Effect of Microstructure on Fracture Mechanisms in Galvannealed Coatings

Transcription

1 , pp Effect of Microstructure on Fracture Mechanisms in Galvannealed Coatings A. T. ALPAS and J. INAGAKI 1) The Department of Mechanical and Materials Engineering, University of Windsor, Windsor, Ontario, N9B 3P4, Canada. 1) Materials and Processing Research Centre, NKK Corporation, Kokan-Cho, Fukuyama, Hiroshima-Prf., Japan. (Received on December 7, 1998; accepted in final form on October 19, 1999) Fracture mechanisms in galvannealed coatings have been studied by performing draw bead tests on galvannealed Ti stabilized interstitial free and Aluminum killed low carbon steel sheets and by investigating coating microstructures by scanning electron microscopy. Galvannealing treatments, on samples galvanized using an industrial hot-dip galvanizing process, were conducted at 450, 500 and 550 C for several time periods between 1 and 360 s in a laboratory induction furnace. In the coatings with low Fe content (up to 5 g/m 2 ), the amount of powdering during the draw bead test was minimal. Growth of cracks nucleated within the d 1 phase was arrested at the steel-coating interfaces where only a limited amount of decohesion occurred. A steep increase in the amount of powdering was observed in coatings with Fe contents between 6 9 g/m 2. In these coatings, cracks originating from the d 1 phase reached G G 1 d 1 phase boundaries, which provided preferential crack growth paths and thus facilitated fracture within the coating. A fracture mechanics model was proposed to account for the powdering resistance of galvannealed coatings. KEY WORDS: galvanizing; galvannealing; draw bead test; Ti stabilized interstitial free steel; aluminum killed steel; Fe Zn intermetallic phases; powdering resistance; fracture mechanisms. 1. Introduction Galvannealed low carbon steel sheets possess good corrosion resistance, 1) weldability 2) and paintability. 3) The drawback of galvannealing process is due to the Fe Zn intermetallic phases which are brittle, and which can fracture and detach from the coating during the sheet forming operations, thus reducing the coating quality and causing damage to the forming tools. Various tests have been developed in order to assess the formability response of galvannealed coatings. These include relatively simple tests such as U- bend, 4,5) 60 V bend type tests 6,7) and Luniaxial tensile tests. 8) More sophisticated tests, such as Double-Olsen, 9) cylindrical-cup 10) and draw bead tests 11 14) provide more realistic simulations of the actual stamping processes in laboratory environment. Inagaki et al. 12) performed draw bead and cylindrical cup tests on galvannealed low carbon steel sheets and found that the amount of powdering of galvannealed coatings (containing around 10% Fe) increased when the coating weight increased from 40 to 60 g/m 2. Also, for a fixed coating weight, the amount of powdering increased with increasing the Fe content of the coating from 8 to 15%. Cylindrical cup tests of Nakamori et al. 10) on galvannealed Al killed and Ti added low carbon steels gave the similar results, i.e., in the coated sheets galvannealed between 450 and 530 C, the amount of powdering was proportional to the Fe content of the coating. It was suggested that an increase in the volume fraction of delta 1 (d 1 ) phase with high Fe concentration was responsible for the deterioration of the powdering resistance. Hisamatsu 15) correlated X-ray intensity ratios of zeta (z ) and d 1 phases with the powdering resistance and suggested that powdering resistance increased with this ratio. Urai et al. 16) supported this view by proposing that the ductility of z would cause relaxation of the surface stresses and thus would improve the powdering resistance. Urai et al. 16) also showed that the powdering resistance increased when the Al content of the galvanizing bath was increased. On the other hand, Sakurai et al. 11) sought a correlation between the ratio of the gamma 1 (G 1 ) phase and the total gamma amount (i.e., the sum of G and G 1 ) and concluded that the increase in the percentage of G 1 phase in the coating improved the powdering resistance. Sakurai et al. 11) also noted that an increase in the thickness of G phase layer would deteriorate the powdering resistance. Jordan et al. 6) proposed that the powdering resistance should decrease when the thickness of the interfacial layer (presumably G ) exceeded 1 mm. Mercer, 17) in agreement with Nakamori s observations, 10,18) noted that the amount of powdering would decrease if the Fe percentage of coating exceeded 15%. This improvement in powdering resistance at high Fe levels was attributed to the increased amount of G phase. 17) Shi et al. 19) indicated that the most brittle region in the coating is the region between the G and d 1 layers. According to Lin et al. 20) on the other hand, a coating microstructure with G 1 interfacing with the steel would give rise to a maximum resistance against powdering ISIJ 172

