New Material Model Predicts Insulator Switches Happen Sooner Than Thought

Researchers are increasingly focused on understanding the behaviour of correlated materials when driven away from equilibrium, a crucial step towards designing novel electronic devices. Tommaso Maria Mazzocchi, Markus Aichhorn, and Enrico Arrigoni, all from the Institute of Theoretical and Computational Physics at Graz University of Technology, present a new application of the mixed-configuration approximation (MCA) to investigate these nonequilibrium properties. Their work details a steady-state approach to modelling multiorbital systems, benchmarking the method against strontium vanadate and demonstrating its ability to capture bias-driven orbital charge transfer under realistic, nonequilibrium conditions. This advancement is significant because it offers a computationally efficient route to study complex materials’ responses to external stimuli, potentially circumventing the limitations of more demanding techniques like quantum Monte Carlo calculations.

This work introduces the mixed-configuration approximation (MCA), an innovative approach embedded within dynamical mean-field theory (DMFT), to model multiorbital correlated systems under both equilibrium and nonequilibrium conditions.

The research details a benchmark assessment of MCA applied to strontium vanadate (SrVO3), a material exhibiting intriguing electronic properties, and demonstrates its capability to simulate realistic scenarios involving voltage biases. MCA successfully reproduces the metallic state of bulk SrVO3 at moderate interaction strengths, although it slightly overestimates the weight of the lower band when compared to more computationally intensive methods like quantum Monte Carlo (QMC) and fork tensor product state (FTPS) solvers.
Notably, the MCA method predicts an earlier metal-to-insulator transition than QMC and FTPS as the strength of electron-electron interactions increases. Investigations into layered SrVO3 reveal that MCA partially captures the orbital polarization favouring the in-plane orbital, and a one-shot calculation, initialised with DMFT-QMC results, further enhances the accuracy of orbital occupation predictions.

This suggests the method is well-suited for analysing multiorbital impurity problems, particularly when focusing on property assessment without requiring a complete self-consistent DMFT loop. Furthermore, under simulated voltage bias conditions, the study observes a clear redistribution of orbital occupations, confirming MCA’s ability to model bias-driven orbital charge transfer in realistic, nonequilibrium environments.

This achievement is crucial for understanding the behaviour of materials used in advanced electronic components and opens avenues for designing devices with enhanced performance characteristics. The MCA method offers a computationally efficient pathway to explore the complex interplay of electron correlations in materials, potentially accelerating the discovery of novel electronic functionalities.

Benchmarking mixed-configuration approximation against dynamical mean-field theory for strontium vanadate

A mixed-configuration approximation (MCA) impurity solver, built upon the auxiliary master equation approach, was implemented to investigate multiorbital correlated systems both in equilibrium and under nonequilibrium conditions within dynamical mean-field theory (DMFT). The study benchmarked this method using bulk and layered SrVO3, subsequently applying it to a system subjected to a voltage bias perpendicular to the layer via reservoirs maintained at differing chemical potentials.

For bulk SrVO3, the MCA successfully reproduced the metallic state at moderate interaction strengths, although it overestimated the weight of the lower band when compared to quantum Monte Carlo (QMC) and fork tensor product state (FTPS) solvers. Notably, MCA predicted an earlier metal-to-insulator transition than both QMC and FTPS as the electron-electron interaction increased.

In layered SrVO3 at equilibrium, MCA partially captured the orbital polarization favouring the in-plane xy orbital, though to a lesser extent than DMFT results converged with QMC. A one-shot impurity calculation, initialised with DMFT-QMC results, improved the MCA’s ability to predict orbital occupations, demonstrating a stronger charge polarization towards the xy orbital.

This suggests the approach is suitable for studying multiorbital impurity problems, particularly when assessing properties without completing the full DMFT self-consistent loop. Under applied bias, a pronounced redistribution of orbital occupations was observed, confirming the method’s capacity to model bias-driven orbital charge transfer in realistic, nonequilibrium materials.

The MCA calculations were performed by solving the impurity Anderson model using a hierarchy of equations of motion, approximating the self-energy with a mixed-configuration ansatz. This involved truncating the equations of motion at a given order, thereby reducing the computational cost while retaining essential physics.

The nonequilibrium simulations employed the Keldysh formalism, enabling the study of systems driven out of equilibrium by an external bias. The method’s performance was rigorously tested against established techniques like QMC and FTPS, providing a comprehensive validation of its accuracy and efficiency.

Mixed-configuration approximation reveals discrepancies in strontium vanadate electronic structure and orbital polarisation

A mixed-configuration approximation (MCA) based on the auxiliary master equation approach impurity solver has been used to study multiorbital correlated systems under both equilibrium and nonequilibrium conditions within dynamical mean-field theory (DMFT). For bulk SrVO3, the MCA reproduces the metallic state at moderate interaction strengths, although it overestimates the weight of the lower band when compared to quantum Monte Carlo (QMC) and fork tensor product state (FTPS) solvers.

Increasing the electron-electron interaction resulted in an earlier metal-to-insulator transition with MCA relative to both QMC and FTPS calculations. In layered SrVO3 at equilibrium, MCA partially captures the orbital polarization favouring the in-plane xy orbital, but to a lesser extent than DMFT-converged results obtained using QMC.

Performing a one-shot impurity calculation initialised with DFMT-QMC results, MCA yields orbital occupations demonstrating a stronger charge polarization in favour of the xy orbital. This suggests the approach is suitable for studying multiorbital impurity problems when assessing properties without completing the full DMFT self-consistent loop.

Under applied bias, a pronounced redistribution of orbital occupations was observed, demonstrating the method’s ability to capture bias-driven orbital charge transfer in realistic materials under nonequilibrium conditions. The work demonstrates that MCA can effectively model the behaviour of strongly correlated materials, particularly in scenarios involving external perturbations.

This capability is crucial for exploring potential applications in transistor-like devices where controlling the metal-to-insulator transition is essential. The pronounced orbital redistribution under bias suggests a pathway for manipulating charge transport in these materials.

Mixed-configuration approximation benchmarks and application to strontium vanadate electronic structure

Researchers have developed a mixed-configuration approximation (MCA) within the dynamical mean-field theory (DMFT) framework to investigate strongly correlated multiorbital systems both in and out of equilibrium. This method addresses the complex behaviour of materials where electron interactions play a significant role, offering a computationally efficient approach to study these systems.

The MCA was benchmarked using strontium vanadate (SrVO3), a material exhibiting interesting electronic properties, and applied to simulate a scenario involving a voltage bias applied across a layered structure. Application of the MCA to bulk SrVO3 successfully reproduced the metallic state observed at moderate interaction strengths, although discrepancies were noted in the relative weighting of the lower electronic band when compared to more computationally intensive methods like quantum Monte Carlo (QMC) and fork tensor product state (FTPS) solvers.

Furthermore, the MCA predicted a metal-to-insulator transition at a lower interaction strength than these benchmark calculations. In layered SrVO3, the method partially captured the orbital polarization, favouring the in-plane orbital, and demonstrated the ability to model bias-driven orbital charge transfer under nonequilibrium conditions.

The authors acknowledge that the MCA overestimates the weight of the lower band in bulk SrVO3 and predicts an earlier metal-to-insulator transition compared to QMC and FTPS. However, they demonstrate that initializing a single impurity calculation with results from DMFT-QMC improves the accuracy of orbital occupation predictions.

Future work could focus on refining the MCA to better align with results from more accurate solvers and extending its application to explore a wider range of correlated materials and nonequilibrium phenomena. These developments offer a promising route towards understanding and predicting the behaviour of complex materials under realistic operating conditions.