Analysis of test methods for UV durability predictions of polymer coatings

F+iYszE@ COATINGS ELSEVIER Progress in Organic Coatings 27 (1996) 95- 106 Analysis of test methods for UV durability predictions of polymer coatings B.W. Johnson, R. McIntyre Courtaulds Coatings Ltd., Stoneygate Lane, Felling, Gateshead, Tyne Br Wear, NE10 OJY, UK Received 17 October 1994; accepted 22 March 1995 Abstract The purpose of this paper is to review procedures which are used for the evaluation of the durability of polymer coatings. In particular, methods of environmental acceleration and techniques of assessment of coating degradation have been examined, with an emphasis upon those which may produce reliable fast answer durability predictions. The advantages and disadvantages of the various exposure regimes currently used (such as Florida exposure, EMMAQUA or artificial light sources) have been discussed in terms of correlation with actual durability data and of degree of acceleration. A general rule of thumb is that the correlation of accelerated methods with natural exposure is inversely proportional to the degree of acceleration used. The common physical methods used to assess the extent of coating durability have been presented, with the general drawbacks to these techniques being highlighted. Finally, the benefits and drawbacks of a number of chemical techniques (in particular electron spin resonance (ESR), Fourier transform infrared spectroscopy (FTIR), hydroperoxide determination and chemiluminescence), which in principal could provide durability information in a fraction of the time of the physical techniques mentioned earlier, have been discussed. Of these, both ESR and FTIR spectroscopy show particular potential because of the short exposure times necessary to obtain significant results under UVA exposure. Keywords: Durability; Polymer coating; Coating degradation; Accelerated exposure tests 1. Introduction The exterior durability of any organic coating can be defined as its resistance to undesirable changes caused by the natural environment to which it is exposed during its service life. The main constituents of the environment which cause the weathering, or degrada- tion of, polymer films are sunlight (in particular UV radiation), temperature, oxygen, water and pollutants. When these factors are combined with other environ- mental variables, such as wind and seasonal periodicity, the magnitude of the problem of producing weather resistant coatings becomes very apparent. Of the above factors, it is usually found that sunlight (that is photodegradation caused by W-Vis radiation) is the most damaging, although the synergistic contribution from the others plays a significant part [l-5]. The degradation of a polymer coating as a result of weath- ering can manifest itself in many ways, for example by loss of gloss, discolouration, chalking, embrittlement 0300-9440/96/%15.00 0 1996 Elsevier Science S.A. All rights reserved SSDI 0300-9440(94)00525-6 etc. As the primary function of a protective coatings (paint) is to protect the substrate and, usually, to fulfil a decorative role, its resistance to outdoor weathering is of paramount importance. The rate at which polymer degradation occurs will vary widely, depending upon temperature, exposure site, time of year, substrate, polymer material etc., all of which makes the evaluation of coating durability difficult. The ideal evaluation of a protective or decora- tive coating would be to expose it in its intended service environment. However, in an industry which is striving to produce coatings of longer and longer lifetimes, this type of evaluation would take an unacceptably long time. Therefore, it is necessary to reduce the time needed to predict coating life. The current method of assessing coating durability is to monitor the gloss loss, colour retention and chalking during natural exposure in areas of high sunlight inten- sity (for example in Florida or Arizona), a procedure which, for more durable systems, can take upwards of 96 B. W. Johnson, R. McIntyre / Progress in Organic Coatings 27 (1996) 95-106 five years to obtain satisfactory results. This length of time is significantly reduced to 12-18 months by the employment of a series of mirrors to intensify the sunlight falling upon the sample, processes known as EMMA and EMMAQUA [6]. Although the results obtained from EMMA and Florida exposure generally give rise to reliable predictions of polymer durability, the techniques themselves are both time consuming and expensive. It is clear that, if successful, the development of a reliable fast answer test which could determine polymer durability in significantly shorter times would reduce, or even obviate, the necessity to send panels to Florida/ Arizona for natural weathering thus saving both time and money. Furthermore, such a fast answer methodol- ogy would accelerate the coating development process and facilitate the rapid transfer of new coating tech- nologies to the market place. The two general ways to accelerate durability testing are either: (a) Use higher intensity and/or shorter wavelength radiation, such as QUV, xenon arc or carbon arc. The advantage of these techniques is that gloss loss and colour retention characteristics can be acquired in times of l-2 months. However, for such accelerated exposure media to provide data which can be extrapolated to long times and which will correlate with natural expo- sure results, there must be no change in either the relative rates of degradation processes occurring or the type of degradation chemistry. Unfortunately it is often the case that ‘unnatural chemistry’ occurs due to higher energy radiation [7,8], which is consequently reflected by a poor correlation between accelerated and natural exposure data. (b) Use more sensitive techniques to evaluate the degradation process. The standard methods of assess- ment are gloss loss, chalking etc., which are essentially measurements of the physical manifestation of degrada- tion. Whilst it is recognised that it is possible for a coating to fail due solely to physical effects, in general, before the physical properties change the chemical properties (especially at the coating surface) will change. Hence, the use of techniques which can detect and quantify chemical changes ought to provide durability information in a fraction of the time. In order that a fast answer test be implemented within the coatings industry, a certain number of crite- ria must be satisfied. By definition the test method must be significantly faster than EMMAQUA and Florida exposure. The technique must not be labour intensive. Florida exposure involves the measurement of gloss, colour and chalking, all of which are very quick and easy, meaning that a high number of samples may be examined in a short time period. Any new method should aim to have a similar measurement time in order that sample turnover remains high. No more than a low level of expertise should be required. If a new method necessitates skilled personnel with a high level of training, then its introduction to, and acceptance by, the coatings industry will be more difficult to achieve. A good correlation with natural/Florida exposure should be apparent, otherwise any time savings will be offset by uncertainty and lack of confidence in the results (in the same way as with QUVB exposure). The running costs should be low (cheap). However, if the technique satisfies the above criteria, the expense may not be too critical. A sensitive technique is required as differences which manifest themselves over long time periods are being sought in much shorter times. Reproducibility of results is important, however re- producibility compared with a known standard would be sufficient. It is the aim of this paper to consider the current, and possibly future, techniques of assessing the extent of polymer degradation, with an emphasis upon those which operate upon much accelerated timescales. The test methods discussed will be illustrated by a series of examples involving a range of polymers and polymer blends. 