- Email: [email protected]
Interfacial microstructure and properties of copper clad steel produced using friction stir welding versus gas metal arc welding
Materials Characterization 104 (2015) 1–9
Contents lists available at ScienceDirect
Materials Characterization journal homepage: www.elsevier.com/locate/matchar
Interfacial microstructure and properties of copper clad steel produced using friction stir welding versus gas metal arc welding Z. Shen a, Y. Chen a, M. Haghshenas a,⁎, T. Nguyen b, J. Galloway c, A.P. Gerlich a a b c
Mechanical and Mechatronics Engineering, University of Waterloo, Waterloo, Canada Mechanical Systems Engineering, Conestoga College, Kitchener, Canada Welding Engineering Technology, Conestoga College, Kitchener, Canada
a r t i c l e
i n f o
Article history: Received 20 November 2014 Received in revised form 31 January 2015 Accepted 26 February 2015 Available online 27 February 2015 Keywords: Copper cladding Friction stir welding Gas metal arc welding Metallurgical bonding
a b s t r a c t A preliminary study compares the feasibility and microstructures of pure copper claddings produced on a pressure vessel A516 Gr. 70 steel plate, using friction stir welding versus gas metal arc welding. A combination of optical and scanning electron microscopy is used to characterize the grain structures in both the copper cladding and heat affected zone in the steel near the fusion line. The friction stir welding technique produces copper cladding with a grain size of around 25 μm, and no evidence of liquid copper penetration into the steel. The gas metal arc welding of copper cladding exhibits grain sizes over 1 mm, and with surface microcracks as well as penetration of liquid copper up to 50 μm into the steel substrate. Transmission electron microscopy reveals that metallurgical bonding is produced in both processes. Increased diffusion of Mn and Si into the copper cladding occurs when using gas metal arc welding, although some nano-pores were detected in the FSW joint interface. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Overlay welding or cladding is often used to provide corrosion or wear resistance to steel components. However, this encounters some challenges in the case of copper claddings. Dissimilar welding of these materials is hampered by the fact that copper and iron form an immiscible binary system, while another key issue is liquid metal embrittlement that arises from the molten copper in contact with the steel, which is able to wet the steel austenite grain boundaries at high temperature [1]. This may lead to sub-surface cracking [2], and cracking of the copper cladding can also readily occur upon cooling due to thermal stress caused by the drastically different conductivities and thermal contraction coefficients of the two materials. Furthermore, high heat input is required to overcome the high thermal diffusivity of the copper, and processes such as electron beam welding (EBW) offer such capability [3]. However, poor mixing of the materials and some defects such as porosity and micro-cracking were observed in EB welded copper–stainless steel [4]. Meanwhile, the interface diffusion control is a key point in Cu/steel dissimilar metal welding process with high heat input, because the mechanical properties of the weld will be deteriorative due to enhanced interface diffusion capability [5]. Due to the problems encountered with cladding copper on steels, solid state processes such as explosive welding are more generally preferred [6,7]. A few studies have shown that laser welding of copper and ⁎ Corresponding author. E-mail addresses: [email protected], [email protected] (M. Haghshenas).
http://dx.doi.org/10.1016/j.matchar.2015.02.022 1044-5803/© 2015 Elsevier Inc. All rights reserved.
steels is feasible [8]. However, the need for higher productivity and material deposition rates has also driven the study of hot-wire tungsten inert gas welding [9]. Although the deposition rates were reported to be 3.1 kg/h, poor wetting to the steel substrate occurs with low arc current, and excessive iron dilution into the copper layer was noted at higher arc currents. It has also been shown that iron dilution into copper will further promote microcracking in the solidified material [10]. Since many issues stem from the interaction of the molten copper with the steel, solid state welding processes may provide a beneficial alternative for joining of these materials. Friction stir welding (FSW) was developed in 1991 by The Welding Institute (TWI) for joining heat treatable aluminum alloys and other materials which are difficult to join by arc welding [11]. The process of FSW involves plunging a rotating cylindrical tool into the interface between two plates, and generating sufficient frictional heat and deformation to consolidate the two materials together [12,13]. FSW has been shown to be applicable for joining dissimilar materials such as Al/Mg alloy [14], Al/Fe alloy [15] and Al/Cu alloy, because the formation of low melting point intermetallic compound (IMC) was limited, and metallurgical bonding was confirmed at the interface [16]. The strength of the weld is drastically reduced when IMCs involving AlCu, Al2Cu and Al4Cu9 form at the Al/Cu interface [17–19]. Meanwhile, defects such as microcracks and voids were observed in the weld [17]. The use of FSW has been demonstrated to join pure copper in sections up to 50 mm thick [20–23], where defect-free welds were produced, and tensile strength of the FSW copper joint was higher than that made by EBW [22], with a joint efficiency of 94% reported [22,24]. FSW also has been developed for sealing copper canisters for the storage of
2
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
nuclear waste, since the copper provides excellent corrosion protection [25]. Furthermore, FSW has been applied to join copper to steel in both lap and butt weld configurations with no defects, where mechanical mixing and metallurgical bonding conformed at the Cu/St interface in lap weld configuration [26,27]. The present work provides the first comparison between FSW lap welding of copper cladding sheets to a steel substrate versus cladding by pulsed-current gas metal arc welding (GMAW). The FSW process provides a method of potentially applying a cladding layer by staggering a series of lap-joints on a sheet of cladding material. This can potentially be highly productive since a copper sheet with quite high thickness can be employed, however the bonding microstructures and bonding mechanisms have yet to be examined and compared to conventional overlay fusion welding methods.
2. Experimental details For the case of FSW, lap welding was performed on a 2.1 mm thick C10200 copper sheet with a hardness of 70 HV, and a 25.4 mm thick A516 Gr. 70 steel substrate with a hardness of 155 HV. The tooling during FSW was a Co–WC cermet, consisting of a 12 mm shoulder, and a smooth 2.1 mm long pin with a tapered geometry (where the diameter increased from 4 to 5 mm), as shown in Fig. 1a. A second ‘3-flat’ tool with the same geometry was also investigated, with the exception of three 0.5 mm flat surface ground into the pin at equal 120° spacings (see Fig. 1a). The 3-flat geometry was considered in order to compare the effect of enhanced mechanical deformation and strain compared to a smooth pin tool. Co-axial alignment of the FSW in the holder was measured to be within 0.035 mm. A tool rotation speed of 1120 RPM and travel speed of 31.5 mm/min was used, with a tool tilt angle of 2.5°, and the tool penetration ranged from 0.06 to 0.60 mm into the steel substrate. It should be noted that in preliminary trials, parameters were also compared using different combinations of 900 RPM and travel speeds of 90 mm/min, however only superficial bonding was produced. The parameters applied appeared to be the minimum required for bonding using the FSW equipment used. No preheating was applied during FSW. Initially two single pass welds were reproduced using varied penetration depths. The microstructures of both single-pass FSW lap welds were compared multiple pass FSW joints staggered at 4 mm intervals, with 5 passes of 100 mm length were produced adjacent to each other using the smooth pin tool with a penetration of 0.25 mm. No tool wear was apparent after the 5 adjacent passes.
Fig. 1. (a) Schematic illustration of the tool geometry, (b) photo of the Cu/steel FSW multipass lap joint surface.
