Highly conductive and transparent Ga-doped ZnO thin films through data-guided optimization of nonlinear PLD process space

Group-III-doped ZnO thin films are promising transparent conducting oxides for optoelectronic and microelectronic applications because of their high optical transparency, tunable electrical conductivity, and chemical stability. Pulsed laser deposition is a versatile technique for producing high-quality thin films with controllable composition and microstructure. However, multiple interacting deposition parameters, including substrate temperature, laser fluence, and oxygen partial pressure, create a nonlinear multidimensional process space that complicates optimization. Here, a data-guided process-space reconstruction approach was used to optimize Ga-doped ZnO thin-film growth using a sparse experimental screening dataset. A Random Forest model trained using rapid measurements of electrical resistivity and average visible transmittance reconstructed multidimensional process–property relationships and identified optimized growth conditions. Films synthesized near the predicted optimum exhibited low resistivity (∼5 × 10−4 Ω cm), high transparency (>90%), strong c-axis-oriented growth, and degenerate n-type transport behavior confirmed through Hall-effect and Seebeck measurements. These results illustrate how sparse experimental screening combined with machine-learning-assisted process-space reconstruction can provide useful guidance for optimization of complex nonequilibrium thin-film growth conditions.