2 Table 1. Chemical composition of Ti stabilized interstitial free and Al killed steel substrates. (mass%) C Ti Si Mn P S Al sol. N O Ti-IF AK From the above literature survey it is clear that, despite the large body of work, the microstructural basis of the fracture processes in galvannealed coatings is not well established. However some general trends emerge, namely, powdering resistance decreases with increasing the coating weight, galvannealing temperature and decreasing the Al content of the bath. The differences in the interpretation of the mechanism of powdering arise from the fact that the proposed correlations were often not supported by direct metallographic evidence. In this study the micromechanisms fracture in the galvannealed coatings have been determined by metallographic observation of fracture paths within the coating microstructure. 2. Experimental 2.1. Materials and Processing Galvannealed coatings were produced on two types of low carbon steel substrates. These were: i) a Titanium stabilized interstitial free steel (Ti-IF); ii) an Aluminum killed steel (AK). The chemical compositions of the steels are shown in Table mm thick cold rolled sheets were electrolytically degreased, annealed at 830 C and galvanized at a continuous hot dip galvanizing line (NKK Corporation, Fukuyama, Japan). The bath temperature was 460 C. The aluminum content of the bath was 0.15 mass%. The coating weight was around 55 g/m 2. The galvanized sheets were cut into rectangular plates of mm and galvannealed in an induction-heating furnace with a horizontal heating chamber in a N 2 atmosphere. Isothermal annealing treatments were carried out at 450, 500 and 550 C for several time periods between 1 and 360 s. To minimize phase transformations in the coating during the heating and cooling cycles, samples were heated to the galvannealing temperature at a rate of 40 s 1. Galvannealed plates were cooled by applying pressurized air uniformly on their surfaces. The coating weight i.e., the total mass of the coating layer was determined from the difference in mass of the galvannealed steel measured before and after dissolving the coating layers in a 3% HCl solution in water. The relative mass percentages of Al, Fe and Zn in the coating were quantified by analyzing the dissolved material using an Inductively Coupled Plasma Atomic Emission Spectroscope. The Fe content of the coating was obtained by multiplying the mass percent of Fe by the coating weight Draw Bead Tests A schematic of the draw bead test assembly is given in Fig. 1. The assembly was attached to the lower cross head of a conventional screw driven tensile testing machine. Test pieces in the form of rectangular strips of mm cut from the galvannealed samples were cleaned in benzene and weighted (W 0 ). A mineral-based lubricant was applied Fig 1. Schematic of the draw bead testing assembly. A) bead; a) die; B) load cell to measure the holding force; C) oil cylinder; c) rod; R) bead tip radius; h) bead height 4 mm ; D) sample (pulled in the direction v). on both surfaces (1 g/m 2 ) prior to the tests. The strip was forced against a bead made of a WC base sintered alloy using a hydraulic cylinder. A load cell attached to the ram in front of the hydraulic cylinder was used to measure the holding force, i.e., the contact force between the bead and coating surface. The drawing force was measured using the load cell attached to the testing machine. During the test, the sample attached to the upper grip of the testing machine, bends around the tip of the bead while the draw bead assembly moves down. Thus the specimen in addition to a uniaxial tensile stress is subjected to alternating tensile and compressive bending stresses. Shear stresses are also generated at the contact surfaces due to friction between the tip of the bead and the coating. Polyethylene films were placed on both faces of selected samples to reduce the effect of friction. To prevent tearing of the polyethylene