2. Exposure media 2.1. Natural exposure There are a number of exposure sites around the world, each with a different climate, which can be used for the purpose of predicting the exterior durability of coatings. The site chosen should be the one which is most representative of the general environment of inter- est, however, it is more usual to choose sites of extreme climates and compare the resulting degradation with the degradation which occurs under the service environ- ment. In this way ‘natural’ exposure is achieved, even though the UV, temperature and humidity levels may be vastly different from those met in service. The expo- sure sites are as follows: (a) Miami, Florida. This region of Southern Florida has been the benchmark for environmental testing for a number of years. High levels of UV radiation, humidity and temperature provide a sub-tropical environment which is itself an acceleration of European climates (an acceleration of approximately four times [9]). The key element of Florida exposure is that, although it may be an ‘artificial’ test (because it cannot replicate all the variables found in different climates around the world), it does use natural sunlight in a defined environment that provides reproducible results. Florida exposure is an industrially accepted standard and is widely specified for outdoor exposure testing. B. W. Johnson, R. McIntyre 1 Progress in Organic Coatings 27 (1996) 95-106 91 (b) Wittmann, Arizona. With an annual rainfall of tion [12]. It is important to note that the temperature of 200 mm and an average high temperature of 26 “C this EMMAQUA samples may vary over a wider range site provides a hot dry climate for durability testing. than the temperature of Florida samples which is Arizona exposure is the second most accepted site for highly likely to contribute somewhat to any lack of environmental testing. correlation. (c) Hook of Holland. This site provides the combined conditions of an industrial and a marine environment. It is generally used for the evaluation of coil coatings and is specified in the European Coil Coaters Associa- tion T19 test method. There are also a number of other exposure sites around the world, including Australia, Israel, France and Saudi Arabia, each with its own particular climatic characteristics. It is important, during natural exposure evaluations, to monitor the exposure conditions (for example, rain- fall patterns, pH of rain, radiation exposure (either spectral density with time or at a given wavelength, e.g. 340 nm, with time), pollutant concentrations and time of initial exposure. The latter is particularly important because the exposure sites themselves are subject to seasonal changes and the results are likely to be depen- dent upon the time of year of initial exposure. The major assumption of EMMAQUA is that gloss loss (extent of degradation) is directly proportional to the total cumulative exposure medium energy. Thus, EMMAQUA results are usually represented as gloss loss versus total energy of exposure, the total energy normally being converted to equivalent Florida years (1 Florida year = 308 MJ rns2 radiation [13]). It has been clearly demonstrated [14] that this assumption is not strictly true and that the gloss loss is in fact propor- tional to both the total energy and the energy flux. In other words, a seasonal variation occurs in which sam- ples exposed will exhibit quite different gloss loss char- acteristics depending upon the time of year of initial exposure. 2.3. Art$cial exposure methods 2.2. EMMA /EMMAQ UA The weathering of materials can be further acceler- ated by a more efficient use of outdoor exposure. An example of this is the widespread use of EMMA (Equa- torial Mounts with Mirrors for Acceleration) and EMMAQUA (EMMA with a water spray cycle). These machines have ten highly polished moveable aluminium mirrors which enhance the rays of sun focused upon a test sample. Although the total amount of radiation reaching the surface with EMMA and EMMAQUA is around nine times that of ‘natural’ exposure, the deteri- oration rate is found, on average, only to be in the region of 5-6 times faster. This is a consequence of the fact that the aluminium mirrors reflect the more dam- aging UV components of sunlight less efficiently than the visible and infrared components. As a result of the high radiation intensity focused upon a sample, temper- atures of up to 150 “C at the sample surface may develop, hence the application of air cooling is neces- sary. A drawback of these techniques is that a continu- ous clear sky is required for their most efficient use, hence, Arizona, with over 4000 hours sunshine per annum, is an ideal test site. The main reason for using artificial weathering methods is to accelerate degradation, under controlled conditions, in order that the durability characteristics of a particular polymer can be assessed in an acceptably fast time. There are many disadvantages in the use of artificial acceleration regimes for durability evaluation. For example, an accelerated test is only viable if all degradation processes in a sample are accelerated to the same extent, a condition which, unfortunately, is rarely achieved [7,15]. Also, if the wavelength of the light source is of higher energy than solar radiation (e.g. QUV-b) then photochemistries which would not occur under natural exposure may result [8]. There are a large number of commercial artificial acceleration procedures available, generally based on the same two or three light sources, as follows: 2.3.1. Carbon arc sources The correlation of results with natural exposure is often relatively good [lo, 1 l] but difficulties can arise due to the fact that polymer coatings have different levels of a number of constituents, the degradation of each likely to be accelerated by different degrees. For example, a correlation of gloss results has previously been observed for a series of paint films, yet colour retention for the same samples exhibited a poor correla- The enclosed carbon arc lamp was the original artifi- cial weathering source, being first employed in 1918 to evaluate textiles. Its spectral output has little compari- son with natural sunlight, the main radiation taking the form of three high intensity peaks between 350- 450 nm. Furthermore, it has a significant amount of radiation below 295 nm (the approximate solar cut off) although this can be removed by the use of suitable filters. Test procedures specifying the use of enclosed carbon arc are becoming fewer in number, but it is still used for some US (AATCC) and Japanese (JIS) standards. The sunshine carbon arc is also being progressively phased out of international test specifications but is still widely used in industry for company specifications. Its output is closer to that of sunlight, but it still has 98 B. W. Johnson, R. McIntyre 1 Progress in Organic Coatings 27 (1996) 95-106 significant differences, in particular being more intense in the 350-450 nm region and less intense below 350 nm. Despite this, the sunshine carbon arc is still commonly used to reproduce the effect of daylight exposure on textiles. Control of temperature, humidity and water spray is possible with these systems. 