- Email: [email protected]
aluminium–silicon interfaces by pushout
ARTICLE IN PRESS
International Journal of Adhesion & Adhesives 27 (2007) 417–421 www.elsevier.com/locate/ijadhadh
Mechanical testing of steel/aluminium–silicon interfaces by pushout O. Dezellusa,, B. Digonneta, M. Sacerdote-Peronneta, F. Bosseleta, D. Roubyb, J.C. Vialaa a
Universite´ Claude Bernard Lyon 1, LMI—UMR CNRS No 5615, 43 Bd du 11 novembre 1918, 69622 Villeurbanne Cedex, France b INSA-Lyon, GEMPPM—UMR CNRS No 5510, 20 av. Albert Einstein, 69621 Villeurbanne Cedex, France Available online 25 October 2006
Abstract The functionality of structural light alloy castings can be improved by inserting into them, upon moulding, local iron base reinforcements. To acquire a better knowledge of such bimetallic assemblies, samples were prepared by immersing a mild steel bar (5 mm in diameter) in aluminium base Al–Si alloy melts held at 730 1C. After melt solidification, the bimetallic samples were cut into 5 mm thick slices and pushout testing was performed on these slices. Characterization of damaging corresponding to different load level before complete debonding allows the determination of the failure mode. Crack initiation occurs at the specimen bottom face in the intermetallic reaction layer, important damage occurs before complete debonding and no brittle failure is observed. The results highlight the necessity of analysing pushout tests with a more integrated approach taking into account shear stress distribution along the interface such as interfacial crack growth. r 2006 Elsevier Ltd. All rights reserved. Keywords: Aluminium and alloys; Titanium and alloys; Adhesion by chemical bonding; Push-out tests
1. Introduction There is major interest (weight saving and low-cost production) in replacing cast iron and steel automotive components by light weight aluminium castings to improve vehicle performance and efficiency [1]. However, a key problem is to form a sound bond between the steel insert and the aluminium casting alloy. Indeed, when using conventional die casting the insert is simply embedded in the light alloy after its solidification. Hot dipping process is a way to produce a sound metallurgical bond at the interface [2,3]. Despite the usefulness of bonding steel to aluminium alloys, some difficulties may arise from the possible development of brittle intermetallic compounds at the steel/aluminium interface [4]. Moreover, little work has been published on the mechanical properties of such interfaces [5]. This paper is an attempt to link the chemistry of the reaction zone at steel/aluminium silicon alloys interface with mechanical behaviour. Therefore, special care was Corresponding author. Fax: +33 4 72 44 06 18.
E-mail address: [email protected] (O. Dezellus). 0143-7496/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijadhadh.2006.09.003
taken to control and characterize (before and after loading) the metal/metal interfaces. As no ASTM test exists for such assemblies, pushout testing was chosen to investigate their mechanical strength. Combining pushout test results with the characterization of the interface zone before and after loading has led us to propose a failure mode for the steel/light weight aluminium alloy assembly. Mechanical testing of joined materials being a hot topic in many fields of materials science, the relevance of pushout test to characterize bimetallic assemblies is also discussed. 2. Experimental procedure Bimetallic samples were processed from an AS-13 foundry alloy (composition in wt%: 12.6 Si, 0.42 Fe, o0.02 Cu, Mg, Mn, Ni, Ti, Zn) and from inserts rods made of mild steel (composition in wt%: 0.2 C, 0.85 Mn, 0.4 Si, 0.045 P, 0.045 S and Fe balance). The steel rods (5 mm in diameter) all received the same surface preparation by mechanical abrasion leading to a mean surface roughness of 3 mm. Moreover, just before the insertion process, steel and aluminium alloy pieces were degreased in an ultrasonic bath of dichloromethane. Bimetallic sample
ARTICLE IN PRESS O. Dezellus et al. / International Journal of Adhesion & Adhesives 27 (2007) 417–421
418
Thermocouple
Thermocouple
Steel insert
Steel insert Al-13 at % Simatrix half saturated in iron
AS 13 matrix
furnace
Induction turn
transfer after 10 mindipping at 730°C aluminizing bath
40 mm
(a)
insertion bath (b)
Fig. 1. Manufacturing process of bimetallic sample: (a) aluminizing step, (b) transfer to the insertion bath.
