Substrate temperature effects on structure and thermoelectric transport in DC-sputtered Bi2Te3 thin films

Abstract

The thermoelectric performance of Bi₂Te₃ thin films is highly sensitive to deposition temperature because substrate heating simultaneously controls microstructure, stoichiometry, and charge transport. Here, ~ 450 nm Bi₂Te₃ films were deposited on glass by single-target DC magnetron sputtering while varying the substrate temperature from room temperature to 300 °C (RT, 100 °C, 200 °C, 300 °C). FESEM reveals progressive grain coarsening and improved grain connectivity with increasing temperature, while cross-sectional imaging confirms a comparable film thickness across all samples. EDS shows a monotonic Te-loss trend at elevated substrate temperatures, indicating increasing deviation from stoichiometric Bi₂Te₃. XRD confirms crystalline Bi₂Te₃ formation for all conditions and shows systematic peak shifts/lattice-spacing trends with temperature; no distinct elemental Bi or Te peaks are detected within measurement limits. Hall and transport measurements demonstrate that electrical conductivity increases with substrate temperature, whereas the Seebeck response decreases in magnitude, reflecting the expected conductivity–thermopower trade-off as composition and carrier transport evolve. The power factor ($$PF = S^{2} \sigma$$) reaches a maximum of ~ 4 µW cm⁻¹ K⁻² for films deposited at 200 °C, identifying an intermediate-temperature window that optimizes the balance between conductivity and thermopower for this DC-sputtering route. UV–Vis reflectance analysis indicates a substrate-temperature-dependent increase in the apparent optical bandgap extracted from Kubelka–Munk/Tauc-type treatment in the 300–1100 nm window, which is reported as an apparent optical metric rather than the intrinsic Bi₂Te₃ bandgap. Overall, these results establish ~ 200 °C as the most favorable substrate temperature in this study for achieving high PF Bi₂Te₃ thin films on glass and provide practical processing guidance for thermoelectric thin-film optimization.