Abstract
Friction stir spot welding has been shown to be a viable method of joining ultra high strength steel, both in terms of joint strength and process cycle time. However, the cost of tooling must be reasonable in order for this method to be adopted as an industrial process. Recently a new tool alloy has been developed, using a blend of polycrystalline cubic boron nitride (PCBN) and tungsten rhenium (W–Re) in order to improve the toughness of the tool. Wear testing results are presented for two of these alloys: one with a composition of 60% PCBN and 40% W–Re, and one with 70% PCBN and 30% W–Re. The sheet material used for all wear testing was 1·4 mm DP 980. Lap shear testing was used to show the relationship between tool wear and joint strength. The Q70 tool provided the best combination of wear resistance and joint strength.
Introduction
Joining of high strength steel sheets is a challenge that automakers are currently dealing with, because resistance spot welding (RSW) of these alloys results in brittle microstructures and microcracks. Some studies have postulated that shrinkage stresses in the weld nugget, along with the presence of zinc, were responsible for the forming of cracks. 1 1,2 Another issue that occurs when RSW is used to join high strength steel sheets is the type of fracture mode which appears during testing of the weld. One study showed that DP 600 exhibited interfacial failures, or failures at the weld interface, when joined by RSW.1– 3 It was found that using specific weld currents and weld times could reduce interfacial failures and cause the desired ‘button pullout’ failure.1 The ‘button pullout’ failure is viewed within the automotive industry as an indication that the weld has the capability of absorbing energy during a crash, with ductile tearing occurring around the periphery of the weld, rather than a brittle fracture occurring at the weld interface. While DP 600 steel can be welded by RSW in the laboratory under controlled conditions, welding process control is a challenge for the production environment and more refined control strategies are required.4 Resistance spot welding of DP 980 alloy has also been studied and the results were mixed, with some failures occurring at the weld interface and some failures occurring in ‘button pullout’ mode. In particular, when expulsion of the weld occurred, which is common in an RSW production process, interfacial failures were the most common type.5
Besides RSW another process used for spot joining of automotive sheet alloys is self-piercing riveting (SPR). This process has been employed primarily for joining of aluminium sheets, as in the production of the aluminium body Jaguar X350. 6 6,7 Self-piercing riveting involves forming the sheets to be joined around a steel rivet, thus forming a purely mechanical joint, without the need for predrilling a hole. Self-piercing riveting works well for joining sheets with reasonable ductility. It can be used for joining dissimilar combinations of aluminium and steel with moderate strength, for example. Studies on the use of SPR include those on joining of aluminium alloys,8– 13 the joining of dissimilar combinations of aluminium and steel,10,13– 15 as well as a limited amount of work on joining of steel alloys.13 The vast majority of work carried out on SPR involves the joining of aluminium sheets, or dissimilar combinations of aluminium and steel sheets. There has not been any published work on the joining of similar combinations of alloys like DP 980, and the likely reason is that SPR does not work well for less ductile, very high strength metals, because of the need to form the sheets around the rivet.
Friction stir spot welding (FSSW) is another process for spot joining of steel sheets. Prior work has produced promising results in high strength steels, with good joint strength, an absence of cracks in the weld, and a ‘pullout’ type failure mode. 5 5,16 All of these characteristics are facilitated by a solid state bonding mechanism which avoids the problems that result from melting and solidification of the weld pool. However, one of the issues that must be studied for FSSW of steels especially is the life of the tool. Tool life affects the number of tool changes made during production and also affects the cost of the process. There are several tool materials which have been used for FSSW of high strength steels, including silicon nitride (SiN), polycrystalline cubic boron nitride (PCBN), tungsten rhenium (W–Re), tungsten carbide (WC), and other tungsten based alloys.16– 25 While these materials have been studied for FSSW in regard to joint properties, there has been no published study of the wear properties of these materials for spot joining of high strength steel like DP 980, and therefore the tool life and effect of tool wear on joint properties is not known. The present paper will present studies on the wear of tools composed of a combination PCBN and W–Re. The effect of tool wear on joint properties will also be studied and discussed.
