Thermal and photochemical ageing of epoxy resin – Influence of curing agents

Polymer Degradation and Stability 91 (2006) 1247e1255www.elsevier.com/locate/polydegstab

Thermal and photochemical ageing of epoxy resin e Influenceof curing agents

F. Delor-Jestin*, D. Drouin, P.-Y. Cheval, J. Lacoste

Ecole Nationale Superieure de Chimie de Clermont-Ferrand, Laboratoire de Photochimie Moleculaire et Macromoleculaire e UMRCNRS 6505, Ensemble Universitaire des Cezeaux e 63177 Aubiere Cedex, France

Received 4 July 2005; received in revised form 12 September 2005; accepted 13 September 2005

Available online 2 November 2005

Abstract

The thermal and photochemical ageing of epoxy resin was studied using photoacoustic-FTIR spectroscopy. This technique was satisfactoryfor both unfilled resin and glass fibre filled epoxy composite. The influence of the curing agent (anhydride or amine) was significant for ageing.The durability of anhydrideeepoxy system was the best for both thermal and photoageing.� 2005 Elsevier Ltd. All rights reserved.

Keywords: Epoxy resin; Curing agents; Ageing; Photoacoustic-FTIR; Thermal analysis

1. Introduction

Epoxy resins are widely used in composite materials andfor different activities such as electrical engineering and aero-nautics. When considering composites for long-term applica-tions, it is necessary to know how the materials will behaveduring the intended service life. A lot of recent papers are fo-cussed on the characterization of epoxy resins after ageing.Several techniques have been used to estimate and to explainthe changes of these polymers. The main studies are summa-rized below.

A study of phenoxy resin photoageing (DGEBA e digly-cidyl ether of bisphenol A-without curing agent) has beenreported by Rivaton et al. [1,2]. They identified the photoprod-ucts by UV and FTIR spectrophotometries, then by chemicalderivatization reactions. They also proposed some mechanismof photoproducts formation after long and short wavelengthirradiation.

Other papers relate to the case of various cured epoxy ma-terials based on DGEBA. One study proposed by Bellenger

* Corresponding author. Tel.: C33 4 73 40 71 53; fax: C33 4 73 40 70 95.

E-mail address: [email protected] (F. Delor-Jestin).

0141-3910/$ - see front matter � 2005 Elsevier Ltd. All rights reserved.

doi:10.1016/j.polymdegradstab.2005.09.009

and Verdu [3] deals with the influence of curing agents onphoto-oxidation of crosslinked epoxy resins. Three systemsbased on diamines (diaminodiphenyl methane, diaminodi-phenyl ether or diaminodiphenyl sulfone) were compared byUV and FTIR analysis. The structure variation with the meth-ylene, ether or sulfone bridges has a noticeable influence onphotoageing. The comparison shows that the photoinitiatingspecies essentially derive from the phenoxy part, whereasthe propagation essentially depends on amine concentrationand electron density on the nitrogen atom. Other studies[4e8] about photo-oxidation give results on anhydrideeepoxysystems. An FTIR study has been reported for two epoxy sys-tems (DGEBAeHHPA, hexahydrophthalic anhydride andDGEBAeMTHPA, methyl tetrahydrophthalic anhydride) byOllier-Dureault and Gosse [4]. This paper reports modifica-tions of the chemical structure and the influence of the addi-tion of a diacidediester type flexibilizer. Monney et al.[5e7] have published a photochemical study of DGEBAeMTHPA system by using different analytical techniques.ATReFTIR study [5] gives the molecular changes and con-firms mechanisms proposed in other literature. Electronbeam X-ray microanalysis [6] allows characterization ofthe changes of C/O ratio and it shows the relative changesof the different oxidation levels of the organic matrix on the

1248 F. Delor-Jestin et al. / Polymer Degradation and Stability 91 (2006) 1247e1255

surface. Another study with DGEBAeMTHPA and DGEBAeIPDA (isophorone diamine) makes a comparison of the matrixdurability by ablation measurement [8]. The photochemical re-moval of products during ageing has been quantified usinga two-dimensional profile measurement technique.