were produced in two successive steps: aluminizing and insertion as shown in Fig. 1. 2.1. Steel rod aluminizing The first step of the sample manufacturing process consists of dipping (under air) mild steel rods in a cylindrical crucible containing an Al–Si mixture based on a commercial AS-13 (Al–13 wt%Si) alloy. Pure silicon chips (electronic grade purity 99.9995 wt%) were added to the AS-13 liquid bath in order to adjust its silicon content at 13at%. Iron was then added at a level corresponding to half saturation (5.5 wt% at 770 1C) to favour the growth of a uniform reaction layer without excessive iron dissolution, as reported in [6] (final composition of the melt in wt%: 13.46 Si, 5.5 Fe, o0.02 Cu, Mg, Mn, Ni, Ti, Zn). In the following this first liquid will be named aluminizing bath. Dipping temperature was measured with a precision better than 70.2 1C by plunging in the melt a K-type (Ni/Cr) thermocouple. Note that with a bath heated at 770 1C, immersion of the insert led to a 30–40 1C decrease in temperature. Therefore, interface reactions in this first step are considered to develop at a temperature of 73575 1C. 2.2. Transfer to the insertion bath After 10 min dipping at 73575 1C, the hot aluminized insert was pulled out of the bath and immediately transferred in a new crucible, heated at 620 1C by RF coupling and containing a commercial AS-13 alloy to produce the final reinforced bimetallic casting. The transfer duration from one bath to the other was such that the AS13 alloy coating the steel rod had no time to solidify. The bottom of the crucible containing the insert bath was machined and a device put on its top so that the steel insert was perfectly vertical. In the following this liquid will be named insertion bath. The RF power supply was turned off and the aluminium alloy was allowed to cool and solidify
around the insert 20 s after plunging the aluminized steel rod in the second bath. The temperature of the AS-13 melt in this second bath was also measured with a K-type thermocouple. In that case, the temperature did not notably change when the aluminized insert was dipped. After solidification, the bimetallic samples were sawn with a diamond coated wire, into slices with a thickness from 3 to 8 mm, the cross-section of these slices being perpendicular to the rod axis. End sections of each cylinder were diamond polished to a finish better than 1 mm for examination by optical microscopy (OM), scanning electron microscopy (SEM) and electron probe microanalysis (EPMA). Other slices were used for pushout testing as shown in Fig. 2. The tests were performed by using an INSTRON 1195 testing frame equipped with a compression load cell of 100 kN capacity. The insert was pushed by means of a flat-bottomed STUB steel cylinder (ball bearing steel) with a diameter of 4 mm, i.e. 1 mm less than the steel insert. The cross-head displacement rate was of 0.2 mm min1. The relative displacement of the indenter compared to the AS-13 matrix was measured by using a modified extensometer (see Fig. 2). This experimental equipment makes it possible to remove the influence of the machine compliance on the displacement measurements, therefore, only the compliance of the indenter remains. The tested slice is placed on a flat supporting surface, the insert being centred on a drilled hole of 6 mm diameter (Fig. 2). 3. Interface chemistry Examination by optical micrography of the bimetallic samples end sections (labelled zero) indicates that the interfacial reaction zone, formed after the complete process of insertion, consists of two different intermetallic layers (see Fig. 3). EPMA characterization indicates the following phase sequence from steel rod to aluminium alloy matrix: (i) a 3–4 mm thick Z-Al5Fe2(Si) layer and, (ii) a 9–10 mm
ARTICLE IN PRESS O. Dezellus et al. / International Journal of Adhesion & Adhesives 27 (2007) 417–421
419
Mechanical load Steel insert Screw AS 13 matrix
8 7 6 5 4 3 2 1 0
M24-3: loaded until completede bonding
Extenso meter
Steel punch specimen Top Face specimen Back Face
M24-0: characterization of interfacial reaction zone
Load Cell
Fig. 2. Bimetallic sample, slices references and principle of the pushout test.
Fig. 3. Optical micrograph of an end section of bimetallic sample showing the reaction zone.
thick t6-Al9Fe2Si2 layer. Note that the crystal nature of these intermetallic phases was also assessed by X-ray diffraction and that the thicknesses of the intermetallic layers were found constant all along the interface [6]. According to Pontevichi et al. [4,6], the phase in equilibrium with the liquid alloy Al–13 at%Si at 700–800 1C is the ternary intermetallic compound g (Al3FeSi). On the other hand, same author reports that, in the Fe–AS-13 system, the solid in equilibrium with an AS-13 liquid alloy between 650 and 576 1C is b-Al5FeSi also designated as t6-Al9Fe2Si2. Therefore, the manufacturing procedure used to produce bimetallic assemblies in the present work is characterized by the following reaction scheme: (i) formation of the sequence Fe/Z-Al5Fe2(Si)/ g-Al3FeSi/L in the aluminizing bath at 735 1C (at that temperature, the iron content is between half saturation and saturation). (ii) once transferred in the insertion bath at 620 1C, the aluminized steel rod is in contact with a liquid, which is no more in equilibrium with g-Al3FeSi but with b-Al5FeSi. Moreover, this liquid is iron free.
