Weatherability of coated EPDM rubber compound by controlledUV irradiation

Milena Ginic-Markovic, Namita Roy Choudhury *, Maria Dimopoulos, Janis G. Matisons

Ian Wark Research Institute, University of South Australia, The Levels, Mawson Lake, SA 5095, Australia

Received 17 November 1999; received in revised form 11 February 2000; accepted 20 February 2000

Abstract

Polymeric composites for automotive window seal applications are generally exposed to complex service environment conditions.The fundamental physico-chemical processes that control the durability of such composite systems on a micro and macroscopiclevel were evaluated in the present case using thermal analysis (TA), elemental mapping and microscopy. Five di�erent EPDM

rubber compounds (control, three bulk modi®ed and one surface modi®ed), coated with polyurethane coating were exposed tocontrolled UV irradiation. The higher the adhesion, the less the extent of UV penetration through the interface. The photo-oxida-tion mainly takes place from the surface, as the energy gets consumed and the intensity of the UV decreases on its passage throughthe strongly adhered layers to the rubber compound. The coating layer being on the surface, is more a�ected and photodegraded by

UV irradiation than the rubber compound. The sample P1cc, being multiphasic in nature and with a strong interface, can scatter theUV strongly and hence shows best weatherability among all the systems. # 2000 Elsevier Science Ltd. All rights reserved.

Keywords: UV exposure; EPDM rubber; Photodegradation; Viscoelastic properties; Thermal stability

1. Introduction

The automotive window seal section is normallybased on ethylene propylene diene (EPDM) rubbercompound. EPDM is selected for such applicationbecause it has very good ozone and weather resistance.In order to obtain low friction, high abrasion resistanceand release properties, the rubber compound is oftencoated with speci®c coating. Polyurethane (PU) coatingis an attractive coating for the automotive industrybecause it possesses the above mentioned excellentproperties.While the aging of the EPDM rubber [1], its com-

pound [2] and PU coating [3] were studied extensively,there are no reports on the UV degradation of coatedEPDM rubber. As aging can in¯uence the interactionbetween the components of a multicomponent systemand hence the performance, it is important to investi-

gate and understand the degradation behavior of thecoated rubber compound. Due to the increased use ofrubber and polymeric coatings for outdoor applicationsan adequate knowledge of the long-term performance ofthe UV exposed products is essential [4]. UV weatheringusually involves a complex sequence of chemical reac-tions. This knowledge also allows one to evaluate theproduct performance and predict lifetime.Various methods exist for characterization of weath-

ered polymers such as chemical, physical or mechanical.Several authors studied degradation of elastomers dueto the UV irradiation. Govorcin-Bajsic et al. [5] studiedthe e�ect of molecular weight of polyol (soft segment)and the concentration of urethane (hard segment) insegmental polyurethane elastomers on their photo-degradation. The comparison of photo-oxidation reac-tion of untreated and peroxide-vulcanized EPDMrubbers was presented by Guzzo et al. [6]. Decker et al.[7] reported the discoloration and chemical modi®cationoccurring on accelerated aging of UV-cured coatings.Sommer et al. [8] introduced a new method for evaluat-ing the weather resistance of polymers, by detecting theincrease of radical concentration in polymers during

0141-3910/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.

PI I : S0141-3910(00 )00053-7

Polymer Degradation and Stability 69 (2000) 157±168

* Corresponding author. Tel.: +61-8-8302-3719; fax: +61-8-8302-

3683.

E-mail address: namita.choudhury@unisa.edu.au (N.R. Choudhury).

irradiation with an intense light source (with the time)by electron paramagnetic resonance (EPR). Accordingto Decker et al. [9] the weathering resistance of organicmaterials can be substantially increased by protectingtheir surface with photo-cured coating containing bothUV absorber (UVA) and a radical scavenger (HALS).Dudler et al. [10] used chemiluminescence for studyingthe photostability of automotive coatings. An integrat-ing sphere-based ultraviolet exposure chamber designwas introduced by Martin et al. [11] claiming reductionin errors caused by previous designs.Blending of polymers often stabilizes or destabilizes

the degradation behavior of the resulting blend, how-ever, that depends mostly on the structure and type ofdegradation products. Deuri et al. [12] studied blends ofbutyl and EPDM rubbers. They noticed that EPDMwhen present in even a small amount signi®cantlyenhances the service life of butyl rubber. In our earliercommunication, we have reported the kinetics ofdegradation of EPDM rubber and its blends by modu-lated thermogravimetry (MTGA) and High-resolutionTGA [13].Despite many studies in the area of photodegrada-