The thermal ageing of epoxy resin has also been investigated.Most of papers relate the case of cured DGEBA polymers[9e17]. A review was published in 1983 on thermo-oxida-tion of DGEBAeamine systems [9]. Some mechanisms ofoxidation are given which depend on the temperature of age-ing. The origin of the colouring after thermal degradation isthen explained for the DGEBAeTETA (triethyl tetramine)system. A kinetic study on DGEBAeMTHPA system [10]allowed the distribution of oxidation products to be assessedby IR and the rate of oxygen consumption by modellingtechnique. Another paper [11] presents a thermogravimetricstudy of DGEBA cured with different amines (aniline, phe-nylene diamine, etc.) in the absence of oxygen. Some corre-lation between thermal stability and cure-agent structures areexplained. The thermal decomposition under nitrogen ofDGEBA/4,4-methylene-dianiline (MDA) with rubber (buta-dieneeacrylonitrile) modified MDA occurred in one stage.Lee et al. [12] showed that the thermal stability of such asystem increased when the content of rubber was higher(because of the four rings with high thermal resistance dueto the resonance structure). The thermal stability and the cur-ing behaviour (at 150 and 220 �C) were also investigated forthree systems with DGEBA and multifunctional aromaticamine used as flame retardant [13]. These special curingagents contain phosphine oxide. We can also note a kineticstudy for DGEBAeDDM (diamine diphenyl methane) [14].

DGEBA

CHCH2 CH2

CH3

CH3

CH2 CH2

CH3

CH3

CH2CH2

O

CH O C O

OH

O

O

CHOC[ ]n

Amine (DETA) H2N-CH2-CH2-NH-CH2-CH2-NH2

Anhydride (MNA)

OH3C

O

OCH2

Damian et al. give the order of the reaction as a functionof the oxygen pressure and the limiting thickness of homoge-neous oxidation. Another approach on epoxy ageing is theuse of elevated pressure to accelerate thermo-oxidative deg-radation in composites (carbon fibre and epoxy resin) [15].We can finally emphasize a study about a specific epoxy ma-trix based on tetraglycidyl-4,4#-diaminodiphenylmethane(TGDDM) with excellent thermal stability [16,17]. This resincured with diaminodiphenyl sulfone can be mixed with

bismaleimide (BMI) to give an interpenetrating polymernetwork. The TGA and DMA analyses [16] showed anincrease in the thermal stability when BMI was added. Atwo-dimensional FTIR spectroscopy study gave the detailsabout the reaction mechanism of thermal degradation [17].The method provided information about intra- and inter-molecular interactions by selective correlations of peaks.

The studies performed to date indicate the complexity of thepossible degradation routes during epoxy resin ageing. Thepurpose of our study is to analyse structural modifications afterageing on specific crosslinked epoxy system by photoacoustic-FTIR spectroscopy, then to give oxidation kinetics. We com-pare the thermal and photoageing of two crosslinked DGEBA(first formulation with amine, second one with anhydride).We would also like to analyse the influence of curing agentand the differences observed for both, because there has beenno similar previous approach. Thermal analysis in the presenceof oxygen with TGAeFTIR was an additional analytical toolwhich was used to improve the understanding of epoxydegradation.

2. Experimental

2.1. Materials

The prepolymer used in this study was an epoxy systemDGEBA (diglycidyl ether of bisphenol A), DER 331, productof Dow Plastics. Typical properties are: epoxide equivalentweight 186 g eq�1, percentage epoxide 23%. Two curingagents were used: diethylene triamine (Ciba) and methyl nadicanhydride (Dow).

The reagents were used as received, without further purifica-tion. The crosslinking reaction with anhydride was initiated withan imidazole-type accelerator (1-methyl imidazole suppliedby Dow). After DSC (Differential Scanning Calorimetry eMettler Toledo DSC 30) measurements, we have chosen thecuring time and the temperature for each formulation. Optimalcuring temperature is obtained when conversion rate is maxi-mum, i.e. we cannot obtain a higher peak height on DSC thermo-gram. Formulation 1 was done with 40 g of amine for 60 g of

1249F. Delor-Jestin et al. / Polymer Degradation and Stability 91 (2006) 1247e1255

epoxy prepolymer. Formulation 2 was done with 90 g of anhy-dride for 100 g of epoxy. The curing cycle was 60 min at100 �C then samples were cooled in the open air. Glass fibrefilled epoxyeanhydride system prepared by pultrusion tech-nique was also studied.