2.3.2. Xenon arc sources This light source provides a much better simulation of sunlight than carbon arc sources, which explains its growing use in the testing of a wide range of materials. Xe arc contains UV radiation of wavelengths shorter than the cut off level of solar radiation, however this is removed by use of a filter system. Other problems with Xe arc are that a high intensity of infrared radiation is emitted which must be removed in order to prevent overheating of the samples and that both the Xe burn- ers and filters deteriorate with use; thus their perfor- mance must be regularly checked [16]. Apart from its close simulation of the solar spectrum, other advan- tages of Xe arc include automatic control of light intensity, temperature, light/dark periods and humidity giving rise to a consistent and reproducible testing method. Furthermore, the incorporation of atmo- spheric pollutants into the system make these instru- ments very useful in relating the behaviour of materials to the aggressive factors of the outdoor environment. 2.3.3. Fluorescent tube lamps These sources are much cheaper to run than Xe arc and have the advantage that they do not produce any unwanted heat. Tubes are available with different spec- tral outputs, the most common being UVB and UVA bulbs. The UVB spectrum has a peak intensity at around 313 nm, with a high intensity at wavelengths below the solar cut off (i.e. 270-290 nm). Consequently they provide strong acceleration, but are also likely to induce unwanted or unnatural chemical reactions [4]. UVA bulbs provide a much closer reproduction of the solar spectrum up to around 350 nm and can therefore produce a much closer correlation with natural expo- sure. As only a negligible heating effect is produced with these lamps, a separate heating source is necessary which imparts reasonable temperature control to the user. Again, with these systems, temperature, humidity and light/dark cycles can be controlled. The correlation of results between artificially and naturally weathered samples is often quite poor [6,17- 191, The main reason for this is the non-agreement between the solar spectrum and that of the artificial light source. Differences in the visible region can lead to poor correlation with coloured materials, although it is generally differences in the UV region which are most likely to cause deviation. It can be taken as a general rule of thumb that the correlation of accelerated meth- ods with natural exposure is inversely proportional to the degree of acceleration used. 3. Physical measurement techniques The standard methods of assessment of the perfor- mance of polymers exposed in Florida are gloss loss, chalking, colour retention, cracking and crazing, and dirt retention. Dirt retention is not strictly a measure of the extent of degradation, but obviously can be very important, as for architectural/decorative coatings. The remainder, along with several others, measure physical changes due to weathering and will be briefly discussed below. 3.1. Gloss loss and chalking As a pigmented film degrades, the surface gradually roughens due to the body of the binder being lost, which is seen as a loss of gloss. This disappearance of the binder eventually leaves behind the more stable pigment particles, which remain upon the surface of the film in the form of non-bound powdery particles, a process termed chalking. Chalking and gloss measure- ments generally go hand in hand, with gloss loss be- coming evident before chalking (chalking will not usually be seen until a significant loss of gloss has occurred). Gloss is assessed by monitoring the reflection from the surface of a visible beam of light in relation to a standard back glass surface, e.g. Fig. 1. Chalking is measured either by performing a ‘tape pull’ or simply by visual assessment. It is common for many coating specifications to indicate the minimum time to which the coating will maintain a gloss level of greater than 50% of its original value. The advantages of gloss and chalking measure- ments are that they are very fast and easy to perform and are accepted as being an accurate indication of the overall durability of a polymer film. Moreover, whether the polymer is degrading mainly via photodegradation 100 80 20 0 0 200 400 600 ml lotHI QW Time I Houn Fig. 1. Gloss retention vs. QUV exposure time for a series of poly(ester)/acrylic blends. - 0-, 100% acrylic; -A-, 80:20; -m-, 60:40; -0-, 40~60; -A-, 20:80; -!I--, 100% PE. B. W. Johnson, R. McIntyre / Progress in Organic Coatings 27 (1996) 95-106 99 0.1 1 I 0 so0 1ooo 1500 QUV Time / Hours Fig. 2. Change in M, with QUVB exposure time for a crosslinked poly(ester) coating. or mainly via mechanical stressing, gloss loss will still detect the changes. The disadvantages are that they are a measurement of the physical manifestation of degra- dation and, in most cases, will not detect the changes due to photodegradation as quickly as the more sensi- tive chemical techniques discussed later. 3.2. Colour retention For a pigmented coating system there are three possible effects upon the colour; either the colour fades, it darkens or there is an overall yellowing due to oxidation products. This is usually assessed visually and any change from the initial state graded. Colour reten- tion in paint coatings is obviously very important, hence, it is necessary to record such data. The value of this information would be highlighted in a situation whereby the polymer itself was stable to UV and the pigment unstable. Changes due to polymer degradation would not be seen by the conventional gloss and chalk- ing measurements, yet fading/darkening could still occur. 3.3. Cracking and crazing The cracking and crazing of most solvent-based clear polymer systems has been well known for many years [5,20]. Exposure of a clearcoat to UV radiation results in a gradual transformation of the longer, more flexible, molecules into shorter, more brittle, ones meaning that the film in general hardens and cracks begin to appear, which can eventually penetrate the whole film. It is possible for an abrupt failure due to cracking to occur before any appreciable changes in coating gloss become apparent, hence this is an important parameter to monitor. 