Composites Science and Technology 69 (2009) 523–530

Contents lists available at ScienceDirect

Composites Science and Technology

journal homepage: www.elsevier .com/ locate/compsci tech

Rapid evaluation of long-term thermal degradation of carbon fibreepoxy composites

J. Wolfrum, S. Eibl *, L. LietchBundeswehr Research Institute for Materials, Explosives, Fuels and Lubricants, Institutsweg 1, D-85435 Erding, Germany

a r t i c l e i n f o a b s t r a c t

Article history:Received 30 July 2008Received in revised form 19 November 2008Accepted 22 November 2008Available online 3 December 2008

Keywords:A. Polymer matrix composites (PMCs)A. Carbon fibresB. Thermal propertiesD. Infrared (IR) spectroscopyE. Heat treatment

0266-3538/$ - see front matter � 2008 Elsevier Ltd. Adoi:10.1016/j.compscitech.2008.11.018

* Corresponding author. Tel.: +49 8122 9590 3217;E-mail addresses: JohannesWolfrum@bundeswehr

Eibl@bundeswehr.org (S. Eibl).

Two commercially available carbon fibre reinforced composites (8552/IM7 and M18-1/G939) wereexposed to heat above maximum operational temperature at various durations. Mass loss and mechan-ical properties were measured over time. A chemical analysis was also performed on these composites.The two primary components of each matrix, the epoxy resin and the thermoplastic, were observed todegrade at different rates under various thermal loading conditions. The epoxy resins degrade predom-inantly as measured by IR spectroscopy and thermal desorption/gas chromatography mass spectrometry.By using mass loss, strength, and IR spectroscopic data, a correlation was made between strength char-acteristics of each composite and the relative amount of the two primary matrix components. The devel-oped relationship can be used to estimate rapidly the mechanical properties from the intensity ratio of IRbands characteristic of the two components.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

Knowledge of long-term behaviour of high-performance com-posites is essential for the application in a higher temperatureenvironment. Especially epoxy resin based composites are widelyused as laminates for aerospace, ballistic, engineering components,etc. because of their excellent bonding, thermal and mechanicalcharacteristics. Similar to any other material, mechanical proper-ties depend on various parameters. Environmental influences suchas moisture absorption are the most crucial, because they can leadto a plastification of the matrix material. Fundamental investiga-tions have been done on the number of factors influencing mois-ture uptake of fibre reinforced composites [1]. While thesematerials have been in use for many years, only limited data isavailable concerning the influence of the long-term thermalloading.

When composites are heated to temperatures in the range ofglass transition temperature, thermo-mechanical effects due tosoftening and/or decomposition of the polymer may occur. Degra-dation is even more critical when the material is heated in an oxi-dising environment than in a vacuum or inert atmosphere [2,3]. Ithas been found that degradation is associated with mass loss fromdifferent surfaces at different rates [4–6]. Matrix cracking anddelamination have been shown to occur after thermal exposureby many authors [7–12].

ll rights reserved.

fax: +49 8122 9590 3602..org (J. Wolfrum), Sebastian

Skourlis used fibre fragmentation tests at different tempera-tures to show that the interphase has a lower glass transition tem-perature than the surrounding resin [13]. This results in a fasterdegradation of the interphase when the composite is exposed tohigher temperatures. These investigations were predominatelyperformed in temperature ranges above the glass transitiontemperature.

A decrease in mechanical strength, a severe thermal damage bycracks and delamination, and high mass losses have been reportedfor 8552/IM7 when the material was heated at 340–450 �C for30 min [14].

For this investigation, the goal was to obtain an initial under-standing of the temperature effects on two different composites(8552/IM7, M18-1/G939) that are common within the aerospaceand other industries. Therefore specimens were exposed to iso-thermal conditions in the range of the glass transition/maximumoperational temperature for various durations, comparable toother studies [15], also including accelerated aging methods[16,17]. Mechanical tests and chemical analyses were performedto characterise the degradation. The thermal aging of 8552/IM7and comparable model type systems were previously investigatedby IR spectroscopy with respect to the changes of the matrix [18–20]. Nevertheless the focus of these studies has been to give a de-tailed mechanism for the thermal degradation of the epoxy resin inthe matrix. Also, we consider the tougheners and provide informa-tion on volatile degradation products additionally. Furthermore anempiric method is developed to predict mechanical compositeproperties on the basis of changes in the matrix composition, rap-idly analysed by IR spectroscopy.

524 J. Wolfrum et al. / Composites Science and Technology 69 (2009) 523–530

2. Material

All tests were carried out with the carbon fibre reinforced epoxysystems 8552/IM7 and M18-1/G939 from Hexcel CompositesGmbH (Stade, Germany). The laminates were cured in an autoclaveaccording to the manufacturer’s recommended conditions [21].

