J Perrett1, J Martin1, J Peterson2, R Steel2 and S Packer3
1TWI Technology Centre (Yorkshire)
2Smith MegaStir, USA
3Advanced Metal Products, West Bountiful, USA
Paper presented at TMS Annual Meeting 2011. 27 Feb. - 3 March 2011, San Diego, CA., USA
Keywords: Friction stir welding, steel, high temperature, PcBN, tungsten rhenium, tool wear
Abstract
Friction Stir Welding (FSW) of steel has been in development for over a decade, but only in recent years have the strength and wear characteristics of tool materials improved extensively. Tool materials have been developed that have excellent high-temperature wear resistance, and consistently achieve weld lengths in excess of 45 meters in steel. These newly developed material characteristics comprise of a fine balance of high-temperature strength, hardness and ductility. Shipbuilding, bridge decking, pipe seam welding, or applications which require long uninterrupted steel welds, are now attractive target markets for the friction stir process. It is the process's ability to create high-strength, low-distortion welds that make it potentially attractive to industry.
Introduction
Invented by TWI in 1991, FSW has been applied successfully to the joining of aluminium alloys across many industry sectors, and is now regarded as relatively mature process for these materials. With the development of a new generation of tool materials, FSW has potential applications for the joining of high-softening temperature materials such as steels, nickel alloys, and titanium alloys. In order to survive high temperatures during the welding operation, it is essential that FSW tool materials possess a challenging combination of properties such as high-temperature strength and toughness, thermal and chemical stability, and oxidation resistance. For FSW of ferrous metals such as steels, polycrystalline cubic boron nitride (PcBN) material offers many advantages over conventional ceramic materials. Written out in full in first instance?
This paper presents tool wear and life data for FSW stainless steel and mild steel and the effect upon tool wear of multiple welding operations is briefly described. The latest tool materials are discussed, and a full performance assessment of the tool materials is provided, using 6mm 304L austenitic stainless steel plate and A36 cold rolled mild steel plate as test materials.
Objective
Evaluate the performance of the latest PcBN and composite PcBN/W-Re tool materials for the FSW AISI 304L austenitic stainless steel and A36 mild steel in 6mm thickness plate.
Friction stir welding tool materials
PcBN materials consist of ultra-hard cBN particles and ceramic or metallic binder phases. In some cases, metallic binders may be used to assist the sintering process under high pressure and high temperature (HPHT) conditions. The cBN particles are finely dispersed in the binder phases. Bonding of cBN to the binder phases is desirable to increase the strength of such materials. PcBN is a brittle material in nature and fracture is usually the dominant failure mode for PcBN FSW tools. In addition to reaction sintering (that stimulates the formation of tightly bonded network of cBN and the binder phases), control of the composition of the sintered PcBN material is critical in improving fracture toughness. Recent developments in designing cBN based metal composite materials have resulted in a significant improvement in FSW tool performance for high temperature welding operations. These material developments are represented by the new tool grades Q60, Q70 and Q80.
MS80 grade is a PcBN material with Al-based ceramic binder phases including aluminium nitride (AlN) and aluminium diboride (AlB2). The volume percentage of the ultra-hard cBN phase is approximately 80%. Manufacturing MS80 starts with mixing cBN powder and Al-based binder. The mix progresses through a series of furnace treatments to promote pre-sintering reactions to obtain the desired phase compositions. The powder is then assembled in a can and sintered at high temperature and pressure using either a cubic, belt, or pistol cylinder press. The sintering temperature is typically between 1000 and 1600°C, and the pressure between 2 and 7GPa. HPHT sintering promotes the formation of a tightly bonded network of cBN and binder phases.
As shown in Figure 1, cBN crystals are uniformly distributed among the binder phases. Individual cBN particles appear to be well separated and the binder is continuous. By adjusting various process parameters, such as powder mixing, pre-sintering treatments, and sintering pressure and temperature, control of the phase composition of the PcBN materials can be achieved. Reaction sintering is the key to producing PcBN materials suitable for FSW tools. Although MS80 material has excellent wear resistance properties, fracture is the predominant failure mechanism for this tool, and is the limiting factor with respect to its use in industry.