tion of coatings and rubber compounds independently,not much work was published on the accelerated agingof coated rubber compounds. About 4% of the totalradiation (290±450 nm) from solar energy reaching theatmosphere in the form of the UV energy is primarilyresponsible for most of di�erent damage/changes inthe products during their service life. The rate atwhich such changes occur depend on a number offactors including geographic locations and season.Therefore, it is important to understand the degrada-tion behavior of a particular product under suchexposure.In the present study the accelerated UV degradation

of coated EPDM windowseal rubber compound wasundertaken, using controlled irradiation in a weathe-rometer. The EPDM rubber was surface and bulkmodi®ed to achieve better adhesion with PU coating.Surface modi®cation was performed by halogenationwith trichloro isocyanuric acid [14]. Bulk modi®cationwas achieved by partially substituting EPDM with twodi�erent rubbers. These rubbers were selected on theground of their structural similarity with EPDM. Theyare maleic anhydride grafted EPDM in which case thebackbone is similar to EPDM and the polynorbornenerubber (PNR) whose backbone is similar to the diene inEPDM.The e�ect of photodegradation on the viscoelastic

properties, surface characteristics and composition ofthe EPDM rubber compound and coating was studiedusing dynamic mechanical analysis (DMA), scanningelectron microscopy (SEM) coupled with energy dis-persive X-ray analysis (EDAX) and thermogravimetricanalysis (TGA).

2. Experimental

2.1. Materials

EPDM rubber (JSR EP 103AF), ethylene content63%, iodine number 15 was supplied by Japan SyntheticRubber. Grafted EPDM Royaltuf 485, ethylene content75%, iodine Number 10, total maleic anhydride 0.5%and Royaltuf 490, ethylene content 55%, iodine number17, total maleic anhydride 1.0%, were supplied by Uni-royal, USA. Polynorbornene rubber (PNR- Norsorex Nwith 60 phr oil), was supplied by Australian Vinyl, Vic-toria. Continental Carbon black (N660) used was man-ufactured by Cabot, Australia. Primer (Chemlock236X) was received from Lord Chemical Products. Atwo pack polyurethane coating formed from a mixtureof diol and diisocyanate was used. The component A isbased on various polyester polyols (Repol 560 and 650supplied by Townsend Chemicals, Australia) and Des-mophene A450 (polyacrylate) supplied by Bayer, Aus-tralia. The component B is Desmodur L75 based ontrimethylol propane and toluene diisocyanate, suppliedby Bayer, Australia. The detail of the coating prepara-tion is reported elsewhere [3].In the present study, EPDM was mainly bulk and

surface modi®ed. Blends of EPDM rubber with graftedrubber and PNR were prepared at 75/25 ratio (Table 1).The details of mixing and compounding are reportedelsewhere [15].For surface modi®cation, the control rubber sample

was immersed in a stirred solution of 3% w/v tri-chloroisocyanuric acid (TCICA, Sigma-Aldrich) inanalytical grade ethyl acetate (ACE Chemicals), fol-lowed by immersing the samples in a fresh solution ofethyl acetate. Immersion time of 120 s was used and thesample is designated as T3c. After treatment, the rubbersamples were air dried, placed in a dessicator and refri-gerated for 24 h. All samples were primed and coatedusing a spray gun supplied by Charmans Spray andPowder Equipment. The coated samples are designatedwith an additional c in the subscript of sample identi®-cation such as S1cc. Once samples were coated they wereair-dried and then cured in an oven for a speci®ed timeand temperature (100� C). All tests were performed after24 hours of curing.

Table 1

Blend composition

Components %

®nal/cured coated

100/0 85/15 75/25 65/35 50/50 0/100

JSR EP 103AF S1cc ± ± ± ± ±

JSR/Royaltuf 485 ± ± M1cc ± M2cc M4cc

JSR/Royaltuf 490 ± ± P1cc ± P2cc P4cc

JSR/PNR ± N1cc N2cc N3cc ± N4cc

158 M. Ginic-Markovic et al. / Polymer Degradation and Stability 69 (2000) 157±168

2.2. Accelerated aging

Accelerated aging was performed over di�erent timein Atlas Ci 3000 Xenon Weatherometer, according toSAE-J-1985 standard, providing automatic control oftemperature (38�C at dark and 89�C at light cycle),relative humidity- RH (95% RH in dark, 50% RH inlight cycle) and irradiation of 0.55W/m2 at 340 nm overdi�erent time. This test method is designed to simulateextreme environmental conditions encountered inside avehicle due to sunlight, heat and humidity for the purposeof predicting the performance of automotive components.

2.3. Dynamic mechanical analysis (DMA)

DMAwas carried out on the primed and coated EPDMrubber compounds, using a DMA 2980 (TA Instruments)operating in tension mode from ÿ100 to 100�C at 1 Hzfrequency and 0.2% strain amplitude, at programmedheating rate of 2�C/min. Liquid nitrogen was used toachieve sub-ambient conditions. Storage modulus and theglass transition temperature (Tg) from tan� were studied.