2.2. Accelerated ageing

Photoageing was carried out in an SEPAP 12-24 device. Thisirradiation system has been described previously [18]. It ischaracterized by the source, medium pressure Hg lamps filteredwith borosilicate envelope (l O 300 nm) and by careful controlof the temperature with a thermocouple in close contact withone of the samples. Samples are rotated at a constant distance(20 cm) from the sources. The epoxy samples were irradiatedat 60 �C in an SEPAP chamber equipped with four lamps.Epoxy samples were prepared in aluminium photoacoustic cru-cibles then placed in front of the lamps in the SEPAP device.

For thermo-oxidation, aluminium photoacoustic crucibleswith epoxy system were exposed in an aerated oven at100 �C (Memmert UM100).

2.3. Analytical tools

Photoacoustic-FTIR has been used in particular for thestudy of cured epoxy resin degradation. Our materials are ther-mosetting. We also studied glass fibre filled epoxy and wefound that this particular accessory for spectroscopy was thebest for each kind of samples. Micro-ATR did not give goodresults. Photoacoustic (PAS) was the only technique whichwe can use for filled epoxy materials. The PAS accessoryfrom MTEC is coupled with a Nicolet 860 FTIR spectropho-tometer. The surface of the sample is lit by an intense IRbeam. The rise of temperature at the sample surface provokesan increase of pressure in the PAS cell which is full of helium.The pressure wave spreads inside the cell and reaches theacoustic membrane of the detector which transforms the

received signal [19]. The analysed thickness (around 15e20 mm) is function of the frequency of absorbed IR beamand thermal properties of materials. TGAeFTIR used a cou-pled system Mettler Toledo TGA/SDTA851 e Nicolet Nexus.The gazes coming from the TGA are transferred by a heatedtube for analysis in the gas cell of the IR spectrometer. Thetemperature program in the TGA part was an increase of tem-perature by 10 �C min�1 from 20 to 800 �C, under oxygen(40 ml min�1).

3. Results and discussion

3.1. Analysis of initial materials

Firstly we have identified the initial spectra of the twocrosslinked systems obtained by PAS-FTIR spectroscopy.We compared the spectra of the two formulations with onespectrum of phenoxy resin (uncrosslinked DGEBA), as illus-trated in Fig. 1. We have given in Table 1 the peak frequenciesand tentative assignments after curing protocol on DGEBA. Itgives information about the difference between the two cross-linking networks. The presence of curing agent is clearly iden-tified. Amine links are characterized specially at 3320 and1650 cm�1. The bands at 1860 and 1740 cm�1 are attributedto anhydride carbonyl groups.

3.2. Analysis of photo-oxidation

The IR analysis as a function of the irradiation time showsa rapid spectral evolution of the two systems under acceleratedUV-ageing. It can be observed from the initial few hours ofexposure as illustrated in Figs. 2A, B and 3A, B. The modifi-cations are essentially in the hydroxyl and carbonyl absorptionareas.

For the epoxyeamine system, the appearance of severalcarbonyl groups absorbing in the range of 1670e1800 cm�1

is given in Fig. 2A. The maxima are obtained at 1760, 1740

A

B

2960

2850

3320

3050

3500

3050 28

70

1880

1780

1600

1506

1650

1610

1510

1460

1250

1190 10

30

830

580

830

1040

907

580

1740 12

40

1860

2930

-40

-30

-20

-10

0

10

20

30

40

50

60

70

80

5001000150020002500 300035004000Wavenumber (cm-1)

PA

Fig. 1. FTIR spectra of initial materials; A e DGEBAeamine after curing; B e DGEBAeanhydride after curing.

