Interpretation of the role of silicon on the galvanizing reaction based on kinetics, morphology and thermodynamics

Scripta METALLURGICA et MATERIALIA

Vol. 28, pp. 1195-1200, 1993 Printed in the U.S.A.

Pergamon Press Ltd. All rights reserved

INTERPRETATION OF THE ROLE OF SILICON ON THE GALVANIZING REACTION BASED ON KINETICS, MORPHOLOGY AND THERMODYNAMICS

J. Foot, P. Perrot, G. Reumont Laboratoire de M6tallurgie Physique, URA CNRS 234 Universit~ de Lille I, F-59655 Villeneuve d'Ascq Cedex (Received February 25, 1993)

Introduction Although industrial steel obviously differs from pure iron by being an alloy, it is frequently observed that during galvanizing at 450°C the coating that results from this reaction nearly obeys the Fe-Zn binary diagram. According to this diagram (1), from steel to zinc the following succession of phases is observed: F, 81, ~ and TIZn. In this case the kinetics of growth roughly vary as ~/t and thus corresponds to diffusion. This ideal situation is completely altered, however, when silicon is added to steel and the well known "Sandelin" behaviour is observed (2). With the widespead killing of steel by silicon, the industrial consequences of the Sandelin phenomenon are considerable: the layer may be too thick or too thin, the appearance poor, corrosion resistance diminished and delamination etc.., may occur. Microscopic studies of the Sandelin phenomenon led to the following results: the "reactivity" of the steel depends on the silicon content, causing the thickness to vary with the percentage of Si, which reaches a maximum at 0.07 wt% Si (figure 1). the kinetics of growth changes with Si content from a parabolic diffusion-type relationship to a linear one (3). the morphology of the coating is characterized by a reduction of the thickness of the 81 phase when the silicon content increases and the appearance of a two-phased layer 51+FeSi, which is responsible for the thickening of the coating. This constituant has been called "diffuse A" (4). Up to now, no clear interpretation of the Sandelin peak has been proposed. However, it has been suggested that this behaviour is related to the supplementary degree of freedom corresponding to the introduction of an alloying element and that the determination of the diffusion paths could lead to a correct interpretation. Further knowledge of the Fe-Zn-Si ternary diagram was needed and this has therefore been recalculated between 450°C and 900°C (5). Another striking element to explain the Sandelin phenomenon was the identification by transmission electron microscopy of FeSi polyhedral particles in the galvanizing coating by Pelerin et al (6). The interpretation given here is based on all these former results and on a careful analysis of the evolution of the microstructure as a function of the Si content.

1195 0956-716X/93 $6.00 + .00 Copyright (c) 1993 Pergamon Press Ltd.

1196

GALVANIZING

Experimental

REACTION

Vol.

28, No.

procedure

Five industrial steels of which the silicon content varies from 0.01 wt% to 0.4 wt%, have been studied (table 1). These steels are identified as A to E with increasing silicon content. Steel A with less than 0.07 wt% Si, is a "hypo-Sandelin" steel. Steels B and C, containing about 0.07 wt% Si, are "Sandelin" steels (2). Steels D and E are characterized as "hyper-Sandelin" steels. 1 : Chemical 1ran

"--~.iiIL~I,I __.I~I___.EI___~I

: I

IoYNI

PILY,~

~ | I

I' %~i i i . ~ ~olU m ~oP~

of steels I i i i

IkIIi~l,I

!~__i I I I I IOl~I

~gI,~

Ik,II~II

I~

___lli___l) ~llOSI

IO~|:I

IIS~!