The GMAW claddings were produced using pulsed current deposition with a 1.14 mm diameter AWS A5.7 ERCu composition wire, travel speeds of 7.45 mm/s, heat input of 725 J/mm, wire feed speeds of 7.37 m/min, and an inter-bead spacing of 4.0 mm. The initial preheat during GMAW was 100 °C with inter-pass temperatures maintained below 300 °C. Standard metallographic preparation techniques were applied, with final polishing using 1 μm diamond slurry. Specimens were first etched using 2% Nital to reveal the steel microstructures, then etched with 60% Nital to reveal the copper grain structure. The joint microstructures were examined by a combination of optical metallography, scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The SEM analysis was performed on a JEOL 6460 instrument equipped with energy-dispersive X-ray (EDX) spectroscopy and operated at 30 keV. TEM analysis involved using a focused ion beam to extract the copper/steel interfacial region in both claddings, and analysis of the thinned foil was performed using a Titan LB instrument. In addition, Vickers microhardness measurements were conducted in both samples to detect any possible embrittlement of the heat affected zone (HAZ) of the steel. 3. Results 3.1. FSW Cu/steel lap joint The copper sheet was adhered to the substrate with minimal flash ejected from the surface as shown in Fig. 1b, and multiple FSW passes could be staggered adjacent to each other without severe accumulation of the flash on the surface. In an individual pass, it is clear that a wavelike or ‘hook’ feature is formed in which the steel has been displaced into the upper copper sheet (see Fig. 2a), which provides strong mechanical interlocking at the Cu/steel interface. The bonded area occurs across the interface which spans this intermixed region, roughly corresponding to the diameter of the tool pin. It is clear that the Co–WC FSW tool penetrated into the steel, and the bonding occurs only along the width of the pin in this displaced region. The cross-section of the Cu/steel interface in the bonded region is shown in Fig. 2b when using the smooth pin tool and a penetration of 0.25 mm. A smooth interface is observed near the centerline of the joint, while a series of ridges or ripples with a 10 μm-amplitude occurs near the sides of the interface, as shown in high magnification in Fig. 2c & d. These ridges appear to promote interlocking between the two materials, similar to that observed in explosion welds [28]. The origin of these ridges appears to be related to the oscillations imposed during the rotation of the tool since their spacing increases towards the outer edges of the bonded interface [29]. The spacing of the ridges becomes finer towards the outer periphery of the stir zone (Fig. 2c), which is likely due to overlapping material flow staggered at finer spacings near the sides of the pin. This can be visualized in a prior work explaining the recirculating flow during FSW [30]. Multiple FSW passes (all beginning at the same side) were traversed in the same direction on the lapped Cu/steel, at a spacing of 3 mm between passes with varying pin penetration depths. The first passes were made using both the smooth pin tool, and then later passes were made using the 3-flat tool pin geometry in order to determine how tool penetration and pin geometry influence the microstructure of the bonded interface. Fig. 3 shows the interfacial region between the two passes made using the smooth pin tool with 0.15 and 0.19 mm of penetration, respectively. It can be noted that 0.15 mm of pin penetration did not promote complete bonding with the steel substrate, and in both cases a large number of steel fragments were noted in the stir zone of the welds. A void can also be noted between the boundaries of the two passes. In order to determine if bonding could be improved with different tool geometry, a 3-flat pin was investigated. Welds were made at progressively increasing plunge depths from 0.06 to 0.60 mm using the 3-
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
3
Fig. 2. (a) Macrograph of transverse section in a single-pass FSW Cu/steel lap weld, and (b, c and d) interface microstructures.
flat tool, see Fig. 4. It was found that partial bonding between the sheets occurred only when at least 0.23 mm of penetration was applied using the 3-flat tool, and the interface contained steel fragments and voids (which are evident in Fig. 4a). The voids and steel fragments remained when the penetration was increased to 0.49 mm (see Fig. 4b). This issue remained when 0.60 mm penetration was applied using the 3flat tool, however, when a weld was made with the smooth-pin tool directly adjacent to this weld with only 0.27 mm penetration, no voids were produced as shown in Fig. 4c. This appears to be a result of the increased cutting action of the 3-flat pin, which caused the steel fragments to be displaced into the copper layer as indicated in Fig. 4. Due to the compliance of the FSW equipment frame, the cutting action allowed further penetration into the steel substrate under the high axial tool loads. Considering the improved bonding and void-free interface microstructures produced using the smooth-pin FSW tool with a pin penetration of 0.25 to 0.27 mm, only these conditions are examined further in terms of microscopy at the interface. The interface microstructures observed by SEM microscopy reveal that the interface of the FSW lap joint was flat and free of microcracks, microvoids and intermetallics as seen in Fig. 5. However, EDX quantification produced inconsistent quantification since there appeared to be non-uniform etching by-products deposited near the copper/steel interface. These produced high copper EDX quantifications of 12 wt.%
copper at location 1 in Fig. 5, while 22.8 wt.% copper was detected at location 2 farther away from the copper interface, and location 3 (the same distance from the interface as location 2) showed 1 wt.% copper. The etching by-products appear to correspond to the light patches on the steel near the interface in Fig. 5. Consequently, further analysis of the interface structure and chemistry was carried out using a combination of ion milling and TEM analysis in order to determine the extent of mutual diffusion at the interface. It is also worth noting that the recrystallized grain size was comparable in the copper sheet before and after FSW. The as-received microstructure of the copper sheet is shown in Fig. 6a, which shows an average grain size of 21 μm. In comparison, the grains in the copper had an average size of 25 μm after FSW, see Fig. 6b. This is consistent with prior work involving FSW of pure copper, where typical grain size of less than 100 μm were observed [22]. Typical annealing twins can be noted in both microstructures in Fig. 6, clearly indicating that recrystallization has occurred. 3.2. GMAW cladding The copper cladding produced by GMAW is shown in Fig. 7, with several layers staggered adjacent to each other. The surface was found to contain numerous microcracks at 45° angles to the welding direction.
Fig. 3. Optical micrograph of the bond interface in multi-pass FSW Cu/steel lap joints made using 0.15 and 0.19 mm.
4
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
Fig. 4. Optical micrograph of the bond interface in FSW Cu/steel lap joints made using 3-flat tool with (a) 0.23 mm and (b) 0.49 mm, and (c) 0.60 mm penetration (compared to a smooth pin tool with 0.27 mm penetration on right side).
Evidence of these microcracks can be observed in the cross-section of the overlay, as evident in the surface crack indicated by the arrow in Fig. 8a, and nearly all such cracks took an inter-granular path. The copper cladding grain size was well over 1 mm, and there also appears to be scattered porosity as indicated by dark circular spots in Fig. 8b. However it should be noted that many finer spots in Fig. 8b correspond to surface pits produced by etching. Near the steel interface, the copper grains are more refined (see Fig. 8b). 3.3. Steel HAZ hardness and microstructures The microhardness values along the through-thickness direction are shown in Fig. 9. The hardness values in the steel HAZ of the FSW joint did not exceed a value of 200 HV, and the copper microhardness slightly increased compared to that of the as-received sheet. The highest hardness value achieved was 225 HV within the steel adjacent to the fusion line in the GMAW cladding. It is worth noting that the hardness values in the steel for both processes remained below the critical value of
248 HV (or 22 HRc), which is a key requirement for many corrosion applications [31]. This is usually required in order to avoid cracking or delamination of the cladding in corrosive environments. The microstructures of the steel HAZ at the locations corresponding to −0.65 and −0.15 mm (i.e., below the bonded interface) are shown in Fig. 10. These indicate that the banded fine-grained ferrite and pearlite structure in the steel base material is refined by the FSW tool, as shown in Fig. 10a. This banded structure is progressively refined further closer to the interface, and is completely broken up into an equiaxed fine grained structure of ferrite and pearlite immediately below the interface as shown in Fig. 10b. This indicates that the temperature in this region remained below the critical A1 steel transformation temperature (727 °C) in the FSW cladding. In the case of the GMAW overlay, the steel HAZ microstructures were mainly acicular ferrite, with a small fraction of lower bainite, see Fig. 11. This is consistent with the hardness values observed in Fig. 9, and suggests that no martensite was formed even close to the copper/ steel interface. The grain structure in the steel at the interface appears
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
5
Fig. 7. Cu cladding produced by GMAW.
occur in various copper cladding processes on steel substrates. The depth of over a dozen copper penetrated zones was measured and these varied in length from 5 to 50 μm. 3.4. TEM studies
Fig. 5. SEM micrograph of FSW joint Cu/steel interface, and corresponding EDX quantifications obtained from regions 1, 2 and 3 in the micrograph.
to remain b5 μm. Several instances of liquid copper penetration could be noted in the interface of the GMAW cladding, as noted by the arrow in Fig. 11. As mentioned previously, this has been noted to
Fig. 6. SEM micrographs showing (a) as-received copper sheet grain structure, and (b) stir zone copper grains in the FSW joint.
