The feasibility of friction stir welding steel (February 1999)

Slightly modified version published in Science and Technology of Welding and Joining, 1999, Vol. 4, No. 6, pp 365-372

1. Introduction

The earliest reference to the use of frictional heat for solid-phase welding and forming appeared over a century ago in a United States patent ^[1] . A period of fifty years then passed before any significant advancement in friction technology took place namely a British patent in 1941 that introduced what is now known as friction surfacing ^[2] . Yet another fifty years went by before friction stir welding (FSW) was invented at TWI ^[3] . This comparatively recent innovation has permitted friction technology to be used to produce continuous welded seams for plate fabrication, particularly in light alloys. Friction stir welding (FSW) is a process for joining workpieces in the solid phase, using an intermediate non-consumable tool see Fig.1.

According to the invention the method comprises a FSW tool of harder material than the workpiece material being welded. The marked difference between the elevated temperature properties of the tool and the workpiece, together with a suitable cyclic movement between the tool and the workpiece, generates sufficient frictional heat to cause plasticised (third-body) conditions in the workpiece material. Thus friction stir welding is a continuous hot shear process that involves slowly plunging a portion of a specially shaped rotating tool between and then along the abutting faces of the joint. The contacting surface of the shoulder of the tool, and the length of the probe below the shoulder essentially allows the probe to maintain penetration to the required through-thickness depth.

Although initially FSW was confined to relatively soft workpiece materials such as lead, zinc, magnesium and a range of aluminum alloy materials, the feasibility of joining copper and in this paper, low carbon chromium steel, and carbon steel has been demonstrated. This range of harder workpiece materials has proved possible by continuing to maintain a suitable differential between the hardness and the elevated temperature properties of the tool compared with the workpiece materials.

Friction stir welding can be regarded as an autogenous keyhole joining technique without the creation of liquid metal. The consolidated weld material is thus free of typical fusion welding defects. No consumable filler material or profiled edge preparation is normally necessary. Already FSW is a practical technique for welding aluminium-based materials, ranging in plate thickness from 0.8mm to 75mm and is in commercial production. Low distortion, cost effective, FSW joints are produced, with excellent mechanical properties being achieved in several aluminium alloys.

Continuing investigations suggest that the FSW of steel will also become commercially attractive for such applications as ships, pipe fabrication, trucks and railway wagons and hot plate fabrication.

This paper describes some results with a plain low carbon steel, 12% chromium alloy steel, and dissimilar 12% chromium steel/carbon steel combinations. These dissimilar welds served to indicate the characteristic flow pattern associated with the FSW process.

1.1 Background

The characteristics of the FSW technique can be compared with other friction process variants, some of which are shown in Fig.2 ^[4] . For example, when Continuous Drive Rotary, Inertia, Linear, Orbital and Arcuate friction welding variants are used to join two bars of the same material and same diameter or aligned cross-section, axial shortening (consumption of the bars) occurs equally from each bar to form a common plasticised 'third-body'. However, differences in diameter or section, lead to preferential consumption of the smaller component. Differences of material in one of the parts to be joined also lead to preferential consumption of the comparatively softer material ^[5] . The unequal consumption and temperature distribution in Rotary friction welding between different diameter bars has already been studied ^[6,7] . This preferential consumption and reprocessing of one component in a friction system has been put to good use in the development of Friction Surfacing, Friction Hydro Pillar Processing and Friction Pillaring, Radial Friction Welding and Friction Plunge Welding. Friction Stir Welding is a further development in that only the workpiece weld region is processed, to form a solid-phase welded joint.

Friction Extrusion and Friction Third-body are exceptions to the latter variants in that the consumed and reprocessed material is introduced into the friction system. This introduced material, which has a comparatively lower thermal softening temperature than the components being welded or the dies used to extrude is frictionally treated to provide a 'third-body' material. Suitably conditioned, this 'third-body' material can be harnessed either as an extruded product or be used as a joining medium.

The lateral movement in Friction Surfacing ^[7,8] and FSW, by introducing new workpiece material at nominally ambient temperature, modifies the already unequal temperature distribution between a comparatively small diameter rotating consumable bar in Friction Surfacing and the rotating tool in FSW, as shown in Fig 2b&c. Both these techniques rely on producing suitable temperature and shear conditions within the 'third-body' transient region that exists in Friction Surfacing between the consumable bar and the substrate, and between the tool and the workpiece in FSW.