Therefore, g-Al3FeSi and a part of Z-Al5Fe2(Si) produced at 735 1C during the first step are dissolved during the second step upon dipping at 620 1C and a new layer of b-Al5FeSi forms upon cooling between 620 and 576 1C. The final reaction layer sequence is then Fe/Z-Al5Fe2(Si)/ b-Al5FeSi/Al–Si. 4. Push-out tests More than 150 test pieces with different thickness were submitted to pushout testing. The typical load–displacement response for an AS-13 matrix reinforced with a mild steel insert in the case of a 4 mm thick slice is reported in Fig. 4. Zero displacement ( ¼ 0) corresponds to the first contact of the indenter with the steel insert, i.e. to the rise of the load measurement (F). Mechanical fitting being quasistatic, the first loading stage (from e ¼ 0 to 0.05 mm) consists in removing misalignments and is therefore characterized by a gentle increase of the load. Next a quasi-linear response with a steep slope is observed until
ARTICLE IN PRESS O. Dezellus et al. / International Journal of Adhesion & Adhesives 27 (2007) 417–421
420
800
Fmax
700 Load (daN)
600 500
FL
400 300 200 100 0 0
0.1
0.2
0.3 0.4 0.5 0.6 Displacement (mm)
0.7
0.8
0.9
Fig. 4. Typical load–displacement curve for an AS-13 matrix reinforced with a mild steel insert in the case of a 4 mm thick slice.
Table 1 Maximum load and load corresponding to the beginning of linear deviation obtained for 5 mm thick slices belonging to different bimetallic assemblies Sample
Fmax (Da N)
FL (Da N)
FL/Fmax
M20-3 M24-3 M25-3 M26-2
749 738 718 672
380 394 405 380
0,51 0,53 0,56 0,57
Average value Standard deviation
719 29
390 10
0,54 0,02
the load has reached a limit value FL ¼ 405 daN. Afterwards, from F ¼ FL to the maximum value Fmax ¼ 718 daN (emax ¼ 0.81 mm), deviation to the linearity occurs. Once the maximum value is attained, the load is observed to decrease gently (no load drop is observed). Table 1 summarizes the main results (maximum load and load corresponding to the beginning of linear deviation) obtained for 5 mm thick slices belonging to different bimetallic assemblies. The results appear to be quite reproducible and it is interesting to note that deviation to linearity occurs with a good accuracy at 50% of the maximum load. After mechanical loading, slices were embedded in resin and sawn with a diamond coated wire to prepare a polished vertical section, i.e. parallel to the axis of the insert. Optical micrograph of such a section is shown in Fig. 5 in the case of the M24-3 sample. Various signs characterize the damage that led to failure of the bimetallic assembly. First, because of indenter misalignment, displacement of the insert and damage are not fully symmetric: on the left side the steel rod has been pushed of 350 mm compared to the matrix (this shift is observed on the top and bottom faces of the assembly—see Fig. 5a), whereas on the other side, steel rod and matrix seem to remain interdependent all along the interface (except a 200 mm interfacial crack initiated on the top
Fig. 5. Optical micrograph of the sample M24-3 after mechanical loading: (a) top left side, (b) top right side, (c) bottom left side, (d) bottom right side.
face—see Fig. 5b) and are both deformed by the load (see Fig. 5d). On the left side two cracks are observed: the first one was initiated at the upper face of the assembly and propagated in the AS-13 matrix from top to bottom (see Fig. 5a). The second one in the lower part of the assembly is situated at the interface, in the intermetallic reaction zone (see Fig. 5c). Moreover, all along the interface, many transverse cracks (perpendicular to vertical symmetrical axis) are observed in the intermetallic reaction zone. Finally, the preferred orientation of the silicon crystals in the AS-13 matrix in the vicinity of the interfacial zone clearly indicates the occurrence of an important plastic flow of the AS-13 alloy (see Fig. 5d). It is obvious that interfacial cracking, matrix cracking and plastic flow did not occur simultaneously during loading of the assembly. 5. Failure mode To better understand the scenario of failure it is necessary to establish the temporal sequence of damaging. Therefore a new series of pushout tests were interrupted at different load level. For that purpose, 5–7 slices with a thickness of 5 mm were sawn in a bimetallic assembly. The
ARTICLE IN PRESS O. Dezellus et al. / International Journal of Adhesion & Adhesives 27 (2007) 417–421
421
AS-13 bimetallic assemblies tested by pushout is not brittle and damaging is allowed. Three different damage mechanisms are observed: (i) crack propagation in the intermetallic reaction layer, (ii) crack deviation towards the AS-13 matrix, (iii) and plastic flow of the matrix (see Figs. 4 and 5). Determination and understanding of the spatial and temporal sequence of damaging require further work and more information, especially concerning the shear stress distribution and its evolution with time along the interface. 6. Conclusion
Fig. 6. Optical micrograph after pushout tests interrupted for 50% of the maximum load reference—initiation of interfacial cracking at the specimen bottom face.