The Cu/steel interface of the FSW joint was extracted using the Focused Ion Beam (FIB) milling approach, and the TEM bright field image of a FSW joint interface is shown in Fig. 12. The TEM images show that there is metallurgical bonding between the steel and copper. Some nano-sized pores (100 to 200 nm) were observed at the interface. However, it is not clear whether these originated from sample preparation or possibly residual entrapped oxides which fell away during sample extraction. The grain structure in the steel was much finer (b 1 μm),
Fig. 8. Optical micrograph of microstructure produced in Cu cladding of the GMAW overlay (a) showing grain structure through the thickness and location of a grain boundary crack at the arrow, and (b) at the Cu/steel interface with larger pore indicated by an arrow.
6
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
Fig. 11. Optical micrograph showing the steel HAZ microstructure in the Cu cladding produced using GMAW, with the location of a liquid Cu penetration indicated. Fig. 9. Comparison of through-thickness microhardness distribution across the cladding interfaces.
compared to the grains in the copper cladding. The dislocation density in the copper was also much higher than that of the steel grains. The EDX mapping shown in Fig. 12b & c shows that there was negligible diffusion of copper into the steel grains, since the interface appears to be sharp and delineated. Also, no Mn or Si could be detected by EDX in the copper side of the interface. The GMAW cladding interface which was extracted by FIB was also studied by TEM, and the micrographs are shown in Fig. 13. The micrograph indicates that metallurgical bonding is also produced, with no pores being observed at the interface of the GMAW cladding. It is clear that the grain size of the steel at the interface is much larger in the
Fig. 10. Steel microstructures near indents at (a) −0.65 mm and (b) −0.15 mm away from the copper interface in the FSW Cu/steel joint (referenced to positions in Fig. 9).
GMAW joint shown in Fig. 13 (compared to the FSW joint interface shown in Fig. 11), and the dislocation density in the GMAW is much higher as well. The dislocation density in the copper grains is also rather high, and it appears that a dislocation cell structure has developed suggesting more recovery has occurred than in the FSW joint. The EDX analysis in Fig. 13c indicates that the some diffusion of Si and Mn towards the copper grains have occurred in the GMAW cladding. 4. Discussion It is clear that metallurgical bonding between copper and steel is produced using both processes. The FSW joint appears to have a minor contribution from mechanical interlocking as a result of fine ripples formed in a part of the interface between the materials. These are likely formed by the oscillations of the tool, since their spacing decreases towards the outer edges of the weld, much like the spacing of typical ‘onion ring’ patterns observed in FSW joints [32]. Since the coaxial alignment of the FSW tool with the tool holder can never be absolutely perfect, there will always be some cyclical component to the tool motion even when a cylindrical tool is used. In the present case, there was 0.035 mm of run-out measured, and this would lead to a slight vertical oscillatory movement of the tool when it is tilted (2.5° in this case), thus promoting the formation of slight ripples at the interface as shown in Fig. 2c. This aspect could potentially be exploited to some degree further by employing higher tilt or having lower tolerances on tool alignment. The selection of FSW tool geometry and penetration depth appears to play a major role in establishing a uniform bond across the Cu/steel interface when multiple tool passes are used to cover a large surface. As shown in Fig. 3, the FSW lap welding process appears to be sensitive to the depth achieved in the steel substrate when a smooth pin tool is used, where at least 0.19 mm of pin penetration appears to be required to achieve bonding. The use of a tool with 3-flats appears to have led to much higher penetrations (0.60 mm) when similar processing parameters were used (see Fig. 4). However, this led to the formation of extensive defects and voids near the sheet interface. The corners on the tool pin of the 3-flat tool promote cutting of the steel, and the steel fragments were broken away and dispersed into the stir zone of the copper cladding. This material removal allowed much higher tool penetration, but also led to the formation of excessive flash by the tool shoulder on the surface of the tool. The most significant microstructural difference between the FSW and GMAW cladding processes is the grain size of the copper layer, which can be expected based on the prevailing literature on FSW which typically reports grain sizes in the stir zone of b 10 μm. These grain sizes do vary with parameters during FSW, and it has been shown that the grain size in pure copper increases from 3.5 to 9 μm as the tool rotation speed increases from 400 to 800 RPM [33]. However, it should be noted that the FSW travel speed was only 31.5 mm/min, and so the cladding grain size may be refined at higher travel speeds.
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
7
Fig. 12. TEM micrographs of (a) FSW joint interface between copper and steel, (b) EDS map location and (c) EDS element maps for Fe, Cu, Mn.
In any case, the FSW cladding grain size observed here (25 μm) was much finer than that observed in the GMAW copper cladding. This has important implications for the desired final properties and applications, for example where creep properties are concerned [34]. One challenge posed by the extremely coarse grain sizes of the GMAW cladding is the difficulty during subsequent ultrasonic inspection. The large grain size will attenuate the ultrasonic beam by increasing the scattering losses thereby reducing the ability to detect defects [35]. This suggests that the inspection of the FSW joint will be more convenient, since the grain size was well under 50 μm in the stir zone. The grain structure of the steel substrate was initially a fine ferrite plus pearlite mixture, as shown in Fig. 10a. In the case of the FSW interface, this structure only became more refined and deformed, as shown in Fig. 10b, which suggests that the peak temperature was below the A1 steel transformation temperature. This readily explains why no liquid copper penetration was observed in the FSW joint, since the mechanism for liquid penetration depends on the formation of austenite in the steel and the FCC structure is very readily wetted by the liquid copper [1]. From the phase diagram in Fig. 14 [36] liquid copper and austenite phase co-exist above 1096 °C. Thus, the temperature at the interface during GMAW cladding was at least 1096 °C. Many have focused on alternative cladding processes in order to help avoid this liquid penetration mechanism, such as solid state methods like explosion welding. Other strategies such as buttering layers using Ni–Cu alloys may also help avoid this issue by providing a
more gradual interface and separating the steel austenite grains from the liquid copper. This can be expected since Savage et al. noted that liquid copper embrittlement could be avoided in alloys with extensive copper solubility, such as Ni, Mn, Pd, and Pt [37]. The formation of microcracks was rather prevalent across the surface of the GMAW cladding shown in Fig. 7. This microcracking is likely mitigated in the FSW joint due to the fact that the temperatures were lower, based on the difference in the steel microstructure at the interface. It has been shown that residual stresses are drastically lower in FSW joints compared to fusion welds [38]. While it may be possible to mitigate the microcracks in the GMAW cladding by increasing the preheating temperature, and hence reducing the thermal gradient, this would also likely promote further liquid copper penetration into the steel substrate since the peak temperature and solidification time would increase. At the interface, there appear to be some nano-sized voids formed in the FSW joint (see Fig. 12a). The voids could have possibly formed due to inadequate mixing of the two materials during FSW, considering their drastically differing properties at high temperature. However, it is also possible that these voids could be the remnants of residual surface oxides which were not fully dispersed during FSW, and this has been known to produce kissing bond structures in similar FSW joints as well [39]. These could potentially be preferentially removed due to differences in ion milling during sample preparation, or become dislodged during TEM sample preparation and manipulation.
8
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
Fig. 13. TEM micrographs of (a) GMAW joint interface between copper and steel, (b) EDS map location and (c) EDS element maps for Si, Fe, Cu, Mn.
