Climbing Jacob’s ladder: A density functional theory case study for Ag2ZnSnSe4 and Cu2ZnSnSe4

Daniel Fritsch,Susan Schorr

doi:10.1088/2515-7655/abc07b

Abstract

Recent advances in the development of exchange and correlation functionals to be employed in density functional theory calculations combined with the availability of ever more powerful high-performance computing facilities, let predictive computational materials science become reality. In order to assess the quality of calculated material properties, Jacob’s ladder provides an informal classification, where exchange and correlation functionals of similar capabilities are placed at the same rung of this ladder, while improved and more accurate ones are placed at higher rungs. Climbing Jacob’s ladder, i.e. employing more accurate exchange and correlation functionals, increases the quality of the results and the computational demands, and provides some guidance as to what accuracies and computational costs to expect from specific calculations. This is particular important for materials whose electronic ground state properties are incorrectly described, e.g. small band gap semiconductors, and materials, where system sizes for subsequent investigations, like defect properties or band offsets in heterostructures, become prohibitively large for more accurate exchange and correlation functionals.Here, we provide a systematic density functional theory study on the ground state properties of Ag2ZnSnSe4 and Cu2ZnSnSe4 for the lowest four rungs of Jacob’s ladder. Cu2ZnSnSe4, and in particular its alloys with Ag, is a promising candidate material for future thin-film solar cell absorber layers. In the present work, the obtained material properties are compared to available experimental data, allowing to benchmark the accuracy of the employed exchange and correlation functionals. We also provide a comparative study for subsequent quasiparticle calculations. Therein, the influence of differently obtained eigenvalues and orbitals as starting points are critically assessed with respect to available experimental data. Our results show that, structural properties based on the SCAN functional show overall best agreement with available experimental data, whereas additional hybrid functional calculations are necessary for satisfying results on electronic and optical properties.

Highlights

Novel photovoltaic materials based on the kesterite crystal structure are ingredients for third generation thin-film solar cells based only on earth-abundant and non-toxic elements
Structural properties based on the SCAN functional show overall best agreement with available experimental data, whereas additional hybrid functional calculations are necessary for satisfying results on electronic and optical properties
As a first approximation to the electronic band gaps of the materials, we look at the Kohn-Sham eigenvalue differences between the valence and conduction bands for the local density approximation (LDA), PBEsol, SCAN, and HSE06 functionals, and the quasiparticle energy differences in case of G0W0 calculations, respectively

Summary

Introduction

Novel photovoltaic materials based on the kesterite crystal structure are ingredients for third generation thin-film solar cells based only on earth-abundant and non-toxic elements. One way to circumvent this particular band gap problem is to employ more accurate exchange and correlation functionals, i.e. hybrid functionals, which incorporate some fraction of Hartree–Fock exact exchange and have been shown to yield better electronic properties of semiconducting materials Another way would be to perform additional quasiparticle investigations based on many-body perturbation theory, i.e. subsequent GW calculations. While those two approaches can routinely be applied to the bulk properties of kesterite type materials, an increased demand of computational resources prohibits their use for subsequent material properties investigations, e.g. formation energies of defects and defect complexes or the calculation of band offsets in modelled device heterostructures To this end, a well tested and proven theoretical approach would be highly desirable, providing a guide towards a useful combination of available exchange and correlation functionals together with their overall accuracy and their demands on computational resources.

Background and computational details

Results and discussion

Conclusions