Combining electronic structure and many-body theory with large databases: A method for predicting the nature of 4f states in Ce compounds

Abstract

Recent progress in materials informatics has opened up the possibility of a new approach to accessing properties of materials in which one assays the aggregate properties of a large set of materials within the same class in addition to a detailed investigation of each compound in that class. Here we present the first large scale investigation of electronic properties and correlated magnetism in Ce-based compounds accompanied by a systematic study of the electronic structure and 4f-hybridization function of a large body of Ce compounds. We systematically study the electronic structure and 4f-hybridization function of a large body of Ce compounds with the goal of elucidating the nature of the 4f states and their interrelation with the measured Kondo energy in these compounds. The hybridization function has been analyzed for more than 350 data sets (being part of the IMS data base) of cubic Ce compounds using electronic structure theory that relies on a full-potential approach. We demonstrate that the strength of the hybridization function, evaluated in this way, allows us to draw precise conclusions about the degree of localization of the 4f states in these compounds. The theoretical results are entirely consistent with all experimental information, relevant to the degree of 4f localization for all investigated materials. Furthermore, a more detailed analysis of the electronic structure and the hybridization function allows us to make precise statements about Kondo correlations in these systems. The calculated hybridization functions, together with the corresponding density of states, reproduce the expected exponential behavior of the observed Kondo temperatures and prove a consistent trend in real materials. This trend allows us to predict which systems may be correctly identified as Kondo systems. A strong anti-correlation between the size of the hybridization function and the volume of the systems has been observed. The information entropy for this set of systems is about 0.42. Our approach demonstrates the predictive power of materials informatics when a large number of materials is used to establish significant trends. This predictive power can be used to design new materials with desired properties. The applicability of this approach for other correlated electron systems is discussed.

A high-throughput data analysis and materials discovery tool for strongly correlated materials

“…A search for strongly correlated actinides We have computed frequency-dependent hybridization functions for all the compounds in our database to supplement our ground state DFT calculations. Our earlier work 44 has shown that hybridization is a good descriptor for detecting localization trends, although an accurate description of localization phenomenon will require self-consistent DMFT calculations. 9,10 However, our findings 44 indicate that a high-throughput, local Green's function based hybridization can at least qualitatively capture physicallyrelevant trends in the localization of the f-states.…”

“…Our earlier work 44 has shown that hybridization is a good descriptor for detecting localization trends, although an accurate description of localization phenomenon will require self-consistent DMFT calculations. 9,10 However, our findings 44 indicate that a high-throughput, local Green's function based hybridization can at least qualitatively capture physicallyrelevant trends in the localization of the f-states. Our analysis with the present f-electron database using ≈350 Ce-based binary compounds shows that the maximum hybridization value near the Fermi energy decreases with increasing lattice spacing.…”

Interplay of electron correlations and lattice distortions in transition metal coordination compounds

“…group F m3m and a structure similar to N aCl. We reproduce previously reported results on ∆(E) [189] for CeX and the established trends in similar compounds [191], indicating the high fidelity of our calculations. Our analysis also identifies a number of similarities and trends in ∆(E) features in various 4f and 5f monopnictides as follows.…”

“…This step captures frequency-dependent hybridization functions for all the compounds in our database. An earlier work has [189] shown that hybridization is a good descriptor for detecting localization trends, although an accurate description of localization phenomena will require self-consistent DMFT calculations. However, earlier findings [189] indicate that a high-throughput, one-shot DMFT based hybridization can at least qualitatively capture interesting trends.…”

“…An earlier work has [189] shown that hybridization is a good descriptor for detecting localization trends, although an accurate description of localization phenomena will require self-consistent DMFT calculations. However, earlier findings [189] indicate that a high-throughput, one-shot DMFT based hybridization can at least qualitatively capture interesting trends. Our analysis with the present f-electron database using ≈ 300 Ce based binary compounds show that the maximum hybridization value near the Fermi energy decreases with increasing lattice spacing.…”