A new generation of cBN-based metal composite FSW tool materials have been developed for applications where toughness is required, and predictable wear is desirable. Like the MS80 material these materials are produced using HPHT technology. However, instead of a ceramic binder, tungsten and rhenium metals are used as the matrix. The ultra-hard cBN particles are finely dispersed within the metal matrix, forming a tightly bonded structure that takes advantage of both the toughness of the metal matrix and the hardness of the cBN phase. Tungsten and rhenium are relatively brittle metals with melting points in excess of 3000°C. Alloying of rhenium in tungsten significantly improves the room temperature material toughness, this is known as the 'rhenium effect'[1]. Alloying between W and Re is evident under HPHT conditions and the alloying temperature is much lower than the temperature required for densification of W-Re alloys[2]. In addition, reaction bonding between cBN and the metal components can be realised by adjustment of powder composition. When a uniform dispersion of cBN is achievable, an increase in the amount of the ultra-hard phase can improve the wear resistance of the composite material. Figures 2 and 3 show the Q60 and Q70 grades of W-Re/cBN composite material where the volume percentage of cBN
Workpiece materials and equipment
All of the welding trials carried out in this study were made as bead-on-plate (BoP) welds to allow multiple weld lengths to be made alongside one another. The workpiece materials were both 6mm in thickness. The stainless steel test plates were 1.5m in length and the mild steel test plates were 3.6m in length. Argon gas shielding was used to prevent oxidation of the weld and the FSW tools. Parameters were chosen to suit each FSW tool material, with the criteria for a successful weld being a flaw-free weld section and a good quality surface finish with little or no expelled flash. Welding parameters were determined prior to the trials, using spare FSW tools manufactured from each material. Tool profile measurements were taken every one metre for the stainless steel and every three metres for the mild steel.
The AISI 304L austenitic stainless steel welding trials were performed as 1m length welds on the FSW machine at MegaStir Technologies laboratory in the USA. This machine is a converted Kearney and Trecker 1957 universal milling machine fitted with a state-of-the-art FSW control system. All of the A36 cold rolled mild steel welding trials were made as 3m length welds using the purpose built TTI FSW machine located at Brigham Young University, USA. The FSW parameters employed during the trials for each tool material are shown in Tables I and II. Parameters were adjusted during welding in order to maintain the required FSW tool temperature while maintaining weld quality.
Table I. Stainless steel welding parameters
| Tool material | Down force*, kN | Rotation speed**, rev/min | Traverse speed, mm/min | Ramp length, mm | Dwell, s | Tool tilt, ° | ax tool temperature, °C |
| MS80 | 40 | 1100/260 | 100 | 1 | 0 | 0 | ~785 |
| Q60 | 42 | 1100/200 | 125 | 1 | 0 | 0 | ~780 |
| Q80 | 40 | 1100/250 | 100 | 1 | 0 | 0 | ~775 |
*Downforce was adjusted as necessary to maintain the welding temperature.
**Where two figures are quoted, the first figure represents rotation speed during the plunge cycle. Second figure represents rotation speed during traverse.
Table II. Mild steel welding parameters
| Tool material | Down force*, kN | Rotation speed**, rev/min | Traverse speed, mm/min | Ramp distance, mm | Dwell, s | Tool tilt, ° | Max tool temperature, °C |
| MS80 | 29 | 800/200 | 100 | 1 | 3 | 0 | ~775 |
| Q60 | 29 | 800/200 | 100 | 1 | 3 | 0 | ~740 |
| Q70 | 29 | 800/200 | 100 | 1 | 3 | 0 | ~770 |
*Downforce was adjusted as necessary to maintain the welding temperature.
**Where two figures are quoted, the first figure represents rotation speed during the plunge cycle.
Results and discussion
The PcBN (MS80) FSW tool was in good condition after the first weld, apart from a single crack running across the shoulder. Also there was slight roughening of the surface in the area where the shoulder meets the probe; this is believed to be the hottest part of the weld where heat is provided by both the probe and shoulder. As further welds were made, the cracks increased in length and number, and began to meet up on the surface around the base of the probe before propagating into the probe. The probe did not fail until the ninth weld, despite exhibiting five large cracks on the shoulder and probe from the fourth weld onwards. The probe eventually failed during the extraction after completion of the ninth weld, leaving the probe in the exit hole.
Figure 4 shows the wear profile of the tool taken after each metre length of weld. The figure shows that little wear to the tool occurred, with the majority of wear occurring on the probe, near to the probe base. The weld surface finish using the PcBN tool was excellent throughout. Trials performed by MegaStir had concluded that, in order to prolong PcBN tool life, the tool shoulder temperature needs to be limited to approximately 750°C, as temperatures higher than this would cause the aluminium nitride-based binder to soften. The tools are designed to accommodate a thermocouple located at the side of the shoulder. The parameters were adjusted to keep the temperature below this figure. As the tool features began to wear, the temperature increased by 20-30°C, although it is not thought that the increase in temperature was responsible for the tool failure, as the initial crack had appeared during the first weld.