1250 F. Delor-Jestin et al. / Polymer Degradation and Stability 91 (2006) 1247e1255

Table 1

Main absorption (wavenumbers, cm�1) observed for raw materials (uncrosslinked DGEBA, amine, anhydride) and the two formulations DGBAeamine and

DGEBAeanhydride, then assignments of IR bands

DGEBA Amine Anhydride DGEBAeamide (Fig. 1A) DGEBAeanhydride (Fig. 1B) Tentative assignment

3500 3500 3500 n OeH

3300 3320 n NeH

3050 3050 3050 n phenyl-H

2960e2930e2870 2920 2930e2850 2690e2930e2870 n CeH and n CH2

1860e1780 1780e1740 n C]O

1650 1650 n NeH

1610e1580e1510 1610e1580e1510 1600e1580e1510 Aromatic ring

1450 1460 1460 d CH2

1230 1230 n CeOeC

1250e1190 1250e1190 n phenyleCeph, n CeOeC

907 907 d C]C

830 830 830 d phenyl-H

n Z stretching vibration; d Z in-plane deformation.

5101520253035404550556065707580859095

15501600165017001750180018501900

50 hrs23 hrs7 hrs3 hrsT=0

1740

Wavenumber (cm-1)

A100

1670

PA

68

10121416182022242628303234363840

PA

15501600165017001750180018501900

50 hrs26 hrs7 hrs3 hrsT=0

B

Wavenumber (cm-1)

Fig. 2. FTIR-PAS spectra upon photo-oxidation; A e carbonyl absorption area for epoxyeamine. B e carbonyl absorption area for epoxyeanhydride.

1251F. Delor-Jestin et al. / Polymer Degradation and Stability 91 (2006) 1247e1255

293032

70

2820

1670

1740

1670

150016

10

1740

2811-60

-55-50-45-40-35-30-25-20-15-10-505

1015202530

PA

150020002500300035004000

A

Photochemical ageing

Thermal ageing

Wavenumber (cm-1)

-80

-60

-40

-20

0

20

40

60

80

100

120

140

160

PA

150020002500300035004000

Photochemical ageing

B

Thermal ageing

Wavenumber (cm-1)

3520

1780

1780

1760

1710

1690

1740

1510

3450

Fig. 3. Subtraction result: aged material� initial material; A e photochemical and thermal ageings for epoxyeamine. B e photochemical and thermal ageings for

epoxyeanhydride.

and 1670 cm�1. We can simultaneously observe the decreaseof the bands at 1610 and 1510 cm�1, which are characteristicof aromatics. We can notice a small increase at 3250 cm�1 onthe subtraction result (Fig. 3A e first spectrum).

The evolution of the DGEBAeanhydride system is quitedifferent and less developed than on the DGEBAeaminesystem. We can observe the appearance of new bands suchas hydroxyl group absorption at 3450 cm�1 (Fig. 3B e firstspectrum), carbonyl absorption at 1780, 1760 and1710 cm�1 (Fig. 3A e first spectrum). The main band for an-hydride at 1740 cm�1 decreased. The main decrease is thennoticed at 1570 cm�1 (Fig. 3B e first spectrum) due to aro-matic C]C bands. The disappearance or modification of nu-merous absorption bands, associated with the strong increaseof the hydroxyl and carbonyl bands or the broadening of thecarbonyl band in formulation 2, is evidence of an important

phenomenon in both systems. The results show that UV irradi-ation leads to a specific attack upon the aromatic functions(disappearance of aromatic C]C and CeH bonds).

It is well known that polymer degradation in the presence ofoxygen leads to a complex mixture of oxidation products, suchas alcohols, hydroperoxides, lactones, esters, carboxylic acidsand ketones.. The hydroxyl and carbonyl bands are oftenvery broad and result from the overlapping of bands that arecharacteristic of numerous oxidation products. The results ob-tained for two formulations are very different so we try to un-derstand the influence of the curing agent on photo-oxidation.

