Materials Science and Engineering A363 (2003) 53–61
Kinetic interactions between solid iron and molten aluminium
A. Bouayad a,b , Ch. Gerometta a,∗ , A. Belkebir c , A. Ambari b
LPMI/Laboratoire Industriel de Fonderie, ENSAM CER d’Angers, 2 Bd du Ronceray, 49035 Angers Cedex, France b LPMI/EMT, ENSAM CER d’Angers, 2 Bd du Ronceray, 49035 Angers Cedex, France c Fonderies du Poitou Fonte, TEKSID Group, 86220 Ingrandes sur Vienne, France Received 15 October 2002; received in revised form 3 June 2003
Abstract In order to improve the metallurgical bond between ferrous inserts and aluminium matrix in castings, the interaction between molten aluminium and solid iron is studied by immersion tests. Morphology evolution and growth kinetics of the intermetallic layers Fe2 Al5 and FeAl3 were established as a function of time and temperature. For medium interaction time (<45 min), a diffusion regime controlled the Fe2 Al5 growth, while FeAl3 grew under a kinetic regime. The apparent activation energy was evaluated in the range of 700–900 ◦ C for the diffusional growth of the Fe2 Al5 layer. A theoretical approach based on a reaction diffusion model of simultaneous growth of the two intermetallic layers was successfully used to describe growth kinetics. The parabolic and linear rate constants of Fe2 Al5 and FeAl3 were identiﬁed. This approach rationalizes the experiments. © 2003 Elsevier B.V. All rights reserved.
Keywords: Kinetic interactions; Intermetallics; Reaction diffusion; Solid iron; Liquid aluminium
1. Introduction The incorporation of reinforcements in aluminium alloy castings is a technique increasingly used to obtain parts combining resistance and lightness [1–3]. Bimetallic pieces, which combine the good mechanical properties of a ferrous metal with the low density of aluminium alloys, are of particular interest for automotive applications. However, achieving high performance requires good bonding. A continuous metallurgical bond, made up of intermetallic layers, is ideal to guarantee the sealing and heat transfer and to ensure a perfect mechanical connection. The study of the formation and the growth of intermetallic compounds in the Fe–Al system is also important for other processing techniques, such as hot dip aluminizing, cementing, CVD and liquid metal corrosion [4–6].
Corresponding author. Tel.: +33-2-4120-7374; fax: +33-2-4120-7394. E-mail addresses: a email@example.com, firstname.lastname@example.org (A. Bouayad), email@example.com (Ch. Gerometta), firstname.lastname@example.org (A. Belkebir), email@example.com (A. Ambari). 0921-5093/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0921-5093(03)00469-6
Reaction diffusion laws govern the formation and the growth of intermetallic layers at the interface between a solid and a liquid metal. Several authors have dealt with the theoretical models that describe these phenomena [7,9]. In a strict ‘diffusional’ perspective, the growth of these intermediate phases follows parabolic kinetics. However, this ‘diffusional’ approach is not sufﬁcient to explain certain experimental observations. The consideration of a chemical reaction step allowed Dybkov  to propose more realistic models for binary systems. These models clarify the reasons for the absence of certain phases expected from the phase diagram and which are not observed experimentally. The properties of the entire intermetallic layer depend on the types of phases formed, their morphologies and thicknesses. Consequently, the knowledge of these phenomena and the associated growth mechanisms are important. The aim of this work is to study the interaction of pure solid iron with pure liquid aluminium in order to characterize the types of interfacial phases formed, their morphology and growth kinetics. The model of Dybkov  is used to quantify the growth kinetics. The Al–Fe system has been selected because of its importance in the ﬁeld of insertion in die casting. Denner et al.  and Bahadur and Mohanty  undertook studies to ensure a perfect