In Friction Surfacing any increase in temperature differential (by the intrusion of cold substrate material) enhances the deposition mechanism and allows comparatively harder materials to be deposited onto nominally softer materials ^[8,9&10] . The inherent temperature gradient leads to minimal dilution. However, in FSW the intrusion of cold workpiece material can, in some cases, hinder the welding performance.

1.2 Previous work on the FSW of steel

At the time of publishing only four references in the open literature introduce the feasibility of friction stir welding of steel ^{[11,12,13&14]} . The following section describes recent work at TWI on this topic.

2. Experimental

2.1 Friction stir welding equipment

The friction stir welding trials were carried out on a modified vertical heavy duty-milling machine. The machine frame is robust, avoiding any significant deflection during the FSW trials. Ample power for the steady rotation at 64 spindle speeds between 90 - 1400 rev/min is provided by a two speed reversing motor that develops 22kW at 1430 rev/min and 15kw at 960 rev/min. The available traverse rate ranged between 0.5mm and 15mm/sec (0.03 and 0.9m/min). A hydraulic force system, which had a maximum capability of 250 kN was also incorporated into the machine to provide the downward welding load.

Thermal imaging of selected welding trials was carried out using an Agema Thermovision 900 series infrared imaging system, with an accuracy of ± 1°C or ± 1% which ever is greater.

2.2 Materials

The workpiece material selected was 12 mm and 15mm thick, low carbon steel grade BS970: Part 1: 1983 07M20 (BS EN 10083-1) and 12mm thick, 12% chromium alloy steel grade DIN 1.4003 (X2CrNi12), (EN 10083-3) with nominal compositions as follows:

Table 1: Chemical analysis of low carbon 12% chromium alloy parent steel (TWI analysis ref: S-98-153)

	Element, wt%
	C	S	P	Si	Mn	Ni	Cr	Mo	V	Cu	Nb	Ti	Sn	Co	N
Parent steel	0.01	0.003	0.027	0.38	1.09	0.55	11.2	0.03	0.07	0.05	0.27	<0.01	<0.01	0.03	0.016
DIN 1.4003* (X2CrNi12) EN 10088-3	0.03	0.015	0.040	1.00	1.50	0.30- 1.00	10.5- 12.5	NS	NS	NS	NS	NS	NS	NS	0.030

* All single figures are maxima
NS = Not specified

Table 2: Chemical analysis of low carbon 12% chromium alloy parent steel (TWI analysis ref: 9B-Rev 6)

Sample Number	Element, % (m/m)
	C	S	P	Si	Mn	Ni	Cr	Mo	V	Cu	Nb	Ti	Al	Co
Low carbon E3 steel EN 10083	0.10	0.023	0.008	0.16	0.69	0.08	0.06	0.01	<0.002	0.25	<0.002	<0.002	<0.003	0.0004
	Element, % (m/m)						Comments
	Sn	Co	As	Ca	Pb	Zr	Direct spark analysis ** Not determined
	0.016	0.012	0.018	0.0003	**	<0.005

The tool geometry was substantially that of the Whorl ^TM type as described elsewhere ^[15] .

Detailed weld parameters, dimensions of the tool and tool material are still being further optimised. These initial feasibility studies however did achieve welding travel speeds of 1.7 to 4 mm/sec.

2.3 Welding procedure

The steel workpiece plates were secured with work holding fixtures onto the machine traverse table. A pilot hole of smaller diameter than the probe was drilled between the abutting plates at the start of the weld seam. Touchdown conditions were set to minimise the stress on the tool. Traversing was initiated after a sufficient time period to plasticise the workpiece material in contact with the shoulder and probe. The friction stir welding operation was carried out at ambient temperature, with no auxiliary pre-heat or interpass heating of the workpiece being used. For double sided welds the first weld surface was dressed flat before turning the workpiece over and repeating the above procedure on the second weld pass.

2.4 Weld assessment

All welds were visually examined for surface roughness, presence of surface breaking defects, and side flash. A number of welds were tensile and bend tested, together with metallurgical examination. All sections were prepared in the direction looking towards the start of the weld, and for clarity all macrographs are marked 'advancing side' and 'retreating side'. Carbon steel sections were prepared to a 1µm finish and etched in nital. The 12% chromium alloy steel was etched in an ethanol solution containing 2.5% picric acid, and 2.5% hydrochloric acid. Vickers hardness traverses, using a 10kg indenting weight, were taken across a number of welds at both mid thickness and quarter thickness.