Although some mutual diffusion across the bond was suggested the EDX line scan of the FSW joint in Fig. 5b, the values were under 1 wt.%. This value is rather low and explains why the element mapping conducted by TEM and EDX reveals a rather sharp interface in Fig. 12c. It should be noted that no intermetallics are possible in the Cu–Fe system as shown in Fig. 14. The TEM results of GMAW are consistent with the grain structures observed by optical and SEM, since the size of the cells shown in Fig. 13a do not represent the grains shown in Fig. 8 or 11, but rather the cells reflect a substructure that is formed within the individual grains. The grains in the steel and copper in the GMAW are actually larger than the field of view in the sample extracted from the GMAW cladding. Convergent beam electron diffraction patterns were obtained from the copper and iron grains in both TEM samples. However, no orientation relationship could be identified between the two phases. In FSW, severe plastic deformation and recrystallization establish the grain structure and orientation. In the case of the GMAW cladding it is possible that epitaxial growth of copper grains from the molten state could occur from the steel surface. However, the steel would be heated to above the A1 temperature by the liquid copper, and so the subsequent orientation of the steel grains will be determined
by the solid state transformation of austenite into ferrite and carbide, thus obscuring any orientation relation. 5. Conclusions A preliminary investigation has shown that cladding of steel with pure copper is possible using either FSW or GMAW, where both processes achieve metallurgical bonding with the substrate. The FSW process yielded a significantly finer grain structure in the copper, with no evidence of cracking in the cladding or liquid copper penetration into the steel substrate. The FSW joint appeared to contain some evidence of nano-pores at the interface of the steel and copper. The steel microstructure adjacent to the fusion line suggests that the temperature remained below the austenite temperature of the steel. The copper cladding produced by GMAW contained many surface cracks which are attributed to high thermal stresses. There was evidence of liquid penetration of the copper into the steel, and diffusion of Mn and Si into the copper cladding due to the higher temperatures imposed. The steel adjacent to the fusion line maintained a hardness value below 248 HV when using either process, since no martensite was detected in both cases.
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
9
Fig. 14. Binary phase diagram of iron and copper, after Ref. [36].
Acknowledgments The authors would like to thank the Natural Sciences and Engineering Research Council (NSERC) of Canada, and Nuclear Waste Management Organization of Canada for their support. The EM research described in this paper was performed at the Canadian Centre for Electron Microscopy at McMaster University, which is supported by NSERC and other government agencies. Support from Linamar (CAMTAC Manufacturing) in Guelph, Ontario for providing the WC tools is also appreciated. References [1] A.R. Cox, J.M. Winn, JISI 203 (1965) 175–179. [2] R. Eborall, P. Gregory, Inst. Met. 84 (1955) 88–90. [3] R.A. Andrews, Friction stir welding — an alternative method for sealing nuclear waste storage canisters, SKB Technical Report TR-04-16, Svensk Kärnbränslehantering AB, 2005. [4] I. Magnabosco, P. Ferro, F. Bonollo, L. Arnberg, Mater. Sci. Eng. A 424 (2006) 163–173. [5] T. Yang, H. Gao, S. Zhang, L. Wu, Acta Metall. Sin. 26 (2013) 328–332. [6] G.R. Cowan, A.H. Holtzman, J. Appl. Phys. 34 (1963) 928–939. [7] K. Raghukandan, J. Mater. Process. Technol. 139 (2003) 573–577. [8] T.A. Mai, A.C. Spowage, Mater. Sci. Eng. A 374 (2004) 224–233. [9] S.X. Lv, Z.W. Xu, H.T. Wang, S.Q. Yang, Sci. Technol. Weld. Join. 13 (2008) 10–16. [10] M. Turna, M. Sahul, J. Ondruska, J. Lokaj, Annals of DAAAM for 2011 & Proceedings of the 22nd International DAAAM Symposium, 2011. Vienna, Austria, EU. [11] W.M. Thomas, E.D. Nicholas, J.C. Needham, M.G. Murch, P. Temple-Smith, C.J. Dawes, International Patent Application PCT/GB92/02203 and GB Patent Application 9125978.8, UK Patent Office, London, December 6, 1991. [12] R.S. Mishra, Z.Y. Ma, Mater. Sci. Eng. R 50 (2005) 1–78. [13] P.L. Threadgill, A.J. Leonard, H.R. Shercliff, P.J. Withers, Int. Mater. Rev. 54 (2009) 49–93. [14] Y.J. Kwon, I. Shigematsu, N. Saito, Mater. Lett. 62 (2008) 3827–3829. [15] T. Watanabe, H. Takayama, A. Yanagisawa, J. Mater. Process. Technol. 178 (2006) 342–349. [16] P. Xue, B.L. Xiao, D. Wang, Z.Y. Ma, Sci. Technol. Weld. Join. 16 (2011) 657–661. [17] T. Saeid, A. Abdollah-Zadeh, B. Sazgari, J. Alloys Compd. 490 (2010) 652–655.
[18] A. Elrefaey, M. Takahashi, K. Ikeuchi, Weld. World 49 (2005) 93–101. [19] J. Ouyang, E. Yarraparredy, R. Kovacevic, J. Mater. Process. Technol. 172 (2006) 110–122. [20] C.G. Andersson, R.E. Andrews, Proceedings of the First International Symposium on Friction Stir Welding, Thousand Oaks, CA, USA, June, 1999, 1999. [21] L. Cederqvist, R.E. Andrews, 4th International FSW Symposium, Park City, UT, USA, May 14–16, 2003, 2003. [22] W.B. Lee, S.B. Jung, Mater. Lett. 58 (2004) 1041–1046. [23] C.-G. Andersson, R.E. Andrews, B.G.I. Dance, M.J. Russell, E.J. Olden, R.M. Sanderson, Proceedings, 2nd International Symposium on FSW, Gothenburg, Sweden, June, 2000, 2000. [24] D. Avula, R.K.R. Singh, D.K. Dwivedi, N.K. Mehta, World Acad. Sci. Eng. Technol. 74 (2011). [25] L. Cederqvist, FSW to manufacture and seal 5 cm thick copper canisters for Sweden's nuclear waste, 6th International Symposium on Friction Stir Welding. SaintSauveur, Canada, 10–13 October, 2006. [26] M. Shamsujjoha, B.K. Jasthi, M. West, C. Widener, Frict. Stir Weld. Process VII (2013) 151–160. [27] Y. Imani, M.K. Givi, M. Guillot, Adv. Mater. Res. 409 (2012) 263–268. [28] M. Hammerschmidt, H. Kreye, Microstructure and Bonding Mechanism in Explosive Welding in Shock Waves and High Strain Rate Phenomena in Metals, in: M.A. Meyers, L.E. Murr (Eds.),Plenum Press, 1981. [29] B. Yang, J. Yan, M.A. Sutton, A.P. Reynolds, Mater. Sci. Eng. A364 (2004) 55–65. [30] P. Su, A. Gerlich, T.H. North, G.J. Bendzsak, Metall. Mater. Trans. 38A (2007) 584–595. [31] NACE MR0175/ISO 15156-3:2003(E), Petroleum and Natural Gas Industries — Materials for Use in H2S-containing Environments in Oil and Gas Production — Part 3: Cracking-resistant CRAs (Corrosion-resistant Alloys) and Other Alloys, NACE/ANSI/ISO, 2003. (©2003). [32] R.D. Haley, A.S. Judy, C.N. Arthur, Metall. Mater. Trans. A 45A (2014) 4411–4422. [33] G.M. Xie, Z.Y. Ma, L. Geng, Scr. Mater. 57 (2007) 73–76. [34] R. Sandström, H. Östling, L.Z. Jin, Modelling of creep in friction stir welded copper, Mater. Res. Innov. 17 (2013) 350–354; T.B. Massalski, Binary Alloy Phase Diagrams, ASM International, New York, 1990. [35] I.N. Ermolov, B.P. Pilin, NDT Int. 9 (1976) 275–280. [36] Alloy phase diagrams, 10th ed., ASM Materials Handbook, volume 03, ASM International, December 1, 1992. [37] W.F. Savage, E.P. Nippes, M.C. Mushala, Weld. J. (1978) 145s–152s. [38] O. Hatamleh, I.V. Rivero, A. Maredia, Metall. Mater. Trans. 39 (2008) 2867–2874. [39] Y.S. Sato, H. Takauchi, S.H.C. Park, H. Kokawa, Mater. Sci. Eng. A405 (2005) 333–338.
