Bandgap energy prediction of senary zincblende III–V semiconductor compounds using machine learning. (July 2023)

• Record Type: Journal Article
• Authors: Alsalman, Mohammed; Alqahtani, Saad M.; Alharbi, Fahhad H.
• Is Part Of: Applied geography. Volume 161(2023)
• Publication Date: 2023-07
• Subjects: Machine learning (ML) -- Bandgap (Eg) -- Zincblende (ZB) -- III–V semiconductor compound -- Regression analysis; Geography -- Periodicals; Human geography -- Periodicals; Human ecology -- Periodicals; 910
• DOI: 10.1016/j.mssp.2023.107461
• Languages: English
• ISSNs: 0143-6228

Abstract

Abstract: Two complimentary machine learning (ML) models are developed to estimate the bandgap energy ( E g ) of senary zincblende (ZB) III–V semiconductor compounds Al x 1 Ga x 2 In x 3 P y 1 As y 2 Sb y 3 from molar fractions. The models work – as well – for any compound with smaller number of constituent elements from binaries to quinaries. Some of these compounds exhibit transition between a direct and an indirect E g as a function of composition. This is due to the fact that the bandgap for this class of materials is either the direct gap at Γ point ( E g Γ ) or the indirect one between the maximum valance band at Γ point and the minimum conduction band around X point ( E g X ). Hence, to capture precisely the E g and its state, the ML models estimate separately the E g Γ and E g X where the lowest energy gap is the effective E g . For model training, 18 ternary ZB III–V semiconductor compounds are used, which contain all possible ternary configurations of III-type cations (Al, Ga, In) and V-type anions (P, As, Sb). The achieved testing accuracy using cubic regression with ridge regularization – in terms of root mean square error (RMSE) – are 7.583 and 0.021 meV for E g Γ and E g X, respectively. These correspond to R 2 of 0.99995 and 1.00000 (8 nines), respectively. To validate the models, they are benchmarked against experimental measurements of four different quaternary ZB III–V semiconductor compounds. The agreement is very noticeable demonstrating the validity and the excellent accuracy of the developed models. Graphical abstract: Highlights: Considering physical insights and constraints are crucial for proper development of the ML models. 2 complimentary ML models are developed to predict the bandgap of zincblende III-V semiconductors. For proper ML modeling, we need reliable data, suitable model space, and relevant attributes.