2.5 Microstructural assessment of 12% chromium alloy and low carbon steel friction stir welds

Studies of a number of materials indicate that there are three primary microstructural regions to consider in friction stir welds, although these may be further sub-divided for certain materials ^[16] . These regions are:

1. Unaffected parent material.
2. Material that has been affected by heat, but not mechanically deformed. This is defined as the heat affected zone (HAZ).
3. Material that has been affected by heat and mechanically deformed. This is defined as the thermo-mechanically affected zone (TMAZ).

The microstructures of the steels examined in this work can be categorised in this way.

3. Results and discussion

3.1 General characteristics

Unlike aluminium and most non-ferrous materials, which show little or no visible change during welding owing to increase in temperature, a colour change was distinctive in the FSW of both grades of steel, which gave approximate indications of temperature. The tool shoulder reached a bright orange colour which indicated an approximate temperature of over 1000°C, within a few seconds of making contact. Also as the tool travels along the seam, the ensuing weld track behind the trailing edge of the rotating tool appeared orange/bright red (900-1000°C). This colour changed to a darker cherry red (about 600°C) 25mm from the tool. The tool shoulder maintained its bright orange colour throughout a 1 metre length of weld.

For low carbon 12% chromium, alloy steel thermal imaging measurements gave a maximum welding temperature close to the tool of around 1090 degrees C. A typical temperature profile along the weld is shown in Fig.3. The temperature was also partly dependent on rotational speed increasing with high speed and falling with lower speed.

The surface of the steel welds showed, a uniform surface ripple (caused by the final sweep of the trailing edge of the rotating tool) which visually was not unlike that of a steel friction surfacing deposit, see for example Fig.4 for the hot work tool steel deposited on to carbon steel. 

However, unlike a deposit that clads the top of the workpiece the FSW weld appeared essentially flush with the surface as shown in Figs 5 & 6. Apart from being a little coarser, the almost semicircular ripple in the weld tracks for steel was essentially the same as those for aluminium FSW welds. The closed part of the ripples are part of a continuous cycloidal motion that characterises all friction stir welds. 

Transverse macrosections reveal HAZ profiles that correspond with the shoulder and probe geometry and reflect the degree of friction treatment received. Frictional contact at the shoulder produces a wide but relatively shallow HAZ which deepens in the centre region, and extends through-the-thickness to a depth and breadth governed by the probe. Typical overall HAZ profiles for double sided welds are shown in Figs 6, 7, and 8.

A marked difference was found in the welding speed possible for the two types of steel. Acceptable welds could be produced at up to 4 mm/sec traverse rate for 12% chromium steel, but only at a slower 1.7 mm/sec traverse rate in carbon steel.

The dissimilar 12% chromium/carbon steel weld specimens differed from normal in that certain regions of the weld profile protruded above the plate surface. Some undercut was noticeable but essentially the surface appearance was marked with a shallow bulge 0.6 mm above the plate, which ran along the entire length of the weld. This bulge lay on the retreating side and mainly comprised of the 12% chromium material that came from the original plate of the joint as shown in Fig.9.

The cyclic nature of the rotary friction stir welding process is revealed in detail in the macrophotograph ( Fig.10) of a 20° tapered transverse section. This confirms that the cyclic pattern is consistent longitudinally and through-the-thickness and also shows that both the 12% chromium and the carbon steel have been moved across the original abutting plate interface.

3.2 Weld integrity low carbon 12% Chromium alloy steel

Cross-weld tensile tests recorded an ultimate tensile stress of around 539 to 541 N/mm² with failure occurring in the parent metal well away from the joint or the HAZ region as shown in Fig.11. Acceptable transverse face and root bends (bends with first pass in tension and bends with second pass in tension), typically achieved 180°, see Fig.12.

Metallographic examination of selected sections, typically as shown in Fig.6 shows a reasonably uniform shaped double-sided weld profile with no evidence of buried defects.

Average hardnesses were:

Parent metal HAZ TMAZ

- - -

158 HV10  280 HV10  230 HV10

3.3 Weld integrity, low carbon steel

The cross-weld tensile tests gave an ultimate tensile stress of around 453 to 457 N/mm with failure occurring in the parent metal well away from the joint or HAZ region, broken tensiles are shown in Fig.13. Transverse 180° side bend tests are shown in Fig.14.

Metallographic examination of selected sections revealed a flow configuration (see Fig.7) not unlike those found on certain aluminium alloy welds. A reasonable shaped weld profile with no evidence of buried defects, but some tool wear debris was observed.