Numerous oxidation sites are possible because of the struc-ture of crosslinked DGEBA systems. We compare our FTIRresults on DGEBAeDETA with phenoxy resin study (uncros-slinked epoxy [1,2]) and other works on DGEBAeamine sys-tems [3]. The main evidences are as follows: the general

1252 F. Delor-Jestin et al. / Polymer Degradation and Stability 91 (2006) 1247e1255

change of DGEBAeamine resembles the ageing of phenoxyresin. The change in the carbonyl region seems the samewith main maxima at 1740 and 1760 cm�1. Another maximum(shoulder) at 1670 cm�1 is different and is due to amidegroups absorption. Hence processes involved in photo-oxida-tion (l O 300 nm) of DGEBA (cured or uncured) have com-mon routes. After phototransformation of chromophoricimpurities, the resulting radical species initiate photoageingby abstraction of hydrogen from the polymer backbone. Thisphotoinduced oxidation leads to the formation of macroradi-cals which can be oxidized to hydroperoxide groups. Due tothe chemical structure of DGEBA, three possibilities (Ha, Hb

and Hc) are noticed as the origin of oxygen abstraction [2].

CH C CH2 O C

CH3

CH2

Hb

OH

Hc

Ha

O

The photoreactivity of aliphatic polyethers and bisphenol-Apolycarbonate has been previously studied. On the basis ofthese different works it results that the photo-oxidationproceeds through a primary abstraction of Ha, labile secondaryhydrogen. The new radical reacts with oxygen and leads tohydroperoxide. The thermal and photochemical decompositionof hydroperoxides leads to alkoxy radicals. After a b-scissionwe can have phenyl formate end groups (identified at1740 cm�1) and another macroradical can also form. Formatesare the most important decomposition products of hydroperox-ides formed through chain breaking. A cage reaction of alkoxyradical may occur leading to phenyl alkylate (1760 cm�1).

O C

O

O

H

1740 cm-1

O C CH

OH

CH2 1760 cm-1

The main photoproducts for DGEBAeamine are carbonylsand amides. The oxidation of amine links is then investigated.The evolution of DGEBAeamine proceeds as quickly as theevolution of uncrosslinked DGEBA resin. At the same timewe get the degradation of DGEBA structure and we can alsoobserve the modifications of amine links, attested by thechanges observed on FTIR spectra at 1670 and 1540 cm�1.

From the literature [3] we know that oxidation yieldsdepend on amine concentrations. The comparison of photo-oxidation rates for different curing agents such as aliphatic(DETA), cycloaliphatic (isophorone diamine) or heterocyclic(amino-ethyl piperazine) amines allowed us to find that amideyields are directly related to the initial a-amino methyleneconcentration. With dianiline-type amines (diaminodiphenylmethane e DDM, ether e DDE or sulfone e DDS), the bridgehas a noticeable effect on photoxidation. The stability wasfound in decreasing order DDS O DDM O DDE. The amideformation depends finally on initial amine concentration andnitrogen atom electron density (which is higher for DGEBAeDDE). The initial rate of amide growth is also higher for DDEsystem. The photosensitivity of phenyl amine is thereforeessentially due to chromophores arising from the phenoxypart, whereas the strongly absorbing chromophores arisingfrom the diamine part are considerably less efficient in photo-initiation. With the new results on DGEBAeDETA we cannotice the strong impact on photo-oxidation of the a-aminomethylene. The oxidation of the amine bridges is quicklydetected.

Next we compare our FTIR results on DGEBAeMNA withthe phenoxy resin study [1,2] and work on DGEBAeMTHPA[4e7] or DGEBAeHHPA [4]. The chemical change with theanhydride system is quite different from that of phenoxy resinageing. The formation of photoproducts is the same for thethree phthalic anhydrides, even if the irradiation system isnot the same. These results show that the influence of anhy-dride links is high because the oxidation photoinitiation onphenoxy part with abstraction of Ha is not the main pathwayof degradation. The anhydride links brings a slight stabiliza-tion of the cured DGEBA system. The mechanism of hydro-gen abstraction is probably due to different pathways Ha,Hb, Hc. No main route can be underlined as for DGEBAeamine. The main photoproducts observed can be identifiedas phenyl alkylate end groups at 1760 cm�1 and acids at1710 cm�1.

