Paper presented at ANTEC 1999, New York, USA, 2-6 May 1999
Vibration welds were made in polyetheretherketone (PEEK) using different pressures, with a variety of welding times, in order to produce welds with a range of mechanical properties. Polarised Fourier transform infrared (FTIR)-microspectrometry was used to measure crystallinity and molecular orientation, and transmitted light microscopy was used to study morphology. Tensile and tensile-impact tests were carried out, and the broken test specimens were examined by scanning electron microscopy (SEM) and transmitted light microscopy to establish where failure occurred, and to examine the fracture surfaces of broken test specimens.
Polyetheretherketone (PEEK) is a high performance thermoplastic with many attractive properties, including exceptional chemical resistance and high thermal stability. It is used in critical and hostile environments, in applications such as chemically resistant coatings, and pump and aircraft components. As PEEK is a semicrystalline material, the mechanical properties will be affected by crystallinity  , molecular orientation  and morphology  , which are dependent on the thermal history, and may be affected by welding. Previous work at TWI has examined the crystallinity and orientation in nylon 6,6 welds  and in PEEK hot plate welds  .
Material and Welding Conditions
Vibration welds were made using injection moulded plaques made from ICI 450G PEEK granules: an unreinforced high molecular weight grade (about 45,000). The plaques were cut into two, machined to size, 75x150x3.2mm, then dried for 12 hours in a vacuum oven at 150°C. The dried material was kept in a desiccator before welding. The original outer moulded edges, were joined by a plain butt joint in the longitudinal axis of the specimen. Welding times were 2.5-15sec, welding pressure vacuum oven -2, welding frequency 200Hz, vibration amplitude 1.4mm, and cooling time 1.5sec. The mechanical test results (given later) showed that the optimum welding pressure was probably 2 or 3Nmm -2, and the subsequent testing programme therefore concentrated on welds made using these conditions.
Sampling and analysis
Microtomed sections for FTIR (25µm) and transmitted light microscopy (10µm) were cut at ambient temperature and -35°C, respectively.
Six sections from selected welds were analysed by FTIR at the weld centreline, at the centre of the parent material thickness. Each analysis area was 188x50µm.
Scanning electron microscopy was carried out on the etched [5,6] microtomed surfaces which remained after removal of the sections for light microscopy. The fracture surface of one half of each broken tensile specimen was examined using the SEM. The other half was sectioned perpendicular to the fracture face, and examined by transmitted light microscopy.
Tensile and tensile-impact specimens had a cross-sectional area of 7x3.2mm, and the flash was removed for the tensile-impact tests, but not for the tensile tests.
Results and Discussion
Crystallinity and molecular orientation
The mean crystallinities of the welds made at pressures of 2 and 3Nmm -2 were typically 22-23%, ie around 8% lower than in the parent material (31%). This was perhaps due to a faster cooling rate in the weld than had been seen by the parent material. This reduction in crystallinity was apparent regardless of the precise welding time or welding pressure. From the welds studied, the crystallinity did not appear to affect the mechanical properties, but the effect on the chemical resistance was not studied. This reduction in crystallinity supports the results of earlier work, where the weld crystallinity in PA6,6 vibration welds was found to be 3 to 4% lower than in the parent  .
The levels of orientation were similar at 2 and 3Nmm -2, but there was more scatter at 3Nmm -2. The mean amorphous orientation was around 2% lower than the mean crystalline orientation for all welds, which was around 13% parallel to the flow direction of the weld at 2Nmm -2, and 11% at 3Nmm -2. The orientation was unaffected by welding time. The amount of orientation was just over twice as much as in hot plate welds  , indicating that the faster cooling rate in the vibration welds meant that polymer chains which had become aligned during welding had less time to relax. However, the amount of orientation was still less than the level of 43 to 53% found in PA6,6 vibration welds  .
Transmitted light microscopy showed that welds with reasonable mechanical properties generally had one or more bulges (see Fig.1), or even double bulges (probably marking the transition from laminar to turbulent flow  ), separating the central weld region from the region of flow lines leading out to the flash, Fig.1 (weld 126, 2Nmm 2, 6sec). The central weld region contained a well defined centreline, but was otherwise relatively featureless and perhaps non-spherulitic: this would correlate with the lower crystallinity than in the parent, and would be in common with the non-spherulitic appearance of vibration welds in other semicrystalline thermoplastics such as PP, PA 6, and PA 6,6. As the welding time increased, the width of the region containing flow lines increased at the expense of the central weld region. At lower welding pressures, lack of fusion defects (1Nmm -2) or welds lacking the characteristic bulges (1.5Nmm -2) were evident. Using too high a welding pressure (4Nmm -2) produced a weld which, although it had bulges, was very narrow at the centre (4 sec), or which failed at the weld centreline on sectioning (8 sec).
