On the Prediction of α-Martensite Temperatures in Medium Manganese Steels

Abstract

A new composition-based method for calculating the α-martensite start temperature in medium manganese steel is presented and uses a regular solution model to accurately calculate the chemical driving force for α-martensite formation, $$ \Delta G_{\text{Chem}}^{\gamma \to \alpha } $$ . In addition, a compositional relationship for the strain energy contribution during martensitic transformation was developed using measured Young’s moduli (E) reported in literature and measured values for steels produced during this investigation. An empirical relationship was developed to calculate Young’s modulus using alloy composition and was used where dilatometry literature did not report Young’s moduli. A comparison of the $$ \Delta G_{\text{Chem}}^{\gamma \to \alpha } $$ normalized by dividing by the product of Young’s modulus, unconstrained lattice misfit squared (δ 2), and molar volume (Ω) with respect to the measured α-martensite start temperatures, $$ M_{\text{S}}^{\alpha } $$ , produced a single linear relationship for 42 alloys exhibiting either lath or plate martensite. A temperature-dependent strain energy term was then formulated as $$ \Delta G_{\text{str}}^{\gamma \to \alpha } \left( {{\text{J}}/{\text{mol}}} \right) = E\varOmega \delta^{2} (14.8 - 0.013T) $$ , which opposed the chemical driving force for α-martensite formation. $$ M_{\text{S}}^{\alpha } $$ was determined at a temperature where $$ \Delta G_{\text{Chem}}^{\gamma \to \alpha } + \Delta G_{\text{str}}^{\gamma \to \alpha } = 0 $$ . The proposed $$ M_{\text{S}}^{\alpha } $$ model shows an extended temperature range of prediction from 170 K to 820 K (−103 °C to 547 °C). The model is then shown to corroborate alloy chemistries that exhibit two-stage athermal martensitic transformations and two-stage TRIP behavior in three previously reported medium manganese steels. In addition, the model can be used to predict the retained γ-austenite in twelve alloys, containing e-martensite, using the difference between the calculated $$ M_{\text{S}}^{\varepsilon } $$ and $$ M_{\text{S}}^{\alpha } $$ .