Average hardnesses were:

Parent metal HAZ TMAZ

- - -

131 HV10  149 HV10  158 HV10

3.4 Low carbon 12% Chromium alloy steel

The weld region exhibited two distinct microstructural zones. One of these is the central weld thermomechanically affected (TMAZ) zone, which had transformed, with associated recrystallisation and grain growth. On both sides of the (TMAZ) central weld zone a HAZ region showed some transformation close to the weld but showed no evidence of grain growth. Further out, towards the parent material, the HAZ still showed a degree of tempering, but had not transformed. The HAZ zones on either side were similar in all features. Typical of this type of steel, the parent material showed a very fine ferritic/martensitic structure.

Within the TMAZ a range of ferrite and martensite structures had developed, a typical example is shown in Fig.15. Some light etching bands were present, towards the top of each weld pass. Energy dispersive x-ray micro analysis of these bands indicated the presence of some tool debris. Longitudinal weld sections, however, confirmed that no measurable reduction in weld depth had occurred after steady state welding conditions had been established. There was no evidence of buried defects within the weld region.

3.5 Low carbon steel

The microstructure of the parent steel consisted of the expected ferrite/pearlite grains. The subcritical HAZ showed the expected spheroidization of pearlite, and the intercritical HAZ showed substantial grain refinement resulting from the partial transformation to austenite. The higher temperature supercritical HAZ in the centre of the weld had transformed to a bainitic/ferritic microstructure, with no evidence for martensite formation.

The boundary between the HAZ, and the TMAZ (i.e. the point of which plastic deformation owing to the welding process first occurred) was difficult to identify, as no evidence of deformation without recrystaliisation was observed. This suggests that the HAZ/TMAZ boundary occurred in the region heated above the AC₃ temperature. This is in accordance with the peak temperature reached during friction stir welding of low carbon steel, which is probably around 1000°-1100°C. A typical microstructure from the centre of the weld (in the TMAZ) is shown in Fig.16(a), while Fig.16(b) shows the HAZ in the region of the AC₃ isotherm, where areas of partial and complete transformation to austenite during the welding cycle are apparent.

Unlike the 12% chromium alloy steel, there was some evidence of tool debris within the weld region of the carbon steel examined, although examination of the tool after welding showed little wear. The debris existed as fine inclusions, exerting no apparent influence on the microstructure of the weld. It is, however, clear that further work on the wear resistance of the tool, especially for carbon steel, is needed.

The parent hardness of about 131 HV10 increases across the weld to a maximum of 160HV at the centre. These low values suggest the material was in the hot rolled condition before welding.

4. General considerations

4.1 Welding mechanisms

In friction joining and forming, the process is akin to a fluid layer of high viscosity between solid components in relative motion and under significant compressive loading. The thixotropic properties and fluid flow features that occur in conventional friction welding have been reported (along with friction induced 'third body' conditions and superplasticity that occurs as a result of extreme plastic deformation). Thus the science of these processes, in some respects, is probably closely allied to that of rheology. ^{[17,18,19&20]}

The relative motion between the tool and the substrate generates sufficient frictional heat to reduce the yield strength of the material. As the temperature rises the yield strength falls below the applied shear stress so that a 'third-body' region of highly deformed plasticised material forms around the immersed and contacting regions of the tool as illustrated in Fig.1.

The outer edges of the weld track only experience limited friction from the periphery of the tool shoulder. In contrast, and depending on the degree of tool tilt, most of the shoulder acts upon the central region of the weld track. Inevitably, it is the central region that receives most friction as well as the stirring due to the probe.

This highly plasticised 'third-body' material provides some hydrostatic effect. As the rotating tool moves along the joint, this hydrostatic effect helps the plasticised weld material to flow around the tool. The plasticised weld material then coalesces behind the tool, to form a solid phase joint as the tool moves away. Evidence that hydrostatic pressure leads to displacement of plasticised material and recovery of the through-thickness dimension is also shown in the dissimilar metal weld sections. Figs.9 & 10 show that even where the trailing edge (heel part of the shoulder) is sunk below the plate surface during the operation, recovery in plate thickness is possible. In the case of dissimilar materials, preferential recovery occurs with the more plasticised material, especially when positioned on the retreating side of the weld. The presence of a shallow bulge above the plate surface, as shown in Fig.8, confirms this effect.

