMT. (b) Drude-Lorentz fit of σ1(ω) at 0 and 10.7 GPa.


GaTa4Se8 is a lacunar spinelchalcogenide undergoing an IMT under pressure. A resistive switching induced by electric pulses was discovered in these compounds, making this material promising for Resistive Random Access Memory applications. It was argued that GaTa4Se8 may belong to a new class of Mott insulators where the relevant entity for electronic correlation is a cluster of transition metals Ta4 rather than a single atomic site, and that its striking resistive switching properties are related to a Mott IMT.
We employ the high brightness of the SISSI infrared beamline to measure the pressure dependent optical conductivity σ1(ω) of GaTa4Se8. A single broad peak [in red in Fig. 1(b)] centered at 4000 cm-1 is required to fit the infrared ambient pressure σ1(ω). By contrast, the simplest fit of the high pressure σ1(ω) requires two additional contributions: a narrow Drude peak [in green in Fig. 1(b)] and a mid infrared band at 1500 cm-1 (blue). The change from one mode to three modes at the IMT suggests that GaTa4Se8 is a Mott rather than a band insulator. In a one-band Mott insulator, a single excitation is expected between the LHB and the UHB, centered at the energy U of the onsite Coulomb repulsion, as shown in Fig.1(a). In GaTa4Se8, this excitation is centered at U=4000 cm-1. On the metallic side of the IMT, theory predicts the appearance of a quasiparticle (QP) at the Fermi level, yielding two new optical excitations: a Drude peak corresponding to QP excitations [in green in Fig. 1(a)] and a contribution at U/2 assigned to excitations from the LHB to the QP, or the QP to the UHB [in blue in Fig. 2(a)]. Note that from our data, U does not depend on pressure: this is consistent with the very small compression with pressure of the Ta4 tetrahedral clusters, since the size of this entity determines the U value in this compound.

Overall the above analysis rules out a band insulator scenario and provides several striking evidences that GaTa4Se8 is a Mott insulator undergoing a bandwidth-controlled IMT under pressure. The bare uncorrelated D can be calculated through conventional band structure calculations. This allows us to pinpoint our data within a typical T/D vs U/D phase diagram from Dynamical Mean Field Theory (DMFT). From Fig. 2, we establish that our room temperature data between 0 and 10 GPa are located in the crossover regime (i.e., aboveT/D=0.04). The onset of metallicity on σ1(w) spectra at 6 GPa is observed at U/Dexp=2.05 (see Fig. 2), in very good agreement with DMFT, predicting U/Dtheo=2.35 on the right side of the crossover domain.

Figure 2: Theoretical T/D versus U/D phase diagramfor a one-band Mott insulator predicted by the DMFT. Data points from σ1(ω) are indicated by white circles.

Such good agreement between data and theory results from the almost ideal one-band Mott insulator character of GaTa4Se8. The canonical Mott insulators considered so far (Cr-V2O3 or NiS2-xSex) are comparatively more complex with several active bands around the Fermi energy. In this context, GaTa4Se8 could become the new archetypal Mott insulator ideally suited for future comparisons between experiments and theory.


This research was conducted by the following research team:

  • Vinh Ta Phuoc, Miriam Chligui, GREMAN, CNRS, Université F. Rabelais, Tours, France
  • Christian Vaju, Benoit Corraze, Etienne Janod, Laurent Cario, IMN, Université de Nantes, France
  • Rodolphe Sopracase, Carlo Marini, Paolo Postorino, Stefano Lupi, CNR-IOM, Università Sapienza, Roma, Italy
  • Andrea Perucchi, Elettra–Sincrotrone Trieste S.C.p.A., Trieste, Italy


V. Ta Phuoc, C. Vaju, B. Corraze, R. Sopracase, A. Perucchi, C. Marini, P. Postorino, M. Chiliqui, S. Lupi, E. Janod and L. Cario, "Optical Conductivity Measurements of GaTa4Se8 Under High Pressure: Evidenceof a Bandwidth-Controlled Insulator-to-Metal Mott Transition", Phys. Rev. Lett. 110 ,  037401 (2013), DOI: 10.1103/PhysRevLett.110.037401.

Last Updated on Tuesday, 05 March 2013 13:43