Nee Joo Teh, Helen Goddin, Andrew Whitaker
Micro friction stir welding (µFSW) is the adaptation of the friction stir welding process (FSW) to materials with thicknesses of 1000μm or less. Applications such as thin walled structures, electrical, electronic and micro-mechanical assemblies can benefit greatly from MSFW’s ability to join a wide range of materials without the use of fluxes, shielding gases, and usually without post-weld cleaning. It is especially useful in joining dissimilar materials. However, downscaling to achieve µFSW presents some significant challenges. This article outlines the developments and applications of µFSW at TWI, together with some of the results achieved.
The basic concept of friction stir welding (FSW) as invented and developed by Wayne Thomas and colleagues at TWI is summarised in Figure 1, and it is now a widely used and successful technology capable of welding a wide range of materials and sections.
Figure 1: Schematic of the friction stir welding process
FSW has inherent advantages that are potentially very attractive for applications at the micro scale:
Joints are made at lower temperatures compared to fusion processes so may be more compatible with temperature sensitive components such as batteries
Aluminium alloys especially can be welded without the use of shielding gases, which are difficult to apply at small scales
Joints are produced without the need for fluxes, which removes the need for cleaning delicate components or the risk of contamination
The process is very reproducible and consistently results in flaw free joints suitable for products requiring guaranteed hermetically sealed packages (eg electronics)
Because the joint is formed in the solid state there is no solidification segregation and reduced formation of intermetallics, which can affect mechanical and electrical performance and lifetime. Friction stir welds can show grain refinement and even improved material properties
There is potential to join a wide range of plasticisable materials, both in similar and dissimilar combinations which are commonly required in micro-assemblies
The work described below successfully demonstrates FSW in micro applications – micro friction stir welding (µFSW). For example Figure 2 shows a weld in progress in just 300μm thick aluminium alloy sheet. However, there are some serious challenges to overcome to optimise the down-scaling of FSW for such micro applications.
Figure 2: µFSW in progress through 300μm thick aluminium alloy sheet
Challenges of down-scaling friction stir welding
The main limitations to down-scaling may be understood directly from the established theory of the process which can be summarised as follows.
A hard and chemically inert tool is used which normally has a flat shoulder and central probe. The shoulder bears mainly on the weld surface, whilst the probe penetrates into the material to be welded. The tool is rotated, the probe plunged into the material and a dwell period initiated to allow the thermal profile of the weld to be established. After the dwell the tool is traversed along the weld path and withdrawn at the end of the weld, so the weld may have three distinct zones along its length:
As µFSW is a keyhole process, an exit hole is left at the end of the weld length where the tool is withdrawn. In most FSW applications the exit hole may be parked in a non-critical location on the assembly or run-off on to a discard tab. It should be noted that the exit hole normally has a full annular weld around its periphery providing structural integrity.
The solid state joint is formed by the massive plastic shear and mixing of the material in a zone surrounding the tool, with material being swept from the leading edge of the tool and deposited at the trailing edge. The very high density and integrity of the weld zone material is maintained by the constraint of the flowing material by the surrounding parent material, the tool shoulder and the anvil (backing plate) supporting the weld root (where the joint does not include integral root side support).
To an extent this prevents atmospheric contamination of the weld. Any oxides or minor contamination on the faying surfaces of the substrates are finely dispersed into the plastic weld zone by the action of the rotating tool. The key to a successful weld is therefore the provision of energy to establish a temperature profile adequate to soften the material, allowing tool traverse with a correctly sized plastic zone around the probe to lower the loads on the tool sufficiently and contribute towards minimising tool wear, but not to over-soften the material.