first slice corresponding to the bottom of the assembly was labelled 0 and used for characterization of the reaction zone. The other slices were increasingly numbered from bottom to top. Number 3, corresponding to the centre of the assembly was usually loaded until failure occurred at a maximum load noted Fmax-ref. Next the other slices were loaded, but pushout test was interrupted when the applied load reached a fraction of Fmax-ref increasing from 10% to 90%. Afterwards, the slices were embedded in a resin, sawn with a diamond coated wire and polished to a finish better than 1 mm to characterize the damage along a vertical section, i.e. parallel to the axis of the insert. Optical characterizations of samples after interrupted mechanical loading indicate that damaging of the specimen does not start in the vicinity of the indenter but systematically at the specimen bottom face by cracking in the intermetallic reaction layer. Such cracking occurs for an applied load corresponding to 50% of the reference Fmax-ref (see Fig. 6). This behaviour is in disagreement with classical analysis of pushout test, which is based on the assumption that, upon loading, debond crack propagates from specimen topface [7]. However, some authors relate the occurrence of bottom face debonding, in some cases for metal matrix composites (MMCs). [8–10] In the present study, bottom face debonding could be due to bending effect on the bottom face [11] or to thin slice geometry (thickness limited to less than 20 insert radii) [12]. Next, no brittle failure due to crack propagation (characterized by an abrupt load drop on the load versus displacement curve) is observed (Fig. 4). Therefore, despite the intermetallic reaction layer, the failure mode of steel/
Reinforcement of AS-13 matrix by steel insert has been performed with an experimental procedure, allowing the control of the interfacial reaction layer. Pushout tests until complete debonding associated with interrupted tests and metallographical characterization, allow us to conclude that for steel/AS-13 bimetallic assembly crack initiation occurs at the specimen bottom face and in the intermetallic reaction layer. Next, no brittle failure is observed and the bimetallic assembly is able to support important damage (crack propagation, crack deviation, plastic flow of the matrix) before failure. The results highlight the necessity of analysing pushout tests with a more integrated approach taking into account shear stress distribution along the interface such as interfacial crack growth. Fracture energies (Gi ) for Fe/ AS-13 interfaces are not yet determined, but crack deviation indicates that Gi exceeds the fracture energy of the AS-13 matrix. Aluminium alloys matrix is too soft compared to the toughness of the interfacial reaction zone to easily determine interface properties by pushout testing. References [1] Clyne TW, Withers PJ. An introduction to metal matrix composites. Cambridge: Cambridge University Press; 1993. [2] Barbeau F. PhD thesis, University Claude Bernard Lyon 1, Lyon, 1999. [3] Stucky M. Fonderies–Fondeur d’aujourd’hui 1997;168:80. [4] Pontevichi S, Bosselet F, Barbeau F, Peronnet M, Viala JC. J Phase Equilib Diffus 2004;25(6):528. [5] Durrant G, Gallernault B, Cantor B. J Mater Sci 1996;31(3):589. [6] Pontevichi S. PhD thesis, University Claude Bernard Lyon 1,Lyon, 2004. [7] Kerans RJ. J Am Ceram Soc 1991;74(7):1585. [8] Ghosn LJ, Eldridge JI, Kantzos P. Acta Mater Metall 1994;42(11): 3895. [9] Koss DA, Hellmann JR, Kallas MN. J Met 1993;46(3):34. [10] Majumdar BS, Miracle DB. Key Eng Mater 1996;116–117:153. [11] Kallas MN, Koss DA, Hahn HT, Hellmann JR. J Mater Sci 1992;27(14):3821. [12] Galbraith JM, Rhyne EP, Koss DA, Hellmann JR. Scripta Mater 1996;35(4):543.