Both friction stir welding and friction surfacing processes show some lack of symmetry. The use of optimised conditions however, virtually ensures that differences between the advancing side and retreating side do not cause any adverse effects. However, with less suitable conditions, the asymmetric nature of the process can lead to defects. In friction surfacing lack of symmetry can lead to excess expulsion of material at the retreating edge of the deposit, as shown in Fig.17. In FSW, defects can be found such as buried voids, or a surface-breaking groove that usually runs along the advancing side. The inherent lack of process symmetry causes a differential pressure around the probe such that the rotating tool tries to veer away from the retreating side of the weld towards the advancing side. Secure fixturing and robust machine tool equipment prevents any noticeable sideways deflection. ^[21]

4.2 Thermal management

Investigations at TWI are continuing to study the value of preheating for the FSW of ferrous and other comparatively high temperature materials to improve welding speed and minimise tool wear.

Before frictional contact is made the workpiece material will be at its hardest, and, therefore, be more likely to wear or to damage the FSW tool. Thus, of the entire friction stir welding operation touch down condition is regarded as the most severe.

It can also be beneficial, for the higher temperature materials, to preheat the touch down region of workpieces so as to condition this region before plunging the probe into the work piece. The welding process can then progress without further additional heating. It is expected that this simple procedure will significantly reduce tool wear at touch down.

Depending on the properties of the workpiece material and its thermal diffusivity, it can also be beneficial to continue the preheating throughout the welding operation. Conversely, cooling or even welding certain materials underwater is found to be beneficial.

Preheating of the tool is also recommended for certain tool materials which are brittle at room temperature, so that they become more ductile and thus better suited to carrying out the welding process. It is considered that any suitable heating process can be adopted for heating the workpiece including heating techniques such as flame, coherent or incoherent radiation, friction, induction resistance or arc/plasma. High frequency induction heating and high frequency resistance heating may be of particular advantage since they can achieve heating through the thickness of the workpiece, rather than just surface heating.

Work is continuing at TWI to investigate the use of hybrid process to fill substantial gaps between the plates to be welded. Processes such as TIG, MIG, sub-arc and hot wire welding methods as well as resistance hot wire can be used to fill gaps between plates just in front of the FSW tool. This hybrid approach effectively allows the FSW technique to become a gap filling and a post fusion welding process to refine and improve the weld from the prior fusion process.

In some cases, where the FSW process is used at high temperatures, a non-oxidising gaseous atmosphere may be needed to protect the joint from atmospheric contamination and to prevent certain tool and workpiece materials becoming oxidised.

The FSW process seems ideally suited to the welding of hot plate where the entire plate or product is raised to a higher temperature e.g. hot plate welding in the steel mills or hot strip tube manufacture in pipe mills.

5. The advantages & disadvantages of the FSW process

5.1 Advantages

Generically friction welding and its related process variants are characterised by being thermomechanically energy efficient solid-phase joining techniques. Friction stir welding is no exception and in addition the welding operation is simple and operator friendly. The following lists some of the advantages of the process at present: -

• The process is machine tool technology based, which can be semi-automatic or fully automated.
• The surface appearance approaches to that of a rough machined surface. In most cases this reduces production costs in further processing and finishing.
• For most materials the process does not normally require a shielding gas.
• High integrity welds are produced for an increasing range of materials.
• Parent metal chemistry is retained without any gross segregation of alloying elements.
• The process is essentially an autogeneous non-consumable keyhole technique. (Therefore, eliminating the problems associated with the selection and storage of consumables).
• Plain low carbon steel and 12% chromium alloy steel can be welded in a single pass in thickness from 3-12mm.
• Steel thickness up to 25mm can be welded from two sides. (Similarly to arc welding, the double sided weld joint is more process tolerant).
• Welding is carried out without spatter, ozone formation, or visual radiation associated with fusion welding techniques.
• The process is relatively quiet.
• The process is solid-phase, with process temperature regimes much lower than fusion techniques, thus avoiding problems which can occur with the liquid phase, such as alloy segregation, porosity and cracking.
• The process can be carried out in all positions - vertical and overhead.
• No special edge preparation is required (only nominally square edged abutting plates are needed for a butt joint), so it saves consumable material, time and money.
• A feature associated with FSW weldments is the comparatively reduced distortion levels.
• FSW is easy to automate, and user friendly.
• Equipment is simple with relatively low running costs.
• Once established optimised process conditions can be pre-set and subsequent in-process monitoring can be used as a first line check that weld quality is being maintained.
• Like most friction techniques the process can be operated underwater.