The composite (Q60) FSW tool performed very well in the trials. The design was similar to the PcBN tool but the material showed less vulnerability to cracking than the PcBN. The Q60 tool did not crack at all, and completed a total of 30m of weld length before it was decided to end the trials. Figure 5 shows the wear profile of the Q60 tool with measurements plotted after every 1m of weld up to 10m and then every 5m up to 30m length. The majority of wear occurs at the threads at the base of the probe, and once these features wear away, the tool is less effective at stirring the material resulting in a small flaw in the weld root on the advancing side of the probe (counter-clockwise rotation). The small flaw began to open up towards the end of the tenth weld (10m) when the tool probe features had started to wear away. The features continued to wear as shown by the wear profile graph and the tool profile became smooth and featureless. Once this smooth shape had formed, the profile wore relatively evenly. The tool would have continued to weld but the features required to stir the material sufficiently to prevent defects had been lost. Once the features had worn away completely, the size of the flaw increased substantially. The surface finish was excellent using the composite tools even after the tool began to wear.
The composite (Q80) FSW tool performed in a very similar manner to the Q60 FSW tool apart from a hairline crack on the shoulder (running along the radius) after the first weld. This crack stayed on the tool shoulder for the total weld length of 20m and did not change in size or shape. This is likely to be due to the additional 20% cBN content reducing the overall toughness of the tool. The wear performance was very similar to that of the Q60 tool. Also as with the Q60 tool, once the scroll features wore away, a small flaw opened up on the advancing side of the probe at the weld root. Due to the presence of the flaw, the welding trials were ended at 20m.
During welding of the A36 steel the MS80 PcBN FSW tool exhibited similar cracking as previous experience. The tool was in good condition apart from a single crack running across the shoulder at 6m. However, as welding continued in mild steel the crack did not propagate and the probe did not fail. 45m of BoP were achieved before weld defects were observed. Figure 6 shows that little wear to the tool occurred, with the majority of wear occurring on the probe, near to the probe base. The weld surface finish using the PcBN tool was excellent throughout.
The composite Q60 FSW tool performed very well in the mild steel trials. The design was similar to the PcBN tool but the material showed less vulnerability to cracking than the PcBN. The Q60 tool did not crack at all, although the profile of the Q60 material wore more rapidly it achieved 42m of weld length before weld defects were observed (Fig.7).
The composite Q70 performed similarly to the Q60 albeit with a less rapid wear progression (Fig.8). 42 meters of weld length were achieved prior to weld defects.
Conclusions
Friction stir welding trials were carried out in 6mm thickness AISI 304L austenitic stainless steel and 6mm thickness A36 cold rolled mild steel and tool wear was monitored. The materials tested in the A36 steel (Q60, Q70, MS80) exhibited linear welds lengths previously not possible in steel using FSW. The Q-series materials are only a selection of the materials that use a common material platform and can be tailored to suit unique applications. These materials experiments show promising results with regard to roughness. The main conclusions from this work are:
- The PcBN FSW tools cracked and failed after 9m welding in 6mm stainless steel. PcBN FSW tools developed cracks at 6m in the A36 steel but continued to produce good welds up to 45m in the mild steel without failing. Voids in theweld were observed after 45m. This tool material showed excellent wear resistance during the trials.
- The cBN-based metal composite FSW tool materials showed significant improvements over the PcBN tool with respect tool lifetimes in the stainless steel, with distances in excess of 30m achieved without tool probe failure. The Q60and Q70 grades each achieved 42 meters of BoP weld before weld defects were observed.
- The cBN-based metal composite FSW tool materials wore significantly more in the stainless steel material than in the mild steel material.
References
- Savitskii, E.M., Tylkina, M.A., 1959. "Constitution diagram and properties of alloys of the tungsten-rhenium system," Izv. Akad. Nauk SSSR, Otd. Tekh. Nauk, Metall. Toplivo, No. 3, pp 99-102.
- Skorokhod, V.V., Radchenko, O.G., Uvarova, I.V., Panichkina, V.V., 1983. 'Production of a sintered tungsten-rhenium alloy at low temperatures', Powder Metall. Metal Ceram. 22, 900-903.