2.4. Scanning electron microscopy/energy dispersive X-ray analysis (EDAX)

As the appearance of the surface generally changesduring weathering, therefore, microscopic informationis very useful to examine the degree of degradation.Also such study is helpful to determine the mechanismof failure. Unexposed and exposed coated rubber sur-faces were examined by scanning electron microscopywith energy dispersive X-ray analysis to determine theelemental composition. Energy dispersive X-ray analysiswas conducted using a Cam Scan CS44 scanning elec-tron microscope operating at an acceleration voltage of20 kV with EDAX detector.

2.5. Compositional analysis by TGA

Thermogravimetric analysis (TGA) was carried outusing cured polyurethane coating, and vulcanized con-trol EPDM sample (S1cc) before and after exposure toUV irradiation, under non-isothermal conditions toanalyze their compositions. TGA analysis was con-ducted using a TGA 2950 thermal analyser (TA Instru-ments) connected to a Thermal Analyst 2200 Controllerusing conventional (heating rate) operation from roomtemperature to 600�C at di�erent heating rates (3±20�C/min). In all analyses, a nominal 20 mg sample was usedunder a nitrogen atmosphere at a ¯ow rate of 50 mlminÿ1. The temperature, at which the rate of mass loss(Tmax) is at maximum, was evaluated from the di�er-ential thermogravimetry curves (DTG). The decom-position kinetics were evaluated using software basedon the Flynn and Wall method.

3. Results and discussion

3.1. Morpholgy and viscoelastic properties

The most important features of photo-oxidation arethe changes, which result in either loss of initial physicalproperties or discoloration or lack of adhesion etcwhich eventually causes loss of product integrity for anyuse. Thus, physical methods allow one to investigate thechange in morphology of the composition after expo-sure. In order to determine the e�ect of photodegrada-tion on the morphology and viscoelastic properties ofcoated rubber compounds, DMA was carried on thesamples before and after exposure to UV. Table 2 com-pares the storage modulus at di�erent temperatures ofthe coated samples, bulk and surface modi®ed rubbercompounds, before and after UV exposure of 800 h. Anincrease in storage modulus is observed in all cases. Thisis due to hardening caused by photodegradation in thecontrol (S1cc), bulk modi®ed P1cc and M1cc samples inthe entire temperature range. However, T3cc does notshow the same trend in the region of ÿ25 and 0� C. Fig.1 shows the representative plot of the e�ect of UVexposure on the storage modulus for the P1cc (beforeand after 800 h of exposure). From the ®gure, the char-

Fig. 1. E�ect of temperature on storage modulus for sample P1cc.

Table 2

Comparison of storage modulus (E0) values at di�erent temperatures

before and after exposure to UV

Sample Exposure Storage modulus (MPa) at(h)

ÿ50�C ÿ25�C 0�C 25�C 50�C

S1cc 0 1215 103.2 48.9 24.8 17.5

800 1270 128.5 63.2 41.7 27.9

M1cc 0 1257 141.6 59.9 31.6 18.8

800 1159 178.1 78.1 51.5 34.6

P1cc 0 1089 70.9 27.4 15.4 11.8

800 1426 122.9 54.3 33.3 25.5

N2cc 0 1060 180.3 91.2 58.3 41.1

800 1082 178.6 90.6 63.9 50.9

T3cc 0 1228 154.2 73.9 29.9 18.8

800 1318 144.4 71.3 49.2 36.9

M. Ginic-Markovic et al. / Polymer Degradation and Stability 69 (2000) 157±168 159

acteristic sigmoidal variation of the elastic modulus withtemperature is observed for the sample. It passes fromthe glassy to the rubbery state through di�erent regionswith the decrease in the sti�ness of the samples. Afterexposure the more pronounced increase of the storagemodulus and, therefore, hardening e�ect are noticedfrom ÿ25�C onwards. This temperature range corre-sponds to the environmental conditions, which will beapplicable for a particular automotive component-win-dow seal section. Fig. 2 shows the plot of storage mod-ulus at three di�erent temperatures (ÿ25, 0 and 25�C)for S1cc and P1cc with time of UV exposure. Two distinctregions are clearly noticed on the plot. An initial steadyincrease in modulus up to 800 h is followed by a levelinge�ect. The observed early increase in storage modulusfor all the samples after exposure could either be due tooil/volatile loss from the composition or due to changein network structure. While TGA can monitor thevolatile loss e�ciently, the minor change in crosslinkdensity or hardness could be best understood from theelastic modulus, for which DMA is a powerful probe.The change in modulus in the early part thus could beaccounted for by the change in EPDM network struc-ture in the form of post curing, consequently an increasein crosslink density. In order to monitor the oil loss, theexposed and unexposed rubber and the coating werealso analyzed for composition using TGA.As UV exposure of di�erent samples could lead to

