You are on page 1of 6

PROCEEDINGS OF THE IV INTERNATIONAL CONFERENCE/HIGH TEMPERATURE CAPILLARITY

J O U R N A L O F M A T E R I A L S S C I E N C E 4 0 (2 0 0 5 ) 2263 – 2268

Theoretical model of the interfacial reactions between solid iron and liquid zinc-aluminium alloy
M.-L. GIORGI ∗ , J.-B. GUILLOT ´ Ecole Centrale Paris, Grande Voie des Vignes, 92 295 Chatenay-Malabry ˆ cedex, France E-mail: Marie-lawrence.giorgi@lgpm.ecp.fr R. NICOLLE IRSID, Voie Romaine, BP 30320, 57283 Maizieres-l ` es-Metz ` cedex, France Steel strip is often coated with a layer of zinc in order to protect it against corrosion. One of the most commonly used coating processes is continuous hot dip galvanizing. In this process, the steel strip is immersed in a molten zinc bath containing small amounts of aluminium (less than 1 wt%). A model has been developed describing the kinetics of the galvanizing reactions that occur at the steel/liquid zinc interface (dissolution of iron, heterogeneous nucleation and growth of the intermetallic phase designated Fe2 Al5 Znx ). The model has been validated using experimental data available in the literature for a classical galvanizing treatment that lasts three seconds. C 2005 Springer Science + Business Media, Inc.

Nomenclature acrystal Side of the crystals assumed to be in the form of cubes (m) Activities of iron, aluminium and zinc aFe , aAl , aZn in the supersaturated bath sat sat sat , aAl , aZn Activities of iron, aluminium and zinc aFe in the saturated bath Concentrations of iron and aluminium cFe , cAl in the bath (mol · m−3 ) ◦ liq ◦ liq cFe , cAl Concentrations of iron and aluminium in the liquid phase in equilibrium with Fe2 Al5 Znx (mol · m−3 ) meta Concentration of iron in the liquid cFe zinc in metastable equilibrium with pure iron (mol · m−3 ) comp DFe Diffusion coefficient for iron in the compound (≈ 10−14 m2 · s−1 ) [12, 29] Zn(L) Zn(L) DFe , DAl Diffusion coefficient for iron (= 9.8 × 10−10 m2 · s−1 ) [29] and aluminium in liquid zinc (m2 · s−1 ) Thickness of the interface layer (m) h diff I Nucleation rate (m−2 · s−1 ) Boltzman’s constant (1.38 × 10−23 kB J · K−1 ) Dissolution rate constant (m · s−1 ) kdiss Rate constant for the growth of kgrowth Fe2 Al5 Znx (m · s−1 )

Nucleation rate constant (knucl /η = 1031 Pa · m−2 ) [26, 27] comp Al comp Zn comp Molar weight of FexFe Mcomp xFe xFe −3 −1 (39 × 10 kg · mol ) Number of moles of compound per unit n nucl surface area contained in the nuclei (mol · m−2 ) Total number of nuclei per unit Nnuclei surface area (≈4×1014 m−2 ) [12] ∗ Mean critical radius of the hemispherrmean ical embryos (m) T Temperature (K) t Galvanizing time (s) Time when nucleation begins (s) tnucl Volume of a molecule of Vm comp Al comp Zn comp ( ≈ 1.374 × FexFe xFe xFe −29 3 m ) [12, 31] 10 comp comp xFe , xAl , Mole fractions of iron (= 0.252), alucomp minium (= 0.623) and zinc (= 0.125) xZn in the interface compound Fe2 Al5 Znx [18] xcov Surface area fraction covered by crystals Surface area fraction in direct contact xfree with liquid zinc (1− xcov ) z Space variable in the direction perpendicular to the plane of the sheet (m) β Degree of supersaturation knucl

∗Author to whom all correspondence should be addressed.