3.4. Contact angle The surface tension of most polymer coatings would be expected to increase in proportion to the extent of degradation which has occurred at the surface, due to an increase in polar groups (i.e. oxidized species) and to suface roughening. This can be estimated by recording the contact angle that a drop of deionized water makes with the polymer film. For lower surface tension mate- rials (undegraded), the contact angle made by water is high whereas for higher surface tension materials (de- graded) a lower angle results. This method has been used to demonstrate that pigmenting a poly(urethane) top coat results in reduced durability when exposed to QUVB radiation [21]. Similar work has been performed upon a range of Courtaulds poly(ester)/acrylic blends with similar results [22]. However, this method did not distinguish between the performance of the different blends any faster than did standard gloss loss measure- ments. Although contact angles can give a general idea as to whether a polymer film is degrading, they do not give any comparative durability information in timescales significantly less than do standard gloss measurements. Therefore, it is unlikely that contact angle measure- ments can be used in a fast answer role; however useful support data can be obtained from this quick and easy technique. 3.5. Weight loss and film thickness As described before, there is a gradual erosion of a paint film surface with time due to weathering. Thus, the degradation of a coating can be monitored with time by measuring its dry film thickness (DFT) or its actual weight loss. Most polymers show an average loss in DFT of between 5 and 25 um/year [20,23-251, except for a few more stable polymers such as poly(methacry- lates) and poly(vinylfluorides). The loss in DFT has been used [20] to help demonstrate the influence of molecular weight upon durability. For co-polymers with the same chemical composition but an order of magnitude difference in molecular weight (100 000 com- pared to over 1 000 000) the higher molecular weight systems possessed outdoor durabilities of over twice their lower molecular weight counterparts. 3.6. SEA4 and optical microscopy Scanning electron microscopy can be used to take a snapshot of surface morphology, at magnifications of up to 100 000 and resolutions of around 8 nm, at any time during a weathering test. If suitable times are chosen then this may provide support evidence for other more sensitive durability tests, as well as giving an excellent visualisation of the physical results of degradation. For example, previous work [26] has used SEM to demonstrate that there is little change in the surface appearance of an alkyd, a poly(urethane) or an acrylic latex top coat during the early stages of degra- dation, even though more sensitive techniques, such as X-ray photoelectron spectroscopy (XPS), were able to pick out changes in chemical composition. In the same 100 B. W. Johnson, R. McIntyre / Progress in Organic Coatings 27 (1996) 95-106 work, after longer exposure times, the magnitude of the chemical changes quantified by XPS had increased and it was possible to observe surface changes using SEM. Work performed in our laboratories [22] has used SEM to monitor the surface of clear poly(ester) films with QUVB exposure time. This work demonstrated the gradual growth of spots of degradation product, which eventually developed into cracks penetrating into the coating. However, the actual quantification of this degradation using SEM was very difficult, as has been shown before [22,27]. Apart from the capital expense and the necessity for a skilled operator, the main disadvantage of SEM is that it is a destructive technique, hence the surface of one particular piece of polymer cannot be monitored with time. All of these problems can be overcome with the use of optical microscopy, the penalty being a reduction in the degree of magnification possible. 3.7. Dynamic mechanical analysis Dynamic mechanical analysis (DMA) has been used [21] to investigate the effect of QUVB exposure upon the storage modulus and the Tg of clear and pigmented poly(urethane) coatings. Large differences in both the Tg and the storage modulus with weathering time were observed in these systems over periods of up to 6000 hours exposure. DMA experiments have been performed in this laboratory upon a series of poly- (urethane)-acrylic systems with similar results [22], however, great care had to be taken when interpreting results to consider the contribution residual solvent loss made to the increase in Tg (as compared to further curing of the system). In the same work, DMA was used to show that very little change in the value of the Tg of a range of crosslinked poly(ester) coatings after 1200 hours QUVB exposure occurred [22], whereas a significant reduction in the rubbery modulus and a decrease in the height of the tan 6 peak associated with the glass transition were observed. Analysis of these results indicated that the degradation of the poly(ester) coating was predomi- nantly by chain scission as opposed to oxidative crosslinking (Fig. 2). Although DMA tests the proper- ties of the whole of a polymer film, it is still a fairly sensitive technique towards extent of degradation. It would not form the basis of any fast answer technique, but can lead to information as to the physical changes upon weathering. 3.8. Other physical methods There are a number of standard physical and me- chanical tests such as impact, tensile strength, elonga- tion at break, solvent swelling, water vapour transmission and laser stylus profilometry which have been applied in order to assess the extent of weathering [5,21,26,28-301. These measurements generally do show a rough correlation with exposure time but are difficult to reproduce and do not have the sensitivity to give more than a qualitative indication of durability. 4. Chemical measurement techniques The mechanism of polymer photodegradation is widely proposed [5,31,32] to follow a general scheme which involves the production, and further reaction, of free radicals formed as a result of exposure to UV radiation: Initiation: Propagation: R+hv-2R’ R’ + O2 - ROO’ ROO’ + RH - ROOH + R’ Chain Branching: ROOH- RO’ + HO’ 2ROOH - ROO’ + RO’ + H,O RO’+RH-ROH+R’ HO’+RH-R’+H,O Termination: 2 radicals - products From this general model, several chemical tech- niques which would monitor such reactions become immediately obvious and will be discussed in the fol- lowing sections. Surface analysis techniques, such as FTIR-ATR and XPS, would appear particularly suit- able as the UV light intensity is highest at the polymer surface and oxygen is not diffusion limited. There are also a number of other chemical methods which may fulfil the role of a fast answer test and these will also be considered. 