Unsteady carbon diffusion in steel carburizing/decarburizing
Alwin Johnson, Roll No. 22b2455
Abstract: Carbon carburizing/decarburizing are surface diffusion processes governed by Fick’s 2nd
law. In typical heat-treatment conditions (850–950 °C), the steel can be modeled as a semi-infinite
medium with either a fixed surface concentration or a constant flux of carbon. For a constant
diffusivity D, the well-known analytical solution involves the complementary error function (erfc).
In practice, carbon diffusivity in austenite follows an Arrhenius form, but experiments show strong
dependence on alloying and concentration. Where closed-form solutions fail, finite-difference
numerical schemes are used. This report reviews analytical and numerical solutions, discusses
concentration-dependent diffusivity models (e.g. Darken’s theory), and outlines how diffusion
coefficients D(C,T) can be fitted using Levenberg–Marquardt regression. Literature results from
EPMA/GDOES profiles are summarized, and a modeling approach for carburizing kinetics is
proposed.
1. INTRODUCTION
Carburizing (adding C) and decarburizing (removing C) are central to case hardening of steels.
These diffusion-controlled processes at 850–950 °C are described by Fick’s 2nd law. The
semi-infinite model assumes the steel bulk initially has uniform C0, and the surface is suddenly set
to Cs. The resulting case depth scales as √(Dt) and controls hardness depth. Accurate modeling
requires knowledge of the diffusion coefficient D. Classical solutions (Crank, 1975) assume
constant D, but modern steels require concentration- and composition-dependent models.
2. FICK’S LAW SOLUTIONS
Analytical solution: For constant D and fixed surface concentration, C(x,t) = C0 + (Cs – C0)
erfc(x/2√Dt). This captures the steep near-surface gradient and defines case depth. For constant flux,
similar solutions exist. Numerical solution: Finite-difference discretization integrates Fick’s law
with D(C,T). Explicit and implicit schemes handle variable diffusivity and evolving boundaries.
These methods enable modeling of alloy effects and concentration dependence.
3. DIFFUSIVITY MODELING
Diffusivity D(C,T): In austenite, D follows Arrhenius D0 exp(-Q/RT). Constant-D models use fixed
parameters, while Darken’s theory incorporates a thermodynamic factor ∂ln a/∂ln c, reducing D at
high C. Advanced fits include polynomial dependence on C. Parameters can be fitted using
Levenberg–Marquardt regression on depth profiles.
4. LITERATURE RESULTS
Literature profiles (EPMA, GDOES) show a high-C surface plateau and diffusion tails. Thibaux et
al. (2007) measured C diffusion in Fe–Ni alloys, finding Arrhenius D = 1.23×10■■ exp(–15050/T).
Lee et al. (2011) modeled alloy effects empirically. Gegner et al. (2012) confirmed alloying reduces
D. These results emphasize that constant-D approximations can mispredict case depths.
5. PROPOSED WORK
We propose implementing a 1D finite-difference solver for semi-infinite carburizing, validating
against the erfc solution. D(C,T) will be fitted using LM regression to synthetic or published
profiles. We will compare constant-D vs concentration-dependent models and assess boundary
condition sensitivity. This approach combines analytical and numerical methods to build a robust
model for carburizing kinetics.
REFERENCES
[1] Thibaux P., Métenier A., Xhoffer C., Metall. Mater. Trans. A 38 (2007) 1169–1176.
[2] Lee S.-J., Matlock D.K., Van Tyne C.J., ISIJ Int. 51 (2011) 1903–1911.
[3] Gegner J., Vasilyev A.A., Wilbrandt P.-J., Kaffenberger M., MMT Conf. Proc. (2012).
[4] Darken L.S., Trans. AIME 180 (1949) 430.
[5] Crank J., The Mathematics of Diffusion, Oxford Univ. Press (1975).