0022–2461

C

2005 Springer Science + Business Media, Inc.

2263

which can cause defects in the final coating.1 µm thick (Fig. growth is assumed to be limited by the diffusion of iron through the compound in the course of formation. involving solution of the diffusion equations for iron and aluminium in the bath. which was considered to be equal to the number of nucleation sites. One of the most commonly used coating processes is hot dip galvanizing. Introduction Steel is often coated with a layer of zinc in order to protect it against corrosion. and in particular to limit as far as possible the formation of dross. 1). frequently performed in a continuous treatment line.025 to 0. Figure 2 Layer of solid Fe2 Al5 Znx (after selective dissolution of the layer of zinc). it is necessary to be able to evaluate the variation in iron flux at the steel/zinc bath interface as a function of process parameters.PROCEEDINGS OF THE IV INTERNATIONAL CONFERENCE/HIGH TEMPERATURE CAPILLARITY δ cFe comp G∗ η ρcomp σ Zn(L)/comp Thickness of the boundary diffusion layer (≈ 20 µm) [12] Maximum difference in the concentration of iron in the intermetallic compound Fe2 Al5 Znx (≈ − 1100 mol. 6] proposed two modifications to Tang’s model: (1) the nucleation rate allows for the surface roughness of the steel and the average number of iron atoms involved in the formation of an embryo.2 wt% aluminium. with 0. 2. The boundary conditions at the steel/zinc bath interface involve the dissolution of iron according to zero order kinetics (dominant term) and growth of the intermetallic compound at the interface. Subsequently. about twenty microns in diameter. In order to facilitate the management of industrial galvanizing baths. it has been sought to develop a model for the kinetics of the galvanizing reactions. 2]. 31] Interfacial tension between interface compound and liquid zinc (J · m−2 ) Part of the iron dissolved diffuses into the bath where it contributes to the formation of Fe2 Al5 Znx precipitates. Linar` es [3] proposed the first overall model for the galvanizing process. More recently. Growth was assumed to be controlled by the diffusion of aluminium in the bath. Galvanizing reactions described in the literature The literature survey was limited to the galvanizing conditions most commonly employed (bath saturated in iron. corresponding to dissolution of iron and nucleation and growth of an intermetallic phase designated Fe2 Al5 Znx (0 < x < 1). The steel strip is immersed in a bath of molten zincaluminium alloy [1. (2) growth is controlled by the diffusion of iron through the compound in the course of formation. Toussaint et al. (2) construction of a model for the kinetics of the galvanizing reactions. at 450 to 480◦ C). Since measurements are difficult to make in industrial plants. [5. 19] Gibbs free energy of formation of a hemispherical embryo of critical size (J) Dynamic viscosity of the liquid zinc (Pa·s) comp Al comp Zn comp (4720 Density of FexFe xFe xFe −3 kg · m ) [12. together with published experimental data concerning growth of the intermetallic compound Fe2 Al5 Znx . 2. based on the results of the literature review. 1. In this case. Tang [4] considered the nucleation and growth of the Fe2 Al5 Znx intermetallic phase layer. 2264 . 2). Figure 1 Schematic representation for the galvanizing reactions. The approach employed involved the following steps: (1) a survey of existing models for the galvanizing process. covered with a 10 µm layer of Zn.16 to 0. (3) validation of the model based on the experimental data available in the literature. two reactions occur at the interface between the steel and the liquid zinc alloy (Fig. The results depend on time t and on a space variable in the z direction perpendicular to the plane of the steel sheet.1. The nucleation rate employed depends in particular on the mole fraction of aluminium at the steel/liquid zinc interface and on the number of iron atoms per unit area on the steel surface.m−3 ) [12. The liquid zinc bath thus contains 0. called dross. The final coating is therefore composed of a thin layer of solid Fe2 Al5 Znx about 0.06 wt% of unwanted iron from the steel. which are the most frequently employed. Modelling In 1993. The present study considered baths containing less than 1 wt% Al. In this model.