4.1. Electron spin resonance spectroscopy Probably the most obvious technique which suggests itself to monitor photodegradation is electron spin reso- nance (ESR) spectroscopy. The above mechanism x a “““““““““““““’ 0 1 2 3 4 5 Nitroxide Concentration Fig. 3. The relative photoinitiation rate (PIR) of three different coating systems after several hours exposure to UVA irradiation. - * -, epoxy; - 0 -, poly(ester); - A - , lumiflon. B. W. Johnson, R. McIntyre / Progress in Organic Coatings 27 (1996) 95-106 101 shows that free radicals are involved in both the initia- tion and the propagation steps of degradation. The advantages of using ESR are that it is chemically specific, sensitive and quantitative, thus would lend itself well to the early quantification of photodegrada- tion reactions. There are two approaches to using ESR. The first [33-361 involves the measurement of free radicals pro- duced by in situ UV irradiation. However, as free radicals are generally highly reactive and very short lived at room temperature, it is usually necessary to perform the experiments at 77 K. In doing so, there is a level of uncertainty introduced as there is no guarantee that the free radicals produced and observed at such low temperatures will be the same as those which occur at ambient temperatures. Previous work, using low temperature ESR [35], has successfully ranked twenty different clear 2-pack poly(urethane) coatings, correlat- ing the rate of radical formation with the time to failure (cracking in this case) under different accelerated weathering regimes. The effect of the addition of differ- ent stabilizers and different curing chemistries was suc- cessfully evaluated. Other work [36] has used low temperature ESR to highlight the effect of different crosslinker levels in epoxy-melamine coil coatings, sug- gesting that higher melamine levels lead to lower durability. This method has been used on a series of Courtaulds poly(ester) systems comprising co-polymers of poly(esters) of different natures [37] to demonstrate successfully a correlation between outdoor durability and acyl radical concentration (formed in the primary photolysis step). In all of the above work, the exposure time necessary to generate results was no more than ten hours. The second method, pioneered by Gerlock and co- workers [38-421, involves the infusion of a persistent nitroxide into a coating which scavenges the free radical intermediates formed during photolysis. In the early work, stable nitroxide radical was homogeneously added, during preparation, to the polymer film in differ- ent concentrations (in order to eliminate the effect of side reactions) and its ESR intensity monitored with time. From such experiments the initial photoinitiation rate (PIR) of the coating can be determined, which has been shown to be related to its overall durability. For example, the PIRs of a series of acrylic-melamines [38] and acrylic-urethanes [39] have been determined in timescales of around 1 h and found to correlate very well with gloss loss measurements determined after over several hundred hours QUVB exposure. This work was extended to examine post-weathering nitroxide infu- sion, rather than pre-weathering, with similar results being obtained [40]. In order to examine post-weathering films, the ni- troxide could not be added in the preparation stage, hence a second method was developed which involved * 0 0 ...,,,...,,,..........*” 0 200 400 600 806 1000 1200 Expomm (QUV houn) Fig. 4. Changes in gloss retention for the above coatings when exposed to QUVB irradiation. -A-, lumiflon; - 3 -, epoxy; --o-, poly(ester). the diffusion of gaseous nitroxide (with the aid of a transfer agent) into the polymer. The determination of the PIR of acrylic-melamines and poly(ester)- urethanes after ‘near ambient’ exposure, i.e. UVA type light, has shown that the acrylic-melamine system reaches a steady state after around 10 h exposure which is sustained for times of over 1000 h, whereas the poly(ester)-urethanes display a gradual increase in PIR over the first 50-100 h before the photooxidation turns autocatalytic and the PIRs increase rapidly [41]. The lack of auto-catalytic peroxide decomposition in the melamine-acrylics was attributed to the decomposition of hydroperoxides by the melamine crosslinker before they could reach a high enough concentration to go auto-catalytic. A correlation between the PIRs of a series of aircraft urethane systems and their Florida exposure performance has been achieved [21], highlight- ing the great potential of the ESR technique for com- parative fast answer data. Similar work has been performed [43] using a phe- noxyl radical instead of the nitroxide radical. This work evaluated the effects of the addition of various UV stabilisers to an acrylic clearcoat and also correlated the ESR results with time to failure under QUVB exposure. Work performed in this laboratory has used ESR to compare systems of known relative durabilities, namely lumiflon, a poly(ester) and an epoxy. Fig. 3 displays the PIRs of each of the systems when irradiated with UVA radiation for periods of up to 10 h, 0 loo 200 WV Tie I Hours Fig. 5. FTIR-ATR assessment of the degradation of poly(ester)/ acrylic blends exposed to QUVA. -0--, PE 100%; -A-, PE 75%; - 0 -, PE 50%; ~ 0 -, PE 25%. 102 B. W. Johnson, R. McIntyre / Progress in Organic Coatings 27 (1996) 95-106 whilst Fig. 4 gives the data obtained from gloss read- ings over a considerably longer period of exposure to QUVB. It should, of course, be mentioned that these three coatings are vastly different in their UV durability and, therefore, achieving a good correlation could be consid- ered to be an easy test of the ESR method. For this approach to be a valuable fast answer method it really ought to be capable of discriminating between coatings with only lo-20% difference in their actual external exposure performance, for different polymer chemistries and also for materials consisting of blends of different chemistries. To this end, further work was performed which was successful in ranking the relative durabilities of a range of poly(ester), poly(ester)/acrylic and acrylic systems using both the phenoxyl radical [44] and the nitroxide radical [45,46] techniques. 0 so 100 150 mo QLwnme/Hour Fig. 6. FTIR-ATR assessment of the degradation of polyester/acrylic blends exposed to QUVB. - 0--, PE 100%; -A-, PE 75%; - ??-, PE 50%; - ??-, PE 25%. As a fast answer method for the prediction of coat- ing durability ESR has great potential, chiefly because a coating can be fully tested in a mater of hours. The nitroxide infusion method is at a disadvantage to the low temperature irradiation method in terms of time as it requires a number of experiments at different nitrox- ide concentrations in order to eliminate the effect of side reactions. However, the nitroxide method is per- formed at room temperatures, removing any uncertain- ties regarding the validity of measuring degradation processes at lower temperatures. The main disadvan- tages of using ESR are that it is a labour intensive operation requiring highly skilled personnel, the capital expenditure is relatively high and it requires a great deal of effort in optimization before any meaningful measurements can be made. tion in the C-H vibration with exposure time, a result which was reproducible. For the same coating systems, the effect of adding either UV absorbers or hindered amine light stabilizer (HALS) upon the FTIR charac- teristics has been clearly demonstrated, with the HALS- containing systems showing the least changes during 3000 h UVB exposure [54]. Work performed upon poly(urethane)-acrylic topcoats has used transmission FTIR to illustrate the effect of crosslinker type upon the rate and type of degradation [53]. 