3.3. Analysis of thermo-oxidation

The PAS-FTIR analysis as a function of the exposure timein oven shows changes after an induction period. The changesare illustrated in Fig. 3A, B (second spectrum) and Fig. 4A andB. For DGEBAeDETA, the induction period is around 400 h.The main photoproducts are observed in the hydroxyl and car-bonyl regions. We can notice the increase of a band centered at3400 cm�1 (hydroxyl groups). The main absorption in the car-bonyl area is at 1670 cm�1 probably due to amide and ketoneend groups. We also observed the 1740 cm�1 band. But theratio DDO(1670)/DDO(1740) is quite different if we comparephotochemical and thermal ageing. The amide formation isfavoured with thermal oxidation at 100 �C. Different hypoth-eses can be proposed; some amides are unstable under UVirradiation or the oxidation pathway to produce amides ispreferred under thermal conditions.

The change of DGEBAeMNA is less significant (Fig. 4B)after 1980 h in the oven. The induction period is around 900 h.

1253F. Delor-Jestin et al. / Polymer Degradation and Stability 91 (2006) 1247e1255

THERMOOXIDATIONInitial material T=O to 1980hrs

-020406080

100120140160180200220240260280300320340360380

PA

1550160016501700175018001850

A1660

1740

T=0

Wavenumber (cm-1)

Wavenumber (cm-1)

50

100

150

200

250

300

350

400

450

PA

15501600165017001750180018501900

1745

1960 hrs

1790 hrs1460 hrs

1270 hrs1150 hrs910 hrsT=0

B

1610

Fig. 4. FTIR-PAS spectra after thermo-oxidation; A e carbonyl absorption area for epoxyeamine. B e carbonyl absorption area for epoxyeanhydride.

New bands with low intensity are detected at 3400, 1780 and1680 cm�1.

We can compare the results on epoxyeDETA with IRstudies on DGEBAeTETA [9]. Visible colour changes canbe detected in both cases. The development of colour isdue to the formation of polyenyl structures with the possibleformation of quinonoid or cyclised conjugated nitrogen com-pounds. In our case, the colouring is enhanced during thermalageing. We also estimate that amine cured epoxy resin tendto have excess amine on material surface, and it leads toenhanced degradation by the mechanism proposed by Patterson-Jones (given in the review [9]).

The results obtained with epoxyeMNA were comparedwith the thermal degradation of DGEBAeMTHPA [10].Both systems gave some anhydride products identified at1850 and 1785 cm�1. With MNA the two ageings also givevery different results.

With TGAeFTIR (Fig. 5) we can see the difference ofthermal stability between uncrosslinked DGEBA and amine

0

10

20

30

40

50

60

0 200 400 600 800 1000Temperature (°C)

Wei

ght L

oss

(%)

DGEBADGEBA-anhydrideDGEBA-amine

Fig. 5. TGA of three unaged samples.

1254 F. Delor-Jestin et al. / Polymer Degradation and Stability 91 (2006) 1247e1255

CO

2

H2O

830

890

1050

CO

2H2O

-0,03

-0,02

-0,01

0,00

0,01

0,02

0,03

0,04

0,05

0,06

Abso

rban

ce

5001000150020002500300035004000Wavenumber (cm-1)

Fig. 6. TGAeFTIR spectra of unaged DGEBAeamine system for two temperatures (300 �C and 500 �C).

or anhydride cured DGEBA. The weight loss starts at 110 �Cfor DGEBA. For DGEBAeamine the weight loss is significantat 200 �C. For DGEBAeanhydride it is at 240 �C. A smalldelay is noticeable. The DGEBAeamine is degraded at a lowertemperature than DGEBAeanhydride. The analysis by FTIR(Fig. 6) after TGA shows that the thermal decomposition underoxygen of DGEBA (crosslinked or uncrosslinked) gave mainlytwo products, water and carbon dioxide (at 300 �C). The spec-trum obtained at the end of the weight loss at 500 �C showsnew bands at 1060 and 890 cm�1. The assignment of thesebands was not conclusive.

