Flame spraying thermoplastics for plastics/metal joining

TWI Bulletin, January 1988

Part 2 - Results and discussion

by Martin Day

Martin Day, BSc, was a Research Metallurgist in the Materials Department.

Part 1

The concluding part of this article reports results of tests of coating soundness and bond strength for a range of plastics, and considers the implications for plastics/metal joining.

Results

Spraying procedure

Best results in terms of surface finish were achieved when a sweeping side-to-side motion, similar to that used in paint spraying, was adopted to apply the coating evenly. Each sweep was initiated and completed beyond the edge of the substrate so that a consistent spraying speed of approximately 150 mm/sec was maintained over the sample. The torch was held at an angle of about 70° to the substrate surface [11] and at a stand-off distance of approximately 150mm.

All consumables showed a similar response to changing the substrate preheat temperature. As summarised in Table 3, neither particle coalescence nor bonding was achieved without preheat. Best results were obtained with a preheat somewhat above the material melting point, as given in Table 1. At higher temperatures increased degradation occurred and the emission of black smoke was observed, but with optimum preheat conditions, the consumables gave similar surface appearance.

Table 3 Effect of preheat on sprayed coatings

Coating properties

It was found that each sprayed single layer had an average thickness of between 200 and 250µm ( Table 4). There was, however, greater variation between the thicknesses of five layer coatings, these spanning 0.80-1.25mm. The range of individual thickness measurements relative to the average value was greater for single layers than for five pass coatings.

Table 4 Coating thickness measurements

This presumably reflects operator variations during manual spraying, variations in travel speed for example being evened out in building up five layers.

Visual inspection of all the coatings indicated that each material had been put down with a good surface finish and no defects were apparent. Metallographic examination and spark testing showed that there was a variable incidence of through-thickness and sub-surface defects ( Table 5).

Table 5 Coating integrity

In the sections examined, the plastics coatings appeared to be in good contact with the substrates, with interface defects only in association with porosity. Porosity was observed in all coatings except the PPA31, and was most marked with the Macropol and Rilsan. The pores were spherical, the largest being over 200µm diameter although more typically 50-100µm diameter. Typical micrographs of transverse sections are shown in Fig.8-10. In addition to porosity, sections showed that some of the coatings had a fine cell-like structure, as in Fig.9 and 10. Individual layers were not detectable on the sections.

The results of the adhesion tests are shown in Table 6. Particular problems were encountered when attaching the loading fixtures to the coatings, especially with Rilsan and PPA, resulting in a high proportion of invalid tests. Of the 105 adhesion tests performed, only 63 were valid adhesion failures. The value of adhesion given in Table 6 is the average of the valid tests on each material. As can be seen from these results, surface treatment was found to be important; grit-blasting the substrate surface increased the average pull-off strength for all the plastics tested.

The highest adhesion values were consistently achieved by Levasint and Macropol although even with these materials there was a wide scatter in results.

Scanning electron micrographs of the surfaces of material adhering to testing dollies in the adhesion tests are shown in Fig.11-13. Figure 11 illustrates adhesive failures without and with grit-blasting. The extreme irregularity of the latter fracture face is evident: in this case, there was some visual evidence of residual coating on the substrate, presumably mechanically keyed-in, but otherwise failure seemed to have occurred actually at the interface.

Figure 12 shows a cohesive failure and the micrograph in Fig.13 is of a mixed failure. In the cohesive failure of the Macropol, a high density of porosity can be observed confirming the findings of sectioning this coating.

Welding

Figure 14 shows an ultrasonic weld between aluminium and Propathene using a flame sprayed PPA coating interlayer and wire mesh energy director. The heat generated in ultrasonic welding was only sufficient to melt the PPA material and not the Propathene since flash was only produced in the former ( Fig.15). In hot plate welding, the heating times could be varied thus to allow more equal melting of the components to be joined ( Fig.16) and flash was produced in both materials. Too much melting of the coating would cause the bare substrate to be exposed to the hot plate so this had to be avoided.

Tables 7 and 8 give the results of shear tests on the hot plate and ultrasonic welds between PPA-coated sheet metal and Propathene sheet. Strengths were in the range 4-6 N/mm ² but one anomalous value of below 2 N/mm ² was recorded. In the majority of cases, shear failure occurred at the metal/coating interface indicating that coating cohesion was the limiting factor in overall joint strength.

Table 7 Hot plate welding - lap shear test results

Table 8 Ultrasonic welding - lap shear test results

Discussion

Flame spraying procedure

When flame spraying, it was found that the selection of 'optimum' flame characteristics depended on operator judgement, and skill in adjusting gas pressures and flow rates was required to apply an even coating consistently with good surface finish. It has been recommended that the flame should at least be neutral and a slightly carburising condition is preferred. ^[12] This is presumably because oxidising conditions may cause degradation of the consumable, but as with gas welding or brazing, for example, experience was needed to adjust the flame to obtain best results.