t ) = cAl (4) 2265 . Initial conditions At the moment when the sheet penetrates the bath. the concentrations of iron and aluminium in the boundary diffusion layer are assumed to be constant and correspond to the composition of the liquid phase in thermodynamic equilibrium with the Fe2 Al5 Znx surface dross [19].9 wt% Zn). 15]. The concentration of aluminium does not vary as long as nucleation has not begun: cFe (z ≤ δ. 5] and none of them allows for the growth kinetics of the Fe2 Al5 Znx phase. i. At the end of the galvanizing process (3 s) [8. the model enables calculation of the concentration profiles cFe and cAl as a function of time t and of a space variable in the direction z perpendicular to the plane of the sheet. The value of z lies between 0. 9. x ≈1 [18]. 8. Toussaint [6] showed that the growth of the interface layer is slower under turbulent conditions than in a laminar flow regime. for example. Using a laboratory device.16]. Construction of the model for the kinetics of the galvanizing reactions The model for the kinetics of the galvanizing reactions is based on solution of the generalized equations for the diffusion of iron and aluminium in the bath (the notation used here is defined above): Zn(L) DFe ∂ cFe ∂ 2 cFe = 2 ∂z ∂t ∂ 2 cAl ∂ cAl = ∂ z2 ∂t (1) (2) 2. forms in less than a second. since the composition of the interface layer is not accurately known. and δ the thickness of the boundary diffusion layer (Fig. 1). 10.1. the measured zinc content varies in the different studies from 4 to 28 wt% for similar galvanizing conditions (0. The simplifying assumptions for modelling the growth will be based on these observations (Section 3. 9] or moving steel samples [6. since the currently available data are insufficient to define them more accurately.PROCEEDINGS OF THE IV INTERNATIONAL CONFERENCE/HIGH TEMPERATURE CAPILLARITY In conclusion. for either stationary [6. It was therefore sought to improve and complete the kinetic laws for the reactions at the steel/liquid zinc interface.10 to 0. the abscissa of the interface between the steel and the liquid zinc. In all cases. since studies of the effect of temperature on the galvanizing reactions are still rare. 11. His experimental points for the turbulent regime are in good agreement with measurements performed on industrial samples [17] and will therefore be used for validating the model (Section 4). existing models for the galvanizing process do not take into account all the phenomena occurring at the steel surface. which will be written comp Al comp Zn comp in the equations describing nucleFexFe xAl xZn ation (Equations 6–9) and growth (Equations 11 and 14). t < tnucl ) = cAl (3) 3. a prior assumption is made concerning the diffusion process that controls the reaction rate [3–5]. The differential equations are solved using the well-known finite difference method.e.23 wt% Al. and (2) larger elongated crystals (300 to 1000 nm) between which smaller crystals about 50 nm in diameter Zn(L) DAl With the initial and boundary conditions described below. For example. 14. 14. including both laboratory materials [7–13] and industrial products [14–17]. t = 0) = cFe ◦ liq ◦ liq cAl (z ≤ δ. appear in places. 3. temperature from 450 to 470◦ C) [7. in order to identify the rate controlling step in the growth of the interface layer at short times. Boundary conditions at the boundary diffusion layer/zinc bath interface It is assumed that the composition of the liquid phase does not vary during galvanizing (steady state regime). Thus. rich in iron and aluminium. Furthermore. During the initial stages of nucleation and lateral crystal growth. The precise composition of this compound. a continuous layer of Fe2 Al5 Znx compound.12–14. in the present work it was assumed to be uniform and equal to the composition of Fe2 Al5 Znx in equilibrium with the galvanizing bath (20. composition and morphology of the interface layer. The thickness of the interface layer is generally estimated from the weight of aluminium present in the compound per unit surface area (g. which the majority of authors identify as Fe2 Al5 Znx [7–10. two do not consider the kinetics of iron dissolution [4.m−2 ). The boundary conditions at the interface between the boundary diffusion layer and the molten zinc bath can therefore be written: cFe (z = δ. It is assumed that the diffusion coefficients for iron and aluminium in liquid zinc do not vary with concentration. composed of small crystals with sizes varying from 20 to 50 nm. Growth then continues at the grain boundaries of the underlying ferrite and then in the grain centres. is given in nomenclature. 16. the analysis was essentially restricted to isothermal galvanizing conditions (the temperatures of the bath and the immersed sheet are taken to be equal). 17]. 3.9.2.2.3). t ) = cFe ◦ liq ◦ liq cAl (z = δ. These investigations reveal a thin layer of intermetallic compound. 17]. The precise composition of this interface layer is still the subject of controversy. Experimental data for the Fe2 Al5 Znx compound formed at the interface Numerous studies have been made on galvanized coatings to determine the nature. the interface layer is composed of different colonies of crystals: (1) small equiaxed crystals with sizes ranging from 50 to 300 nm. Recent experiments [12] have shed new light on the interface layer growth mechanisms.