4.2. Fourier transform infrared spectroscopy Fourier transform infrared spectroscopy is both a chemically specific and a very sensitive technique which renders it potentially a very powerful tool in the areas of fast answer methodology and photodegradation product identification. Early work to assess the rate of degradation [ 1,31,47,48] using FTIR was generally per- formed upon non-carbonyl containing polymers, such as linear low density poly(ethylene) or poly(propylene) to monitor quantitatively the growth of carbonyl inten- sity formed as a result of photooxidation. In order to gain information upon the products of photodegrada- tion, derivatization reactions have been performed [49- 51] whereby the products will further react with a chosen reactive gas (such as NO or SF,) to give prod- ucts which can be very clearly identified. As weathering is a surface phenomenon, FTIR-ATR (which samples the top 2-3 urn) would suggest itself very strongly as a method of following quantitatively the chemical changes which occur during degradation. A significant amount of work has appeared in the literature using this technique with some success [7,19,28,47,49,52,56,57]. An example of its use, not only in monitoring degradation but also following the chem- ical changes, has been provided for acrylic-urethane coatings [19]. It was shown that, although the samples degraded during both carbon arc and Florida exposure, the chemistry of degradation was different, meaning that the carbon arc could not be used as a valid accelerated exposure medium for the coatings in ques- tion. Other work [7,49] carried out upon poly(styrene) films indicated that the products of photooxidation resulting from exposure to either 253.7 mn or > 300 nm radiation were the same but that their distribution was different, that is the acceleration factors for the various processes, resulting from irradiation at shorter wave- lengths, were not the same. The FTIR-ATR technique has been used upon a series of Courtaulds poly(ester)/acrylic blends [22] ex- posed to either UVA or UVB radiation. The results obtained highlighted that, although a faster rate of degradation occured under UVB exposure, the tech- nique could predict reproducibly their relative durabili- ties in the order expected from known durability data after less than 300 h irradiation (Figs. 5 and 6). Much of the previous work [5 1 - 551 has been centred Similar work [22] was performed upon a poly(ur- upon transmission FTIR through thin polymer films ethane)-acrylic system to demonstrate the effect of (Z 5 pm). It has been shown, for example [52], that adding a hindered amine light stabilizer (HALS) to the melamine-acrylic coatings exhibit a quantitative reduc- coating. In the case of the non-HALS containing B. W. Johnson, R. McIntyre / Progress in Organic Coatings 27 (1996) 95-106 103 0.4 c 1 0300600900 1200 lsoo QUV TIME I HOURS Fig. 7. The relative degradation (arb. units) of a HALS and a non-HALS containing poly(urethane)-acrylic coating exposed to QUVB radiation. 0, HALS Sample 1; A, HALS Sample 2; I, HALS Sample 3; x , No HALS; 0, No HALS; 0, No HALS. coating the onset of degradation was immediate and continued at a constant rate for up to 1000 h UVB exposure, whereas, for the HALS containing film, there was essentially no change in the infrared characteristics for around 500 h, after which the coating degraded at approximately the same rate as the former coating (Fig. 7). It appears likely that the ‘incubation time’ of 500 h observed for the latter coating could be representative of the effective lifetime of the HALS. Photoacoustic (PAS) FTIR has also been used to study photodegradation [8,21,58]. PAS has the advan- tages that the sample does not have to be removed from the substrate and that not only is it a surface technique but it can sample to different depths into the surface. The disadvantages of PAS are that the signal-to-noise ratio is often high making the acquisition of high quality spectra difficult, saturation of the stronger bands can occur at penetration depths of only a few microns and considerable spectral distortion is also often apparent. For fast answer durability assessments, FTIR-ATR appears to be very useful as it examines the polymer surface which is the region of greatest interest. Other advantages are that FTIR spectrometers are relatively inexpensive and easy to use, computer manipulation allows accurate determination of the chemical differ- ences due to weathering, the average time required per LOOEM = T Ezl l.MIE-07 1 I 0 10 20 30 40 QIJV Time I Days Fig. 8. Hydroperoxide levels for an epoxy and a poly(ester) coating exposed to QUVB radiation. 0, epoxy; 0, PE. sample is 5- 10 min, the reproducibility of results is good and pigmented materials can be examined. Draw- backs to the technique are that it could reveal specific chemical information about the coating being tested and a series of measurements are necessary per coating system. 4.3. Iodometric titration The presence of hydroperoxides (ROOH) as stable intermediates in the photodegradation of polymers can be exploited by utilizing standard analytical methods [59-611 to measure thier concentration with time. The most common method of assessing the ROOH concen- tration is via iodometry, in which the iodide (I-) ion reduces the ROOH to water and alcohol and is itself oxidized to iodine (IT): ROOH + 31-(xs) - ROH + H,O + I, (1) As the reaction is one to one, a quantitative determina- tion of the free iodine produced will lead to the original ROOH concentration. The I; concentration can be determined via its UV absorbance, via potentiometry (the electrochemical potential set up at a platinum electrode is proportional to iodine concentration) or via back titration with a suitable reducing agent such as thiosulphate. Much work has been performed in this area in an attempt to use iodometric titration as a cheap and easy fast answer technique [42,50,57,62-641. For example, the ROOH concentration detected in melamine-acrylic coatings has been shown to increase by over two orders of magnitude after 500 h exposure to UVB before falling to a sustained lower level [62]. The durability of these coatings was claimed to be strongly dependent upon the magnitude of this sustained hydroperoxide concentration. Further, the ROOH concentration in these coatings was found to be proportional to the photoinitiation rate detected by ESR spectroscopy [42], a correlation which led to investigations as to how quickly changes in the ROOH concentrations with ex- posure time could be detected. The result of this work [64] indicated that the ROOH levels in acrylic- melamines and acrylic-urethanes increased after only l-2 d ambient exposure. Two very important points were highlighted as a result of this, and other [63], investigations. First, the profile of ROOH with time was different between the two coating chemistries used, with no sustained level after the initial increase being reached with the acrylic-urethanes. Instead, in these systems the hydroperoxide concentration eventually reached a critical level and autoxidation occurred, which then caused a very rapid increase in ROOH with time. Second, the consideration of any additives to a coating is important as, for example, nitroxide itself can oxidise iodide to iodine and can clearly affect any results [63]. 