C

.083

. 10 8

.320

.026

.027

.023

.027

.003

.045

.072

.023

D

.080

.167

.503

.018

.031

.024

.031

.004

.010

.004

.009

E

.196

.367

1.42

.017

.026

.085

.024

.007

.297

.085

.027

Before being dipped in a zinc bath, industrial steels were prepared in a classical galvanizing treatment cycle. Samples (30x10x5 mm) were polished and then successively cleaned in an alkaline solution, pickled in an acid solution (HCI+H20 50%) to which a 2 wt% of a corrosion inhibitor (hexamethylenetetramine) has been added, fluxed in an ammonium chloride and zinc chloride solution and dried at 120°C. Fluxing enables steel samples to be protected before galvanizing treatment and promotes steel attack by zinc during the first seconds of reaction. Samples were then dipped at 450°C for 9 minutes in a zinc bath saturated with iron (about 0.04% Fe), and aircooled. To reveal the galvanizing layers by micrography, samples were cut and polished to 1 ~tm. Interfaces between the different layers of the coating were shown by chemical pickling in a Nital 4% solution for a few seconds. Our main means of investigating the coating were optical and scanning electron microscopy coupled with energy-dispersive spectroscopy system. R¢sult8 Scanning electron micrographs and thickness obtained after 9 minutes galvanizing at 450°C are shown in figure 1. Clearly the "reactivity" of the steels, which can be measured by the thickness of the coating observed for the same duration of immersion and at the same temperature, is strongly related to the morphology, i.e. to the shape and extension of the Fe-Zn intermetallic compounds especially 81 and ~ and to a l e s s e r extent F. For a low silicon content, which corresponds to the "hypoSandelin" steels, the morphology does n o t differ greatly from that obtained with the iron. When the silicon concentration of steel is near 0.07 wt%, the relative importance of the ~ compound dramatically increases and a thick two-phased ~+rl layer is overcoated by the rl zinc layer corresponding to the solidification of the liquid which was wetting the sample when pulled out the zinc pool. This concentration of 0.07 wt% corresponds to the Sandelin peak and the thickness of the coating, which in the present case is near 500 ktm, is roughly 10 times that obtained with pure iron. For "hyper-Sandelin" steels, when the Si concentration is higher than 0.1 wt%

i0

Vol.

28, No. i0

GALVANIZING REACTION

0,2

0,'3 ~

1197

0,#

FIG. 1. Thickness o f the coating and typical m o r p h o l o g i e s depending on Si content Zn 1% si

15

Fe FeSi FIG. 2. The ternary Fe-Zn-Si system at 450°C

1198

GALVANIZING

REACTION

Vol.

28, No. I0

(steels D and E), the morphology is characterized by a rather thick ~ layer and by a fine precipitation of FeSi inside the mixed 81+~ phases. In addition , the ~ domains present a polyhedral shape with well-developed planar interfaces; as already mentioned (6), the fine FeSi particles also display very clear planar interfaces. Usually, the ~ domains are well-developed and do not appear to interact with each other, being separated by a 81+~ "sea". The morphology described as hyper-Sandelin steels clearly differs from that obtained at the Sandelin concentration of 0.07 wt%. As shown by figure 1, this variation of the microstructure of the coating with the silicon content is also the source of changes in the kinetics of the galvanizing reaction. As a first approximation for hypo-Sandelin steels, the reaction obeys rather well a ~ t relationship and, when Sandelin and hyper-Sandelin steels are considered, an initial t variation is observed. Discussion

Although it is frequently accepted that the first intermetallic compound that appears during the Fe-Zn reaction is 4, the reasons for this are not clear. In our opinion, the first obvious cause is that the zinc bath, misleadingly described as ironsaturated, is in fact "t-saturated". Therefore, as soon as the iron concentration increases, an appreciable driving force for ~ precipitation has to be taken into account. Another factor that favours the germination and growth of the ~-phase is related to its crystallographic structure, which is much simpler than that of 81. and to the atomic bonds which compared with those in 81. are more metallic as shown by the hardnesses Hv81=340 and Hv~=ll2 under 50g (7, 8). The question to be asked now concerns the role of silicon in the germination and growth of ~-phase. Electron microprobe analysis shows that the solubility of Si in ~phase is vanishing. On the other hand, the 81 phase is likely to dissolve about 1 at% Si at 450°C and FeSi accepts about 1 at% Zn. Previous experimental data and thermodynamic calculations (5) lead us to propose the Fe-Zn-Si ternary diagram at 450°C shown in figure 2. It is worth noticing that the ~-phase cannot lead to any binary equilibrium with FeSi. The excess of silicon in the bath is likely to be reduced in two ways: (i) precipitation of FeSi particules; (ii) nucleation and growth of 81. From the preceding discussion it is now possible to propose a model for the influence of the percentage of Si on the galvanizing reaction. a) For a low silicon content, hypo-Sandelin steels and pure iron, the growth and the morphology of the coating are described by figure 3a. The silicon concentration does not greatly impede the early formation of the ~-phase in contact with the steel surface which favours a heterogeneous nucleation. When the iron concentration in the t-phase in contact with the steel increases, the conditions for the appearance of 81 are fulfilled. Later, a faint layer of F-phase is likely to be observed between o~ and 81, but the growth of F is limited by the fact that the rate of formation at the iron side is "counterbalanced" by the rate of destruction on the 81 side. All these mechanisms are driven by diffusion in the solid state and the resulting kinetics obeys a ~t-relation. b) At high silicon concentration in the substrate (near 0.07 wt%), silicon builds up in the liquid neighbouring the interface. From the phase diagram (figure 2), the liquid is no more in equilibrium with ~ but with 81 (5). However, the formation of ~-crystals is observed in t h e liquid, where the silicon content vanishes (figure 3b). In this intermediate liquid, the transport of iron is faster and and possibly governed by

Vol.