It is expected that the results will provide some guidance for the interpretation of more complex reactions when Fe-based alloys are in contact with Al melts of industrial purity. Eggeler et al. This tongue-like interface is a result of favorable possibilities for aluminium atoms to diffuse along the c-axis direction of the Fe2 Al5 orthorhombic structure on structural vacancies. The furnace was heated gradually and controlled by a regulator that ensured a temperature of the melt equal to the required temperature (T = 700.5 30 Co 8. 1. Shyam et al.  who observed negative deviations from the parabolic law after long reaction times. both experimentally and theoretically. contrary to Bouché et al. this interface irregularity seems to conform to a planar shape when the temperature of the aluminium melt increases. Experimental Iron samples of high purity (see Table 1) were machined to cylinders of 8-mm diameter and 7-mm height. Eggeler et al.9 – Mn 0. . Heumann and Dittrich  give a plausible explanation for the Fe/Fe2 Al5 interface irregularity. However. 1.  and Bouché et al. 800 or 900 ◦ C). The phases formed in the case of the interaction of pure liquid aluminium (or saturated with iron) with pure solid iron (or a mild steel) are FeAl3 near aluminium and Fe2 Al5 near the ferrous substrate [13.54
A. Bouayad et al. Rings and elastomer seals prevented leaks in the furnace. the iron sample was preheated to 200 ◦ C before dipping. The parabolic kinetics of growth found by Denner and Jones  and Bouché et al. The temperature of the aluminium melt was measured by a thermocouple in contact with the alumina crucible. A thermocouple placed on the specimen holder close to the substrate controlled the tem-
Table 1 Chemical analysis of iron and aluminium specimens used in this investigation Impurity (ppm) Iron Aluminium C ≤5 – Si ≤10 400 Fe Bal. 300 Cu 6. Therefore. The aluminium was melted in an alumina crucible under argon atmosphere as shown in Fig. the present work investigates this interaction for the binary model system Fe–Al. According to the same author.  showed that this tongue-like morphology cannot be explained by a preferential nucleation because no inﬂuence of ferrous substrate’s grain size was observed. then decontamination (neutralization) for 3 min with diluted nitric acid (d = 1. A small quantity (17 g) of pure aluminium (99.14]. In fact. They were pickled in an ultrasonic bath in a chemical solution made up of 50% nitric acid (5%) and 50% hydroﬂuoric acid (2%). He experimentally conﬁrmed the presence of a linear growth stage after the parabolic one (paralinear growth kinetics) of the layers formed during the interaction of this stainless steel with an aluminium melt saturated with iron.4 –
. The average surface roughness Ra was 2.35 – Mg – 4 Cr 0. Dybkov . Samples were ﬁxed to a specimen holder by their tapping.
Fig. / Materials Science and Engineering A363 (2003) 53–61
adherence in the case of hot dip aluminizing of sheets of low and medium carbon steel.  also conﬁrmed this observation. This shows that there is still a need to clarify the morphology and the growth kinetics of the intermetallic phases that form when Fe is in contact with liquid Al. The latter afﬁrm that this deviation is due to the iron enrichment in the aluminium melt by dissolution and not to the spalling of the intermetallic layer as Denner and Jones  concluded. Fixed on a stainless steel holder.  state that this morphology is not time-dependent.
2.33). pointed out the effect of the presence of a chemical reaction step on the layer growth. followed by washing. using the rotating disc technique to study the interaction of a 18Cr–10Ni stainless steel with liquid aluminium. see Table 1) was cleaned of its oxide ﬁlm in a 5% NaOH solution for 1 min before melting. in a more recent investigation. This needs conﬁrmation.5 – Ca – 30 S ≤5 – Ni 0.  and Eggeler et al. This gives a rational explanation for the negative deviation from the parabolic kinetics for long interaction time.9%. Device used for immersion tests.2 m. The interface between the ferrous substrate and the Fe2 Al5 layer is serrated.  contradicts the works of Yeremenko et al.
The morphology of layer 1 (adjacent to pure iron) consisted of a polycrystalline structure in the part close to layer 2. 3). When the temperature of molten aluminium was stabilized (rolled up resistance ensures good temperature homogeneity). The origin of the tongue-like shape remains unclear. 2). The bimetallic samples were examined by both optical and scanning electron microscopy (JEOL 6301F). 2.A.
3. Nital (2%) for pure iron and NaOH (10%) for the intermetallic phases.5%) was used for aluminium.
. independently of immersion time and melt temperature. The cylindrical sample was cut parallel to its axis (i. 4 shows polyhedric crystals peeled off from the intermetallic layer in the aluminium matrix. Morphology and types of intermetallic phases In all experiments carried out. Scanning electron micrograph showing crystals spalling (interaction time 45 min at 800 ◦ C). An X-ray energy dispersive microprobe (Link-Isis-Oxford 1994) was used. Iron enrichment in the aluminium matrix zone close to the interface with the second layer was observed (see Fig.