International Journal of Adhesion & Adhesives 27 (2007) 417–421 www.elsevier.com/locate/ijadhadh
Mechanical testing of steel/aluminium–silicon interfaces by pushout O. Dezellusa,, B. Digonneta, M. Sacerdote-Peronneta, F. Bosseleta, D. Roubyb, J.C. Vialaa a
Universite´ Claude Bernard Lyon 1, LMI—UMR CNRS No 5615, 43 Bd du 11 novembre 1918, 69622 Villeurbanne Cedex, France b INSA-Lyon, GEMPPM—UMR CNRS No 5510, 20 av. Albert Einstein, 69621 Villeurbanne Cedex, France Available online 25 October 2006
Abstract The functionality of structural light alloy castings can be improved by inserting into them, upon moulding, local iron base reinforcements. To acquire a better knowledge of such bimetallic assemblies, samples were prepared by immersing a mild steel bar (5 mm in diameter) in aluminium base Al–Si alloy melts held at 730 1C. After melt solidification, the bimetallic samples were cut into 5 mm thick slices and pushout testing was performed on these slices. Characterization of damaging corresponding to different load level before complete debonding allows the determination of the failure mode. Crack initiation occurs at the specimen bottom face in the intermetallic reaction layer, important damage occurs before complete debonding and no brittle failure is observed. The results highlight the necessity of analysing pushout tests with a more integrated approach taking into account shear stress distribution along the interface such as interfacial crack growth. r 2006 Elsevier Ltd. All rights reserved. Keywords: Aluminium and alloys; Titanium and alloys; Adhesion by chemical bonding; Push-out tests
1. Introduction There is major interest (weight saving and low-cost production) in replacing cast iron and steel automotive components by light weight aluminium castings to improve vehicle performance and efficiency [1]. However, a key problem is to form a sound bond between the steel insert and the aluminium casting alloy. Indeed, when using conventional die casting the insert is simply embedded in the light alloy after its solidification. Hot dipping process is a way to produce a sound metallurgical bond at the interface [2,3]. Despite the usefulness of bonding steel to aluminium alloys, some difficulties may arise from the possible development of brittle intermetallic compounds at the steel/aluminium interface [4]. Moreover, little work has been published on the mechanical properties of such interfaces [5]. This paper is an attempt to link the chemistry of the reaction zone at steel/aluminium silicon alloys interface with mechanical behaviour. Therefore, special care was Corresponding author. Fax: +33 4 72 44 06 18.
E-mail address: [email protected] (O. Dezellus). 0143-7496/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijadhadh.2006.09.003
taken to control and characterize (before and after loading) the metal/metal interfaces. As no ASTM test exists for such assemblies, pushout testing was chosen to investigate their mechanical strength. Combining pushout test results with the characterization of the interface zone before and after loading has led us to propose a failure mode for the steel/light weight aluminium alloy assembly. Mechanical testing of joined materials being a hot topic in many fields of materials science, the relevance of pushout test to characterize bimetallic assemblies is also discussed. 2. Experimental procedure Bimetallic samples were processed from an AS-13 foundry alloy (composition in wt%: 12.6 Si, 0.42 Fe, o0.02 Cu, Mg, Mn, Ni, Ti, Zn) and from inserts rods made of mild steel (composition in wt%: 0.2 C, 0.85 Mn, 0.4 Si, 0.045 P, 0.045 S and Fe balance). The steel rods (5 mm in diameter) all received the same surface preparation by mechanical abrasion leading to a mean surface roughness of 3 mm. Moreover, just before the insertion process, steel and aluminium alloy pieces were degreased in an ultrasonic bath of dichloromethane. Bimetallic sample
ARTICLE IN PRESS O. Dezellus et al. / International Journal of Adhesion & Adhesives 27 (2007) 417–421
418
Thermocouple
Thermocouple
Steel insert
Steel insert Al-13 at % Simatrix half saturated in iron
AS 13 matrix
furnace
Induction turn
transfer after 10 mindipping at 730°C aluminizing bath
40 mm
(a)
insertion bath (b)
Fig. 1. Manufacturing process of bimetallic sample: (a) aluminizing step, (b) transfer to the insertion bath.