104 B. W. Johnson, R. McIntyre / Progress in Organic Coatings 27 (1996) 95-106 Such success prompted Courtaulds Coatings to ex- plore the value of the technique when applied to its own proprietary coatings [22,45,65]. Initial work [22] was performed upon an epoxy (based on Epikote 1004) doped with known amounts of t-butylhydroperoxide (t-BOOH) under a nitrogen atmosphere. Analysis of the films via potentiometry resulted in the expected Nernstian response of 30 mV difference between sam- ples, with over 70% of the t-BOOH added to the coatings being detected in all cases. Later work (Fig. 8) proceeded to demonstrate that this technique could distinguish quite clearly between a more durable polyester and a less durable epoxy coating exposed to QUVB radiation for up to 40 d. similar systems could not separate these latter two systems even after over 1500 h UVB exposure. Investigations upon three general systems of widely different and known durabilities [76] concluded that, after 532 h QUVB exposure, their relative durabil- ity order was Lumiflon > poly(urethane)-acrylic > an Epikote 1001 based epoxy coating. However, this iodometric technique could not clearly separate two poly(ester) systems of different known durabilities when exposed to UVA [45,65] or UVB [22] irradiation, even though, in both cases an increase with time was observed. CL is not particularly suitable for fast answer durability predictions, chiefly because of the long expo- sure times necessary (upwards of 1500 h QUVB for the more durable systems). When UVA exposure is used instead of UVB, there is very little change in the CL characteristics for a low durability poly(ester) system even after more than 3000 h exposure [75]. Other draw- backs include the non-ability of the method to examine pigmented systems and the fact that CL is destructive, hence a large number of samples are necessary. 4.5. X-ray photoelectron spectroscopy (XPS) The benefits of using hydroperoxide titration as a fast answer durability predictor are that it is simple, inexpensive, no highly trained operators are necessary, pigmented systems can be analyzed and specific chemi- cal information relating to the materials tested will not be revealed. However, there are many drawbacks to this technique such as it is destructive, the extent of penetration of the iodide into the polymer matrix is not known and will probably vary between polymer types, any reducible species (e.g. 0,) present in the coating could oxidize the iodide, reproducibility is difficult and the ROOH concentration is assumed constant through- out the polymer (which is not true, hence the thickness of the material tested is of importance). As XPS is a powerful surface technique it may be expected to provide useful data regarding chemical changes caused by weathering and has been suggested as having great potential in this area [27]. In reality, XPS has not been used with any great success because of its very small sampling depth, roughly 5 nm into the surface. Due to the small surface layer examined it is difficult to be sure that the results obtained are repre- sentative of the polymer in question as, for example, the presence of impurities at the surface could act as chromophores and induce changes. Furthermore, if the products of degradation at the surface are friable and drop off the polymer then XPS will be extremely sensi- tive to this, resulting in problems in making repro- ducible quantitative measurements. 4.4. Chemiluminescence The occurrence of very weak chemiluminescence (CL) upon oxidation of organic compounds has been known for many years and is a feature for most com- mon organic polymers [66-711. The source of the lu- minescence is generally agreed to be an excited carbonyl species, but the origin of this carbonyl is under dispute [67-691. However, whatever the mechanism of lumines- cence, it is directly proportional to the concentration of peroxide groups in the polymer. Therefore, CL can be used as a measure of hydroperoxide content and, hence, as an indication of the rate and extent of photodegradation. Despite these drawbacks, XPS has been used to demonstrate that the changes in the oxygen/carbon (O/C) ratio at a poly(urethane) surface are similar after 3000 h QUVB and 18 months Florida exposure [21]. The technique has also been used to correlate the changes in the O/C ratio of a poly(urethane), an alkyd and an acrylic latex during exterior exposure with the changes during a corrosion weathering cycle [26]. Within Courtaulds Coatings XPS has been used to detect an increase in the O/C ratio for a poly(ur- ethane)-acrylic, a poly(ester)/acrylic blend and an alkyd [22]. These changes, in most cases, were small and of a cyclic nature, suggesting the formation and loss of surface oxidation products with time. Attempts to use CL to predict the relative degrada- As a fast answer technique XPS is not particularly tion rates of a series of Courtaulds poly(ester) coatings suitable. The changes that are detected are difficult to have been made [72]. After 203 h QUVB irradiation it reproduce and are often not significant enough to lead was possible to distinguish between a less durable sys- to any overall durability prediction. Moreover, the tem and two more durable systems, however the latter equipment is expensive, the turnover rate of samples two systems could not be separated as the exposure can be low (up to 3 h per sample) and specialized time was not long enough. Later work [73-751 upon personnel are required. B. W. Johnson, R. McIntyre / Progre Table 1 Advantages and disadvantages of ESR and FTIR-ATR methods for providing a fast answer method to quantitatively predict coating lifetimes Requirements Ideal ESR FTIR-ATR Timescales d One point measurement Measurement time Sensitive Reproducible Capital Running costs Skilled operator Fast Yes Fast Yes Yes Low Low No &20 000 Low Yes f20 000 Low Easy training a Depend upon the exposure medium and the material being tested. The figures quoted here are for a 1 -5 year (Florida durability) poly(ester) under UVA exposure. 4.6. Oxygen absorption As can be seen from the general mechanism of polymer photodegradation, the consumption of oxygen is an integral part of the process. The direct measure- ment of this should be proportional to the extent of degradation and, if sensitive enough, could form the basis of a fast answer test. The technique has been used previously to evaluate the durability of an alkyd system containing a range of six different pigments [77]. The results, which were obtained over times of O-600 h UVA exposure, showed clear differences in oxygen consumption which were found to be in agreement with the durabilities predicted from Florida exposure. The sensitivity of the technique has been shown to be in the region of 2 x 105mol kgg’ polymer hr’ [l]. The major disadvantage of this type of measurement is that they are markedly influenced by temperature changes, with a 0.1 “C change likely to cause greater effects than the actual absorption of oxygen [78]. This is particularly important as, depending upon the UV source, the radiation can cause heating of the polymer. Other factors which make this method unsuitable for fast answer predictions are that the method itself is fairly labour intensive, it needs strict control over the variables and skilled personnel are required. 