Fig. Indeed. Microstructure of the interface zone after etching (5% NaOH) revealing the presence of two phases (T = 900 ◦ C. These long grains seemed to grow either in transgranular or intergranular directions (Fig. the wetting surface is large and the interaction time is long. t = 10 min). Fig. Compositions of the phases were obtained by electron probe microanalysis (EPMA). Results and discussion 3. 3. t = 5 min). This layer presented also some irregularities on the side of the aluminium (see Fig. The chemical reagents used for etching depend on the phases to reveal.e.1. The second (adjacent to aluminium) was thin and of darker color because it was etched more by the metallographic reagent. The ﬁrst intermetallic layer (adjacent to iron) was thick and presented a tongue-like shape oriented towards the iron. while long monocrystals were present near the iron.
Fig. two phases were observed at the interface of pure iron with pure liquid aluminium. Microstructure of the intermetallic layer formed: morphology of the interface with iron (T = 800 ◦ C. normal to the Fe–Al interface) with a diamond disc saw in order to minimize damage at the interface. The presence of these crystals can also be explained by the decrease of the iron solubility limit in the aluminium during cooling. the sample was dipped into static molten aluminium for a ﬁxed time t (t = 1–45 min). It was then polished down to 1 m for microstructural examination. Hydroﬂuoric acid (0. Bouayad et al. 2). / Materials Science and Engineering A363 (2003) 53–61
perature. The occurrence of these crystals is more favored when the cooling rate is increased. Anisotropic diffusion (high vacancy concentration in the c-axis of the orthorhombic structure of the formed phase) is a possible explanation for this irregularity. The solid–liquid interaction was interrupted by switching off the heating system and quenching the aluminized substrate in water to room temperature (>400 K s−1 ).
6 shows the variation of the Fe2 Al5 layer mean and maximum thickness with time at 800 ◦ C. as shown in Fig.452 t (m). 5. Omitting the initial stage of interaction (t < 1 min). The evolution of the morphology of the wavy Fe/Fe2 Al5 boundary with interaction time can be characterized by the
Fig. from which a parabolic evolution of the type (dx/dt) = (k1 /x) with a constant k1 = 9.
.217 × 10−6 (m). At 800 ◦ C. Time dependence of the Fe2 Al5 intermetallic layer thickness at 800 ◦ C (maximum thickness Xmax (᭜) and mean thickness Xmean (᭡)).8 6.
Table 2 Experimental data at 800 ◦ C Time (min) Xmean (m) Xmax (m) Z (dimensionless) Y (m) 1 90.4 × 10−9 t + 2. 6. regardless of the interaction time. except in the initial stage of growth. 7). Kinetics of growth of the intermetallic layers In view of the irregularity of the interface between the intermetallic layer and the iron substrate. 3.19. With regard to the Fe √2 Al5 layer. corroborates this ‘directed’ diffusion postulate. Being linear. . The composition of the minor layer corresponded to FeAl3 .4 × 10−9 m s−1 (see Fig. This conﬁrmed that a number of these crystals had become detached from the intermetallic layer.05 × 10−11 m2 s−1 is obtained. Table 2 lists the various thicknesses of the intermetallic layers.21].6 326 0.87 6. the FeAl3 layer thickness is seen to increase linearly with time (see Fig. / Materials Science and Engineering A363 (2003) 53–61
the existence of cavities and pores in the intermetallic layer formed.5 159 1. 7).5 13 45 512 672 0.4 15 287 423 0. This indicates a constant growth rate. which were observed in the aluminium matrix.
Fig. Fe2 Al5 and FeAl3 .67 8 30 422 572 0. Xmax is ﬁtted with the curve Xmax = 13. the EDX microprobe gave a major phase composition in the existing interval of Fe2 Al5 . Bouayad et al. Fig. The intermetallic phases formed were identical to those observed by other authors [15. a linear regression of the curve after the transient stage of growth gives the equation y = 6. This curve demonstrates a parabolic law of growth for the maximum thickness Xmax and mean thickness Xmean . For the polyhedric crystals.44 20
For the FeAl3 layer. Indeed. as shown in Fig. several compositions were found such as FeAl2 .2. Schematic diagram showing the measurement method of the thickness of intermetallic layers on micrographs. the thickness quantiﬁcation of the layers required the determination of a maximum and mean layer thickness. the thickness Y of FeAl3 (layer 2) is very low compared to that of Fe2 Al5 . 4.07 8 5 159. 5. which was also observed by Bouché et al.5 257 0.56
A.5 10 207. the kinetics growth rate is then (dy/dt) = k2 = 6.