change in the level of interaction or morphology to dif-ferent extents, change in modulus on exposure was fol-lowed for di�erent samples. Fig. 3 shows the increase inmodulus (E0 at 25�C) with time of exposure. The bulkmodi®ed rubbers with maleated EPDMs (P1cc and M1cc)show highest increase in modulus of all samples over theentire exposure time. It must be pointed out that thecrosslink density of the unexposed samples was alsofound to be higher. However, N2cc shows least changes.The control sample (S1cc) and the surface modi®edcompound T3cc show intermediate behavior. It is note-worthy that S1cc and P1cc show marginal drop in mod-

ulus at 435 h, which may be due to destruction of somecross linkages to form mono or di sul®dic linkages. Fig.4 is a representative plot of the temperature dependenceof tan d with and without exposure for the sample P1cc.The distinct transition relaxation peaks are obviousfrom the graph. The low temperature transition Tg1

before exposure at ÿ36.8�C changed marginally to thevalue at ÿ36.7�C after exposure. Chailan et al. [16]reported that two important changes occur during agingas observed in mechanical damping such as decrease inthe height of tan � peak and shift of tan � peak tem-peratures to the higher value after an induction period.In the present case only change of peak height wasdetected indicating that UV irradiation a�ects the rub-ber compound marginally. It must be pointed out thatthe rubber compound is covered and protected by pri-mer/coating and, thereby, less a�ected.The sharp peak is followed by a broad transition at

5.3�C assigned to the transition relaxation of coating,the height and position of which change after exposure.The decreasing peak height signi®es that part of thecoating is depleted. The trend of tan d for all other

Fig. 2. Plot of the elastic modulus as a function of exposure time for

the control and bulk modi®ed samples.

Fig. 4. Temperature dependence of tan d for the sample P1cc.

Fig. 3. E0 at 25�C of various samples with the exposure time.

160 M. Ginic-Markovic et al. / Polymer Degradation and Stability 69 (2000) 157±168

samples remains the same for the exposed and non-exposed sample except T3cc in which case the secondtransition for the unexposed sample occurs at a highertemperature (Fig. 5). As surface modi®cation was per-formed with a ®nal batch unvulcanised sample, this mayhave caused limited availability of unsaturation on thesubsurface for curing but better interfacial bonding ofthe rubber compound with the coating. The surfacetreated/coated rubber (T3cc) thus shows an increased tand, which diminishes after exposure (Fig. 5).

3.2. E�ect of UV exposure on the surface morphology ofthe surface and bulk modi®ed rubbers

All the samples were exposed to UV for 112, 435, 800and 1200 h. The samples were withdrawn from the UVchamber at respective time interval. After 24 h storage,the surface morphology was examined using SEM. Figs.

6 and 7 show the surface features of control, coatedbulk and surface modi®ed EPDM rubber compounds atdi�erent magni®cation before and after exposure. FromFig. 6 the surfaces of those compounds appear roughbut no cracks are evident. The appearance of the crackis noticed in the control sample earlier (after 435 h) thanthat on any of the modi®ed samples. The surfaceappearance of all samples is completely di�erent afterexposure of 800 h to UV (Fig. 7). The cracks are morepronounced and deeper in S1cc and M1cc samples. Theleast changes on the surface are observed in sample P1cc.In terms of crack formation and their depth samples canbe put in the following order: S1cc>M1cc>T3cc>N2cc>P1cc, indicating the best performance of the coated bulkmodi®ed rubber with maleated EPDM (1%). Theweathering performance of coated bulk modi®ed com-pound (N2cc) reveals better performance than coatedsurface treated EPDM rubber compound (T3cc). Onlymarginal improvement of coated EPDM rubber com-pound was noticed when rubber was bulk modi®ed with0.5% maleated EPDM. No crack formation is observedon P1cc even after 1200 h. In order to con®rm whethersurface or subsurface events are responsible for damageand disintegration, the ®lm of coating was also exposedto UV for di�erent time periods and is shown in Fig. 8at low and high magni®cations. This ®lm shows crack-ing, pitting and cavitation all over the surface. As Figs.6 and 7 are at di�erent magni®cations (100 and 10 mm),therefore, the uncoated and coated rubber, before andafter 800 h exposure are presented in Fig. 9 at the samemagni®cation (100 mm). While the rubber surface isquite smooth, the coated rubber exhibits a rough sur-face. The cracks can be noticed after prolonged expo-sure on both coated and uncoated rubber surfaces (Fig.

Fig. 5. Temperature dependence of tan d for the sample T3cc.

Fig. 6. SEM images of unexposed coated EPDM rubber compounds.

M. Ginic-Markovic et al. / Polymer Degradation and Stability 69 (2000) 157±168 161

9). However, the UV is damaging the coating more thanthe rubber, the severity and the extent of which vary anddepend on many di�erent factors such as UV transmis-sion, scattering, the multiphase nature of the composite,the rate of di�usion of macroradicals etc.

