Accueil > English > News

How can insulating iridates become conducting ?

For a material to become metallic, some electrons must be “free” to move. And yet, some compounds remain insulating even when they should have such free electrons. Electrons may be localized by Coulomb repulsion, which prevents them from meeting and from interacting too strongly. Such compounds are called correlated systems. A new family of correlated systems appeared recently with iridates, notably Sr2IrO4. In addition to “usual” interactions, spin-orbit coupling is very strong there because of the large mass of the iridium atom. This unusual situation has been little studied up to now. To better understand their properties, one can attempt to induce metallicity by adding or removing electrons to the system. We studied how the electronic structure of these compounds changes across the metal-insulator transition, thanks to angle resolved photoemission experiments on the CASSIOPEE beamline of the SOLEIL synchrotron.

We have synthesized at the LPS Sr2IrO4, which has a lamellar perovskite structure, with initial help from the group of Ian Fisher at Stanford University. We have then used substitutions (Rh on Ir site or La on Sr site) to make it more conducting. Angle-resolved photoemission (ARPES) allows imaging the electronic structure.

On the figure, two different bands are observed. One is centered at Γ (red line) and remains far from the Fermi level EF (white line). The other is centered at X (blue line), where it tops at -0.2eV from EF in the pure compound. These two bands are characterized by an effective total angular momentum Jeff=3/2 and 1/2, as sketched at the bottom of the figure. The Jeff=3/2 band contains 4 electrons and is completely filled, while the Jeff=1/2 band only contains one electron that should be free to move. In reality, this band splits into one band below EF, called LB for Lower Band, and one band above EF, which cannot be observed in photoemission, called UB for Upper Band. There is a gap of 0.6eV between them. This splitting may be due to electronic correlations inducing a Mott insulating state.

Top : Images of dispersions in the ΓX direction of the reciprocal space, obtained by angle resolved photoemission. The color scale is proportional to the number of electrons detected at one energy (y axis) and one momentum k value (x axis). Dispersions from the two bands J=3/2 and J=1/2 are indicated by red and blue lines. 4 different samples were studied : the pure compound, one compound doped with 15% Rh (Rh1), one compound doped with 1% La (La1) and one compound doped with 4% La (La2). Bottom : Sketch of the electronic density integrated in reciprocal space in each case. The dotted line indicates the Fermi level (white line in images).

In principle, there are several ways of evolving towards a metal. The gap could close or electrons could be added to UB or removed from LB or new states could be created within the gap. Our experiments provide information on this point. We observe that the shape of the bands remain the same, but they shift up towards EF in the Rh case and down, away from it, in the La case. This is consistent with a shift of EF towards LB and UB as sketched in the bottom of the figure. The most interesting situation occurs when more electrons are added to the system (“La2“ case). The bands move slightly back towards EF, but they also become hard to detect. In fact, their spectral weight is transferred to other structures, such as a “quasiparticle band” near EF or in-gap states. Such transfer of spectral weight is typical of correlated systems.

Before the discovery of this family, it was believed that 5d transition metals do not exhibit very interesting properties from the electronic correlations point of view. The Coulomb repulsion is much weaker there than in 3d transition metals, for example, because of a larger spatial extension of the orbitals. We understand now that the strong spin-orbit coupling cooperates with Coulomb repulsion to recreate a situation of strong correlations. Spin-orbit coupling is responsible for the formation of the J=3/2 and J=1/2 bands. Without this structure, it is likely that there would not be an insulating state. This is interesting because it “extends” the domain where interesting strongly correlated states could be looked for. In particular, the situation in Sr2IrO4 presents many analogies with cuprates, well known for their high temperature superconductivity. To better understand how far the analogy can be pushed, we are looking for better ways to dope these systems more strongly.

Reference :

Transfer of spectral weight across the gap of Sr2IrO4 induced by La doping
Véronique Brouet, Joseph Mansart, Luca Perfetti, Christian Piovera, Ivana Vobornik, Patrick Le Fèvre, François Bertran, Scott C. Riggs, M. C. Shapiro, Paula Giraldo-Gallo and Ian R. Fisher
Physical Review B 92, 081117(R) (2015).

Contact :

Véronique Brouet