Modeling the correlation between composition and Néel transition temperature in Fe-Mn-Al-Cr-Si alloys

Abstract The Néel transition temperature (T N ) is a critical parameter governing the magnetic and elastic behavior of Fe–Mn–C–Al–Cr–Si alloys. However, previously proposed empirical equations often fail to capture the complex interactions among alloying elements, limiting their predictive accuracy. In this study, an artificial neural network (ANN) model is developed to quantitatively relate the chemical composition (Mn, C, Al, Cr, Si) to T N . The optimized ANN—featuring a single layer with 8 neurons, a momentum value of 0.3, a learning rate of 0.45, and 24,000 iterations—predicted T N with the highest accuracy. The model achieved an exceptional fit with an adjusted R 2 of 0.98 for the training data and a strong generalization capability with an adjusted R 2 of 0.93 for the test data. ANN predictions were further validated through comparison with established empirical equations reported in the literature, confirming the predictive reliability of the model. Sensitivity analyses were performed to evaluate the individual and combined effects of alloying elements, and the relative importance index was used to quantify their contributions to T N variation. ANN predictions for unseen samples showed excellent agreement with experimental values. The results are further interpreted from a metallurgical perspective, offering deeper insight into the compositional dependence of the Néel transition temperature.