6. The Fe2 Al5 layer thickness was found to increase with temperature T according to the law of Arrhenius: Q D = D0 exp − . These distances tend to decrease. in proportion to the mean thickness. Fig. for short immersion times. The thicknesses of the layers formed were measured at 700. 800 and 900 ◦ C. signiﬁcant distances between peaks and valleys. rarely mentioned in literature. . are in agreement with the results obtained by Bouché et al. as a function of temperature. 8. this reduction in tongue size seems to lessen with time and to stabilize around a ratio Z = 0. / Materials Science and Engineering A363 (2003) 53–61
Fig. 8 displays the variation of Z at different times of interaction. This ratio allows us to follow the interface ﬂuctuations evolution proportionally to the mean thickness. spreads continuously during the growth of the layer. 7. with increasing interaction time (Fig. The interface shows.A.
The results of this study on Fe/Fe2 Al5 interface morphology. This shows that the tongue-like shape does not disappear for long interaction times and consequently is characteristic of this interface between pure iron and the Fe2 Al5 layer. Morphology evolution of Fe/Fe2 Al5 interface at 800 ◦ C: Z characterizes the wavy boundary. was studied for a ﬁxed immersion duration of 10 min. appearing from the beginning of the interaction. (1) RT
Fig. However. At the beginning of the interaction the tongue shape presents a lengthened proﬁle. The latter afﬁrm that the disturbance of this interface. 8). The intermetallic layer growth kinetics.
. Variation of the FeAl3 layer thickness with time at 800 ◦ C. Bouayad et al.
non-dimensional ratio Z = (Xmax −Xmin )/Xmoy .
2 with Xmean 74. Fig. it seems that temperature has no appreciable effect on the Fe/Fe2 Al5 interface perturbation. On the other hand. The different values found by these authors cannot only be caused by the purity degree of the metals used (low alloyed steel by Eggeler et al. This model deals with the growth of two chemical compound layers at the interface between two elementary substances. the spalling of the intermetallic layer affects the measurement of layer thickness [8. For the FeAl3 layer.16% C. / Materials Science and Engineering A363 (2003) 53–61
Fig. After an initial stage
Table 3 Values of the apparent activation energy Q obtained by several authors [8.19 for 800 ◦ C and barely reaching 1. whereas the contrary is observed by Heumann and Dittrich . Actually. D0 a frequency factor.1
. This observation conﬁrms the activation energy evaluation for Xmax and Xmean . since these authors used a low alloyed steel containing 0.  and Heumann and Dittrich  (see Table 3). ). Hence.4 m) does not change when the temperature increases from 700 to 800 ◦ C. 1.1 Eggeler et al. 3.13]. the same variation is observed concerning the growth of the intermetallic layer when the temperature increases. Irregularities in the shape of the interface Fe/Fe2 Al5 . present little change with temperature (1. at long interaction times and high temperatures. Moreover. This explains the large difference between the experimental values obtained in this study and those by Eggeler et al. contrary to the observations of Eggeler et al.19 for 700 ◦ C.20] Author Denner and Jones  Heumann and Dittrich  With Xmax Q (kJ mol−1 ) 155 58.  for Q with Xmax is greater than that obtained with Xmean . which describe growth of intermetallic layers formed when a liquid metal reacts with a solid substrate.3.8 With Xmean 76. The apparent activation energies obtained for the maximum and mean thickness of the Fe2 Al5 layer were 73.13. the thickness of this layer extends considerably (17 m) when the temperature rises to 900 ◦ C. These authors explained the difference by the temperature dependence of the ratio of the maximum to the mean thickness. 9.17%. only valid under a diffusional regime of growth.1 kJ mol−1 . One observes also that the value obtained by Eggeler et al. respectively. this may have an important inﬂuence on the apparent activation energy evaluation. characterized by the ratio Z. Our apparent activation energy values obtained for Xmax and Xmean were almost identical.