Within limits, the other examined spraying variables (gas pressures, flame length, stand-off distance and travel speed) were not unduly critical in terms of obtaining a coating of good appearance. Table 2 indicates a reasonable range of usable gas pressure, while the other parameters could be varied by around ±10% of the preferred values. Steadiness in torch movement was, however, needed. A similar technique was found suitable for all the plastics grades used despite appreciable differences in melting point, although it is probable that further optimisation could be achieved with a mechanised spraying system and this might reduce the incidence of defects found with Macropol and Rilsan. Above their melting points, thermoplastics start to degrade, but it was found essential for the substrates to be preheated in excess of the powders' melting points to achieve satisfactory coatings. In this regard, the cell-like structures in the Levasint and Rilsan coatings ( Fig.9 and 10) may be a manifestation of degradation since the cell size approximates to the particle size of the powder. The temperature range between the melting point and the preferred preheat level varies considerably between the coatings ( Table 1) and this may be a result of differences in the effective viscosity of the sprayed coating at the preheat temperature.

Coating integrity and adhesion

Provided that general degradation during spraying is avoided, the principal defects which might be expected in plastics coatings are:

1. Porosity;
2. Breaches in the coating;
3. Lack of bonding.

Traditionally, a spark tester is used to reveal breaches in plastics coatings through to the substrate, but this method cannot detect porosity or lack of bonding.

Hearn ^[8] claims that flame sprayed deposits should be free from porosity but it would appear from the present findings that, because of the nature of the process, it is more difficult to eradicate porosity in some materials than in others. It is likely that the porosity stems mainly from air entrapment, since SEM examination gave no evidence of particular material degradation around the pores observed. Air may well be entrained between the powder particles during spraying, and bubbles frozen into the coating. In the sectioned Levasint and Rilsan coatings, the pores seemed to be associated with the cell boundaries, but this could arise from either degradation or air entrapment. If air entrapment is responsible, some improvement may be feasible by looking at, for instance, the effect of air pressure during spraying, and further study is desirable. Attention could also be paid to the role of particle size and shape.

The origin of the breaches evident on spark testing is not known. Sectioning gave no sign of defects through the coating, whether cold shuts or interconnected porosity. The high voltage used in testing can induce breakdown of paint coatings, for example, actually causing a breach, but this would not have been expected with the present thick coatings.

Lack of bonding between coating and substrate can be caused by surface contaminants on the substrate prior to spraying. Thus, once a coating has been applied, moisture and other corrosive media can be trapped so reducing the resistance to substrate attack and reducing the bond strength. However, this form of defect was not observed in the present work, indicating the chosen surface preparation techniques to have been satisfactory. Uneven preheat could cause lack of bonding but would be more likely to occur when spraying larger samples than those considered here. Grit-blasting was effective in increasing the adhesion of all the coatings ( Table 6). It is believed that the main mechanism controlling the adhesion of thermally sprayed coatings is mechanical bonding. ^[13] Gritblasting increases the effective surface area, and the provision of a mechanical 'key' between the coating and the substrate is clearly significant as in Fig.11. However, Packham ^[14] claims that the evidence for this is contradictory, advising that rough surfaces can hinder efficient wetting of the coating.

There is no completely satisfactory technique for the adhesion testing of coatings. ^[2] Ballard ^[4] claimed that plastics coatings should be able to withstand a straight pull of around 7-8 N/mm ² from brass and light alloys and 10-11 N/mm ² from steel, although he did not state how these figures were obtained. The adhesion values obtained in this study, using a test conforming to ASTM D ₄₅₄₁, ^[10] are substantially less than Ballard's figures. According to ASTM D ₄₅₄₁ adhesion strength increases with substrate thickness. The present data were obtained from coatings on sheet of only about 1mm thickness, and the adhesion values are therefore probably low because of flexing of the thin substrate during the test. On this basis, higher strengths would be anticipated using thicker substrates, but appropriate testing is necessary.

Moreover, Sickfeld ^[15] states that too much reliance should not be attached to absolute adhesion values as obtained from the pull-off method since they are an order of magnitude lower than theoretical adhesion values. He also suggests that differences in coating thickness can explain low adhesion results.

However, initial adhesion is not the only important property when considering metal-polymer bonds. Bond durability, as considered by Brockmann, ^[16] is of great importance in the longer term especially since, for example, water molecules can attack the interface by diffusing through the polymer component. In automobiles the breakdown of these bonds in a wet environment could be catastrophic to the coating system.

The problems experienced with attaching the loading fixtures to the coatings, for subsequent testing, are reflected in the low number of valid tests achieved, especially on Rilsan and PPA 31; polyamide and polypropylene being notoriously difficult materials for adhesive bonding because of their low surface free energy. Therefore, there are few valid test results for these two materials. However, a greater proportion of valid tests was possible with Macropol, Levasint and Polythene coatings. The former two materials gave slightly higher adhesion strengths than the last, but clearly all three displayed superior adhesion to the Rilsan and PPA 31.

Good adhesion is difficult to achieve when coating with polyamide, and the problem is normally overcome with adhesive primers as used in the paint industry. However, when flame spraying, the flame can cause primers to degrade chemically and actually behave as releasing agents. ^[12] Thus primers are not used when flame spraying polyamide and, as a result, flame sprayed polyamide coatings in particular may suffer inferior adhesion compared with coatings applied by fluidised bed dipping or electrostatic spraying, where primers are used. ^[12] Other plastics investigated in this work were all nominally self-adhesive.