Experimental
Two tool materials were tested for wear resistance in FSSW of DP980 steel. One tool had a composition of 60% PCBN and 40% W–Re (referred to in the present paper as Q60), while the other tool had a composition of 70% PCBN and 30% W–Re (referred to in the present paper as Q70). The objective of alloying W–Re with PCBN was to achieve a good combination of hardness and toughness in order to withstand the high stresses and abrasive conditions inherent in FSSW. Unpublished prior work by the authors showed that cracking occurred in tools made only of PCBN, so the addition of W–Re was intended to increase toughness. The use of W–Re as an addition to PCBN was first performed for friction stir welding tools designed for linear welding.26 The microstructure of PCBN/W–Re is shown in Fig. 1a for the tool with 60% PCBN and in Fig. 1b for the tool with 70% PCBN.
a microstructure of 60% PCBN/40% W–Re, where darker material is PCBN (note that while greater fraction of material is PCBN in this case, micrograph can locally show what appears to be region with greater W–Re content) and b microstructure of 70% PCBN/30% W–Re
Both tools had a 10 mm diameter concave shoulder and a pin with three flats, as shown in Fig. 2.
Friction stir spot welding tool with three flat pin and slightly concave shoulder
Spot welds were produced in a tight pattern on two overlapped sheets of DP 980 steel sheet with thickness of 1·4 mm. Spot welding was performed on a Kearney and Trecker three axis mill that had been converted to computer numerical control operation with variable revolutions per minute (RPM) capability and displacement control. Flexing in the frame of the machine accounted for ∼0·36 mm of travel in the z direction for the experiments presented in the present paper, and therefore, the plunge depths were adjusted to account for this flexing over the course of the experimental work. After the first 50 welds were completed in each case six lap shear specimens were produced in order to evaluate joint strength and to prepare cross-sections of the weld. The lap shear specimens were 100 mm long and 25 mm wide, with a 25 mm overlap. The weld was placed in the centre of the overlap, where the weld diameter was ∼10 mm. This process was repeated every 50 welds until 250 welds had been completed. From that point forward, lap shear specimens were made after approximately every 100 additional welds. A picture of the FSSW of a lap shear specimen is shown in Fig. 3.
Friction stir spot welding of lap shear specimen
Tool wear was measured each time wear testing was interrupted to produce lap shear specimens. In order to measure wear, the tool was placed on an optical comparator where its silhouette was photographed, as shown in Fig. 4.
Friction stir spot welding tool photo on comparator
Photo editing software was then used to convert the border of the tool image to a solid coloured line. These lines were superimposed onto each other in order to graphically show the progression of tool wear over time. In addition to the graphical representation of tool wear, lap shear strength was plotted over time as a function of the number of spots produced, in order to evaluate effective tool life. Microhardness maps were also generated for welds made with both tools, in order to show the effect of tool material on joint properties. A Leco LM100 machine with a diamond indenter was employed for microhardness testing, using a 500 μm spacing between each indentation and a 1 kg weight setting.
Results and discussion
Before wear testing was started some welding development was carried out using a Q60 tool. After testing several combinations of RPM, plunge depth, and plunge rate a set of welding parameters was chosen which provided lap shear strength consistently greater than 13·4 kN. This level of strength was chosen to provide a margin of safety beyond the minimum acceptable value of 11·1 kN, which is the AWS standard for DP 980 sheet with 1·4 mm gauge.27 These same parameters were then used for wear testing of both the Q60 and Q70 tools, in order to maintain similar conditions for a comparison of tool life. The welding parameters are shown in Table 1, where the process was divided into two stages: a first stage with a fast plunge rate and a second stage with a slow plunge rate.
Weld parameters for 1·4 mm DP 980 sheet used for wear testing
No shielding gas was used during the welding experiments. The tool holder was cooled, in order to maintain a coolant temperature of 15°C. The Q60 tool was tested first and was found to have an effective tool life of ∼700 welds, as shown in Fig. 5.
