Manuscript received February 11, 2026; accepted March 29, 2026; published April 28, 2026
Abstract—Perovskite solar cells have attracted considerable attention owing to their outstanding photoelectric conversion performance; however, interface defects severely impede further improvements in device efficiency and stability. This work proposes a machine learning-accelerated framework for interface defect identification and passivation strategy optimization in perovskite solar cells. A high-quality defect feature database incorporating Density Functional Theory (DFT) calculation data is constructed, and random forest and Gradient Boosting Decision Tree (GBDT) algorithms are employed for accurate classification of interface defect types, while a Bayesian optimization method is applied to systematically screen passivation molecules. In alignment with existing research on explainable artificial intelligence methods and ensemble learning frameworks, this work introduces machine learning interpretability analysis into the domain of materials defect identification. Simulation results demonstrate that the proposed method achieves a defect identification accuracy of 96.218%. After optimized passivation treatment, the Power Conversion Efficiency (PCE) of the device is improved from 19.347% to 24.186%, the open-circuit voltage (Voc) increases from 1.082 V to 1.156 V, and the Fill Factor (FF) rises from 0.791 to 0.843. This study provides a data-driven paradigm for the rational design of highly efficient and stable perovskite solar cells.
Keywords—perovskite solar cells, machine learning, interface defects, passivation strategy, power conversion efficiency, random forest, Bayesian optimization
Cite: GWei Yan, "Machine Learning-Accelerated Identification and Passivation of Interface Defects in Perovskite Solar Cells," International Journal of Engineering and Technology, vol. 18, no. 2, pp. 31-36, 2026.
Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (
CC BY 4.0).
