Jj. Wang et al., Determination of martensite start temperature for engineering steels part II. Correlation between critical driving force and Ms temperature, MATER T JIM, 41(7), 2000, pp. 769-776
Various physical models are discussed based on different correlations estab
lished between the Ms temperature, the critical driving force, and the stee
l chemistry. These correlations were derived for a group of Fe-(0.2-0.5)C-(
0.5-2.0)Mn-(0.5-2.0)Si-(0.5-2.0)Cr-(0.1-0.7)Mo test alloys, which serve as
a good representative for most low alloy engineering steels. The chemical d
riving force was calculated by thermodynamic software and the Ms temperatur
e was predicted by a validated artificial neural network model. Two basic p
hysical models are discussed: the Ms-dependent model and the chemistry-depe
ndent model. In the Ms-dependent model, the critical chemical driving force
is linearly related to the Ms temperature: the effect of the steel chemist
ry is indirect. The standard error of the simple Ms-dependent model is 51.9
J/mol when the spontaneous Zener ordering of carbon atoms is taken into ac
count. The chemistry-dependent model is based on the hypothesis that the cr
itical driving force con be fully represented in terms of the steel chemist
ry. The critical driving force has been estimated using either linear, expo
nential, Pythagorean or mixed superposition laws. Comparisons of the critic
al driving force predicted by these addition methods with the thermodynamic
result indicate that an exponential addition method, with the optimum expo
nent index value of 2.07 (approximately square) gives the best predictive r
esult. The quality of the linear relation between the critical driving forc
e and the Ms temperature is improved slightly if the critical driving force
is corrected fur die elastic strain energy, estimated by assuming that the
elastic moduli, lattice constants. and molar volumes of ferrite and austen
ite are both temperature- and chemistry-dependent is removed from the criti
cal driving force. Analysis indicates that the simple Ms-dependent model wi
ll be improved after considering the extra minor effect of steel chemistry.
In contrast, the error of the chemistry-dependent model can be hardly rela
ted to the Ms temperature. Based on the above analysis, an accurate mixed M
s-dependent plus chemistry-influence model, in which the elastic strain ene
rgy is considered, is developed, and which yields for the chemical driving
force -Delta G(r)* = 3247 - 4.8446 . Ms (degrees C), with a standard error
of only 40.7 J/mol.