Contents lists available at ScienceDirect
Materials Characterization journal homepage: www.elsevier.com/locate/matchar
Interfacial microstructure and properties of copper clad steel produced using friction stir welding versus gas metal arc welding Z. Shen a, Y. Chen a, M. Haghshenas a,⁎, T. Nguyen b, J. Galloway c, A.P. Gerlich a a b c
Mechanical and Mechatronics Engineering, University of Waterloo, Waterloo, Canada Mechanical Systems Engineering, Conestoga College, Kitchener, Canada Welding Engineering Technology, Conestoga College, Kitchener, Canada
a r t i c l e
i n f o
Article history: Received 20 November 2014 Received in revised form 31 January 2015 Accepted 26 February 2015 Available online 27 February 2015 Keywords: Copper cladding Friction stir welding Gas metal arc welding Metallurgical bonding
a b s t r a c t A preliminary study compares the feasibility and microstructures of pure copper claddings produced on a pressure vessel A516 Gr. 70 steel plate, using friction stir welding versus gas metal arc welding. A combination of optical and scanning electron microscopy is used to characterize the grain structures in both the copper cladding and heat affected zone in the steel near the fusion line. The friction stir welding technique produces copper cladding with a grain size of around 25 μm, and no evidence of liquid copper penetration into the steel. The gas metal arc welding of copper cladding exhibits grain sizes over 1 mm, and with surface microcracks as well as penetration of liquid copper up to 50 μm into the steel substrate. Transmission electron microscopy reveals that metallurgical bonding is produced in both processes. Increased diffusion of Mn and Si into the copper cladding occurs when using gas metal arc welding, although some nano-pores were detected in the FSW joint interface. © 2015 Elsevier Inc. All rights reserved.
1. Introduction Overlay welding or cladding is often used to provide corrosion or wear resistance to steel components. However, this encounters some challenges in the case of copper claddings. Dissimilar welding of these materials is hampered by the fact that copper and iron form an immiscible binary system, while another key issue is liquid metal embrittlement that arises from the molten copper in contact with the steel, which is able to wet the steel austenite grain boundaries at high temperature [1]. This may lead to sub-surface cracking [2], and cracking of the copper cladding can also readily occur upon cooling due to thermal stress caused by the drastically different conductivities and thermal contraction coefficients of the two materials. Furthermore, high heat input is required to overcome the high thermal diffusivity of the copper, and processes such as electron beam welding (EBW) offer such capability [3]. However, poor mixing of the materials and some defects such as porosity and micro-cracking were observed in EB welded copper–stainless steel [4]. Meanwhile, the interface diffusion control is a key point in Cu/steel dissimilar metal welding process with high heat input, because the mechanical properties of the weld will be deteriorative due to enhanced interface diffusion capability [5]. Due to the problems encountered with cladding copper on steels, solid state processes such as explosive welding are more generally preferred [6,7]. A few studies have shown that laser welding of copper and ⁎ Corresponding author. E-mail addresses: [email protected], [email protected] (M. Haghshenas).
http://dx.doi.org/10.1016/j.matchar.2015.02.022 1044-5803/© 2015 Elsevier Inc. All rights reserved.
steels is feasible [8]. However, the need for higher productivity and material deposition rates has also driven the study of hot-wire tungsten inert gas welding [9]. Although the deposition rates were reported to be 3.1 kg/h, poor wetting to the steel substrate occurs with low arc current, and excessive iron dilution into the copper layer was noted at higher arc currents. It has also been shown that iron dilution into copper will further promote microcracking in the solidified material [10]. Since many issues stem from the interaction of the molten copper with the steel, solid state welding processes may provide a beneficial alternative for joining of these materials. Friction stir welding (FSW) was developed in 1991 by The Welding Institute (TWI) for joining heat treatable aluminum alloys and other materials which are difficult to join by arc welding [11]. The process of FSW involves plunging a rotating cylindrical tool into the interface between two plates, and generating sufficient frictional heat and deformation to consolidate the two materials together [12,13]. FSW has been shown to be applicable for joining dissimilar materials such as Al/Mg alloy [14], Al/Fe alloy [15] and Al/Cu alloy, because the formation of low melting point intermetallic compound (IMC) was limited, and metallurgical bonding was confirmed at the interface [16]. The strength of the weld is drastically reduced when IMCs involving AlCu, Al2Cu and Al4Cu9 form at the Al/Cu interface [17–19]. Meanwhile, defects such as microcracks and voids were observed in the weld [17]. The use of FSW has been demonstrated to join pure copper in sections up to 50 mm thick [20–23], where defect-free welds were produced, and tensile strength of the FSW copper joint was higher than that made by EBW [22], with a joint efficiency of 94% reported [22,24]. FSW also has been developed for sealing copper canisters for the storage of
2
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
nuclear waste, since the copper provides excellent corrosion protection [25]. Furthermore, FSW has been applied to join copper to steel in both lap and butt weld configurations with no defects, where mechanical mixing and metallurgical bonding conformed at the Cu/St interface in lap weld configuration [26,27]. The present work provides the first comparison between FSW lap welding of copper cladding sheets to a steel substrate versus cladding by pulsed-current gas metal arc welding (GMAW). The FSW process provides a method of potentially applying a cladding layer by staggering a series of lap-joints on a sheet of cladding material. This can potentially be highly productive since a copper sheet with quite high thickness can be employed, however the bonding microstructures and bonding mechanisms have yet to be examined and compared to conventional overlay fusion welding methods.
2. Experimental details For the case of FSW, lap welding was performed on a 2.1 mm thick C10200 copper sheet with a hardness of 70 HV, and a 25.4 mm thick A516 Gr. 70 steel substrate with a hardness of 155 HV. The tooling during FSW was a Co–WC cermet, consisting of a 12 mm shoulder, and a smooth 2.1 mm long pin with a tapered geometry (where the diameter increased from 4 to 5 mm), as shown in Fig. 1a. A second ‘3-flat’ tool with the same geometry was also investigated, with the exception of three 0.5 mm flat surface ground into the pin at equal 120° spacings (see Fig. 1a). The 3-flat geometry was considered in order to compare the effect of enhanced mechanical deformation and strain compared to a smooth pin tool. Co-axial alignment of the FSW in the holder was measured to be within 0.035 mm. A tool rotation speed of 1120 RPM and travel speed of 31.5 mm/min was used, with a tool tilt angle of 2.5°, and the tool penetration ranged from 0.06 to 0.60 mm into the steel substrate. It should be noted that in preliminary trials, parameters were also compared using different combinations of 900 RPM and travel speeds of 90 mm/min, however only superficial bonding was produced. The parameters applied appeared to be the minimum required for bonding using the FSW equipment used. No preheating was applied during FSW. Initially two single pass welds were reproduced using varied penetration depths. The microstructures of both single-pass FSW lap welds were compared multiple pass FSW joints staggered at 4 mm intervals, with 5 passes of 100 mm length were produced adjacent to each other using the smooth pin tool with a penetration of 0.25 mm. No tool wear was apparent after the 5 adjacent passes.
Fig. 1. (a) Schematic illustration of the tool geometry, (b) photo of the Cu/steel FSW multipass lap joint surface.