The SEM photomicrograph taken from the centre of etched weld 126 (2Nmm -2, 6sec) is shown in Fig.2a, where the weld centreline, visible heat affected zone (VHAZ), and the parent material adjacent to the VHAZ were revealed. The attack on the parent material had taken the form of pitting, the weld region was smoother and contained much finer pits, and the weld centreline had undergone preferential attack. The VHAZ consisted of deformed spherulites. Figure 2b shows the 'flow' region to the left of the left hand bulge in Fig.1: the photomicrographs are mirror images, so the relationship between the areas in each are shown in Fig.3. Region 'a' in Fig.2b was similar to the weld region in Fig.2a, and the centreline was present, and the outer edges of the flow region, 'b', were comprised of oriented material.
At a weld pressure of 1Nmm -2 maximum weld strength was 33% of the parent tensile strength (107Nmm -2) with welding times of 13 to 15 seconds, while three tensile-impact specimens fell apart during machining, and the fourth gave 16% of parent strength. At a pressure of 1.5Nmm -2, increasing the welding time from 5.25 to 8 seconds increased the tensile strength from around 20% to 80%, and tensile-impact strength from 20% to 49-72%. Figure 4a indicates the optimum welding conditions (based on tensile results) to be 5 seconds at a pressure of 3Nmm -2 which gave welds which were 92 to 95% of the parent strength. This was a higher pressure than would typically be used for PP (around 0.5 to 1.0Nmm -2) as more energy is required to melt PEEK than PP. Figure 4b shows that these welding conditions were also near optimum for the tensile-impact behaviour. At a pressure of 4Nmm -2, 4 seconds gave tensile strengths of 57% and 83%, and tensile-impact strengths between 52 and 63%; and welds produced at 8 seconds fell apart during machining. For the weld made at 5Nmm -2, with a time of 3 seconds, tensile strengths were 44% and 79%, and tensile-impact strengths were 64%, and 90%.
Examination of the fracture surfaces of the broken mechanical test specimens showed the lowest tensile and tensile-impact strengths were associated with failure in the VHAZ. Highest strengths were also obtained when failure occurred in the VHAZ, but with a more irregular fracture surface appearance, Fig.5a. In the tensile tests, failure occurred predominantly in the VHAZ, whereas under tensile-impact testing the majority of samples failed at the weld centreline, Fig.5b. The presence of inclusions on the fracture surface did not appear to affect the strength of the welds, unless these were large enough to cause beach marks. The fracture surfaces did not show the zones of vertically structured material along the outside edges which were observed in PA 6,6 vibration welds and were associated with failure initiation within the flow lines leading out to the edge of the welds [4,8] .
Summary and conclusions
Work carried out on PEEK hot plate welds, using a variety of techniques to provide information on the structural, morphological, tensile, and tensile-impact properties, has shown:
Mechanical property results showed that welding pressures of 2 or 3Nmm -2 were necessary to produce satisfactory welds, and that (on the basis of tensile results) the optimum welding condition was 5 seconds at 3Nmm -2, where the strengths were 92 to 95% of parent strength.
Mean weld crystallinities were 22 to 23% (8% lower than in the parent), and were unaffected by the welding time or welding pressure within the range observed.
At 2Nmm -2, around 13% of the molecules in the crystalline phase were aligned parallel to the flow direction of the weld, and 11% at 3Nmm -2. The amorphous phase orientation was about 2% lower than that of the crystalline phase.
The central weld region contained much smaller spherulites than the parent, and had therefore undergone a faster cooling rate, and was separated from the parent by a VHAZ of deformed spherulites. The flow region also contained a central region which was similar to the weld and contained a centreline, and had therefore been fully molten and had undergone similar cooling conditions.
Failure generally occurred in the VHAZ under tensile testing, and at the weld centreline in tensile-impact tests.
This work was funded by Industrial members of TWI and the Information and Manufacturing Technology Division of the UK Department of Trade and Industry.
- Cebe P et al: 'Effect of thermal history on mechanical properties of polyetheretherketone below the glass transition temperature'. Journal of Applied Polymer Science, 1987, 33, 487-503.
- Wu G-M and Schultz J M: 'Fracture behaviour of orientated poly(ether-ether-ketone) (PEEK)'. Polymer Engineering and Science, 1989, 29, 405-414.
- Mijovic J and Gsell T C: 'Calorimetric study of polyetheretherketone (PEEK) and its carbon fibre composite'. Sampe Quarterly, 1990, 42-46.
- Stevens S M: 'Characterisation of nylon 6,6 vibration welds'. ANTEC 97, 55th Annual Technical Conference of SPE, April 1997, 1228-1232.
- Stevens S M: 'Structure Property relationships in polyetheretherketone hot plate welds'. ANTEC 98, 56th Annual Technical Conference of SPE, April 1998, 1067-1071.
- Peacock J A et al: 'Examination of the morphology of the aromatic composite (APC-2) using an etching technique'. Composite Interfaces, ed Ishida H and Koenig J L, Elsevier Science Publ Co Ltd 1986.
- Schlarb A K and Eherenstein G W: 'The impact strength of butt welded vibration welds related to microstructure and welding history'. Polymer Engineering and Science, 1989, 29, 1677-1682.
- Moody S M and Atkinson J R: Linear friction welding of polypropylene and polycarbonate-Pt II: The microstructure and mechanical properties of PP welds made using optimum conditions. Plastics, Rubber and Composites Processing andApplications, 1992, 17, 211-217.