To provide the welding energy the FSW machine applies three basic forces to the tool:
Down force, which is a significant large force required to:
Ensure constant contact of the shoulder with the workpiece
Maintain correct penetration of the probe
Provide load for generation of friction at the shoulder as a contribution to energy generation in combination with the other forces
Rotational force (torque on the tool) which:
Produces friction between the tool probe and weld material and internal friction (shear) in the weld material plastic zone
Produces friction at the shoulder
Transports weld material around the tool
Traverse force which:
The essential heating of the material can be viewed to arise simplistically from the following contributions:
Many analyses of FSW use an energy balance approach to link:
Power input (energy per unit weld length at constant traverse velocity) from the FSW machine, mainly from rotational torque
Power dissipation (heat capacity of weld zone and mainly conductive heat transfer to surrounding parent material and backing plate and tool holder as the plastic zone passes through
Power conversion models identifying magnitude and location of frictional heating from tool geometry, tool forces and velocity, and material properties
These considerations can be used to explain the inherent problems in down-scaling for µFSW applications which have been encountered as follows:
The surface area to volume ratio of the plastic zone increases with decreasing zone radius, so the surrounding parent material provides more rapid heat dissipation
Relative heat loss into the backing anvil will increase as the plastic zone size decreases and the weld material thickness decreases. Ideally an insulating but resilient backing anvil should be used (see example below)
Probe length must be limited to prevent risk of root-side over penetration, so its contribution to heating the weld is limited
Complex probe geometries are difficult to produce at small scales and therefore this technique is less useful in generating shear friction heating.
So relatively more energy is required per unit weld volume as size decreases, and relatively more of this must come from the shoulder friction
Increasing the down force would increase the shoulder friction, however there are important limitations which require an optimisation for factors such as tool wear, excessive penetration due to the higher pressure when using small diameter tools, and unacceptable loads on sensitive components.
Therefore the shoulder diameter cannot be scaled-down in proportion to material thickness (there may be other reasons such as practicable tool stiffness)
Higher rotational speeds could also be used to increase heat input power, however, the rotational speed cannot be increased too much with relatively large tool shoulders as there is a risk of tearing with thin sheet parent materials.
Reduced heat power input arising from the above considerations may mean slower relative traverse speeds
Precision clamping as close as possible to the tool path is also required for thin sheet materials, and this may be another compromise with regard to heat sinking and welding speed
We can see that, as with larger scale FSW, there is much work required to optimise the tool design for a particular application, but for micro applications there may be more demanding constraints which require careful consideration to achieve the optimum welding parameters.
Tool design for micro friction stir welding
Figure 3 provides examples of tool design scaled down from larger tools for potential application of µFSW of aluminium alloys. In practice it is found that the grooved detail machining on the finer tool imparts only marginal improvement in stirring compared to the flats shown on the larger example tool, which in turn does offer some improvement over a simple cylindrical probe design.
Figure 3: Examples of tool designs for MFSW
In line with the predictions above, successful tool design is at least equally dependent on the correct geometry of the shoulder as on the pin. For example the radii used to minimise surface damage can be critical in micro applications where it is important to produce no flash.
MFSW Equipment Development at TWI
Figure 4 illustrates some of the earlier CNC programmable micro-milling machines adapted for µFSW at TWI. For thin materials it would appear ideal to maintain surface quality by the use of adaptive load control. However, it has been found that good results are achievable without adaptive load control provided the machine has high stiffness and programmable offset for vertical z-runout on the x-y work tables.
Figure 4: µSFW machines at TWI adapted from CNC micro-milling machines
High rotational speeds (up to 3000 rpm) and high torque from drives with very robust bearing systems are essential, and this has been assisted by recent developments in small format synthesised 3-phase motor drives. Figure 5 shows a large bed µFSW machine at TWI equipped with such a motor drive and programmable z-correction. In this case the µFSW heating is supplemented by a fibre delivered radiant heat source.
Figure 5: Large bed µFSW machine at TWI with supplementary radiant heating delivered by fibre to the µFSW head
Some materials of great interest to micro fabrications, such as polymers and soft alloys, can have very poor ability to be friction stir welded, especially at the micro-scale, due to for example:
low coefficient of surface friction
low thermal conductivity combined with over-rapid softening with temperature
very high thermal conductivity and low softening with temperature
welding of dissimilar materials
Supplementary heating was identified as a useful option early in the development of FSW by TWI, and the latest fibre delivery technology is now being explored to give a better temperature profile in the weld material ahead of the µFSW tool.
A further adaption for thin sheets is the use of vacuum hold downs, and in particular a combination of this with a low thermal conductivity ceramic anvil back plate. This assists the µFSW process by reducing heat extraction from the weld zone. Such an arrangement is shown in Figure 6.
Figure 6: Use of a ceramic vacuum bed during µFSW of 300μm thick aluminium alloy sheet
Micro scale welds have been achieved for example in the following materials and combinations:
Aluminium alloys to self: eg AA1020, AA2024, AA6068, and 3xxx and 5xxx series alloys
Brass to self
Pure copper to self
Aluminium to copper
Welding traverse speeds between 50 and 500 mm/min have been demonstrated. Weld joints in butt, lap and spot formats have all been produced, as exemplified below, with lap-welded and spot-welded joints in material thicknesses of less than 300 μm, and butt welds in thicknesses down to 300 μm.