Lap shear strength as function of number of welds, for Q60 tool
A drop in lap shear strength for the Q60 tool occurred at ∼220 welds and was caused by a dimple in the anvil supporting the sheets during welding. Since lap shear specimens were produced after every 50 welds, the drop in lap shear load was not noticed immediately, and that is why ∼200 welds were completed before the anvil was replaced. When the anvil was replaced the lap shear strength returned to a level above the minumum standard of 11·1 kN, then eventually dropped below this standard at ∼800 welds. After this point the strength continued to drop with more spots, finally reaching ∼4 kN after 1100 welds. It should be pointed out that the welding parameters remained constant throughout the test, where displacement control was applied to the tool according to the data shown previously in Table 1.
Testing of the Q70 tool was carried out in the same manner. The lap shear strength as a function of the number of spots produced showed an effective tool life of over 1100 welds, as presented in Fig. 6.
Lap shear strength as function of number of welds, for Q70 tool
The hardness of the tool increases in proportion to the content of PCBN, so the greater life of the Q70 tool can be attributed to greater hardness. All other things equal, a harder tool will perform better from a life perspective as long as it is not too brittle. Brittleness is something that was seen in prior unpublished work by the authors to limit life in tools composed of 100% PCBN, where cracking tended to result in failure before abrasive and chemical tool wear caused a drop in joint strength. For the current experiments neither tool was limited by cracking or catastrophic failure. Both were limited by a drop in joint strength caused by a progressive wear of the tool, which reduced both the pin diameter and the shoulder diameter. The profiles of the Q60 and Q70 tools over time are shown in Fig. 7.
Profiles of a Q60 tool and b Q70 tool over course of welding experiments
Tool profiles were taken every 50 welds, up to 250 welds. After that, profiles were taken every 100 welds. So the first 6 profiles are 50 welds apart, while all subsequent profiles are 100 welds apart. For the Q60 tool the top diameter, or the smallest diameter, of the pin was reduced from 2·57 to 2·14 mm, or 17% after ∼1000 welds. The bottom diameter of the pin, or largest diameter, was reduced from 3·89 to 2·93 mm, or 25%. For the Q70 tool, the pin diameter reductions were 17% for the top and 20% for the bottom after ∼1160 welds. The maximum depth of wear on the shoulder was 0·82 mm on the Q60 tool and 0·66 mm on the Q70 tool. The greater wear resistance of the Q70 tool is attributed to greater PCBN content, where PCBN has a hardness of about 3000–3600 kg mm−2 compared to a hardness of 600–650 kg mm−2 for W–Re. The hardness of Q60 was found to be 1800 kg mm−2 while the Q70 had a hardness of nearly 2000 kg mm−2. The welding process was controlled for each weld, such that the zero position of the tip of the tool was determined by touching off on the top surface of the sheet in order to ensure that the depth of plunge was consistent. Since the plunge was maintained at a constant level, the drop in joint strength is attributed principally to the smaller potential bond area that could be obtained by the dimishing size of the tool. Other factors which may have played a role in reducing joint strength over time include a change in the surface condition of the tool, as well a change in shape of the tool, as was seen in Fig. 7.
In terms of the wear mechanisms present during FSSW of steel, there have been studies that investigated the effect of wear mechanisms in linear friction stir welding (FSW). One study found that wear in a PCBN cutting tool could be correlated to wear during linear friction stir welding, both chemical and mechanical wear were observed.28 Another study show that FSW of stainless steel using a PCBN tool showed that wear depended on welding temperatures, where a temperature range of 1100–1200°C caused migration of boron and nitrogen from the tool to the weld.29 While the current results show the effect of wear on the shape of the tool and the mechanical properties of the weld, the wear mechanisms have not been documented. They are likely a combination of chemical and mechanical mechanisms, seen in prior work on FSW, and are likely a function of welding temperature.
The hardness of the welds was measured in order to study the impact of tool composition on weld properties. Hardness maps of weld no. 54 produced by both tools are shown in Fig. 8.