5. Conclusions The more common techniques which can be used to evaluate the rate and extent of polymer degradation, which occurs as a result of weathering, have been reviewed in this paper. The limitations, in particular with respect to time, of the industrially accepted weath- ering standard of Florida exposure in conjunction with gloss loss, colour retention, chalking and cracking and crazing monitoring have been highlighted. Orgunic Coatings 27 (1995) 95- 106 105 Of the other techniques considered, the two which appear most likely to form the basis of a fast answer method to predict quantitatively coating lifetimes are ESR and FTIR-ATR spectroscopy, chiefly because of their significant time advantages over EMMAQUA and Florida exposure. In both cases, further work would be necessary in order to establish their correlation with Florida and EMMAQUA. Table 1 summarizes the pros and cons of the two techniques. References [II 121 [31 [41 [51 [61 [71 PI 191 UOI 1111 [I21 [I31 [I41 [I51 [161 [I 71 [ISI [I91 WI PII P21 ~231 ~241 I251 WI v71 w ~291 [301 [311 [321 [331 A. Davis and D. Sims, Weathering of Polymers, Applied Science, London, 1983. H. HotImann and K. Vogel, Polym. Paint Colour J., 175 (1985) 802. SM. Halliwell, Rap. Rev. Rex, 5 (5) (1992) Report 53. N. Pate], J. Oil Colour Chem. Assoc., 3 (1991) 104. V. Schmid, Polvm. Pain? Colour J., 179 (1989) 206. G.A. Zerlaut, Proc. Inst. Emiron. Sci. (1974) 153. B. Mailhot and J.L. Gardette, Macromolecules, 2.5 (1992) 4127. D.R. Bauer, M.C. Paputa-Peck and R.O. Carter III, J. Coat. Technol., 59 (1987) IO. Accelerated and Natural Testing of Materials, Alpas Technology, Oxford. E. Oakley, J. Paint Technol., 43 (555) (1971) 43. J.L. Scott and T.E. Anderson, J. Oil Colour Chem. Assoc., 59 (1976) 404. H. Hoffmann and K. Vogel, Polym. Paint Coluur J.. 175 (4156) (1985) 849. A.B. Wooton, Polym. Paint Colour J.. 183 (4332) (1993) 346. J. Reid, personal communication. J. Sims, J. Coat. Technol., 59 (748) (1987) 45. D. Kockott, Angew. Makromol. Chem., 137 (1985) 1. M. Camina, Tech. Rep. No. TR,‘lWB, Paint Research Associa- tion, Middlesex, 1988. P.J. Pagan, MQAD Tech. Paper No. 787, Material Quality Assurance Directorate, Woolwich, London, 198 1. D.R. Bauer. J.L. Gerlock and R.A. Dickie, Prog. Org. Coar., 15 ( 1987) 209. E.V. Schmid, Estrrior Durability of’ Orgunic Coatings, FMJ, Surrey, UK, 1988. S.G. Croll, Prog. Org. Coat., 15 (1987) 223. B.W. Johnson, Tee/z. Rep. NCT 93/104, Courtaulds Coatings. H.S. Ritter, J. Paint Technol., 43 (552) (1971) 74. J.E. Fullard, J. Oil Colour Chem. Assoc., 52 (1969) 334. C.J. Berg, W.R. Jarosz and G.E. Salathe, J. Paint Technol., 39 (510) (1967) 436. G.R. Wilson, C.H. Wilson and B.S. Skerry, Proc. SSPC Coat. Eual. Durability Co@, Pittsburgh, PA, 1991. G. Kampf, XVIth FATIPEC Congr., Lugano, Italy, 1984, Vol. II, p. 81. J. White and A. Turnbull, NPL Rep. DMM(A) 75, National Physical Laboratory, Division of Materials Metrology, 1992. S.W. Shalaby, J. Polym. Sci. Macro. Rev., 14 (1979) 419. D.J. Rutherford and L.A. Simpson, Polym. Paint Colour J., 177 (4199) (1987) 653. J.F. Rabek, Mechanisms of Photophysical Processes and Photo- chemical Reactions in Polymers, Wiley, Chichester, UK, 1987. J.F. Mckellar and N.S. Allen, Photochemistry of Man Made Pol.vmrrs, Applied Science, New York, 1979. E. Butcher, UV degradation of polymer films as investigated by ESR, 1st Year Rep. AGNY, Courtaulds Coatings, 1992. 106 B. W. Johnson, R. McIntyre 1 Progress in Organic Coatings 27 (1996) 95-106 [34] S. Storp and M. Bock, Proc. XIth Int. Conf Organic Coatings Science and Technology, Athens, Greece, 1985, p. 385. [35] A. Sommer, E. Zirngiebl, L. Kahl and M. Schonfelder, Prog. Org. Coat., 19 (1991) 19. [36] M.R. Binns, C.A. Lukey, D.J.T. Hill, J.A. O’Donnell and P.J. Pomery, Polym. Bull., 27 (1992) 421. [37] E. Butcher, 2nd Year Rep. AGNY, Courtaulds Coatings, 1993. [38] J.L. Gerlock, D.R. Bauer and L.M. Briggs, Am. Chem. Sot. Symp. Ser., 243 (1984) 285. [39] J.L. Gerlock, D.R. Bauer, L.M. Briggs and R.A. Dickie, J. Coat. Technol., 57 (722) (1985) 37. [40] J.L. Gerlock, D.F. Mielewski and D.R. Bauer, Poiym. Deg. Stab., 20 (1988) 123. [41] J.L. Gerlock, D.R. Bauer and D.F. Mielewski, Free Rad. Res. Commun., 10 (1990) 123. [42] D.R. Bauer, D.F. Mielewski and J.L. Gerlock, Polym. Deg. Stab., 38 (1992) 57. [43] S. Okamoto, K. Hikita and H. Ohya-Nishiguchi, Eur. Coat., 3 (1990) 239. [44] A. Dolby, Tech. Rep. RES 89/077, Courtaulds Coatings, 1990. [45] A. Dolby, Tech. Rep. RES 9O/OOl, Courtaulds Coatings, 1990. [46] D.J. Holmes, Tech. Rep. RES 90/118, Courtaulds Coatings, 1990. [47] D.J. Carlsson and D.M. Wiles, Polym. Lett., 8 (1970) 419. [48] D.J. Carlsson and D.M. Wiles, Macromolecules, 2 (1969) 597. [49] B. Mailhot and J.L. Gardette, Macromolecules, 25 (1992) 4119. [50] D.J. Carlsson and J. Lacoste, Polym. Deg. Stab., 32 (1991) 377. [51] D.J. Carlsson, R. Brousseau, C. Zhang and D.M. Wiles, Chemi- cal Reactions on Polymers, American Chemical Society, San Francisco, CA, 1988. [52] L.M. Briggs, D.M. Bauer and R.O. Carter III, Ind. Eng. Chem. Res., 26 (1987) 667. [53] L.G.J. Van der Ven and P.J.A. Geurink, J. Oil Colour Chem. Assoc., II (1991) 401. [54] D.R. Bauer, M.J. Dean and J.L. Gerlock, Ind. Eng. Chem. Res., 27 (1988) 65. [55] D.R. Bauer, R.A. Dickie and J.L. Koenig, Ind. Eng. Chem. Prod. Res. Deu., 25 (1986) 289. [56] D.J. McEwen, M.H. Verma and R.O. Turner, J. Coat. Technol., 59 (755) (1987) 123. [57] R. McIntyre and C. Regan, Tech. Rep, NCT 94/027, Courtaulds Coatings, 1994. [58] P. Delprat and J.L. Gardette, Polymer, 34 (1993) 933. [59] I. Vogel, Textbook of Quantitative Chemical Analysis, Longman, Harlow, UK, 5th edn., 1989. [60] Kolthoff and Sandell, Textbook of Quantitative Inorganic Analy- sis, Macmillan, London, 1950. [61] C.D. Wagner and R.H. Small, Anal. Chem., 19 (1974) 976. [62] D.F. Mielewski and J.L. Gerlock, Polym. Deg. Stab., 33 (1993) 93. [63] D.F. Mielewski, D.R. Bauer and J.L. Gerlock, Polym. Mater. Sci. Eng., 63 (1990) 642. [64] J.L. Gerlock, D.R. Bauer and D.F. Mielewski, Proc. 18th Conf Organic Coatings, 1992, p. 225. [65] R. Howell, Tech. Rep. NCT92/20, Courtaulds Coatings, 1992. [66] G.E. Ashby, J. Polym. Sci., 50 (1961) 99. [67] M.P. Schard and C.A. Russell, J. Appl. Polym. Sci., 8 (1964) 985. [68] G.D. Mendenhall, Angew. Chem. Int. Ed., 16 (1977) 225. [69] G.A. George, Deu. Polym. Deg., 3 (1981) 279. [70] N.C. Billingham, J.W. Burdon, I.W. Kaluska, E.S. O’Keefe and E.T.H. Then, Proc. Int. Symp., Lucerne, Vol. 2, 1988, p. 11. [71] N.C. Billingham, E.S. O’Keefe and E.T.H. Then, Polym. Mater. Sci. Eng., 58 (1988) 431. [72] NC. Billingham and E.T.H. Then, Tech. Rep. MISC/9l/SUS- SEX (AGGX), Courtaulds Coatings, 1991. [73] J.C. Strong, Chemiluminescence of polymer films, Tech. Rep. AJAO, Courtaulds Coatings, 1992. [74] J.C. Strong, Chemiluminescence of poly(ester) coatings, Tech. Rep. AJDN, Courtaulds Coatings, 1993. [75] J.C. Strong, Chemiluminescence of poly(ester) coatings, Tech. Rep. AJDO, Courtaulds Coatings, 1993. [76] N.C. Billingham and E.T.H. Then, Tech. Rep. MZSC/9I/SUS- SEX (AFWN), Courtaulds Coatings, 1991. [77] A. Mackor and T.P.M. Koster, Polym. Paint Colour J., 179 (4248) (1989) 760. [78] T.M. Wills, personal communication.