ISSN print edition: 0366-6352 ISSN electronic edition: 1336-9075 Registr. No.: MK SR 9/7 Published monthly

First-principles and machine learning-assisted exploration of the structural, electronic, optical, and thermoelectric properties of AuBX₂ chalcopyrite compounds

N. Boucerredj, R. Khanata, H. R. Jappor, H. Khachai, Devraj Singh, Z. Chouahda, S. Ghemid, H. Oughaddou, S. Bin-Omran, and S. Goumri-Said

Radiations Physics Laboratory, Physics Department, Faculty of Sciences, Badji Mokhtar Annaba University, Annaba, Algeria

E-mail: noureddine.boucerredj@univ-annaba.dz

Received: 7 January 2026  Accepted: 23 April 2026

Abstract:

In this investigation, boron-based chalcopyrite semiconductors AuBX₂ (B = boron; X = S, Se, and Te) are explored to assess their potential for optoelectronic and photovoltaic applications. First-principles calculations were performed using the WIEN2k code within density functional theory, employing the generalized gradient approximation (GGA), local density approximation (LDA), and Wu–Cohen GGA (WC-GGA). The structural, elastic, elastic anisotropy, electronic, optical, and thermoelectric properties were systematically investigated, including the effects of spin–orbit coupling (SOC) due to the presence of heavy elements (Au, Te). Structural optimization reveals minimal deviations in lattice and internal parameters, which matches well with available literature. The calculated elastic constants confirm the mechanical stability of the studied chalcopyrites. The computed electronic band structures using the Tran–Blaha modified Becke–Johnson (TB-mBJ) potential and WC-GGA, with and without SOC, indicate a direct band-gap semiconducting nature for all compounds, with band gaps ranging from 0.88 to 2.51 eV. SOC induces a slight reduction in the band gap (0.05–0.15 eV) but preserves the direct gap character at Γ. The optical properties indicate strong absorption and favourable photonic characteristics. Thermoelectric parameters including the Seebeck coefficient and transport properties were evaluated using the BoltzTrap code. To complement the first-principles results, machine learning models based on Random Forest and XGBoost regressors were developed using combined elemental and structural descriptors. The models achieved high predictive accuracy (R² ≈ 0.9) and were employed to propose new chalcopyrite compounds AuBAs₂, AuBSb₂, and AuBPo₂. The predicted materials exhibit direct band gaps (0.94–1.28 eV), high refractive indices, and sizable Seebeck coefficients, highlighting their promise for optoelectronic and thermoelectric applications.

Keywords: Au-based chalcopyrite; DFT; Elastic features; Electronic features; Optical features

Full paper is available at www.springerlink.com.

DOI: 10.1007/s11696-026-04953-9

Chemical Papers 80 (8) 9295–9317 (2026)