Nucleation [23] Because of the dissolution. falling to zero when the bath is saturated in the dissolving species [20. The crystals are assumed to grow in the form of cubes (3b) and to extend laterally until they completely cover the surface (3c). so that the boundary condition at the interface between the steel and the molten zinc can be written (Fig. 3a): Zn(L) − DFe ∂ cFe ∂z z =0 = meta kdiss cFe (z = 0. The time tnucl is then given by: Nnuclei = 0 tnucl and (8) † 2 σ Zn(L)/comp Vm r∗ = kB T ln β I (t ) dt (9) To simplify the problem. t ) (5) meta is the concentration of iron in the liquid where cFe zinc in metastable equilibrium with pure iron. where G ∗ is the Gibbs free energy of formation of a hemispherical embryo of critical radius r ∗ .2). In fact. it is justified by the shape of the nucleation rate curve (Equation 7). and is given by: β= aFe sat aFe xFe comp aAl sat aAl xAl comp aZn sat aZn xZn comp (6) The determination of β involves two calculations: the composition of the bath at saturation using the equilibrium diagram [19]. the concentration of iron at the interface between the steel and the liquid zinc increases and exceeds the saturation limit with respect comp Al comp Zn comp . 2 8 π Vm σ Zn(L)/comp G = 3 (kB T ln β )2 ∗ 3 steps fully separated in time.3. t ) − cFe (z = 0. 21]. since it neglects the growth of the first nuclei formed. is the driving force for heterogeneous nucleation of the intermetallic compound on the surface of the sheet. Nucleation ceases at the instant tnucl when the total number of nuclei per unit surface area becomes equal to Nnuclei (≈4.3.2. using the model for the Gibbs free energy of formation of the ternary liquid phase [22]. and the activities of iron. In agreement with theoretical and experimental studies of nucleation in liquid metals [24–27]. Theoretical studies of the dissolution of a solid metal in a liquid metal always consider a first order reaction rate. The degree of supersaturation. and is calculated using the model for the Gibbs free energy of formation of the ternary liquid zinc-aluminium-iron phase described previously [22].1014 m−2 [12]) † . . 3c) nucleation of Fe2 Al5 Znx . it is assumed that nucleation and growth of the compound represent two distinct 2266 Nnuclei is estimated assuming that the surface is covered with a layer of crystals in the form of cubes of side 50 nm (first stage of lateral growth Section 2. aluminium and zinc in the saturated and supersaturated baths. Boundary conditions at the steel/liquid zinc interface 3. to FexFe xAl xZn designated β . 3. Before nucleation of the Fe2 Al5 Znx compound The dissolution of iron is the only reaction that occurs at the interface between the steel sheet and the galvanizing bath.PROCEEDINGS OF THE IV INTERNATIONAL CONFERENCE/HIGH TEMPERATURE CAPILLARITY 3. This assumption might appear restrictive.3.1. This approach will therefore be adopted for the dissolution of iron in liquid zinc. the expression chosen for the rate of heterogeneous nucleation is as follows: G∗ knucl exp − I = η kB T (7) Figure 3 Schematic representation for the boundary conditions at the steel/liquid zinc interface before (3a) and after (3b.