3.3. Chemical changes on the surface by SEM-EDAX

The chemical nature of degradation is often probed toobtain sensitive indication of the material's change.Elemental mapping of the above mentioned compoundswas thus performed with EDAX to get insight into thechemical changes on the surface and sub surface. Figs.

10 and 11 show the EDAX of the S1cc before and S1ccand P1cc after exposure to UV irradiation. Table 3represents the results of elemental mapping and theirconcentration. From the table the di�erence in the ele-mental structure of the coated bulk modi®ed and sur-face modi®ed samples before and after exposure isevident. The elemental level was determined by ratioingthe particular peak to reference (C peak). An increasingtrend of oxygen level is found in all samples after expo-sure to UV due to the photo-oxidation. Oxygenabsorption is normally followed from carbonyl andhydroxyl groups using infra red absorption spectro-scopy (IR). However, in the present case, the amount of

Fig. 7. SEM images of UV exposed coated EPDM rubber compounds.

Fig. 8. SEM images of coating after exposure.

162 M. Ginic-Markovic et al. / Polymer Degradation and Stability 69 (2000) 157±168

carbon black present in the sample does not allow us touse IR. It is interesting to note that the amount of oxygenin all the samples before exposure varies marginallybetween 0.8 and 1.16. However, after exposure the oxygen

level is increased by a factor of 2±3 in di�erent samplesexcept P1cc in which case the change is negligible.The amount of silicon present in di�erent samples due

to di�erent components in the coating is almost similar,

Fig. 9. SEM images of control and coated control rubber sample before and after 800 h exposure.

Table 3

Elemental composition from SEM-EDAX

Element detected S1ccbefore

S1ccafter

M1cc

before

M1cc

after

P1cc

before

P1cc

after

N2cc

before

N2cc

after

T3cc

before

T3cc

after

Oxygen (O) 0.8 2.62 1.16 2.61 1.0 1.29 1.19 2.27 1.15 2.5

Silicon (Si) 3.2 0.91 3.37 0.44 3.2 3.77 3.05 5.82 3.2 0.5

Sulfur (S) 0.25 3.05 0.21 2.69 0.25 0.29 0.24 2.0 0.3 3.1

Zinc (Zn), Zn Ka1 ÿ 1.19 ÿ 1.26 ÿ 0.12 ÿ 0.36 ÿ 1.4

Zinc (Zn), Zn Kb1 ÿ 0.19 ÿ 0.17 ÿ ÿ ÿ ÿ ÿ 0.2

Zinc (Zn), Zn La1 ÿ 1.95 ÿ 2.17 ÿ 0.12 ÿ 0.73 ÿ 2.5

Chlorine (Cl), Cl Ka1 ÿ ÿ ÿ ÿ- ÿ ÿ ÿ ÿ 0.2 0.4

Fluorine (F), F Ka1 0.55 ÿ 0.74 ÿ 0.38 ÿ 0.48 ÿ 0.7 ÿ

M. Ginic-Markovic et al. / Polymer Degradation and Stability 69 (2000) 157±168 163

however, it was found to decrease over time of exposurein all samples except P1cc and N2cc. The change in Si inP1cc remains almost invariant whereas N2cc showsincreasing trend. The EDAX analysis shows the pre-sence of silica (O/Si=2) in these samples in additionto silicon. The excessive amount of Si in N2cc is pre-sumed to be due to local accumulation and enrichmentof Si in that particular region. Also the SEM imagesof the samples except P1cc and N2cc show lot of sur-face cracking indicating photo degradation of the

coating in those samples. The SEM micrographs ofthose two samples rather show stabilized silica particlesin the coating, which shielded the subsurface from thee�ect of UV radiation. Generally, the UV rays lead tobrittleness and as a result surface cracks appear. Inextreme case, the material can disintegrate. Such surfacedamage caused by UV light is physico-chemical in nat-ure. The resistance to the photoaging of P1cc and N2cc isalso clearly noticed from the SEM images with marginalcracking in the coating layer. In contrast an increasing

Fig. 10. EDAX of S1cc before exposure.

Fig. 11. EDAX of S1cc and P1cc after 800 h of exposure.