films during the test, steel foils of 0.1 mm in thickness were placed over these films. Tests were conducted at two different load levels, namely at 200 kg and 750 kg and samples were pulled through the bead at a rate of 200 mm/min. The amount of powdering was evaluated by measuring the weight difference of the strips before (W 0 ) and after the test (W 1 ). Prior to the weight measurement following the test (W 1 ), samples were cleaned in benzene and an adhesive tape was applied to remove the powders attached to the surface. The mass loss (W 0 W 1 ) was normalized by dividing it to the surface area of the strip and this quantity is referred as mass loss from coating. The coating microstructures before and after the mechanical tests were observed by a field emission type scanning electron microscope (FE-SEM) and X-ray diffraction (XRD). The scanning electron microscope was equipped with an energy dispersive electron spectroscope (EDS). X- ray diffraction was performed using a diffractometer with CuK a source, on the top surfaces of 28 mm diameter ISIJ

3 Fig. 2. Mass loss from coating versus Fe content of the coating on Ti-IF steel substrate galvannealed at: a) 450 C; b) 500 C. DB1: P 200 kg, R 5.0 mm, with polyethylene film sheets on both surfaces; DB2: P 200 kg, R 0.5 mm, with polyethylene film sheets on both surfaces; DB3: P 750 kg, R 0.5 mm, without a polyethylene film. Fig. 3. Mass loss versus Fe content of the coating on AK steel substrate galvannealed at: a) 500 C; b) 550 C. DB1: P 200 kg, R 5.0 mm, with polyethylene films on both surfaces; DB2: P 200 kg, R 0.5 mm, with polyethylene films on both surfaces; DB3: P 750 kg, R 0.5 mm, without a polyethylene film. coupons. Phases were identified using the methods described in refrences 11,21) For cross-sectional examinations samples were polished using diamond paste and etched by immersing in 3% nital solution in alcohol. 3. Results Draw bead tests were performed under the following conditions: i) P 200 kg, R 5.0 mm, with polyethylene films on both surfaces (DB1); ii) P 200 kg, R 0.5 mm, with polyethylene films on both surfaces (DB2); iii) P 750 kg, R 0.5 mm, without the polyethylene film (DB3). Where P is the holding force and R is the radius of curvature of the bead tip. The results of draw bead tests on the Ti-IF steel samples galvannealed at 450 C and 500 C are given in Fig. 2. In coatings with the Fe content up to 5 g/m 2 (about 9 mass% Fe), the mass loss from the coating is low. A steep increase in the amount of material loss occurs in the coatings with the Fe contents between 6 and 7 g/m 2 (about mass% Fe). The amount of material loss remains high for coatings with higher Fe contents (up to 9 g/m 2 ). This relationship between the powdering resistance and the Fe content of the coating was typical and has been observed to occur in all coatings regardless of galvannealing or draw bead testing conditions. However, increasing the severity of the applied stresses by either decreasing the bead radius (compare DB1 and DB2 conditions) or increasing the magnitude of the holding force (compare DB2 and DB3) causes an increase in the amount of mass loss from coating. The results of draw bead tests for the coating on the AK steel are given in Fig. 3. It is seen that for these coatings the general features of the mass loss vs. Fe content curves are similar to those for the coatings on the Ti-IF steel. The higher the galvannealing temperature the steeper the increase is in the mass loss. Cross sectional micrographs of the galvannealed Ti-IF samples after the draw bead tests (DB2) are given in Fig. 4. The microstructures of the coatings with low Fe contents (e.g. coatings with an Fe content of 3.2 g/m 2 as in Fig. 4a), consist of an upper layer of mixture of z columnar crystals 2000 ISIJ 174

4 Fig. 4 Cross sectional views of the galvannealed Ti-IF steel after the draw bead test (Test condition: DB2). a) galvannealed at 450 C for 10 s; Fe content of coating: 3.2 g/m 2. (Arrows indicate location of cracks at the coating-steel interface.) b) galvannealed at 550 C for 8 s; Fe content of coating: 6.3 g/m 2. (Arrows indicate cracks that are in the coating and running parallel to coating-steel interface.) Fig. 5. Cross sectional views of the galvannealed AK steel after the draw bead test (Test condition: DB2). a) galvannealed at 550 C for 6 s; Fe content of coating: 5.3 g/m 2. b) galvannealed at 550 C for 20 s; Fe content of coating: 7.6 g/m 2. (Arrows indicate cracks that are in the