Investigating and forecasting the performance of surface passivated SnO ₂ based electron transport material for dye-sensitized solar cells by machine learning (ML)

Surface passivation has showed to be a successful strategy to enhance the overall performance of the dye-sensitized solar cells (DSSCs) and reduce the charge recombination kinetics. Machine learning (ML), however, has recently made a significant impact on the field of energy by determining secreted patterns and classifying correlations between desired output values and input objects. Organic passivating materials, compared to other passivating materials, play a significant role in reducing the charge recombination rate while enhancing the performance. Herein, an organic passivating compound of 2-aminoterephthalic acid (2ATA) is applied to passivate SnO 2 photoanodes for the first time, and the observed impact on the photovoltaic properties of DSSCs. The SnO 2 photoanode was synthesized using a facile modified-solvothermal technique. The prepared sample was studied by X-ray diffraction (XRD), field-emission scanning electron microscopy (FE-SEM), energy-dispersive X-ray spectroscopy (EDS), Fourier transform infrared spectrometer (FTIR), UV–vis spectrophotometer (UV-Vis), and current density- voltage (J-V). The XRD pattern confirmed the tetragonal rutile phase of SnO 2 structure, with pure and amorphous in nature. The FE-SEM exhibited platelet morphology of pure SnO 2 with more agglomeration. The surface engineering of SnO 2 photoanodes by 2ATA passivation, as confirmed by FTIR, UV-Vis spectra, and finally improving device performance. It was found that passivated SnO 2 enhanced overall performance compared to bare SnO 2 photoanodes and achieved the highest efficiency of ∼ 0.64%. The ML algorithms, XGBoost regression, and random forest (RF) regression were analyzed to predict the efficiency. The three different architectures of the artificial neural network (ANN) algorithm with 100, 1000, and 2000 epochs were also analyzed with different metrics to accurately predict the efficiency and finally compared with this experimental SnO 2 based DSSCs work.