where D is the diffusion coefﬁcient.2 and 74. Denner and Jones  afﬁrm that carbon presence in iron (low alloyed steel) has a signiﬁcant inﬂuence on the apparent activation energy evaluation. The latter in-
creases with increasing carbon content up to 0. X is the maximum Xmax or mean thickness Xmean of Fe2 Al5 . 9 shows the relationships between the logarithm of (X/t1/2 ) and the reciprocal of the absolute temperature 1/T. It is noticed that FeAl3 thickness (6.58
A. These are also dependent on the immersion time at which the effect of temperature is studied. Dybkov’s model  remains most suitable to be applied in the case under consideration.  With Xmax 155 With Xmean 134 Present study with Xmax 73. This corresponds to the growth of Fe2 Al5 and FeAl3 layers at the interface between pure iron and pure liquid aluminium. Q the apparent activation energy of the process. Consequently. T the absolute temperature and R the gas constant. This relationship. Bouayad et al. .16 for 900 ◦ C). Relationship between ln(X/t1/2 ) (Xmax (᭜) and Xmean (᭡)) and 1/T for the Fe2 Al5 layer at a ﬁxed immersion time of 10 min. is indirectly veriﬁed by the evaluation of Q. Growth of the intermetallic phases: theoretical approach Among the models.
05 × 10−11 m2 s−1 . described by a parabolic equation of the type dx/dt = k1 /x. (9)
At this stage. with k1 = 9.85k0B2 . (rg/p)k0B2 rg k0B2 . we note that k1B1 k1A2 because k1A2 is the Fe2 Al5 growth rate by Fe diffusion at the interface with FeAl3 .7. k0B2 : growth rate under conditions of reaction control of the FeAl3 layer by Al diffusion at the interface where the following chemical reaction takes place: Al (diffusing) + Fe2 Al5 → 2FeAl3 . For the second assumption. That is largely satisﬁed by the model validity range. Therefore. which are not obvious initially. Then Eqs. dt (6) (7) (4) (5)
These simpliﬁcations. for the Fe2 Al5 layer where the kinetics of growth is almost parabolic. The second assumption is satisﬁed if x (q/sg)(k1A2 /k0B2 ). one can compare the data found with those of other authors (see Table 4). for which the kinetics is linear. respectively. s = 3. Using the experimental data for the Fe2 Al5 growth kinetics. hence the molar volumes are V(Fe2 Al5 ) = 6 × 10−5 m3 mol−1 and V(FeAl3 ) = 3. The ‘paralinear’ model of Dybkov is in good agreement with experimental observations of the growth kinetics of these layers except for the transient regime of growth (see Figs.6 × 10−5 m3 mol−1 . dt x For large X these two equations are reduced to k1B1 + k1A2 dx = . Hence. dt x p q k1A2 dy = k0B2 − .22]. / Materials Science and Engineering A363 (2003) 53–61
of growth (which was difﬁcult to study experimentally and after which saturation of the melt region near the interface was supposed to reach). (7)).A. 8 and 9).05 × 10−11 m2 s−1 and k1B1 k1A2 . r = 1.394 × 10−10 m2 s−1 for an exponent
. These two elements will then form Fe2 Al5 . Moreover.39 × 10−4 m. whereas Al (which diffuses more quickly) will traverse almost the same distance because of the very small thickness of FeAl3 . (2) and (3) become k1B1 + k1A2 dx = − 0. (6)) is valid only if k1B1 + k1A2 x hence x k1B1 + k1A2 . However. Ap Bq is Fe2 Al5 and Ar Bs is FeAl3 . Heumann and Dittrich  found k1 = 10−10 m2 s−1 at 700 ◦ C. p
If A is Fe. dt sg x (2)
For the FeAl3 layer.4 × 10−9 m s−1 . g = 1. we can justify the validity of assumptions of this theoretical approach (for the general assumptions refer to Dybkov’s study ). p = 2. B is Al. Bouché et al. This condition is satisﬁed because the Fe2 Al5 thickness never reached the value of 1 mm.98 . hence x 1.05 × 10−11 m2 s−1 . we ﬁnd k0B2 = 6. it is difﬁcult to compare the experimental values k1 and k2 with those available in the literature because the models used are not the same. the densities of the compounds are ρ (Fe2 Al5 ) = 4100 kg m−3 and ρ(FeAl3 ) = 3800 kg m−3 . (8)