coating and running parallel to coating-steel interface.) in the etha (h) phase and a lower layer of d 1. (The thin G layer which also exists can be resolved at high magnification micrographs, e.g. Fig. 8.). In the samples with low Fe contents, there is almost no evidence for material loss from the coating (Fig. 4a). Since about 50% of the coating consists of z and h phases with high zinc content these coatings are more ductile compared to those annealed for longer times. Through thickness cracks divide the coating into segments of approximately 20 mm in length (Fig. 4a). These cracks propagate until they reach the coating-steel interface. A decohesion process is initiated at the coatingsteel interface at the root of each through thickness crack. As seen in Fig. 4a, the typical decohesion length seldom exceeds 8 10 mm before the interfacial crack propagation has been arrested. Some crack growth within the steel substrate can also be observed. Decohesion of steel-coating interfaces does not result in the powdering of the coating. Details of the crack propagation and coating microstructures will be discussed later (Fig. 8). In the coatings with Fe contents between 6 and 9 g/m 2, the density of through thickness cracks is higher. As shown in Fig. 4b, in a coating with an Fe content of 6.3 g/m 2, the average distance between the through thickness cracks is reduced to approximately 13 mm. Cross sections of galvannealed AK steels subjected to draw bead tests are given in Fig. 5. As shown in Fig. 5a, coatings with an Fe content of 5.3 g/m 3 mainly consist of a d 1 phase, with a thin G layer at the steel interface. Through thickness cracks occur in the d 1 phase, but the amount of material removed from the coating is small. Cross-sectional examinations of coatings with high Fe contents indicate that a significant degree of fracture and fragmentation occurred inside these coatings. This is shown in Fig 5b. for a coating containing 7.6 g/m 2 Fe. A particularly important aspect of the events leading to the powdering of the coatings with high Fe contents is the propagation of cracks within the ISIJ

5 coating in a direction parallel to the steel-coating interfaces. In these coatings cracks do not normally propagate along the steel interface but at a certain distance, typically 2 3 mm away from this interface. It will be demonstrated later that in this case the fracture occurs along the G G 1 and G d 1 phase boundaries. The variation of the through thickness crack density (i.e., reciprocal of the average distance between the cracks) as a function of the Fe content of coatings is depicted in Fig. 6. In the coatings with Fe contents less than about 5 g/m 2, the crack density increases with increasing the amount of Fe. The crack density then reaches a maximum value of mm 1 beyond which no further increase is observed. Figure 7 shows the top surfaces of coatings after the draw bead tests at 200 kgf. In coatings with low Fe contents, the cracks exhibit large opening displacements and are widely spaced (Fig. 7a). These cracks run perpendicular Fig. 6. Average density of through thickness cracks (i.e., 1/distance between cracks) against the Fe content of coating. to the drawing direction and are parallel to each other. The average crack length varies between mm. It can be seen that the increase of the Fe content of the coating influences the crack morphology in two ways (Fig. 7b); i) by causing a decrease in the average distance between the cracks and ii) by forming a network of secondary cracks, consisting of finer cracks that extent across the first set of cracks. 4. Discussion 4.1. Micromechanisms of Coating Fracture Two distinct types of coating failure modes are identified and classified according to the Fe content of the coating Failure Mode in Coatings with Low Fe Content SEM observations of the as galvannealed coatings frequently showed the existence of pre-existing cracks in d 1 phase which is the product of an outburst reaction. 22) For example, the crack marked as A in Fig. 8 is thought to be such a pre-existing crack. These cracks propagate in the d 1 phase in a direction perpendicular to the coating-steel interface. Cracks at other angles may occasionally develop inside the d 1 phase (a 45 crack is marked as B in Fig. 8). The ductile phases, z and h ahead of the d 1 phase tend to retard the propagation of the cracks to the top surface of the coating (see C, Fig. 8). Nevertheless the majority of through thickness cracks reach the surface and divide the coating into segments of approximately equal length. The damage accumulation events progress with the operation of two additional mechanisms. These are i) Decohesion of steel coating interfaces (D, Fig. 8). A thin layer (0.5 mm) of discon- Fig. 7. Views (low and high magnification) of the top surfaces of galvannealed Ti-IF steel samples subjected to draw bead test (DB1). Galvannealing temperature 500 C. The drawing direction is indicated by the arrows. a) galvannealed for 2 s; Fe content of coating: 3.5 g/m 2. b) galvannealed for 20 s; Fe content f coating: 6.5 g/m ISIJ 176