The GMAW claddings were produced using pulsed current deposition with a 1.14 mm diameter AWS A5.7 ERCu composition wire, travel speeds of 7.45 mm/s, heat input of 725 J/mm, wire feed speeds of 7.37 m/min, and an inter-bead spacing of 4.0 mm. The initial preheat during GMAW was 100 °C with inter-pass temperatures maintained below 300 °C. Standard metallographic preparation techniques were applied, with final polishing using 1 μm diamond slurry. Specimens were first etched using 2% Nital to reveal the steel microstructures, then etched with 60% Nital to reveal the copper grain structure. The joint microstructures were examined by a combination of optical metallography, scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The SEM analysis was performed on a JEOL 6460 instrument equipped with energy-dispersive X-ray (EDX) spectroscopy and operated at 30 keV. TEM analysis involved using a focused ion beam to extract the copper/steel interfacial region in both claddings, and analysis of the thinned foil was performed using a Titan LB instrument. In addition, Vickers microhardness measurements were conducted in both samples to detect any possible embrittlement of the heat affected zone (HAZ) of the steel. 3. Results 3.1. FSW Cu/steel lap joint The copper sheet was adhered to the substrate with minimal flash ejected from the surface as shown in Fig. 1b, and multiple FSW passes could be staggered adjacent to each other without severe accumulation of the flash on the surface. In an individual pass, it is clear that a wavelike or ‘hook’ feature is formed in which the steel has been displaced into the upper copper sheet (see Fig. 2a), which provides strong mechanical interlocking at the Cu/steel interface. The bonded area occurs across the interface which spans this intermixed region, roughly corresponding to the diameter of the tool pin. It is clear that the Co–WC FSW tool penetrated into the steel, and the bonding occurs only along the width of the pin in this displaced region. The cross-section of the Cu/steel interface in the bonded region is shown in Fig. 2b when using the smooth pin tool and a penetration of 0.25 mm. A smooth interface is observed near the centerline of the joint, while a series of ridges or ripples with a 10 μm-amplitude occurs near the sides of the interface, as shown in high magnification in Fig. 2c & d. These ridges appear to promote interlocking between the two materials, similar to that observed in explosion welds [28]. The origin of these ridges appears to be related to the oscillations imposed during the rotation of the tool since their spacing increases towards the outer edges of the bonded interface [29]. The spacing of the ridges becomes finer towards the outer periphery of the stir zone (Fig. 2c), which is likely due to overlapping material flow staggered at finer spacings near the sides of the pin. This can be visualized in a prior work explaining the recirculating flow during FSW [30]. Multiple FSW passes (all beginning at the same side) were traversed in the same direction on the lapped Cu/steel, at a spacing of 3 mm between passes with varying pin penetration depths. The first passes were made using both the smooth pin tool, and then later passes were made using the 3-flat tool pin geometry in order to determine how tool penetration and pin geometry influence the microstructure of the bonded interface. Fig. 3 shows the interfacial region between the two passes made using the smooth pin tool with 0.15 and 0.19 mm of penetration, respectively. It can be noted that 0.15 mm of pin penetration did not promote complete bonding with the steel substrate, and in both cases a large number of steel fragments were noted in the stir zone of the welds. A void can also be noted between the boundaries of the two passes. In order to determine if bonding could be improved with different tool geometry, a 3-flat pin was investigated. Welds were made at progressively increasing plunge depths from 0.06 to 0.60 mm using the 3-
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
3
Fig. 2. (a) Macrograph of transverse section in a single-pass FSW Cu/steel lap weld, and (b, c and d) interface microstructures.
flat tool, see Fig. 4. It was found that partial bonding between the sheets occurred only when at least 0.23 mm of penetration was applied using the 3-flat tool, and the interface contained steel fragments and voids (which are evident in Fig. 4a). The voids and steel fragments remained when the penetration was increased to 0.49 mm (see Fig. 4b). This issue remained when 0.60 mm penetration was applied using the 3flat tool, however, when a weld was made with the smooth-pin tool directly adjacent to this weld with only 0.27 mm penetration, no voids were produced as shown in Fig. 4c. This appears to be a result of the increased cutting action of the 3-flat pin, which caused the steel fragments to be displaced into the copper layer as indicated in Fig. 4. Due to the compliance of the FSW equipment frame, the cutting action allowed further penetration into the steel substrate under the high axial tool loads. Considering the improved bonding and void-free interface microstructures produced using the smooth-pin FSW tool with a pin penetration of 0.25 to 0.27 mm, only these conditions are examined further in terms of microscopy at the interface. The interface microstructures observed by SEM microscopy reveal that the interface of the FSW lap joint was flat and free of microcracks, microvoids and intermetallics as seen in Fig. 5. However, EDX quantification produced inconsistent quantification since there appeared to be non-uniform etching by-products deposited near the copper/steel interface. These produced high copper EDX quantifications of 12 wt.%
copper at location 1 in Fig. 5, while 22.8 wt.% copper was detected at location 2 farther away from the copper interface, and location 3 (the same distance from the interface as location 2) showed 1 wt.% copper. The etching by-products appear to correspond to the light patches on the steel near the interface in Fig. 5. Consequently, further analysis of the interface structure and chemistry was carried out using a combination of ion milling and TEM analysis in order to determine the extent of mutual diffusion at the interface. It is also worth noting that the recrystallized grain size was comparable in the copper sheet before and after FSW. The as-received microstructure of the copper sheet is shown in Fig. 6a, which shows an average grain size of 21 μm. In comparison, the grains in the copper had an average size of 25 μm after FSW, see Fig. 6b. This is consistent with prior work involving FSW of pure copper, where typical grain size of less than 100 μm were observed [22]. Typical annealing twins can be noted in both microstructures in Fig. 6, clearly indicating that recrystallization has occurred. 3.2. GMAW cladding The copper cladding produced by GMAW is shown in Fig. 7, with several layers staggered adjacent to each other. The surface was found to contain numerous microcracks at 45° angles to the welding direction.
Fig. 3. Optical micrograph of the bond interface in multi-pass FSW Cu/steel lap joints made using 0.15 and 0.19 mm.
4
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
Fig. 4. Optical micrograph of the bond interface in FSW Cu/steel lap joints made using 3-flat tool with (a) 0.23 mm and (b) 0.49 mm, and (c) 0.60 mm penetration (compared to a smooth pin tool with 0.27 mm penetration on right side).
Evidence of these microcracks can be observed in the cross-section of the overlay, as evident in the surface crack indicated by the arrow in Fig. 8a, and nearly all such cracks took an inter-granular path. The copper cladding grain size was well over 1 mm, and there also appears to be scattered porosity as indicated by dark circular spots in Fig. 8b. However it should be noted that many finer spots in Fig. 8b correspond to surface pits produced by etching. Near the steel interface, the copper grains are more refined (see Fig. 8b). 3.3. Steel HAZ hardness and microstructures The microhardness values along the through-thickness direction are shown in Fig. 9. The hardness values in the steel HAZ of the FSW joint did not exceed a value of 200 HV, and the copper microhardness slightly increased compared to that of the as-received sheet. The highest hardness value achieved was 225 HV within the steel adjacent to the fusion line in the GMAW cladding. It is worth noting that the hardness values in the steel for both processes remained below the critical value of
248 HV (or 22 HRc), which is a key requirement for many corrosion applications [31]. This is usually required in order to avoid cracking or delamination of the cladding in corrosive environments. The microstructures of the steel HAZ at the locations corresponding to −0.65 and −0.15 mm (i.e., below the bonded interface) are shown in Fig. 10. These indicate that the banded fine-grained ferrite and pearlite structure in the steel base material is refined by the FSW tool, as shown in Fig. 10a. This banded structure is progressively refined further closer to the interface, and is completely broken up into an equiaxed fine grained structure of ferrite and pearlite immediately below the interface as shown in Fig. 10b. This indicates that the temperature in this region remained below the critical A1 steel transformation temperature (727 °C) in the FSW cladding. In the case of the GMAW overlay, the steel HAZ microstructures were mainly acicular ferrite, with a small fraction of lower bainite, see Fig. 11. This is consistent with the hardness values observed in Fig. 9, and suggests that no martensite was formed even close to the copper/ steel interface. The grain structure in the steel at the interface appears
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
5
Fig. 7. Cu cladding produced by GMAW.
occur in various copper cladding processes on steel substrates. The depth of over a dozen copper penetrated zones was measured and these varied in length from 5 to 50 μm. 3.4. TEM studies
Fig. 5. SEM micrograph of FSW joint Cu/steel interface, and corresponding EDX quantifications obtained from regions 1, 2 and 3 in the micrograph.
to remain b5 μm. Several instances of liquid copper penetration could be noted in the interface of the GMAW cladding, as noted by the arrow in Fig. 11. As mentioned previously, this has been noted to
Fig. 6. SEM micrographs showing (a) as-received copper sheet grain structure, and (b) stir zone copper grains in the FSW joint.
