Elsevier

Acta Materialia

Volume 48, Issue 17, 8 November 2000, Pages 4217-4224
Acta Materialia

On some limitations of the Johnson–Mehl–Avrami–Kolmogorov equation

https://doi.org/10.1016/S1359-6454(00)00280-9Get rights and content

Abstract

Some limitations of the Johnson–Mehl–Avrami–Kolmogorov (JMAK) equation used widely for describing kinetics of phase transformation are demonstrated using probabilistic analysis and Monte Carlo simulations. The JMAK equation predicts correctly the real transformed fraction only if the number of the growing nuclei in the controlled volume is large. If the number of growing nuclei is small, the transformed fraction predicted by the JMAK equation deviates significantly from the real transformed fraction, no matter how large the volume of the controlled volume is. As an alternative, another equation should be applied, which for any number of the growing nuclei predicts correctly the true amount of transformed fraction.

Introduction

The question concerning kinetics of a phase transformation characterised by constant rates of nucleation and radial growth is essentially geometrical and was first treated by Kolmogorov [1], Johnson and Mehl [2] and Avrami [3], [4], [5]. Assuming that the growth rate is constant in all directions and that nuclei “β” grow as regular spheres in the matrix “α” (Fig. 1), the equationξβ(τ)=1−exp−Iπ3k3τ4has been derived for the time dependence of the transformed volume fraction ξβ(τ) [1], [2], [3], [4], [5]. In equation (1) “I” is the nucleation rate (number of nuclei per unit volume per second), “k” (m/s) is the radial growth rate and τ (s) is the time since the start of the transformation. The two-dimensional analogue of equation (1) is:ξβ(τ)=1−exp−Iπ3k2τ3which gives the time dependence of the transformed areal fraction during growth of circular nuclei with constant radial rate “k”, which nucleate with a nucleation rate I (number of nuclei per unit area per second). Kinetics of one-dimensional growth is described byξβ(τ)=1−exp(−Ikτ2)where the nucleation rate I has a dimension “number of nuclei per unit length per second”.

Some drawbacks of equation (1) have been analysed in earlier work [6], where an exact kinetics equation was derived. According to this equation, the volume fraction of the transformed zone during phase transformation of type nucleation and growth [7] with a constant radial growth rate is given byξβ(τ)=1−exp0τln[1−ψ(υ)]I(τ−υ)dυwhere υ is a dummy integration variable; I(τυ) is the number of spherical nuclei of volume v(υ)=(4/3)πk3υ3 and volume ratio ψ(υ)=(4/3)πk3υ3/V, (0≤υτ) which have nucleated in the elementary time interval (τυ,τυ+) in the controlled volume V, (for unit volume V=1m3ψ(υ)=(4/3)πk3υ3)). If the nucleation rate I is constant, equation (4) becomesξβ(τ)=1−expI0τln[1−ψ(υ)]dυ

The two- and one-dimensional analogues of equation (5) have the same functional form. The only differences are the expressions giving the volume ratio ψ(τ) of the growing nuclei and the dimension of the nucleation rate I. For two-dimensional nucleation and growth, the areal ratio is ψ(υ)=πk2υ2/S,(0≤υ≤τ),(ψ(υ)=πk2υ2 for unit area S=1m2)). In case of one-dimensional growth, the lineal ratio in equation (5) is ψ(υ)=2kυ/L,(0≤υ≤τ),(ψ(υ)=2kυ for unit length L=1m).

Section snippets

A general equation for random coverage by overlapping objects

It has been demonstrated [6] that the topology and the transformation kinetics of a microstructure formed from growing of nuclei with a constant radial growth rate, are identical to the topology and kinetics of a “microstructure” in which the same nuclei grow by interpenetrating (overlapping) (Fig. 1). Consequently, there exists equivalence between the kinetics of a phase transformation of the type “nucleation and growth with a constant radial rate” and the process of continuous coverage of

Monte Carlo simulations of a phase transformation characterised by constant nucleation and growth rates

Unlike the experimental studies, Monte Carlo simulations permit precise investigation of phase transformation kinetics characterised by constant nucleation and growth rates.

Accordingly, Monte Carlo simulations have been performed in cases of three-, two- and one-dimensional nucleation and growth. In the 3D-case, the system was initially assumed to be a cube with side 1 mm. The coordinates of the centres of the spherical nuclei were produced by a generator of uniformly distributed over the

Discussion

The computer simulation results show that for a large number of nuclei in the controlled volume, the predictions from the JMAK equation (1) agree with the true transformed fraction. In contrast, for a small number of nuclei the predictions deviate significantly from the true values. If the number of growing nuclei does not change, the predicted from the JMAK equation (1) values deviate more from the true values for large volume ratios of the nuclei. For small volume ratios, the predictions from

Conclusions

  • 1.

    If the volume ratios of the growing nuclei are small, the JMAK equation approximates very well the true transformed fraction and is a special case of the exact equation (5). This is the case at the beginning of the transformation and when the system contains a very large number of nuclei so that the transformation is completed before the nuclei can attain large volume ratios.

  • 2.

    If the system contains small number of nuclei with relatively large volume ratios, no matter how large the volume of the

References (12)

  • M.T Todinov

    Acta Mater.

    (1996)
  • A.N Kolmogorov

    Izv. Akad. Nauk. SSSR.

    (1937)
  • W.A Johnson et al.

    Trans. Am. Inst. Min. Metall. Engrs.

    (1939)
  • M Avrami

    J. Chem. Phys.

    (1941)
  • M Avrami

    J. Chem. Phys.

    (1940)
  • M Avrami

    J. Chem. Phys.

    (1939)
There are more references available in the full text version of this article.

Cited by (0)

1

Formerly with the Department of Materials Science, Technical University of Sofia, Bulgaria.

View full text