were produced in two successive steps: aluminizing and insertion as shown in Fig. 1. 2.1. Steel rod aluminizing The first step of the sample manufacturing process consists of dipping (under air) mild steel rods in a cylindrical crucible containing an Al–Si mixture based on a commercial AS-13 (Al–13 wt%Si) alloy. Pure silicon chips (electronic grade purity 99.9995 wt%) were added to the AS-13 liquid bath in order to adjust its silicon content at 13at%. Iron was then added at a level corresponding to half saturation (5.5 wt% at 770 1C) to favour the growth of a uniform reaction layer without excessive iron dissolution, as reported in [6] (final composition of the melt in wt%: 13.46 Si, 5.5 Fe, o0.02 Cu, Mg, Mn, Ni, Ti, Zn). In the following this first liquid will be named aluminizing bath. Dipping temperature was measured with a precision better than 70.2 1C by plunging in the melt a K-type (Ni/Cr) thermocouple. Note that with a bath heated at 770 1C, immersion of the insert led to a 30–40 1C decrease in temperature. Therefore, interface reactions in this first step are considered to develop at a temperature of 73575 1C. 2.2. Transfer to the insertion bath After 10 min dipping at 73575 1C, the hot aluminized insert was pulled out of the bath and immediately transferred in a new crucible, heated at 620 1C by RF coupling and containing a commercial AS-13 alloy to produce the final reinforced bimetallic casting. The transfer duration from one bath to the other was such that the AS13 alloy coating the steel rod had no time to solidify. The bottom of the crucible containing the insert bath was machined and a device put on its top so that the steel insert was perfectly vertical. In the following this liquid will be named insertion bath. The RF power supply was turned off and the aluminium alloy was allowed to cool and solidify
around the insert 20 s after plunging the aluminized steel rod in the second bath. The temperature of the AS-13 melt in this second bath was also measured with a K-type thermocouple. In that case, the temperature did not notably change when the aluminized insert was dipped. After solidification, the bimetallic samples were sawn with a diamond coated wire, into slices with a thickness from 3 to 8 mm, the cross-section of these slices being perpendicular to the rod axis. End sections of each cylinder were diamond polished to a finish better than 1 mm for examination by optical microscopy (OM), scanning electron microscopy (SEM) and electron probe microanalysis (EPMA). Other slices were used for pushout testing as shown in Fig. 2. The tests were performed by using an INSTRON 1195 testing frame equipped with a compression load cell of 100 kN capacity. The insert was pushed by means of a flat-bottomed STUB steel cylinder (ball bearing steel) with a diameter of 4 mm, i.e. 1 mm less than the steel insert. The cross-head displacement rate was of 0.2 mm min1. The relative displacement of the indenter compared to the AS-13 matrix was measured by using a modified extensometer (see Fig. 2). This experimental equipment makes it possible to remove the influence of the machine compliance on the displacement measurements, therefore, only the compliance of the indenter remains. The tested slice is placed on a flat supporting surface, the insert being centred on a drilled hole of 6 mm diameter (Fig. 2). 3. Interface chemistry Examination by optical micrography of the bimetallic samples end sections (labelled zero) indicates that the interfacial reaction zone, formed after the complete process of insertion, consists of two different intermetallic layers (see Fig. 3). EPMA characterization indicates the following phase sequence from steel rod to aluminium alloy matrix: (i) a 3–4 mm thick Z-Al5Fe2(Si) layer and, (ii) a 9–10 mm
ARTICLE IN PRESS O. Dezellus et al. / International Journal of Adhesion & Adhesives 27 (2007) 417–421
419
Mechanical load Steel insert Screw AS 13 matrix
8 7 6 5 4 3 2 1 0
M24-3: loaded until completede bonding
Extenso meter
Steel punch specimen Top Face specimen Back Face
M24-0: characterization of interfacial reaction zone
Load Cell
Fig. 2. Bimetallic sample, slices references and principle of the pushout test.
Fig. 3. Optical micrograph of an end section of bimetallic sample showing the reaction zone.
thick t6-Al9Fe2Si2 layer. Note that the crystal nature of these intermetallic phases was also assessed by X-ray diffraction and that the thicknesses of the intermetallic layers were found constant all along the interface [6]. According to Pontevichi et al. [4,6], the phase in equilibrium with the liquid alloy Al–13 at%Si at 700–800 1C is the ternary intermetallic compound g (Al3FeSi). On the other hand, same author reports that, in the Fe–AS-13 system, the solid in equilibrium with an AS-13 liquid alloy between 650 and 576 1C is b-Al5FeSi also designated as t6-Al9Fe2Si2. Therefore, the manufacturing procedure used to produce bimetallic assemblies in the present work is characterized by the following reaction scheme: (i) formation of the sequence Fe/Z-Al5Fe2(Si)/ g-Al3FeSi/L in the aluminizing bath at 735 1C (at that temperature, the iron content is between half saturation and saturation). (ii) once transferred in the insertion bath at 620 1C, the aluminized steel rod is in contact with a liquid, which is no more in equilibrium with g-Al3FeSi but with b-Al5FeSi. Moreover, this liquid is iron free.
