Artificial intelligence-driven data generation for temperature-dependent current–voltage characteristics of diodes

Abstract

Artificial intelligence has the potential to develop models that accurately predict the behavior of electronic devices under various operating conditions. Such models allow researchers to conduct precise performance evaluations during the design phase, reducing development time and costs. For instance, by analyzing the effects of current, voltage, and temperature on diode performance, these models can shorten the diode development cycle and promote sustainable industry growth. In this study, a nanocomposite diode based on lanthanum-doped polyethyleneimine-functionalized graphene quantum dots was fabricated to investigate the impact of temperature on diode performance. The diode's current–voltage characteristics were measured experimentally over a temperature range of 77–400 K. These measurements were used to train machine learning algorithms. Specifically, K-Nearest Neighbors, Decision Trees, and Gradient Boosting were employed to predict current–voltage characteristics at temperatures lacking experimental data. The performance of these models was evaluated using metrics such as the coefficient of determination, mean squared error, and mean absolute error. Among the models, Gradient Boosting demonstrated the highest accuracy, achieving a coefficient of determination of 0.9998, a mean squared error of 0.0026, and a mean absolute error of 0.0222, though accuracy varied with temperature. To test the accuracy of the predicted values, experimental measurements were repeated for the corresponding temperatures, confirming the model's performance. The findings indicate that artificial intelligence-assisted, temperature-dependent data generation can enhance the development of a sustainable diode industry by reducing energy consumption.