5.2 Disadvantages

• It is necessary to clamp the workpiece materials firmly. Suitable jigging and backing bars are needed to prevent the abutting plates moving apart or material breaking out through the underside of the joint.
• An end of run hole is left as the probe is withdrawn.
• To overcome the latter feature run-on/run-off plates which take the end of the run hole from the substrate joints are sometimes used or the hole can be left in a suitable region. In addition, one of the friction hole fillingtechniques, such as taper plug and friction hydro pillar welding can be considered.
• At present for plain low carbon and 12% chromium steels the welding traverse speed is typically in the order of 1.7 to 4 mm/sec which could be considered comparatively slow for relatively thin plate material.
• For plain low carbon steels and to a lesser extent 12% chromium alloy steels tool wear is a limiting feature.

 

Concluding remarks

Although more development work has to be carried out, primarily on improved tool materials, the feasibility of friction stir welding steel has been demonstrated. Tensile testing and bend testing has confirmed that the mechanical properties of FSW 12% chromium alloy and low carbon steel joints compare well with parent metal properties.

This paper has described some initial feasibility studies on the FSW of both 12% chromium alloy and low carbon steel; the prognosis for continued progress is good.

Acknowledgments

The Authors thank Dr K I Johnson, Dr P Woollin, Dr R E Dolby, Mr J C Needham and Mr P T Smith who contributed to this work.

References

1	Bevington J:	'Spinning tubes mode of welding the ends of wire, rods, etc, and mode of making tubes'. US Patent 463134 January 13, 1891.
2	Klopstock H and Neelands A R:	'An improved method of joining or welding metals'. British Patent specification 572789, October 17, 1941.
3	Thomas W M, Nicholas E D, Needham J C, Murch M G, Temple-Smith P and Dawes C J:	'Improvements relating to friction stir welding'. European Patent Specification 0615 480 B1.
4	Nicholas E D:	'Friction processes technologies'. International Seminar 'Advances in welding and related technology'. 8-9 September 1997, TWI, Cambridge, UK.
5	McMullan D J and Bahrani A S:	'The mechanics of friction welding dissimilar metals'. Second International Symposium of the Japan Welding Society on Advanced welding technology, 25-27 August, 1975, Osaka, Japan.
6	Hasui A et al:	'Effect of the relative difference of bar diameter on the friction welding of different diameter bars'. IIW Doc. III-679-81.
7	Fukakusa K and Satoh T:	'Travelling phenomena of potational plane during friction welding. Application of Friction Hardfacing'. International Symposium on Resistance Welding and Related Welding Processes. 10 th-12 th July 1986, Osaka.
8	Nicholas E D and Thomas W M:	'Metal deposition by friction welding'. Welding Journal, August 1986, pp17-27.
9	Bedford G M:	'Friction surfacing for wear applications'. Metals and Material, November 1990, pp 702-705.
10	Thomas W M:	'Solid phase cladding by friction surfacing'. Welding for the Process Industries, International Symposium, April 1988.
11		'The Al to Zn of friction stir welding - fastest thickest and now in ferrous materials.' TWI Connect, November-December 1997, p3.
12	Thomas W M, Nicholas E D & John R:	'Friction stir welding applications for the transport industry'. Towards 2000 Conference, Melbourne, Australia, 10-14 November 1997.
13		'Friction stir welding of steel and titanium alloy show a reality.' EWI Insights, May-June 1998, pp1-3.
14		'Inching onward - Friction stir steels the lead'. Connect, September/October 1998.
15	Thomas W M:	'Friction stir welding and related friction process characteristics.' Inalco 98 7 th International Conference, Joints in Aluminium, Cambridge, UK, 16 April 1998.
16	Threadgill P L:	'Friction stir welds in aluminium alloys - preliminary microstructural assessment'. TWI Bulletin, March (April 1997), 38 (2), 30-33.
17	Godet M:	'The third-body approach: A mechanical view of wear'. Wear, Vol No 100, (1984) pp 437-452.
18	Singer Irwin L:	'How Third-Body Process Effect Friction and Wear' MRS Bulletin 1998 6.
19	Khazanov I O and Fomin N I, (Tomsk Polytechnical Institute):	'Friction Welding in the Temperature Range of Superplasticity of High-Speed Tool Steel', Welding International 1990 4 8 633-634 (Selected from Svarochnoe 1989 36 4-5, Reference SP/89/3/4, Translation 860.
20	Gulyaev A P:	'Superplasticity of Steel', Publ Metallurigiya Moscow, 1982
21	Thomas W M, E D Nicholas & M Gittos:	'Friction based technology for aluminium', Proceedings of the Aluminium 98 Conference, 23-24 September 98, Essen, Germany.