Inserting the corresponding values gives x 0. faster diffusion of Al than Fe allows us to conclude that k1B1 must be much higher than k1A2 . as we only know the value of the sum k1B1 + k1A2 = 9. this model predicted a ‘paralinear’ stage of growth. we should verify x (q/sg)(k1A2 /k0B2 ). dt x k1A2 dy = k0B2 − 0. k1A2 can be overestimated by k1B1 /100. which decreases with interaction time. we deduce the value of k1B1 + k1A2 = 9. Thus. To ensure this. The diffusing element iron will then combine with FeAl3 to give Fe2 Al5 . one can interpret the various coefﬁcients as follows: k1A2 : rate constant of the Fe2 Al5 layer growth by Fe diffusion at the interface with FeAl3 : Fe (diffusing) + 5FeAl3 → 3Fe2 Al5 . This implies parabolic kinetics of growth for the Fe2 Al5 layer (Eq.05 × 1013 m2 s−1 . Bouayad et al. Denner and Jones  found a value of k1 = 1. Besides. by the equations: k1B1 + k1A2 rg dx = − k0B2 . The kinetic equation (Eq. the higher value achieved by k1A2 is (k1A2 )max = k1B1 /100∼ =9. dt x dy = k0B2 . The Fe2 Al5 and FeAl3 growth kinetics are described. which concerns equation (Eq. This statement is also conﬁrmed experimentally by the perturbation of the Fe/Fe2 Al5 interface. q = 5.  reported a ratio of DAl /DFe = 10−6 at 800 ◦ C in agreement with several bibliographical data mentioned. If we reconsider the deﬁnitions of the different coefﬁcients. k1B1 : rate constant of the Fe2 Al5 layer growth by Al diffusion at the interface with Fe: 2Fe + 5Al (diffusing)→Fe2 Al5 . According to Dybkov. it is sufﬁcient to have x (q/sg)((k1A2 )max / k0B2 ). will be justiﬁed hereafter. g is the ratio of molar volumes of Fe2 Al5 and FeAl3 . (6)) and linear kinetics for FeAl3 (Eq. (7)). Thus. It is more difﬁcult to do because the value of k1A2 is unknown. Thus. All the preceding studies show that the element Fe diffuses more slowly than Al [15. it is clear that iron needs to pass across the entire Fe2 Al5 layer to meet FeAl3 . k1B1 is the Fe2 Al5 growth rate by Al diffusion at the interface with Fe.0166 m.
 D.  G. The acquired knowledge on the interaction of liquid aluminium and solid iron will allow manufacturing of bimetallic pieces (high-pressure die casting of motor bedplates) under well-controlled conditions. Auer. Glasbrenner.
 M. F. hence x = 2k1 t (exponent 0. Matsuura. Smith. Stucky. for ﬁnancial and experimental support. Heo. G. in agreement with the experimental observations and the preceding studies.5. Natanzon. O. Yeremenko. 1985. V. Les Ulis. Kim.  800 99.  S.  A. Parabolic and linear kinetic constants were determined for both layers formed.9% √ x = 2 k1 t 9.B.1 kJ mol−1 for the mean thickness. Fujimura. R. Bouché et al. The theoretical approach of Dybkov. Eggeler. Wedemeyer.45 (instead of 0.I. J.V. H. R. The growth kinetics was determined experimentally at temperatures between 700 and 900 ◦ C. Kaesche.  H.
4.645 × 10−12 Present study 800 99. was successfully applied to the Fe–Al system.T. metal purity.  S.  J.N. Bouché. / Materials Science and Engineering A363 (2003) 53–61
Table 4 Comparison of kinetic constants obtained by several authors [8. Denner.H. H. Conclusion Two intermetallic layers were observed to form during the interaction between pure solid iron and pure liquid aluminium. Bouché.  K. Kim.64 × 10−11 Denner and Jones  771 99. Materials and Design 18 (1997) 285.5). This is also due to the different methods of determining k1 . variation of melt temperatures). The discrepancies observed can be explained by different experimental conditions (iron saturated melt.05 × 10−11
n = 0. M. The major phase was Fe2 Al5 and the other FeAl3 . Jones. Eggeler. Etude thermocinétique de la dissolution des métaux solides (fer et nickel) dans l’aluminium liquide. Furthermore.N.2 kJ mol−1 for the
.Y. while the FeAl3 layer growth is linear.  J. Journal of Materials Science 16 (1981) 1748. the value of k1 can change. Dybkov’s model was used to identify the Fe2 Al5 growth kinetics by taking into account a chemical reaction stage omitted in the other models. O.J. Eggeler et al. Diffusion et transport de matière dans les solides. Journal of Materials Science 21 (1986) 3087.  N. 1995.  V. Brungs.  K. .997% √ x = 2 k1 t 2. Jones. which models the simultaneous growth of two intermetallic layers at the interface of two elementary substances by taking into account both diffusion and chemical reaction processes. Valette.60