Therefore, g-Al3FeSi and a part of Z-Al5Fe2(Si) produced at 735 1C during the first step are dissolved during the second step upon dipping at 620 1C and a new layer of b-Al5FeSi forms upon cooling between 620 and 576 1C. The final reaction layer sequence is then Fe/Z-Al5Fe2(Si)/ b-Al5FeSi/Al–Si. 4. Push-out tests More than 150 test pieces with different thickness were submitted to pushout testing. The typical load–displacement response for an AS-13 matrix reinforced with a mild steel insert in the case of a 4 mm thick slice is reported in Fig. 4. Zero displacement ( ¼ 0) corresponds to the first contact of the indenter with the steel insert, i.e. to the rise of the load measurement (F). Mechanical fitting being quasistatic, the first loading stage (from e ¼ 0 to 0.05 mm) consists in removing misalignments and is therefore characterized by a gentle increase of the load. Next a quasi-linear response with a steep slope is observed until
ARTICLE IN PRESS O. Dezellus et al. / International Journal of Adhesion & Adhesives 27 (2007) 417–421
420
800
Fmax
700 Load (daN)
600 500
FL
400 300 200 100 0 0
0.1
0.2
0.3 0.4 0.5 0.6 Displacement (mm)
0.7
0.8
0.9
Fig. 4. Typical load–displacement curve for an AS-13 matrix reinforced with a mild steel insert in the case of a 4 mm thick slice.
Table 1 Maximum load and load corresponding to the beginning of linear deviation obtained for 5 mm thick slices belonging to different bimetallic assemblies Sample
Fmax (Da N)
FL (Da N)
FL/Fmax
M20-3 M24-3 M25-3 M26-2
749 738 718 672
380 394 405 380
0,51 0,53 0,56 0,57
Average value Standard deviation
719 29
390 10
0,54 0,02
the load has reached a limit value FL ¼ 405 daN. Afterwards, from F ¼ FL to the maximum value Fmax ¼ 718 daN (emax ¼ 0.81 mm), deviation to the linearity occurs. Once the maximum value is attained, the load is observed to decrease gently (no load drop is observed). Table 1 summarizes the main results (maximum load and load corresponding to the beginning of linear deviation) obtained for 5 mm thick slices belonging to different bimetallic assemblies. The results appear to be quite reproducible and it is interesting to note that deviation to linearity occurs with a good accuracy at 50% of the maximum load. After mechanical loading, slices were embedded in resin and sawn with a diamond coated wire to prepare a polished vertical section, i.e. parallel to the axis of the insert. Optical micrograph of such a section is shown in Fig. 5 in the case of the M24-3 sample. Various signs characterize the damage that led to failure of the bimetallic assembly. First, because of indenter misalignment, displacement of the insert and damage are not fully symmetric: on the left side the steel rod has been pushed of 350 mm compared to the matrix (this shift is observed on the top and bottom faces of the assembly—see Fig. 5a), whereas on the other side, steel rod and matrix seem to remain interdependent all along the interface (except a 200 mm interfacial crack initiated on the top
Fig. 5. Optical micrograph of the sample M24-3 after mechanical loading: (a) top left side, (b) top right side, (c) bottom left side, (d) bottom right side.
face—see Fig. 5b) and are both deformed by the load (see Fig. 5d). On the left side two cracks are observed: the first one was initiated at the upper face of the assembly and propagated in the AS-13 matrix from top to bottom (see Fig. 5a). The second one in the lower part of the assembly is situated at the interface, in the intermetallic reaction zone (see Fig. 5c). Moreover, all along the interface, many transverse cracks (perpendicular to vertical symmetrical axis) are observed in the intermetallic reaction zone. Finally, the preferred orientation of the silicon crystals in the AS-13 matrix in the vicinity of the interfacial zone clearly indicates the occurrence of an important plastic flow of the AS-13 alloy (see Fig. 5d). It is obvious that interfacial cracking, matrix cracking and plastic flow did not occur simultaneously during loading of the assembly. 5. Failure mode To better understand the scenario of failure it is necessary to establish the temporal sequence of damaging. Therefore a new series of pushout tests were interrupted at different load level. For that purpose, 5–7 slices with a thickness of 5 mm were sawn in a bimetallic assembly. The
ARTICLE IN PRESS O. Dezellus et al. / International Journal of Adhesion & Adhesives 27 (2007) 417–421
421
AS-13 bimetallic assemblies tested by pushout is not brittle and damaging is allowed. Three different damage mechanisms are observed: (i) crack propagation in the intermetallic reaction layer, (ii) crack deviation towards the AS-13 matrix, (iii) and plastic flow of the matrix (see Figs. 4 and 5). Determination and understanding of the spatial and temporal sequence of damaging require further work and more information, especially concerning the shear stress distribution and its evolution with time along the interface. 6. Conclusion
Fig. 6. Optical micrograph after pushout tests interrupted for 50% of the maximum load reference—initiation of interfacial cracking at the specimen bottom face.