6 Fig. 8. Cross sectional views of the galvannealed IF steel subjected to draw bead test (DB1, galvannealed at 450 C for 10 s, Fe content 3.2 g/m 2 ) displaying characteristic features of fracture in coating with low Fe content: (A) Cracks nucleated at G 1 z interface propagating inside the outburst d 1 phase; (B) Cracks oriented at about 45 to the coating plane are propagating inside the d 1 phase; (C) z and h phases tend to retard the propagation of the cracks; (D) Decohesion of steel G 1 interfaces; (E) Crack nucleation and growth in the steel substrate. Fig. 10. Cross sectional views of the galvannealed AK steel subjected to draw bead test (DB1, galvannealed at 550 C for 20 s, Fe content 7.6 g/m 2 ) displaying characteristic features of failure in coatings with an Fe content between 6 9 g/m 2. (a) Decohesion along the G G 1 d 1 phase boundaries. Interfacial cracks at the coating-steel interface are almost immediately stopped. (b) Removal of G 1 phase layers following the crack propagation at the G G 1 and G d 1 phase boundaries. tinuous G phase, which nucleates simultaneously with d 1, 21) is present at the interface and consequently, interfacial decohesion typically occurs along the G-ferrite phase boundaries; ii) Crack nucleation and growth in the steel substrate which commonly occurs along the ferrite grain boundaries (E, Fig. 8). Cracks at the substrate-coating interfaces could propagate only short distances and seldom reach lengths exceeding 10 mm. This type of coating failure can be regarded as a relatively ductile behaviour not only because of the retardation of the growth of through thickness cracks by zinc rich h and z phases but also, the difficulty of crack propagation at the G- steel interfaces. The type of failure described above is observed to occur in coatings containing less than 5 g/m 2 Fe regardless of the substrate composition, galvannealing temperature, draw bead testing condition and results in a good powdering resistance Failure Mode in Coatings with Higher Fe Content Coatings with Fe contents between 6 and 9 g/m 2 are mainly composed of d 1, G and G 1 phases. The G phase forms a uniform and continuous layer at the steel interface provided that its thickness exceeds mm. The volume Fig. 9. Volume fractions of G and G 1 vs. Fe amount in galvannealed Ti-IF steel. percentages of G and G 1 phases in the coatings have been determined by measuring the area fractions of these phases on the cross-sectional SEM micrographs. 23) The volume fraction of the G phase increases with the Fe content of the coating as shown in Fig. 10. G 1 nucleates at the G d 1 inter ISIJ

7 Fig. 12. A crack of length c, subject to a stress s, will grow spontaneously in the G 1 phase if the strain energy release rate due to crack growth G, exceeds the critical strain energy rate (or plain strain toughness) of this phase, G IC. Fig. 11. Schematic summary of the two main types of coating failure modes in galvannealed coatings. faces and is in the form of columns that penetrate into d 1. As shown in Fig. 9, the volume percentage of G 1 increases rapidly with the Fe content of the coating and reaches a maximum around 6 7 g/m 2. The G 1 phase then dissolves in G. 22) In the coatings with Fe contents between 6 and 9 g/m 2, cracks originating in d 1 reach the steel-coating interfaces (marked as A in Fig. 10a) but do not propagate along these interfaces. Phase boundaries between G G 1 and G d 1 are the preferential crack propagation sites (B, Fig. 10a) probably because these boundaries have a lower interfacial strength compared to that of the G-steel interface. Crack propagation along the G G 1 and G d 1 phase boundaries leads to the detachment of significant amount of material (mainly d 1 ) from the coating (Fig. 10b), which is the main cause of the powdering. In the coatings with Fe contents of 4 6 g/m 2, volume fractions of G and G 1 are small (Fig. 9). These coatings show an intermediate failure behaviour; namely, decohesion at the steel G interfaces occurs where the thickness of G layer is less than 1 mm and crack propagation at G G 1 and G d 1 boundaries takes place at locations where the G layer is relatively thick and a columnar G 1 morphology has been formed. Figure 11 summarizes the two main types of coating failure modes in relation to the microstructural and morphological features of intermetallic phases Micromechanics of Coating Failure In order to gain insight into the problem of fracture in galvannealed coatings under complex stress strain conditions generated during draw bead tests (that simulate actual forming operations) it is of value to examine the micro-mechanical conditions under which crack growth occurs in coatings subjected to uniaxial tensile and compressive stresses Crack Growth under Lateral Tension A pre-existing crack of length c (nucleated during cooling of the coated steel sheet from the galvannealing temperature to room temperature), subjected to a tensile stress s, will grow spontaneously in the d 1 phase (Fig. 12) if the strain energy release rate due to crack growth G, exceeds the critical strain energy release rate (or plane strain toughness) of this phase, G IC, G G IC...(1) where G IC is defined as 2 2 Y c G...(2) IC σ π E E is the Young s modulus of the phase. Y is the geometrical factor which can be taken approximately as 1.12 (single edge notched plate model) for a surface crack, c, such as the one in Fig. 12. Noting that s E e, the critical fracture strain e c is given as ε...(3) c G IC 2 Y Eπc Equation (3) gives the critical strain under which a pre-existing crack of length c will propagate in the d 1 phase. Since the thickness of the galvannealed coating is about mm and the length of pre-existing cracks in d 1 are of the order of 2 4 mm (Fig. 8), it can be assumed that c 0.2t. Consequently, an expression relating the fracture strain, e c, to the thickness of d 1 layer, t, can be obtained IC ε...(4) c G Et In a more general form this expression can be rewritten as a function of coating thickness as GIC ε c...(5) 2 Y Ef() t Equation (5) applies to all brittle intermetallic phases and shows that intermetallic phases such as z, h with lower Fe contents and hence higher toughnesses will provide higher resistance to crack growth than d 1. Equation (5) also shows that higher strains are needed for the propagation of cracks in thinner intermetallic phase layers such as G Crack Growth under Lateral Compression Powdering as a result of decohesion along the steel-coating interface or along G G 1 d 1 phase boundaries is more 2000 ISIJ 178

8 Fig. 13. Two possible ways of coating failure under lateral compression. (a) buckling and (b) wedging of the coating depends on the relative strength of the substrate-coating boundary with respect to the intrinsic strength of the coating itself. In buckling the strength of the interface is less than the coating itself so that decohesion of the substrate-coating precedes through thickness cracking. Wedging occurs by the formation of through thickness cracks (normal or 45 to the interface) followed by the decohesion of the interface. likely to occur during the compression cycle of the draw bead test. Assuming that a crack of size 2c already exists at the interface and that the size of the crack is greater than the thickness of coating, i.e., 2c t, it can be shown that 24) the critical fracture strain under compression is IC ε...(6) c G Et Theoretically there are two possible ways by which coatings can fail under lateral compression. 25) These are the buckling (Fig. 13a) and the wedging (Fig 13b) types of failures. Buckling failure occurs if the strength of the interface is less than that of the coating itself so that decohesion of the substrate-coating interface precedes the formation of the through thickness cracks. Current metallographic observations suggest that fracture in the galvannealed coatings occurs by wedging i.e., by propagation of through thickness cracks in d 1 followed by the decohesion of the interfaces between the G G 1 d 1 phases, provided that the imposed compressive strain exceeds the critical strain given by the Eq. (6). The material properties which need to be known to compute the fracture strain are E and G IC which are difficult to determine experimentally for the individual phases Origin of Cracks in d 1 and along G G 1, G 1 d 1 Boundaries When the coated steel is cooled from