The Cu/steel interface of the FSW joint was extracted using the Focused Ion Beam (FIB) milling approach, and the TEM bright field image of a FSW joint interface is shown in Fig. 12. The TEM images show that there is metallurgical bonding between the steel and copper. Some nano-sized pores (100 to 200 nm) were observed at the interface. However, it is not clear whether these originated from sample preparation or possibly residual entrapped oxides which fell away during sample extraction. The grain structure in the steel was much finer (b 1 μm),
Fig. 8. Optical micrograph of microstructure produced in Cu cladding of the GMAW overlay (a) showing grain structure through the thickness and location of a grain boundary crack at the arrow, and (b) at the Cu/steel interface with larger pore indicated by an arrow.
6
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
Fig. 11. Optical micrograph showing the steel HAZ microstructure in the Cu cladding produced using GMAW, with the location of a liquid Cu penetration indicated. Fig. 9. Comparison of through-thickness microhardness distribution across the cladding interfaces.
compared to the grains in the copper cladding. The dislocation density in the copper was also much higher than that of the steel grains. The EDX mapping shown in Fig. 12b & c shows that there was negligible diffusion of copper into the steel grains, since the interface appears to be sharp and delineated. Also, no Mn or Si could be detected by EDX in the copper side of the interface. The GMAW cladding interface which was extracted by FIB was also studied by TEM, and the micrographs are shown in Fig. 13. The micrograph indicates that metallurgical bonding is also produced, with no pores being observed at the interface of the GMAW cladding. It is clear that the grain size of the steel at the interface is much larger in the
Fig. 10. Steel microstructures near indents at (a) −0.65 mm and (b) −0.15 mm away from the copper interface in the FSW Cu/steel joint (referenced to positions in Fig. 9).
GMAW joint shown in Fig. 13 (compared to the FSW joint interface shown in Fig. 11), and the dislocation density in the GMAW is much higher as well. The dislocation density in the copper grains is also rather high, and it appears that a dislocation cell structure has developed suggesting more recovery has occurred than in the FSW joint. The EDX analysis in Fig. 13c indicates that the some diffusion of Si and Mn towards the copper grains have occurred in the GMAW cladding. 4. Discussion It is clear that metallurgical bonding between copper and steel is produced using both processes. The FSW joint appears to have a minor contribution from mechanical interlocking as a result of fine ripples formed in a part of the interface between the materials. These are likely formed by the oscillations of the tool, since their spacing decreases towards the outer edges of the weld, much like the spacing of typical ‘onion ring’ patterns observed in FSW joints [32]. Since the coaxial alignment of the FSW tool with the tool holder can never be absolutely perfect, there will always be some cyclical component to the tool motion even when a cylindrical tool is used. In the present case, there was 0.035 mm of run-out measured, and this would lead to a slight vertical oscillatory movement of the tool when it is tilted (2.5° in this case), thus promoting the formation of slight ripples at the interface as shown in Fig. 2c. This aspect could potentially be exploited to some degree further by employing higher tilt or having lower tolerances on tool alignment. The selection of FSW tool geometry and penetration depth appears to play a major role in establishing a uniform bond across the Cu/steel interface when multiple tool passes are used to cover a large surface. As shown in Fig. 3, the FSW lap welding process appears to be sensitive to the depth achieved in the steel substrate when a smooth pin tool is used, where at least 0.19 mm of pin penetration appears to be required to achieve bonding. The use of a tool with 3-flats appears to have led to much higher penetrations (0.60 mm) when similar processing parameters were used (see Fig. 4). However, this led to the formation of extensive defects and voids near the sheet interface. The corners on the tool pin of the 3-flat tool promote cutting of the steel, and the steel fragments were broken away and dispersed into the stir zone of the copper cladding. This material removal allowed much higher tool penetration, but also led to the formation of excessive flash by the tool shoulder on the surface of the tool. The most significant microstructural difference between the FSW and GMAW cladding processes is the grain size of the copper layer, which can be expected based on the prevailing literature on FSW which typically reports grain sizes in the stir zone of b 10 μm. These grain sizes do vary with parameters during FSW, and it has been shown that the grain size in pure copper increases from 3.5 to 9 μm as the tool rotation speed increases from 400 to 800 RPM [33]. However, it should be noted that the FSW travel speed was only 31.5 mm/min, and so the cladding grain size may be refined at higher travel speeds.
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
7
Fig. 12. TEM micrographs of (a) FSW joint interface between copper and steel, (b) EDS map location and (c) EDS element maps for Fe, Cu, Mn.
In any case, the FSW cladding grain size observed here (25 μm) was much finer than that observed in the GMAW copper cladding. This has important implications for the desired final properties and applications, for example where creep properties are concerned [34]. One challenge posed by the extremely coarse grain sizes of the GMAW cladding is the difficulty during subsequent ultrasonic inspection. The large grain size will attenuate the ultrasonic beam by increasing the scattering losses thereby reducing the ability to detect defects [35]. This suggests that the inspection of the FSW joint will be more convenient, since the grain size was well under 50 μm in the stir zone. The grain structure of the steel substrate was initially a fine ferrite plus pearlite mixture, as shown in Fig. 10a. In the case of the FSW interface, this structure only became more refined and deformed, as shown in Fig. 10b, which suggests that the peak temperature was below the A1 steel transformation temperature. This readily explains why no liquid copper penetration was observed in the FSW joint, since the mechanism for liquid penetration depends on the formation of austenite in the steel and the FCC structure is very readily wetted by the liquid copper [1]. From the phase diagram in Fig. 14 [36] liquid copper and austenite phase co-exist above 1096 °C. Thus, the temperature at the interface during GMAW cladding was at least 1096 °C. Many have focused on alternative cladding processes in order to help avoid this liquid penetration mechanism, such as solid state methods like explosion welding. Other strategies such as buttering layers using Ni–Cu alloys may also help avoid this issue by providing a
more gradual interface and separating the steel austenite grains from the liquid copper. This can be expected since Savage et al. noted that liquid copper embrittlement could be avoided in alloys with extensive copper solubility, such as Ni, Mn, Pd, and Pt [37]. The formation of microcracks was rather prevalent across the surface of the GMAW cladding shown in Fig. 7. This microcracking is likely mitigated in the FSW joint due to the fact that the temperatures were lower, based on the difference in the steel microstructure at the interface. It has been shown that residual stresses are drastically lower in FSW joints compared to fusion welds [38]. While it may be possible to mitigate the microcracks in the GMAW cladding by increasing the preheating temperature, and hence reducing the thermal gradient, this would also likely promote further liquid copper penetration into the steel substrate since the peak temperature and solidification time would increase. At the interface, there appear to be some nano-sized voids formed in the FSW joint (see Fig. 12a). The voids could have possibly formed due to inadequate mixing of the two materials during FSW, considering their drastically differing properties at high temperature. However, it is also possible that these voids could be the remnants of residual surface oxides which were not fully dispersed during FSW, and this has been known to produce kissing bond structures in similar FSW joints as well [39]. These could potentially be preferentially removed due to differences in ion milling during sample preparation, or become dislodged during TEM sample preparation and manipulation.
8
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
Fig. 13. TEM micrographs of (a) GMAW joint interface between copper and steel, (b) EDS map location and (c) EDS element maps for Si, Fe, Cu, Mn.