2).PROCEEDINGS OF THE IV INTERNATIONAL CONFERENCE/HIGH TEMPERATURE CAPILLARITY For the growth stage. considering the uncertainties in the diffusion coefficients. the crystals grow in the form of cubes of side acrystal and extend laterally until they completely cover the surface (acrystal is then equal to 50 nm). However.7 · 10−4 m · s−1 . as an approximation. strictly speaking. In order to express the flux per unit area of iron consumed by the formation of Fe2 Al5 Znx . the boundary condition for aluminium at the interface therefore depends only on the aluminium flux necessary to form the compound (Fig. were unable to be determined accurately. it should be based on the activities in the compound. Fig. The thickness of the interface layer is then h diff and the boundary condition for any time t is given by: Zn(L) − DFe ∂ cFe ∂z h diff =− comp DFe z =0 comp cFe − kgrowth cFe (z = 0. t ) − cAl xFe z =0 (14) comp cFe comp acrystal xcov The flux of iron diffusing in the intermetallic compound comp depends on the maximum difference cFe in the concentration of iron through Fe2 Al5 Znx [19] and on the comp diffusion coefficient of iron in this compound DFe (whose estimation allows for diffusion in the grains and in the grain boundaries). in direct contact with the liquid zinc. The total quantity of compound n comp (mol · m−2 ) present on unit area of the sheet surface can be used to determine the crystal size and hence the area fraction of surface covered as a function of time t : n comp = n nucl + kgrowth comp xFe 4.e. starting from an initial estimation [12]. t ) xfree ◦ liq − kgrowth cFe (z = 0. The expression proposed for this flux is open to criticism since. t ) − cFe (z = 0. σ justed. 3b and c). the number of moles of intermetallic compound per unit surface area contained in the nuclei). kgrowth . These parameters were adkdiss . Zn(L) − DFe From the equations used for the nucleation stage (Equation 8). σ Zn(L)/comp = 0. i. equal to the mean critical radius rmean nuclei formed (calculated conserving n nucl (mol · m−2 ). so that for any time t : Zn(L) DAl ∂ cFe ∂z z =0 = meta kdiss cFe (z = 0. the compound growth rate is very rapid. and introducing the simplifying assumptions to facilitate the calculations. Four physical parameters involved in the model Zn(L) .3. at any time t . 3b).2 wt% aluminium and saturated in iron (T = 460◦ C). t ) − cFe − DFe comp xcov (10) ∂ cAl ∂z x ◦ liq = Al comp kgrowth cFe (z = 0. Predictions of the galvanizing kinetics model The model was applied to galvanizing in a bath containing 0. by comparing the results of the calculation with Toussaint’s experimental measurements [6]. t ) − cFe ◦ liq (13) The diffusion of aluminium in the interface layer already formed is neglected. The values obZn(L) = 5.4 s of immersion (growth is then controlled by the diffusion of iron through the compound). taking into account the degree of saturation (to the authors knowledge. During growth of the compound. Growth then continues in a uniform manner perpendicular to the plane of the sheet. tained in this way are: DAl −5 −1 kdiss = 1. the boundary condition involves (1) the iron dissolution flux from the sheet surface. the initial surface area fraction covered by the ∗ )2 Nnuclei . the growth of Fe2 Al5 Znx is assumed to occur in two steps. kgrowth = 1. a first order law was chosen. Finally. and (3) the iron flux consumed by growth of the crystals (Fig. 3c). 3b and c): Before the formation of a complete interface layer. First of all.3.245 J · m−2 .1 s are probably 2267 . DAl Zn(L)/comp . × t tnucl cFe (z = 0.7 · 10 m · s . no experimental law for the growth of intermetallic compounds involving an interaction between a solid metal and a liquid metal exists). t ) − cFe dt = ◦liq ρcomp 3 Nnuclei acrystal Mcomp (11) 2 Nnuclei xcov = acrystal and xfree = 1 − xcov (12) 3.. all the hemispherical nuclei are considered to have the same ∗ of all the size. it remains acceptable. After an incubation period corresponding to the establishment of the supersaturation β . crystals is π (rmean The crystals meet in the first layer when xcov = 1 (Fig. These two growth stages correspond to two different boundary conditions at the interface for iron (Fig. The differences between the experimental points and the calculation at the shortest contact times can be the result of both causes: (1) Toussaint’s experimental measurements at 0. Growth In agreement with the observations described above (Section 2. (2) the iron flux diffusing through the crystals already formed. but slows significantly after 0.0 · 10−9 m2 · s−1 . 4 shows the weight of aluminium per unit area of intermetallic compound on the surface as a function of galvanizing time. The calculation is in good agreement with Toussaint’s experimental points [6].