the galvannealing temperature to the room temperature i.e., in the temperature range of DT an elastic mismatch strain (De DTDa) is generated. Because the thermal expansion coefficient (Da) of d 1 phase is two times larger than that of the ferrite 26) and the d 1 phase is inherently brittle, the tensile residual stresses are expected to be present in the d 1. The residual stresses are at least partially relieved by the formation of internal cracks. The pre-existing cracks are generally oriented in a direction perpendicular to the steel interface (Fig. 8), which suggests that the basal planes of the d 1 crystals (with a hexagonal close packed crystal structure) would be the possible crack initiation and growth sites because these planes are normally oriented at 90 to the steel interface. 27) On the other hand, formation of a new morphological features, such as a wavy interface morphology 28) or a new phase with a columnar structure (i.e., G 1 phase) may result in the modification of the local residual stresses as discussed below. Figure 14 a shows the planar growth of a G interface at the expense of G 1 phase. Because the direction Fe Zn interdiffusion path is perpendicular to the G d 1 interface, the displacement of uniformly growing G phase should accommodate strains accompanying solid state phase transformations. Consequently no elastic mismatch strain due to phase growth could be generated. On the other hand, as illustrated in Fig. 14b, the formation of G 1 phase with a columnar morphology at the planar G d 1 interface is accompanied with the development of inhomogeneous residual stresses. Assuming that the growth of columnar crystals of G 1 in d 1 would exert compressive stresses on the portions of d 1 phase remaining between the G 1 columns, the distribution of local residual stresses becomes modified as shown in Fig. 14c. The inhomogeneous distribution of the residual stresses at both sides of the G G 1 and G d 1 boundaries is expected to generate interfacial shear stresses along these boundaries in the vicinity of a through thickness crack perpendicular to this interface. 29) Evans and Cannon 30) have shown that the onset of the micro-cracking in oxide layers is governed by the length of the geometric discontinuity such as the grain size of metal adjacent to the oxide interface where a stress concentration exists. In the galvannealed coatings, the stress intensity at a defect formed at the intersection of a through thickness crack and the longitudinal phase boundary will be governed ISIJ

9 t Γ EG C σ 2...(7b) where E is the Young s modulus of G, G IC is the average toughness (critical strain energy release rate) of the interface, s is the applied stress. The current metallographic observations (e.g. see Fig. 10) suggest that the critical thickness is of G phase that satisfies the interfacial fracture condition is about mm. Fig. 14. Model illustrating events leading to decohesion of phase boundaries in galvannealed coatings. (a) The direction Fe Zn interdiffusion path is perpendicular to the interface. The diffusion process occurs unidirectionally, where upon rigid and uniform displacement of growing G phase accommodates the strain. Consequently no strain is generated as a result of the phase growth. (b) G 1 is located over the planar G interface and grows into the d 1 by forming a columnar morphology. The resulting change in the distribution of local residual stresses are schematically given in (c). (c) The inhomogeneous distribution of the residual stresses causes generation of interfacial shear stresses along the G G 1 and G d 1 phase boundaries near a preexisting through thickness crack perpendicular to this interface. by the thickness of the underlying G phase layer, t G, (see Eqs. (2) (4). No interfacial crack growth, will occur if t Γ EG C σ 2...(7a) However, crack growth along the G G 1 and G d 1 boundaries will occur if 5. Conclusions (1) During the draw bead tests, the mass loss from the galvannealed coatings on Ti stabilized interstitial free and aluminum killed sheet steels increases with increasing the Fe content of the coating. At low Fe contents up to about 5 g/m 2 the coating loss is minimal. A steep increase in the amount of material loss is observed in coatings with Fe contents between 6 and 9 g/m 2. The higher the galvannealing temperature and applied stress the steeper the increase is in the powdering amount. (2) Two distinct types of coating failure modes are identified: i) In coatings with low