164 M. Ginic-Markovic et al. / Polymer Degradation and Stability 69 (2000) 157±168

trend of sulfur level was observed in other samplesexcept P1cc. The level of sulfur in the sample P1cc

remains unchanged due to strong adhesion to the coat-ing. Migration of Zn [17] to the surface represented byK�1 and L�1 caused by photodegradation is obvious inall samples. Signal K�1 is noticed only in S1cc, M1cc andT3cc. It is interesting to note that the level of Zn migra-ted to the surface due to the UV weathering is negligiblein sample P1cc compared to the others. The observationof certain elements like sulfur and zinc in some of thestudied samples strongly indicates that the coating layersof those samples are completely depleted thus exposingthe rubber surface. The degradation of coating layer isnot at all noticed on aging (Fig. 7) in the case of P1cc

thus restricting the migration of Zn to the surface. Fromthe ®gure not many cracks are also observed in sampleN2cc, which does not reveal a signi®cant level of Zn onthe surface. The 2 min surface chlorinated sample showsincreasing level of Cl on the surface after aging indicat-ing migration of Cl to the surface through degradedcoating layer. Disappearance of ¯uorine from the coat-ing was noticed after aging in all samples. The DMA,

SEM and EDAX results indicate substantial amount ofphotodegradation of the coating while no signi®cante�ect of aging on the rubber compound is observed.

3.4. Physico-chemical changes of the compounds duringUV exposure

Although the highest energy wavelength reaching theatmosphere is 290 nm, yet the chemical changes in thepolymeric product induced by such radiation can pro-ceed either by molecular or free-radical mechanism. Asigni®cant proportion of radiation could be absorbed bya coated ®lm on the rubber compound because of thepresence of the aromatic substituent (the chromophore)present in the PU coating. Any surface chemical reac-tion will be thus preceded by physical processes such as(1) absorption of radiation by the surface yielding theexcited species, and dissipation of energy by emitting alow energy photon or by converting to vibrationalenergy/heat and (2) transfer of this radiation energy toanother species and (3) ®nally occurrence of chemicalreaction/degradation from the excited states. The sim-pli®ed photochemical reactions of polyurethane thatoccur during exposure ( reaction 1 or 2 or 3) could berepresented as shown in Scheme 1. It consists of a chainprocess involving a large number of chemical reactionsand species as a consequence of absorption of photonwhich propagate spatially with the formation andtranslation of small radicals. Gardette et al. [18]observed that aliphatic poly(ester-urethane)s, on expo-sure to long wavelength radiation, show a selective lossin the intensity of methylene groups as evident from IRstudy at 2940 and 2860 cmÿ1 indicating oxidation ofcarbon atoms in the alpha position to the ÿNH group.Although several studies have been conducted so far tostudy the exact mechanism of photodecomposition ofaromatic diisocyanate based PU, yet the de®nitemechanism is not clear. Based on an analogy with thestudy made by Beachell and Chang [19] on photo-degradation of aryl monocarbamate model compound,it has been suggested [20,21] that TDI (toluene di iso-cyanate) and MDI (methylene bis isocyanate) basedisocyanates photodecompose to some extent by aphoto-Fries rearrangement. Further photolysis of theprimary products can yield a colored azo compoundwhich is responsible for discoloration in light colormaterial. The formation of the ortho photo Fries rear-rangement product was detected by Hoyle et al. [21] by¯uorescence spectroscopy, and is also shown to bedependent on the degree of hydrogen bonding in seg-mented PU. Other quinonoid type products have beenproposed by other workers [22]. However, according toGardette and Lemaire [22±24] both types of productsmay form on photolysis of aromatic diisocyanate basedpolyurethanes depending on the wavelength of excita-tion. Because the PU system studied here contains an

Scheme 1. Simpli®ed reactions for photodegradation of PU coating

on the coated rubber surface.

M. Ginic-Markovic et al. / Polymer Degradation and Stability 69 (2000) 157±168 165

aromatic component, it can absorb signi®cant propor-tion of UV at 340nm. The photo-oxidation can occureither through urethane linkages (as shown in reactions1 and 2 of Scheme 1) or can form a diimide product(reaction 3 of Scheme 1) [25]. However, the rate ofphoto-oxidation and the termination reaction are con-trolled by the rate of di�usion and restricted motion ofthe macroradicals in the solid phase and also the UVtransmission characteristics of the visco-elastic compo-site. Thus oxidation in the interior rubber matrix islimited due to the rapid reaction near the surface. Itmust be pointed out here that the experimental set upfor xenon arc exposure includes elevated temperature aswell as UV radiation.The stabilizer present in the EPDM composition may

di�use to the surface under such conditions to allowreplenishment of the amount lost during reaction. Alsoduring the experiment, as a period of darkness followsas at night in natural weathering, partial recovery of thematrix oxygen takes place. Thus variation of degrada-tion can occur with the depth of the composite and thelevel of adhesion of the coating to the substrate. Thehigher the adhesion, the less the extent of UV penetrat-ing through the interface. This is due to the fact that asthe photo-oxidation takes place from the surface, theenergy is consumed and the intensity of the UV decrea-ses on passage through the strongly adhered layers tothe rubber compound. Through proper compounddesign, the energy can be absorbed by the stabilizer.However, the depth of UV penetration is controlled bythe extent of absorption and scattering. Often this ismore pronounced on the surface. Hulme et al. [26] cal-culated and modeled the molecular weight degradationas a function of depth. White reported that the scatter-ing is strong in multiphase materials and the UV inten-sity and molecular degradation in a glass-®brereinforced polypropylene is con®ned within a depth of 1mm [27]. In the present case, while the control sample isa single phase EPDM compound, the P1cc sample (ablend of EPDM with maleated EPDM) being multi-

phase in nature and strongly adhered to the coatingdoes not allow enough UV penetration into the inter-face, which diminishes progressively at locations furtherinto the interior. It is apparent that the degradationproducts have no antagonistic e�ect on the elastomercomposition.