3.4. Comparison of ageing and determinationof curing agent influence

Fig. 7 illustrates the comparison of the two studied epoxysystems during thermal or photoageing. We have calculatedthe oxidation rate at 1740 or 1780 cm�1 for each exposure

0

50

100

150

200

250

0 500 1000 1500 2000 2500

DO

PAS

at 1

740

cm-1

EP-amine PhotoEP-anhydride Photo EP-amine ThermoEP-anhydride Thermo

Time (hours)

Fig. 7. Kinetics of oxidation after thermal or photochemical ageing for epoxyeamine (1740 cm�1) and epoxyeanhydride systems (1780 cm�1).

time and each material. Photo-oxidation is quickly detected.No induction period is noticed. For thermo-oxidation theinduction period depends on the curing agent, and is higherfor the anhydrideeepoxy system (respectively 400 and900 h). The epoxy resin crosslinked with anhydride is lesssensitive to ageing than the epoxyeamine system. The choiceof curing agent depends on the required specific properties.The long-term behaviour of crosslinked epoxy system dependson the chosen curing agent. We can conclude that the epoxyeanhydride system was finally more satisfactory towardsageing.

3.5. Analysis of glass fibre filled epoxy composite

The filled system was an epoxyeanhydride material. ThePAS-FTIR analysis was also satisfactory for glass fibre filledresin. The PAS-FTIR spectrum of initial filled material wassimilar to the unfilled DGEBAeanhydride. The photochemi-cal and thermal ageing of the final formulation gave thesame analytical results. We obtained low oxidation and the in-duction periods were almost the same. Under UV irradiationthe induction period was 350 h. It was 900 h during thermalageing. The addition of glass fibre did not affect the materials’behaviour towards oxidative degradation.

4. Conclusion

The need for a better understanding of basic relationshipsbetween resin structure and performance is classical in ageingstudies. The analysis of two crosslinked DGEBA systems(DETA and MNA) after thermal and photochemical ageingallows us to propose a complementary approach in terms ofoxidation reactions, curing agent influence and comparisonof ageing. We can notice the simultaneous chemical changesof the DGEBA structure and of the bonds due to crosslinker.The behaviour of curing agents such as anhydride appears to

1255F. Delor-Jestin et al. / Polymer Degradation and Stability 91 (2006) 1247e1255

be much better both upon thermal and photoageing. It followsthat curing agents play an important role in epoxy resin degra-dation and stability.

References

[1] Rivaton A, Moreau L, Gardette J-L. Polym Degrad Stab 1997;58:

321e32.

[2] Rivaton A, Moreau L, Gardette J-L. Polym Degrad Stab 1997;58:333e9.

[3] Bellenger V, Verdu J. J Appl Polym Sci 1983;28:2677e88.

[4] Ollier-Dureault V, Gosse B. J Appl Polym Sci 1998;70:1221e37.

[5] Monney L, Belali R, Vebrel J, Dubois C, Chambaudet A. Polym Degrad

Stab 1998;62:353e9.

[6] Monney L, Rouge N, Dubois C, Chambaudet A. Polym Degrad Stab

1998;62:367e71.

[7] Monney L, Bole J, Dubois C, Chambaudet A. Polym Degrad Stab

1999;66:17e22.

[8] Guillot L, Monney L, Dubois C, Chambaudet A. Polym Degrad Stab

2001;72:209e15.

[9] Maxwell ID, Pethrick RA. Polym Degrad Stab 1983;5:275e301.

[10] Le Huy HM, Bellenger V, Paris M, Verdu J. Polym Degrad Stab

1992;35:171e9.

[11] Dyakonov T, Mann PJ, Chen Y, Stevenson TK. Polym Degrad Stab

1996;54:67e83.

[12] Lee JY, Shim J, Kim SW. J Appl Polym Sci 2001;81:479e85.

[13] Jain P, Choudhary V, Varma IK. Eur Polym J 2003;39:181e7.

[14] Damian C, Espuche E, Escoubes M. Polym Degrad Stab 2001;72:

447e58.

[15] Tsotsis TK, Keller S, Bardis J, Bish J. Polym Degrad Stab 1999;64:

207e12.

[16] Musto P, Ragosta G, Russo P, Mascia L. Macromol Chem Phys

2001;202:3445e58.

[17] Musto P. Macromolecules 2003;36:3210e21.

[18] Penot G, Arnaud R, Lemaire J. Die Angew Makromol Chem

1992;30:493e9.

[19] Delprat P, Gardette JL. Polymer 1993;34(5):933e7.