Implications for joining plastics to metals

Flame spraying of plastics can clearly be used to produce a relatively thick coating on a metal substrate for subsequent plastics/metal joining. Depending on the grade of thermoplastic used, the coating may be defective and, in principle, porosity or coating breaches may reduce the final joint strength directly or act as initiation sites for failure in service. However, such defects should not necessarily be regarded as unacceptable. Most plastics/plastics welding processes involve some degree of material flow and, when applied to a plastics coating, this may well lead to consolidation of the coating, by either welding the flaws closed or by extruding them from the joint region.

Moreover, the strength of the metal-plastics bond must be considered: despite the presence of coating flaws, the adhesion tests carried out gave some incidences of cohesive as opposed to adhesive failures, indicating coating adhesion rather than defects to be limiting on overall strength. The pull-off strengths achieved ranged from 0.7-4.7 N/mm ², whereas the shear tests on welded coupons consistently resulted in strengths of 4-5.6 N/mm ². This latter strength range was sufficiently high to permit component handling, and could be adequate for lightly loaded fixtures for example. The data also compare favourably with the shear strengths of welds in Propathene made using the same conditions, although using optimum welding conditions, the strength of a Propathene weld would be an order of magnitude greater.

For plastics/plastics welds, strengths of up to 100 % of the parent material strength can be achieved depending on the material, the welding process and conditions. ^[17] The tensile strengths of polyamides are of the order of 50-80 N/mm ², polyethylenes around 5-40 N/mm ² and polypropylenes around 30-40 N/mm ². ^[17] These are well in excess of the values achieved for coating adhesion and weld shear strength in this study, and even the higher results of Ballard, leading to the conclusion that the weak point in a plastics metal joint is the metal-coating interface as confirmed by weld shear tests in this study. This will also be the case with alternative coating methods, and it is difficult to see how it can be overcome apart from specific situations where chemical bonding between the coating and substrate can be used. There may therefore be some limit on the usable strength of plastics/metal joints, although attention to component design to achieve the maximum possible joint area would be of benefit.

A further problem exists and must be borne in mind when considering the joining of plastics to metals using a plastics coating interlayer, namely that of dissimilar materials. Welding dissimilar plastics together may present problems, and powder grades of the major plastics materials such as polyethylene, polypropylene and polyamide do not appear to be compatible with the extruded or moulded grades used in component manufacture.

In this work, the PPA melted at approximately 120°C while Propathene melts at around 220°C. The PPA mmaterial, although nominally a polypropylene alloy, in fact only contained a small amount of PP. ^[18] It is common practice apparently, for plastics powders to contain other polymers ( e.g. polybutylene in PPA), to lower the overall melting point and thus facilitate coating, and other additives to improve the flow performance of the coating material.

Uniform melting was not possible with ultrasonic welding but to achieve it in hot plate welding, the heating times were varied. However, this is a time consuming process and would not lend itself to a production line environment such as in the automotive industry. Thus welding dissimilar materials may be inhibited.

Summary

The flame spraying of thermoplastic materials on to metal substrates has been examined, with the objective of using plastics coatings to achieve joints between plastics and metals. A manual single and multilayer spraying procedure was obtained for a range of plastics coatings deposited on to sheet steel or aluminium. Coating soundness and bond strengths were assessed. Hot plate and ultrasonic welds were produced between sheet metal and polypropylene sheet, using a nominally polypropylene based coating as an interlayer, and shear tested.

The following conclusions were reached:

1. The manual flame spraying of thermoplastics requires a level of operator skill but, once this has been achieved, a number of materials may be sprayed on to metal substrates. The required spraying conditions are not unacceptably critical, and five layer coatings of over 1mm thickness were obtained.
2. To achieve an optimum quality coating the substrate must be preheated to above the melting point of the plastics powders.
3. The sprayed coatings contained some porosity of 50-200µm diameter, the extent varying between materials. No interface defects were observed.
4. Coatings on grit-blasted surfaces gave better adhesion results than were obtained on surfaces which had only been degreased. Adhesive rather than cohesive failure was generally observed.
5. Adhesion strength varied between materials, values around 1-4 N/mm ² being obtained. However, all strengths were substantially lower than typical tensile strengths of the plastics materials in solid form.
6. Welds were successfully produced between the coated metal sheets and polypropylene by both hot plate and ultrasonic methods. Shear strength levels were approximately 2-6 N/mm ²; weld failures mainly occurred at the coating/substrate interface.
7. It is concluded that flame spraying can be used to provide an intermediary for joining plastics and metal components ( Fig.17), but consideration of the coating defects and low coating/substrate bond strength will be necessary. The compatibility of the plastics materials involved must also be considered.

More details on plastics spraying are available from Martin Day or Chris Quy at Abington.

Acknowledgements

The work was funded by Research Members of The Welding Institute. The spray torch used in the programme was supplied by Schori (UK) Ltd.

References