3.5. E�ect of UV exposure on the composition of therubber and the coating

In general the physical and mechanical properties ofthe window seal section are in¯uenced by environmentalconditions such as heat, light and humidity. TGA canbe used to study the compositional changes of the coat-ing and the EPDM rubber compound on exposure.Thus for further investigation on the e�ect of UVexposure on the composition of the control rubber (S1c)and the coating ®lm TGA was undertaken. Fig. 12shows the DTG curves of the coating ®lm before andafter exposure. The Tmax corresponding to the max-imum weight loss is found from the derivative thermo-gravimetry curve (DTG). Degradation behavior ofexposed and unexposed S1c, and the coating, is alsopresented in Table 4. A shift in the decomposition tem-perature (to a lower value) is observed at a heating rateof 10�C/min after UV exposure. The shift is more pro-nounced with the coating than with the rubber com-pound indicating that the coating ®lm being on thesurface is a�ected by UV. This ®nding is in good agree-ment with the conclusion of faster aging of coating thanrubber, obtained with previous experimental techniques.The change in decomposition of the volatile componente.g. oil is more than the elastomer component (EPDMrubber) in the compound S1c. The activation energy ofdecomposition of S1c was calculated in the range of380±500�C (where the elastomer component primarilydecomposed), using the Flynn and Wall method [28].Fig. 13 illustrates a semilogarithmic plot of heating ratevs reciprocal temperature for selected levels of conver-sions. The slopes of the lines represent the activation

Fig. 12. DTG curves of coating ®lm before and after exposure.

Table 4

Compositional analysis and degradation behavior of unexposed and

exposed control rubber S1c and the PU coating

Sample Exposure

(h)

T1max

(�C)T2max

(�C)T3max

(�C)T4max

(�C)Ea (kJ/mol)

Oil Polymer

S1c

weight 0 317.2 460.8 ÿ ÿ 93 228

27% 31.4%

weight 1200 319.5 464.7 ÿ ÿ 84 22019.5% 33.6%

Coating

weight 0 333.1 432.7 545.2 ÿ 169

34.1% 44.6% 12.1%weight 1200 301.4 393.8 460.1 540.9 ÿ 128

22.7% 33.5% 14.7% 15.3%

166 M. Ginic-Markovic et al. / Polymer Degradation and Stability 69 (2000) 157±168

energy of decomposition for weight losses from 20 to90% using the method of Flynn and Wall [28] as follows:

E � ÿR=b d log �=d 1=T� �� � �1�

where E � activation energy (J/mol), R �gas constant,T �temperature at constant conversion, � �heating rate(�C/min) and b � a constant and taken as 0.457. FromTable 4, not a signi®cant drop in activation energy ofdecomposition at 90% conversion (from 228 to 220 kJ/mol) [29] was detected for the rubber compound. In ourearlier communication, we noticed that the degradationof polyurethane occurs in two stages [3]. This occursdue to two consecutive reactions occurring during thecomposite decomposition of PU. After exposure, thedegradation of the main components is shifted to301.4�C (cf 333.1�C) and 393.8�C (cf 432.7�C) with anew peak at 460.1� C. The contribution of these com-ponents (before and after exposure) towards the totalwt% varies within 11 and 12%. Table 4 also shows theTmax and the wt% of di�erent components present inthe control rubber. The Tmax of the components do notvary much but the% oil is reduced by 7.5% after expo-sure with a minor change in the rubber component(2.2%) Such evaporation of oil indicates hardening ofthe sample, ultimately a�ecting the viscoelastic proper-ties by increasing the storage modulus at higher tem-peratures (Fig. 1).It is clear from Fig. 12 that the decomposition of PU

after exposure also does not involve independent reac-tions but rather consecutive steps. Fig. 14 shows theplot of activation energy at di�erent percent decom-position of the exposed and unexposed coating samples.The activation energy increases monotonically as dis-

ruption of chemical bonds takes place in the unexposedcoating sample. In contrast, in the exposed sample theactivation energy decreases up to 60% decompositionincreasing slightly after that. At 5% conversion, theactivation energy of photoaged sample is signi®cantlyhigher than the unexposed one due to the fact that cer-tain percentage of the coating was already degradedduring photoaging and the remaining degradation startsfrom there. It is noteworthy that after exposure theactivation energy of degradation of PU coating isreduced by 41 kJ/mol. This observation again indicatesthat the coating layer on the surface is more a�ected byUV than the rubber.