A. Surface and Coating Technology 124 (2000) 39.A.
Fe2 Al5 maximum thickness and 74. which is of industrial interest in manufacturing bimetallic pieces. Kempster. R. J. Shatynski. Marseille. Kaesche. Barbier. the Fe2 Al5 layer growth obeys a parabolic law. W. Acta Metallurgica 24 (1976) 1071. Thomas. Editions de la physique.49) 5. Denner. M. Dybkov. After a short transient period.  V. France. Bouayad et al. University of Provence. Iron and Steel International 48 (1975) 241. H. Hirth.  786 Low-alloyed steel Iron-saturated melt √ x = 2k1 t n (n = 0. H. R.P. The apparent activation energy for the Fe2 Al5 layer growth process was evaluated and compared with the results of other authors.H. Fonderie Fondeurs d’aujourd hui 168 (1997) 80. W.49. Dybkov’s theory envisaging linear kinetics was conﬁrmed by experimental measurements.78% 99.  A. Materials Science and Engineering A249 (1998) 167. A. Bahadur.20] Author Temperature (◦ C) Purity degree of iron Purity degree of aluminium Fe2 Al5 kinetic ﬁtting method for Xmax k1 (m2 s−1 ) Heumann and Dittrich  700 Pure iron Iron-saturated melt √ x = 2 k1 t 10−10 Eggeler et al.G. For the FeAl3 layer. Rapp.78% 99. For moderate interaction times the kinetics.394 × 10−10 Bouch´ e et al. Fonderie Aluminium de Cléon (TEKSID group) represented by A. is parabolic. Pan. therefore. or we ﬁt the data to an equation of the form x = ktn .  S. A. C. Smith.
Acknowledgements The authors wish to thank the industrial partner of this study.645 × 10−12 m2 s−1 at 800 ◦ C. found k1 = 5. France.G. M.5 corresponding to parabolic kinetics) at a temperature of 771 ◦ C. Journal of Nuclear Materials 257 (1998) 274. Philibert. Surface and Coating Technology 120 (1999) 112. Metals Technology (1977) 167.R. It is. Mohanty. Yoshida. Depending on whether we suppose√ a perfectly parabolic kinetics dx/dt = k1 /x. Materials Transactions JIM 36 (1995) 1170.  found the growth constant k1 = 2.14. where n is close to 0. Coulet. Zeitschrift für Metallkunde 77 (1986) 239. Dybkov.45) 1. Fukunaga. Journal of Materials Science 21 (1986) 3348. thesis. J.I.13. It was found to be 73. This important result is rarely mentioned in the literature because of the low thickness of FeAl3 . Y.D. Scripta Materialia 43 (2000) 155. Shin. for an iron-saturated aluminium melt at a temperature of 786 ◦ C.D.  G.999% 99.99% √ x = 2k1 t n (n = 0.64 × 10−11 m2 s−1 for an exponent of 0. Sasaki. Auer.
Praktische Metallographie 34 (1997) 264. Shabestari. Dybkov. / Materials Science and Engineering A363 (2003) 53–61  V. Ghomashchi.  T. S. Stein-Fechner. Dittrich. Suwas. Koyns.A.R. M. Journal of Materials Science 25 (1990) 3615. J. O. S. J.  A. Hejazi. Journal of Materials and Processing Technology 124 (2002) 345.
.  K. Bhargava. Zeitschrift fur Metallkunde 50 (10) (1959) 617. Heumann. S. Journal of Nuclear Materials 249 (1997) 33. Shahverdi. Wedemeyer. Shyam. Bouayad et al.
 H.R.I. S.