Figure 7 shows butt welds in 300 μm thick aluminium alloy. The welds are fully penetrating with good surface quality if the correct tool design, attack angle and penetration control are used. Figure 7 (a) shows a completed weld run with both the start penetration point and the withdrawal point visible.
Figure7: Butt weld of 300μm thick sheets of aluminium alloy (a) Complete weld run showing penetration and withdrawal; (b) Detail of top surface of a 2.0 mm width weld line; (c) Cross section showing transition from weld metal (left) to parent metal (right); and (d) Bending of test strip cut across butt weld in 500µm thick aluminium alloy
As with larger scale FSW the metallurgical structure of the weld is fine grained and free of the segregation and coarse grains which can degrade fusion welds. Figure 7(c) shows a cross section through the transition from plasticised weld metal (left) to parent metal (right) of a butt weld in aluminium alloy, demonstrating the fine grained defect free structure of the weld material which has parent metal levels of ductility, Figure 7 (d).
The same defect free and fine grained weld structure is produced in lap welds, as shown in Figure 8. A high quality top surface finish can be produced, and the cross section shows ideal depth of penetration.
Figure 8: Lap weld between two 300μm thick sheets of aluminium alloy (a) Top surface of short length of weld (11mm); and (b) Cross section of weld
It can be seen that both butt and lap weld geometries would be applicable to joining thin sections together, or joining thin sections to thicker sections – such as the lid on a hermetically sealed package. This is a common requirement for high reliability electronics, and is shown schematically in Figure 9.
It is essential in this application not only to guarantee hermeticity (ie no pinholes in the weld) but to avoid flux contamination or spatter into the package interior, and not to heat the package excessively. The package wall may not sustain the load required for µFSW in aluminium or nickel, so TWI is exploring the µFSW of pre-tinned layers for a low load and low temperature “friction soldering” solution.
Figure 9: Schematic of hermetic lid sealing of an electronics package
Spot welding for thin materials is not only useful for mechanical assembly but also can provide a sufficient contact area for electrical connection. So there is a special interest for spot welding of aluminium and copper alloys.
Figure 10: Spot welds in thin sheets (a) Copper; (b) Aluminium alloy; and (c) Cross section of deep spot weld in aluminium alloy sheets
A spot weld is simply created by penetrating the lapped materials with the µFSW tool and withdrawing with no traverse. A consistent weld is produced with no use of electrical current and without risk of spatter.
A special problem in joining dissimilar materials for electrical connections is the joining of aluminium to copper. Fusion welds create large amounts of Cu-Al intermetallics which increase resistance and can reduce strength. µFSW produces very much less intermetallic. Figure 11 shows a simple example where an aluminium crimp is additionally spot-welded to a copper cable to improve both mechanical and electrical connection.
Figure 11: µFSW spot welding of dissimilar materials – aluminium connector to copper cable. Crimped only (right). Crimped and Spot Welded (left)
FSW of thermoplastic polymers presents many challenges for the reasons described above, but is potentially attractive if the polymer’s properties can be protected from thermal degradation during welding. Attempting to weld dissimilar thermoplastic polymers without thermal degradation is especially difficult and requires a special combination of thermal management and tool design, which is the subject of current ongoing research at TWI. Figure 12 provides a cross section showing the plastic mixing achieved in a butt weld of thin sheet laminates of polypropylene and polyethylene from this work.
Figure 12: µFSW butt weld of dissimilar thermoplastic sheet materials – cross section showing polymer mixing
There are many potential applications in micro-fabrication which may benefit from micro friction stir welding and its advantages of low temperature, flux and inert gas free processing, and the excellent metallurgical structure of the welds
Very consistent and void free joints can be produced meeting the requirements of structural, hermetic, and electrical connectivity in micro applications
A wide range of materials with dimensions less than 1000μm can be joined using this technique, including aluminium alloys, copper alloys, solders, and thermoplastics
The process is very scale sensitive and careful selection of tool design, micro-jigging, and thermal flow management techniques are required for success
The authors wish to thank Wayne Thomas and the TWI Friction Process group for their advice and encouragement, and recognise the support of the TWI Core Research Programme.