first slice corresponding to the bottom of the assembly was labelled 0 and used for characterization of the reaction zone. The other slices were increasingly numbered from bottom to top. Number 3, corresponding to the centre of the assembly was usually loaded until failure occurred at a maximum load noted Fmax-ref. Next the other slices were loaded, but pushout test was interrupted when the applied load reached a fraction of Fmax-ref increasing from 10% to 90%. Afterwards, the slices were embedded in a resin, sawn with a diamond coated wire and polished to a finish better than 1 mm to characterize the damage along a vertical section, i.e. parallel to the axis of the insert. Optical characterizations of samples after interrupted mechanical loading indicate that damaging of the specimen does not start in the vicinity of the indenter but systematically at the specimen bottom face by cracking in the intermetallic reaction layer. Such cracking occurs for an applied load corresponding to 50% of the reference Fmax-ref (see Fig. 6). This behaviour is in disagreement with classical analysis of pushout test, which is based on the assumption that, upon loading, debond crack propagates from specimen topface [7]. However, some authors relate the occurrence of bottom face debonding, in some cases for metal matrix composites (MMCs). [8–10] In the present study, bottom face debonding could be due to bending effect on the bottom face [11] or to thin slice geometry (thickness limited to less than 20 insert radii) [12]. Next, no brittle failure due to crack propagation (characterized by an abrupt load drop on the load versus displacement curve) is observed (Fig. 4). Therefore, despite the intermetallic reaction layer, the failure mode of steel/
Reinforcement of AS-13 matrix by steel insert has been performed with an experimental procedure, allowing the control of the interfacial reaction layer. Pushout tests until complete debonding associated with interrupted tests and metallographical characterization, allow us to conclude that for steel/AS-13 bimetallic assembly crack initiation occurs at the specimen bottom face and in the intermetallic reaction layer. Next, no brittle failure is observed and the bimetallic assembly is able to support important damage (crack propagation, crack deviation, plastic flow of the matrix) before failure. The results highlight the necessity of analysing pushout tests with a more integrated approach taking into account shear stress distribution along the interface such as interfacial crack growth. Fracture energies (Gi ) for Fe/ AS-13 interfaces are not yet determined, but crack deviation indicates that Gi exceeds the fracture energy of the AS-13 matrix. Aluminium alloys matrix is too soft compared to the toughness of the interfacial reaction zone to easily determine interface properties by pushout testing. References [1] Clyne TW, Withers PJ. An introduction to metal matrix composites. Cambridge: Cambridge University Press; 1993. [2] Barbeau F. PhD thesis, University Claude Bernard Lyon 1, Lyon, 1999. [3] Stucky M. Fonderies–Fondeur d’aujourd’hui 1997;168:80. [4] Pontevichi S, Bosselet F, Barbeau F, Peronnet M, Viala JC. J Phase Equilib Diffus 2004;25(6):528. [5] Durrant G, Gallernault B, Cantor B. J Mater Sci 1996;31(3):589. [6] Pontevichi S. PhD thesis, University Claude Bernard Lyon 1,Lyon, 2004. [7] Kerans RJ. J Am Ceram Soc 1991;74(7):1585. [8] Ghosn LJ, Eldridge JI, Kantzos P. Acta Mater Metall 1994;42(11): 3895. [9] Koss DA, Hellmann JR, Kallas MN. J Met 1993;46(3):34. [10] Majumdar BS, Miracle DB. Key Eng Mater 1996;116–117:153. [11] Kallas MN, Koss DA, Hahn HT, Hellmann JR. J Mater Sci 1992;27(14):3821. [12] Galbraith JM, Rhyne EP, Koss DA, Hellmann JR. Scripta Mater 1996;35(4):543.