Fe contents ( 5 g/m 2 ) failure process involves crack propagation within d 1 and then a limited amount of decohesion along the steel-coating (G ) interfaces or along the ferrite grain boundaries of substrates. Only a limited amount powdering occurs as a result of this process. ii) In coatings with Fe contents between 6 and 9 g/m 2, cracks originating from d 1 propagate along the G G 1 and G d 1 phase boundaries resulting in a significant amount of material loss. (3) A coating stress strain analysis which considers that formation of G 1 crystals with a columnar morphology leads to the modification of local residual stresses has been proposed to account for the possible mechanism of crack formation along the G G 1 and G d 1 phase boundaries. (4) A coating microstructure consisting of a d 1 phase as the main constituent with a z layer on the top surface provides a good fracture resistance. A thin G layer ( 1 mm) is tolerable. Galvannealed Ti stabilized interstitial free, and aluminum killed sheet steels with an average Fe content of 4 6 g/m 2 satisfy these requirements consequently posses good press formability characteristics. Acknowledgement A.T.A. gratefully acknowledges the research scholarship and financial support provided by the NKK Corporation. REFERENCES 1) H. H. Lee and D. Hiam: Corrosion, 45 (1989), ) A. J. C Burghardt, A. van der Heiden, E. B. van Peristein and J. P. Schoen: Proc. of 2nd Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (GALVATECH 92), CRM, Amsterdam, (1992), ) T. Irie: Zinc Based Steel Coating Systems: Metallurgy and Performance. ed. by D. K. Matlock and G. Krauss, TMS, Warrendale PA (1990), ) M. Arimura, M. Urai and H. Sakai: Proc. of 2nd Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (GALVATECH 92), CRM, Amsterdam, (1992), ) M. Arimura, M. Urai., J. Iwaya and M. Iwai: The Use and Manufacture of Zinc and Zinc Alloy Coated Sheet Steel Products Into the 21st Century (GALVATECH 95), ISS, Warrendale PA, (1995), ISIJ 180

10 6) C. E. Jordan, K. M. Goggins and A. R. Marder: Metall. Trans., 25A (1994), ) V. Jagannathan: Proc. of 2nd. Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (GALVATECH 92), CRM, Amsterdam, (1992), ) E. Gallo and D. K. Matlock: The Use and Manufacture of Zinc and Zinc Alloy Coated Sheet Steel Products into the 21st Century (GAL- VATECH 95), ISS, Warrendale, PA, (1995), ) P. Martin, P. Handford, R. Packwood, L. Dignard L. and V. Moore: Proc. of 2nd. Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (GALVATECH 92), CRM, Amsterdam, (1992), ) T. Nakamori. and A. Shibuya: in Corrosion Resistant Automotive Sheet Steels, ASM, Metals Park, (1988), ) M. Sakurai, L. Zhang, Y. Tajiri and T. Kondo: SAE Technical Paper Series, paper no , Warrendale, (1992). 12) J. Inagaki, S. Nakamura, M. Yoshida and A. Nishimoto: SAE Tecnical Paper Series , Warrendale, (1989). 13) H. D. Nine: J. Appl. Metalwork., 2 (1982), ) H. D. Nine: in Zinc Based Steel Coating Systems:Metallurgy and Performance.ed by G. Krauss and D.K. Matlock, TMS, (1990), ) Y. Hisamatsu: Proc. of 1st. Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (GALVATECH 89), ISIJ, Tokyo, (1989), 3. 16) M. Urai, M. Terada, M. Yamaguchi and S. Nomura: Proc. of 1st Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (GALVATECH 89), ISIJ, Tokyo, (1989), ) P. D. Mercer: Proc. of 2nd Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (GALVATECH 92), CRM, Amsterdam, (1992), ) T. Nakamori: CAMP-ISIJ, 2 (1989), ) M. F. Shi, G. M. Smith, M. Moore. and D. J. Mueleman: Zinc Based Steel Coating Systems: Metallurgy and Performance, ed by D. K. Matlock and G. Krauss, TMS, Warrendale, PA, (1990), ) C. S. Lin and M. Meshii: The Use and Manufacture of Zinc and Zinc Alloy Coated Sheet Steel Products into the 21st Century (GALVAT- ECH 95), ISS, Warrendale, PA, (1995), ) J. Inagaki, M. Morita and M. Sagiyama: Surf. Eng., 7 (1991), ) J. Inagaki, M. Sakurai, M. and T. Watanabe: ISIJ Int., 35 (1995), ) A. T. Alpas and J. Inagaki: Proc. of 4th. Int. Conf. on Zinc and Zinc Alloy Coated Steel Sheet (GALVATECH 98), ISIJ, Tokyo, (1998), ) H. E. Evans: Mat. Sci. Technol., 4 (1988), ) M. S. Hu and A. G. Evans: Acta Metall., 37 (1989), ) A. Iost and J. Foct: J. Mater. Sci. Lett., 12 (1993), ) V. Rangarajaran, C. C. Cheng and L. L. Franks: Surf. Coat. Technol., 56 (1993), ) C. H. Hsueh and A. G. Evans: J. Appl. Phys., 54 (1983), ) A. G. Evans, G. B. Crumley and R. E. Demaray: Oxid. Met., 20 (1983), ) A. G. Evans and R. M. Cannon: Mater. Sci. Forum, 43 (1989), ISIJ