Which glass stability criterion is the best?

In order to investigate which glass stability criterion (G-criterion) is the best, we analyzed the following criteria: KH, γ, γm, β, β′, α, Φ, ζ, ω and ω2. All these criteria are expressed through temperature relationships: r = Tx,c/Tg and m = Tl,m/Tg where Tg is glass transition temperature, Tx is the onset of the crystallization peak (or Tc is the maximum of the crystallization peak), and Tl is liquidus temperature (or Tm melting temperature). First, we considered the values of the R2 factor of correlation between G-criteria and critical cooling rate Rc. By expressing the G-criteria via r and m we established the relationships between them. These relations show that there may exist a potential linear dependence between γ and γm, as well as between γ and β′, ω, ω2 and ζ. Further, the sensitivity of G-criteria is a necessary condition for determining the differences between different glasses. We also analyze how big the statistical deviations are in relation to the calculated values for the normalized and non-normalized G-criteria. The linear dependent criteria have a very similar R2 factor of correlation with Rc and in that respect they are equal. Among them, ω2 shows the best value of R2. In relation to the accordance of the calculated and experimental values of the normalized Gn-criteria, the statistical deviation strongly depends upon the choice of the point where it is assumed that Gn is zero. If we compare the experimental values and the values which are calculated from the line determined by two given points, γ is the best criterion. The γ parameter has the lowest sensitivity in relation to the changes of both super cooled region and reduced glass transition temperature. Our conclusion is that from all the analyzed criteria there is not one which can be declared the best. In the case we have to make a choice, ω2 is the best criterion, in addition to γm. The reason for this fact is that ω2 has the biggest R2 factor of correlation with Rc and also high sensitivity.