J . 4. etallurgie—CIT 96 (1999) 381. The model has been validated using experimental data available in the literature for a threesecond galvanizing treatment. 20. Z H A N G . More results and a study of the sensitivity of the model to variations of four important physical paZn(L) . J . overestimated. “Handbook of Industrial Crystallization” (Butterworth-Heinemann Series in Chemical Engineering. 25. M . N . E . M . G I O R G I . R . 1998) p. . in “Zinc-Based Steel Coating Systems: Production and Performance”. 418. 2nd ed. Z . R . Drillet for their enthusiasm and help with this study. edited by A. 194. W . Private communication.82 nm).2 g · m−2 [28]). I . D U R I G H E L L O . L . Soc. Ecole Centrale Paris (1993). 1995. M . P . the nucleation rate. Universit´ e Libre de Bruxelles (1997). ´ R A N C E . M . G. V A N D E N B E R G H E and B . 17. Mater. “The Theory of Transformations in Metals and Alloys—Part 1 Equilibrium and General Kinetic Theory. J . M . E. J . Avt. M cD E V I T T . Trans. B O U T I N . Calphad 25 (2001) 461. D¨ usseldorf. 26. the average critical embryo radius. D U B O I S . M . T U R N B U L L .2 s). Germany. The model is capable of calculating the following parameters as a function of time: the concentration profiles for iron and aluminium in the boundary diffusion layer. 22. Y . P . E. 2268 Received 31 March and accepted 20 October 2004 . in Proceedings of the Fourth International Conference on Zinc and Zinc Alloy Coated Steel Sheet (Galvatech’98) (The Iron and Steel Institute of Japan. S H A R P and J . L I N A R E ´ tinu”. ` l’` etude de la formation de la 6. B . 37 (1997) 776. N . 21. P . R A N D L E . PhD thesis. ¨ NBERGER. 13. PA. 15. C H E N . M. Ecole Centrale Paris (2000). R .B R A Y . T A N G . L E P R E des m´ ecanismes r´ eactionnels de la galvanisation”. 1998) p. PhD thesis. Y . P. R . Mater. kdiss . G R E G O R Y . 39 (2004) 5803–5808. 675. The model is capable of calculating the following parameters as a function of time: the concentration profiles for iron and aluminium in the boundary diffusion layer. 5. Metals and Materials Society. M . P I M M I N G E R and L . T . Metals and Materials Society. 23. Marder (The Minerals.Y . N I C O L L E and J . S C H O in Proceedings of the Second International Conference on Zinc and Zinc Alloy Coated Steel Sheet (Galvatech’92). E . P . D U R I G H E L L O . 21 (2000) 70. Metallurg. C H A L M E R S . (Clarendon Press. Metallurg. 1975) p. M O R E A U . Sci. 8. Trans. Phys. H O N G and F . the average critical embryo radius (rmean degree of supersaturation β (= 2. 28. 9. Z . L . 374. G I O R G I . ` S and J . I . 229. PhD thesis. Forum 14 (1990) 130. 14. A . ISIJ Intern. Phase Equil. in Proceedings of the Third International Conference on Zinc and Zinc Alloy Coated Steel Sheet (Galvatech’95) (The Iron and Steel Society. BE ´ LISLE. D . C H R I S T I A N . edited by CRM (Verlag Stahleisen GmbH. Mater. 33. A . X . the nucleation rate.L . Metallur. M . 10 (1969) 473. S H A R P and J . 