4. Conclusions

The UV rays are damaging to the coating more thanthe rubber. The extent of UV degradation depends onmany di�erent factors such as UV transmission andscattering ability, the multiphase nature of the compo-site, the rate of di�usion of macroradicals through thematrix etc.The degradation thus varies with di�erent composites,

with the depth of UV penetration through the interfaceand the level of adhesion of the coating to the substrate.The bulk modi®ed rubber compound (P1cc) showed thebest UV weathering characteristics.

Acknowledgements

The authors are thankful to Australian ResearchCouncil (ARC) for support of this work through thecollaborative grant scheme. Thanks also go to BSTGfor providing access to the Atlas Ci 3000 XenonWeatherometer and Uniroyal for the supply of EPDMsamples.

Fig. 13. Plot of logarithmic heating rate vs. temperature for PU coating.

Fig. 14. Activation energy vs. % of decomposition for PU coating

before and after exposure.

M. Ginic-Markovic et al. / Polymer Degradation and Stability 69 (2000) 157±168 167

References

[1] Baba M, Gardette J-L, Lacoste J. Polym Degrad Stab

1999;63:121.

[2] Saha Deuri A, Bhowmick AK. J Appl Polym Sci 1987;34:2205.

[3] Markovic MG, Choudhury NR, Matisons JG, Williams DRG. J

Therm Anal and Calorim 2000;59:409.

[4] Lemaire J, Gardette JL, Lacoste J, Delprat P, Vaillant D. In

American Chemical Society Clough R, Billingham, NC, Gillen,

T. editors Polymer durability, Washington: 1996.

[5] Govorcin-Bajsic E, Rek V, Sendijarevic A, Sendijarevic V, Frish

KC. Polym Degrad Stab 1996;52:223.

[6] Guzzo M, DePaoli AMJ. Polym Degrad Stab 1992;36:169.

[7] Decker C, Moussa K, Bendaikha TJ. Polym Sci Part A Polym

Chem 1991;29:739.

[8] Sommer A, Zirngiebl E, Kahl L, Schonfelder M. Progress in

Organic Coatings 1991;19:79.

[9] Decker C, Zahouily KJ. Polym Sci Part A Polym Chem

1998;36:2571.

[10] Dudler V, Bolle T, Rutz GJ. Polym Degrad Stab 1998;60:351.

[11] Martin JW, Chin JW, Emree E, Kraft KMJ. Polym Degrad Stab

1999;63:297.

[12] Deuri AS, Bhowmick AKJ. Appl Polym Sci 1987;34:2205.

[13] Gamlin C, Markovic MG, Dutta NK, Choudhury NR, Matisons

JG. J Therm Anal and Calorim 2000;59:319.

[14] Old®eld D, Symes TEFJ. Adhesion 1983;16:77.

[15] Markovic MG, Choudhury NR, Dimopoulos M, Matisons JG,

Dutta N, Bhattacharya AK. Polym Eng and Sci (accepted).

[16] Chailan JF, Boiteux G, Chauchard J, Pinel B, Seytre GJ. Polym

Degrad Stab 1995;47:397.

[17] Dimopoulos M, Choudhury NR, Markovic MG, Matisons J,

Williams DRG. J Adhesion Sci Technol 1998;12:1377.

[18] Wilhelm C, Gardette J-L. Polymer 1997;38(16):4019.

[19] Beachell HC, Chang IL. J Polym Sci 1972;Pt. A-1:10503.

[20] Hoyle CE, KimK-J. J Polym Sci 1986;Pt A: Polym. Chem. 24:1879.

[21] Hoyle CE, Kim K-J, No YG, Nelson GL. J Appl Polym Sci

1987;34:763.

[22] Gardette JL, Lemaire. J Makromol Chem 1981;182:2723.

[23] Gardette JL, Lemaire. J Makromol Chem 1982;183:2415.

[24] Gardette JL, Lemaire. J Polym Degrad Stab 1984;6:135.

[25] Schollenberger CS, Dinsbergs K. Soc Plast Eng Trans 1961;1:31.

[26] HulmeAJ,Mills NJ. Plast Rubb Compos Proc Applic 1994;22:285.

[27] White JR, Turnbull J. Mater Sci 1994;29:584.

[28] Flynn JH, Wall LA. Polym Lett 1966;4:323.

[29] Kole S, Chaki TK, Bhowmick AK, Tripathy DK. Polym Degrad

Stab 1993;41:109.

168 M. Ginic-Markovic et al. / Polymer Degradation and Stability 69 (2000) 157±168