Although some mutual diffusion across the bond was suggested the EDX line scan of the FSW joint in Fig. 5b, the values were under 1 wt.%. This value is rather low and explains why the element mapping conducted by TEM and EDX reveals a rather sharp interface in Fig. 12c. It should be noted that no intermetallics are possible in the Cu–Fe system as shown in Fig. 14. The TEM results of GMAW are consistent with the grain structures observed by optical and SEM, since the size of the cells shown in Fig. 13a do not represent the grains shown in Fig. 8 or 11, but rather the cells reflect a substructure that is formed within the individual grains. The grains in the steel and copper in the GMAW are actually larger than the field of view in the sample extracted from the GMAW cladding. Convergent beam electron diffraction patterns were obtained from the copper and iron grains in both TEM samples. However, no orientation relationship could be identified between the two phases. In FSW, severe plastic deformation and recrystallization establish the grain structure and orientation. In the case of the GMAW cladding it is possible that epitaxial growth of copper grains from the molten state could occur from the steel surface. However, the steel would be heated to above the A1 temperature by the liquid copper, and so the subsequent orientation of the steel grains will be determined
by the solid state transformation of austenite into ferrite and carbide, thus obscuring any orientation relation. 5. Conclusions A preliminary investigation has shown that cladding of steel with pure copper is possible using either FSW or GMAW, where both processes achieve metallurgical bonding with the substrate. The FSW process yielded a significantly finer grain structure in the copper, with no evidence of cracking in the cladding or liquid copper penetration into the steel substrate. The FSW joint appeared to contain some evidence of nano-pores at the interface of the steel and copper. The steel microstructure adjacent to the fusion line suggests that the temperature remained below the austenite temperature of the steel. The copper cladding produced by GMAW contained many surface cracks which are attributed to high thermal stresses. There was evidence of liquid penetration of the copper into the steel, and diffusion of Mn and Si into the copper cladding due to the higher temperatures imposed. The steel adjacent to the fusion line maintained a hardness value below 248 HV when using either process, since no martensite was detected in both cases.
Z. Shen et al. / Materials Characterization 104 (2015) 1–9
9
Fig. 14. Binary phase diagram of iron and copper, after Ref. [36].
Acknowledgments The authors would like to thank the Natural Sciences and Engineering Research Council (NSERC) of Canada, and Nuclear Waste Management Organization of Canada for their support. The EM research described in this paper was performed at the Canadian Centre for Electron Microscopy at McMaster University, which is supported by NSERC and other government agencies. Support from Linamar (CAMTAC Manufacturing) in Guelph, Ontario for providing the WC tools is also appreciated. References [1] A.R. Cox, J.M. Winn, JISI 203 (1965) 175–179. [2] R. Eborall, P. Gregory, Inst. Met. 84 (1955) 88–90. [3] R.A. Andrews, Friction stir welding — an alternative method for sealing nuclear waste storage canisters, SKB Technical Report TR-04-16, Svensk Kärnbränslehantering AB, 2005. [4] I. Magnabosco, P. Ferro, F. Bonollo, L. Arnberg, Mater. Sci. Eng. A 424 (2006) 163–173. [5] T. Yang, H. Gao, S. Zhang, L. Wu, Acta Metall. Sin. 26 (2013) 328–332. [6] G.R. Cowan, A.H. Holtzman, J. Appl. Phys. 34 (1963) 928–939. [7] K. Raghukandan, J. Mater. Process. Technol. 139 (2003) 573–577. [8] T.A. Mai, A.C. Spowage, Mater. Sci. Eng. A 374 (2004) 224–233. [9] S.X. Lv, Z.W. Xu, H.T. Wang, S.Q. Yang, Sci. Technol. Weld. Join. 13 (2008) 10–16. [10] M. Turna, M. Sahul, J. Ondruska, J. Lokaj, Annals of DAAAM for 2011 & Proceedings of the 22nd International DAAAM Symposium, 2011. Vienna, Austria, EU. [11] W.M. Thomas, E.D. Nicholas, J.C. Needham, M.G. Murch, P. Temple-Smith, C.J. Dawes, International Patent Application PCT/GB92/02203 and GB Patent Application 9125978.8, UK Patent Office, London, December 6, 1991. [12] R.S. Mishra, Z.Y. Ma, Mater. Sci. Eng. R 50 (2005) 1–78. [13] P.L. Threadgill, A.J. Leonard, H.R. Shercliff, P.J. Withers, Int. Mater. Rev. 54 (2009) 49–93. [14] Y.J. Kwon, I. Shigematsu, N. Saito, Mater. Lett. 62 (2008) 3827–3829. [15] T. Watanabe, H. Takayama, A. Yanagisawa, J. Mater. Process. Technol. 178 (2006) 342–349. [16] P. Xue, B.L. Xiao, D. Wang, Z.Y. Ma, Sci. Technol. Weld. Join. 16 (2011) 657–661. [17] T. Saeid, A. Abdollah-Zadeh, B. Sazgari, J. Alloys Compd. 490 (2010) 652–655.
[18] A. Elrefaey, M. Takahashi, K. Ikeuchi, Weld. World 49 (2005) 93–101. [19] J. Ouyang, E. Yarraparredy, R. Kovacevic, J. Mater. Process. Technol. 172 (2006) 110–122. [20] C.G. Andersson, R.E. Andrews, Proceedings of the First International Symposium on Friction Stir Welding, Thousand Oaks, CA, USA, June, 1999, 1999. [21] L. Cederqvist, R.E. Andrews, 4th International FSW Symposium, Park City, UT, USA, May 14–16, 2003, 2003. [22] W.B. Lee, S.B. Jung, Mater. Lett. 58 (2004) 1041–1046. [23] C.-G. Andersson, R.E. Andrews, B.G.I. Dance, M.J. Russell, E.J. Olden, R.M. Sanderson, Proceedings, 2nd International Symposium on FSW, Gothenburg, Sweden, June, 2000, 2000. [24] D. Avula, R.K.R. Singh, D.K. Dwivedi, N.K. Mehta, World Acad. Sci. Eng. Technol. 74 (2011). [25] L. Cederqvist, FSW to manufacture and seal 5 cm thick copper canisters for Sweden's nuclear waste, 6th International Symposium on Friction Stir Welding. SaintSauveur, Canada, 10–13 October, 2006. [26] M. Shamsujjoha, B.K. Jasthi, M. West, C. Widener, Frict. Stir Weld. Process VII (2013) 151–160. [27] Y. Imani, M.K. Givi, M. Guillot, Adv. Mater. Res. 409 (2012) 263–268. [28] M. Hammerschmidt, H. Kreye, Microstructure and Bonding Mechanism in Explosive Welding in Shock Waves and High Strain Rate Phenomena in Metals, in: M.A. Meyers, L.E. Murr (Eds.),Plenum Press, 1981. [29] B. Yang, J. Yan, M.A. Sutton, A.P. Reynolds, Mater. Sci. Eng. A364 (2004) 55–65. [30] P. Su, A. Gerlich, T.H. North, G.J. Bendzsak, Metall. Mater. Trans. 38A (2007) 584–595. [31] NACE MR0175/ISO 15156-3:2003(E), Petroleum and Natural Gas Industries — Materials for Use in H2S-containing Environments in Oil and Gas Production — Part 3: Cracking-resistant CRAs (Corrosion-resistant Alloys) and Other Alloys, NACE/ANSI/ISO, 2003. (©2003). [32] R.D. Haley, A.S. Judy, C.N. Arthur, Metall. Mater. Trans. A 45A (2014) 4411–4422. [33] G.M. Xie, Z.Y. Ma, L. Geng, Scr. Mater. 57 (2007) 73–76. [34] R. Sandström, H. Östling, L.Z. Jin, Modelling of creep in friction stir welded copper, Mater. Res. Innov. 17 (2013) 350–354; T.B. Massalski, Binary Alloy Phase Diagrams, ASM International, New York, 1990. [35] I.N. Ermolov, B.P. Pilin, NDT Int. 9 (1976) 275–280. [36] Alloy phase diagrams, 10th ed., ASM Materials Handbook, volume 03, ASM International, December 1, 1992. [37] W.F. Savage, E.P. Nippes, M.C. Mushala, Weld. J. (1978) 145s–152s. [38] O. Hatamleh, I.V. Rivero, A. Maredia, Metall. Mater. Trans. 39 (2008) 2867–2874. [39] Y.S. Sato, H. Takauchi, S.H.C. Park, H. Kokawa, Mater. Sci. Eng. A405 (2005) 333–338.