11. Y . V . C . M Y E R S O N and R . 2. S P A E P E N . ibid. 1992) p. V . M .PROCEEDINGS OF THE IV INTERNATIONAL CONFERENCE/HIGH TEMPERATURE CAPILLARITY References 1. H A R V E Y and P . Trans. M A R D E R . Acta Metallurgica 31 (1983) 2021. L . Universit´ e Paris XI Orsay (1996). T . 45 (2000) 191. Y . 27. R . F A D E R L . edited par F. L ’ E S P E 30A (1999) 681. GAGNE ´ R A N C E . G U I L L O T and R . A D A M . S. ´ ˆ T R E . G I N D E . G A S T . Conclusions A model has been developed to describe the kinetics of galvanizing reactions that occur at the steel/liquid zinc interface (dissolution of iron. M A Y E S . G U T T M A N N . the establishment of supersaturation and the time necessary to form a continuous layer. Mater. M E S H I I . J. 38 (1998) 985. W . G U I L L O T . R. W I N A N D and M . S P I L L . the ∗ = 0. A . “Etude des cin´ etiques des r´ eactions de galvani´ sation”. C . Sci. M E R C E R . T A N G and G . Goodwin (The Minerals.3 g · m−2 after 3 s. W . The calculation program also predicts the right order of magnitude for the weight of iron dissolved. LINAR E 29. . kgrowth . B . 4 (1973) 619. L’ESPE G O O D W I N . Metallur. S . D . J . 41. Oxford. BARIL. L O M M E L and B . 24. T A N G . Mater. ` S . T O U S S A I N T . M A S C H E K and J . 30.09 s). ´ . A . Mater.3) and the time necessary to form a continuous layer (= 0.B . A 10. M . due to the cooling time of his samples. compared to 0. albeit slightly overestimated (0. Progr. Svarka 5 (1968) 68. 5. T H O M P S O N and F . H E R T V E L D T . D E C O O M A N . S T R U T Z E N B E R G . Guttmann et Dr. (Pergamon Press. Tech. Figure 4 Weight of aluminium per unit area of interface compound on the surface as a function of galvanizing time: The calculation is in good agreement with Toussaint’s experimental points [6]. E . E . J . 1995) p. La revue de M´ ´ 12. Forum 155/156 (1994) 527. Mater. “Mod´ elisation du r´ eacteur de galvanisation en con3. “Contribution a couche d’inhibition de l’acier galvanis´ e en continu”. Sci. S . 18. Warrendale. B A R I L and G .L . C H E N . 16. PhD thesis. 1947) p. J. . P I C H I L I N G I and T . 31. P . and (2) the calculation could be underestimated because some parameters of the model are not well-known. 1993) p. N . L A R I K O V . M O R I M O T O and M . S E G E R S . G . T O U S S A I N T . 46 (1998) 1135. W . R .H U G U E S . M O E L W Y N . AIME 215 (1959) 499.L . G I O R G I . Contempor. N . T .” 2nd edn. Trans. 6 (1990) 1173. . “The Kinetics of Reactions in Solution”. N I C O L L E . Sci. E . G R E G O R Y . ISIJ Intern. “Etude 7. I . . Acknowledgments The authors are extremely grateful to Dr. in “The Physical Metallurgy of Zinc Coated Steel”. F A D E R L . heterogeneous nucleation and growth of the intermetallic phase designated Fe2 Al5 Znx ). the time tnucl (= 0. rameters involved in the model ( DAl σ Zn(L)/comp ) will be given elsewhere. A 26A (1995) 1699. P R I C E . 1993) p. 19. during which the interfacial layer is likely to grow. 190.