28, No.

i0

GALVANIZING

REACTION

1199

500

~"

400

J

ffJ

a)

300

C

.R 200

t~ tO ~)

1O0

% Si 0

0,0

I

!

i

0,1

0,2

0,3

a b

Fe ~

Liq

%Si

i

*/.Si F.e..

..........

°/oSi - - - 1

F~/r,

Liq Fe I~j,.. . . . . . . . . . .

,,0

Liq

%Si

° Fe [;.__~__L|q_ ,

. . . . . . .

;

/

%Si . . . . .

i ~-. . . . . . . . . . . . .

I r~

c

Liq

°

0,4

~l+

;

FeSi %Si

Fe o

:i:

~

Liq

',oS~ . . . . . B : : ~

.

.

*/.Si . . . . . .

.

.

o/oSi -

q .

~1

.

.

.

.

;

.

.

-'.

. .

Liq .

.

~51+

,51+

FeSi ;

FeSi

%Si */.Si .F_e_.

Liq

Fe ~ FIG. 3. Model of the influence of Si on the galvanizing reaction (a: hypo-Sandelin steel, b: Sandelin steel, c and d: hyper-Sandelin steel)

~;

1200

GALVANIZING

REACTION

Vol.

28, No. I0

convection; the kinetics is faster and, as long as this intermediate liquid phase exists, may vary linearly with t. In analogy with other study (9), it is also suggested that the fractal dimension of the t-rl interface changes at the Sandelin peak. When the t-layer formed by heterogeneous germination in the liquid (by comparison with ~-dross precipitation in pure zinc bath) is compact, 81 in equilibrium with the liquid grows between the substrate and the ~-layer. Thus, the reaction slows down and the excess silicon dissolves in 81. c) If the silicon concentration increases, the first stage of the process is very similar to (b), but when the liquid phase is confined between a and a continuous ~ layer, the conditions for the germination of FeSi are favoured by a higher Si supersaturation. The liquid can then solidify into a two-phased 81+FeSi mixture (figure 3c). The period of linear variation of the kinetics with t becomes shorter and the overall kinetics is slower. d) When the Si concentration continues to increase the nucleation and growth of occur further towards the liquid, and the growth of ~ crystals perpendicular to the Znsteel interface is made easier, although the growth parallel to the interface and, therefore the welding of the ~ crystals, is delayed. The resulting thickness of the coating increases in comparison with (c) (figure 3d). The morphology observed is typical of this domain of composition and shows the easy growth of the ~-phase perpendicular to the Fe-Zn interface. The mixture 81+FeSi, "diffuse •", on one side limits the growth of the ~-phase parallel to the interface and on the other is in contact with the liquid zinc, which leads to a larger concentration gradient and therefore to the coating becoming too thick. Conclusion The model that is proposed and summarized in figure 3 is based on the observation that a t Fe-Zn intermetallic compound is observed first and that the nucleation and growth of this phase occur at the steel interface when there is a low silicon content, or further into the liquid phase in the case of Sandelin and hyperSandelin steels. This work suggests many complementary experiments to determine the different kinetics related to the different stages, and the different means of accelerating or impeding the formation of the t-phase inside the liquid. In this way better control of the morphology of the coating during industrial process may be achieved. References 1. 2. 3. 4. 5. 6. 7. 8. 9.

J.Y. Dauphin, P. Perrot, U. Tchissambot, M6m. Sci. Rev. M6t., 84, 6 (1987) 329-336 R. Sandelin, Wire and wire products, 16, 1, 28-35, (1941) D. Horstmann, "Reactions between iron and molten zinc", Zinc Develop. Ass. (1978) H. Guttmann, P. Niessen, Proceedings CRM-ILZRO, Liege (1975) 198-218 P. Perrot, J.Y. Dauphin, Calphad, 12, 1, 33-40 (1988) I. Pelerin, D. Coustouradis, J. Foct, M6m. Et. Sci. Rev. M6t., 4 (1985) 191-198 J. Foct, J. Bouffette-Aryani, A. lost, G. Reumont, P. Perrot, F. Goodwin, J. Wegria, Proceedings of Galvatech'92, Amsterdam, Stahl und Eisen, 468-474 (1992) J. Foot, Scripta Metallurgica et Materialia, 28 (1993) 127-132 G. Reumont, P. Perrot, J. Foot, J. Mat. Sci. Letters, 11 (1992) 1611-1613