Hardness maps of weld no. 54 produce by a Q60 tool and by b Q70 tool
The weld hardness produced by the Q60 tool is greater than that produced by the Q70 tool, indicating that with the same welding parameters and tool design the Q60 material produced a higher welding temperature. A higher welding temperature results in a greater proportion of austenite being formed during the welding process, which is then transformed to martensite on cooling, because DP steels have high hardenability.30 This explains the greater weld hardness for welds made with the Q60 tool. The higher welding temperature using Q60 could be caused by several different factors. One factor could be that the Q60 tool has a higher friction coefficient, because it contains less PCBN, which has a lower coefficeint of friction than W–Re and other hard tool materials. 31 31,32 Other things being equal, the Q60 tool would generate more heat when using the same welding parameters. However, this is a hypothesis which needs further investigation, since it is not possible to isolate friction coefficient as the only factor in heat generation when the tool materials are not the same, and when the surface condition of the tool is not the same. Nevertheless, the notion of generating greater heating using the Q60 tool based on the evidence of the hardness maps of Fig. 8 is confirmed by a lower vertical welding load for the Q60 tool welding experiments, as can be seen in Fig. 9.
Vertical welding load over time for Q60 and Q70 tools: loads decreases over time with tool wear, and Q60 tool had lower loads than Q70 tool
In terms of joint strength, the Q70 tool produced welds that had essentially the same strength as the Q60 tool, but was more consistent over time. Wear was more significant on the Q60 tool, as can be seen in weld cross-sections in Fig. 10, where the hole left by the pin in the Q60 tool is visibly smaller than the hole left by the Q70 tool after a similar number of welds. The greater shoulder wear of the Q60 tool can also be seen in Fig. 10.
a weld no. 1009 produced by Q60 tool and b weld no. 1054 produced by Q70 tool
The drop in joint strength over time can be attrituted to a diminshing tool size, but there can also be an impact from contamination of the weld by the tool itself. For example there is prior work which showed that joint strength in friction stir spot welded 590 steel was decreased by the presence of silicon and nitrogen which was embedded in the weld by wear of the silicon nitride tool.33 A study of the impact of contamination was not done in the present work, but it is likely that cBN particles and W were introduced into the weld, especially as the tool wear reached more advanced stages. While the presence of W is likely to increase weld hardness in the areas where the tool was in contact with the steel, it is not likely that W would be found within the stir zone itself, based on prior study conducted in stainless steel.34
These results have demonstrated that tools composed of PCBN and W–Re can provide good joint performance over a large number of welds in DP 980 sheet material. The Q70 tool had better and more consistent performance, showing less wear than the Q60 tool. This should be expected since W–Re is not as hard as PCNB. Both tools had significant wear on the shoulder, resulting in a smaller potential bond area and subsequent decreased lap shear joint strength over time. The mechanism which limited tool life in both cases is likely a combination of chemical wear and abrasive wear, not cracking of the tool. Therefore, the toughness that was expected by addition of W–Re to PCBN was achieved. However, based on the results of the present work it is anticipated that a greater proportion of PCBN may be beneficial to resist abrasive tool wear, especially for FSSW of very hard alloys like DP 980.
Conclusions
Wear testing results have been presented for two FSSW tools. One tool had 60% PCBN (Q60 tool) while the other had 70% PCBN (Q70 tool), where the balance was W–Re in both cases. The sheet material used for all wear testing was 1·4 mm DP 980. Tool profiles were measured during the testing process and showed significant shoulder wear over the course of testing for both tools. For the welding parameters used in the present study the Q70 tool provided the best combination of wear resistance and joint strength, lasting for ∼1200 welds, versus 700 welds for the Q60 tool. Microhardness maps and vertical load plots indicate that the Q60 tool generated a higher welding temperature than the Q70 tool. This may be caused by a higher proportion of W–Re in the composition of Q60, which would result in a greater coefficient of friction than the Q70 tool, although further study would be needed to confirm this result.
Footnotes
Acknowledgements
Funding for the present work was provided by DOE PNNL contract no. 116126. Friction stir spot welding tools were provided by Megastir Technologies.