Fig. 4: Correlation between the predicted transition temperature and Helmholtz energy difference between β-quartz and β-cristobalite at \({T}_{{\rm{q}}-{\rm{c}}}^{\exp }\), and \({V}_{{\rm{cristobalite}}}^{\exp }\) and \({V}_{{\rm{quartz}}}^{\exp }\). | npj Computational Materials

Fig. 4: Correlation between the predicted transition temperature and Helmholtz energy difference between β-quartz and β-cristobalite at \({T}_{{\rm{q}}-{\rm{c}}}^{\exp }\), and \({V}_{{\rm{cristobalite}}}^{\exp }\) and \({V}_{{\rm{quartz}}}^{\exp }\).

From: Free-energy perturbation in the exchange-correlation space accelerated by machine learning: application to silica polymorphs

Fig. 4: Correlation between the predicted transition temperature and Helmholtz energy difference between β-quartz and β-cristobalite at $${T}_{{\rm{q}}-{\rm{c}}}^{\exp }$$ T q − c exp , and $${V}_{{\rm{cristobalite}}}^{\exp }$$ V c r i s t o b a l i t e exp and $${V}_{{\rm{quartz}}}^{\exp }$$ V q u a r t z exp .

The filled dark green circles show explicitly computed results and the line a linear fit of them. The hollow circles mark extrapolations out of the stability regime of the phases. The light green triangles mark predictions for rung 4 and 5 functionals (labeled in italics) utilizing the fit.

Back to article page