The 8552/IM7 laminates consisted of 16 UD-plies denoting anapplication oriented quasi-isotropic (QI) lay-up with [0/90/+45/�45]2S fibre orientations for the tensile and compression test speci-men. Unidirectional (UD) lay-up was chosen for Interlaminar ShearStrength (ILSS) specimen. The M18-1/G939 laminates were madeof eight plies denoting a QI lay-up with [(0/90)/(90/0)/(+45/�45)/(�45/+45)]S fibre orientations for the tensile and compression, anda [(0/90)]8 lay-up for ILSS test specimen. All laminates were visuallyinspected for any surface defects. Ultrasonic C-scans were per-formed to ensure that the test laminates were free of delamination,voids and fibre orientation errors. After processing, the laminateswere cut into samples of 250 mm � 100 mm with a water-cooleddiamond wheel saw. The samples were redried at 70 �C.

The matrix systems 8552 and M 18-1 consist of aromatic epoxyresins, which are toughened with the temperature resistant ther-moplastics polyethersulfone and polyetherimide, respectively [22].

3. Experimental

The samples were isothermally aged in standard convectionovens at 180 �C, 190 �C and 200 �C. They were removed and cooleddown to room temperature for measuring the mass loss and fortesting at various time intervals up to approximately 500 days.The masses were recorded with an accuracy of 0.0001 g.

Tensile tests were performed in accordance with DIN EN 2561,compressive tests with DIN EN 2850A and ILSS tests with DIN EN2563.

Changes in the composition of the polymer matrix were ana-lysed by micro-attenuated total reflection (ATR) fourier transforminfrared spectroscopy (FTIR). Spectra were recorded from the sur-face and at least 500 lm under the surface (bulk). Volatile productsformed by thermal degradation and oxidation reactions wereexplicitly identified by thermal desorption gas chromatography/mass spectrometry (TD-GC/MS) after desorption at 200 �C.

Fracture surface and cross sections were investigated usingscanning electron microscopy (SEM). For cross-sectional views,samples were embedded in epoxy resin and polished.

4. Results

4.1. Mass loss and mechanical properties

Fig. 1 illustrates the measured mass loss of the two materials atthree aging temperatures. The mass loss increases with higher

0

1

2

3

4

0 100 200 300 400 500

Time t [d]

Mas

s lo

ss Δ

m [%

]

180°C

200°C190°C

8552/IM7

Fig. 1. Mass loss of 8552/IM7 (left) and M18-1/G939 (righ

aging temperatures. The overall mass loss and the mass loss rateof M18-1/G939 are about twice as high as those of 8552/IM7, espe-cially for aging times longer than 200 h.

Tensile, compressive and ILS strength values obtained at differ-ent temperatures after different aging time intervals are comparedas a function of both temperature and mass loss in Figs. 2–4. A de-crease in strength with increasing mass loss is observed, but no sig-nificant influence of the temperatures applied. M18-1/G939 showsa slightly greater decrease in ILS strength (by 40%) compared to8552/IM7 (by 25%, Fig. 2), whereas tensile strength decreases morefor 8552/IM7 (by about 33% compared to by about 13% for M18-1/G939, Fig. 3). The change of compressive strength does not differsignificantly between the two composites (Fig. 4). Typical resultsof the tensile test are shown in the stress–strain diagrams inFig. 5. For both materials, the slopes of the curves do not changewith increasing aging time at 200 �C. In contrast, the ultimate ten-sile stress and strain decrease with increasing aging time.

4.2. Fracture behaviour

In addition to mechanical strength decrease, the appearance ofthe specimens after failure also changed. Fig. 6 shows the represen-tative specimens of 8552/IM7 aged at 200 �C compared to an un-aged specimen. The photograph illustrates increased ‘‘brooming”with increasing thermal aging. This effect was not as obvious forM18-1/G939 since the fibre filaments were locked into positionby the fabric structure. Fig. 7 shows a comparison of the cross sec-tions of unaged and aged 8552/IM7 ILSS specimens after aging at200 �C for 69 days. The entire cross section of the aged specimenexhibits micro cracking, and some of the cracks have developedinto delaminations. Similar results were obtained for M18-1/G939. Fig. 8 illustrates differences of fracture surfaces of an unagedand aged M18-1/G939 tensile test specimen. The aged specimenwas stored at 200 �C for 34 days. On fracture surfaces of unagedspecimens fibres are covered with the matrix representing a goodfibre to matrix adhesion. In aged specimens fibres are separatedfrom the matrix, indicating a low fibre to matrix adhesion. Thesame effects were also observed for 8552/IM7.

4.3. Chemical analysis

An IR spectroscopic analysis of the matrix system 8552 showsintensive bands at 1610 cm�1 and 1510 cm�1, which are attributedto the epoxy resin (Fig. 9). Bands at 1579 cm�1 and 1486 cm�1 arecharacteristic of the polyethersulfone. A detailed interpretation ofthe components’ IR spectra is given in references [18,19,22]. Forthe evaluation of the epoxy resin the 1510 cm�1 band was chosen,as well as the 1486 cm�1 band for the polyethersulfone. The spec-tra of the matrix recorded on the specimen surface are comparedfor different durations of aging at 180 �C (Fig. 9). A decrease in

Time t [d]

180°C

200°C190°C

0

2

4

6

8

0 100 200 300 400 500

M18/G939M18-1/G939

t) versus aging time at different aging temperatures.

0

20

40

60

80

100

120

140

0 1 2 3

Mass loss Δm [%] Mass loss Δm [%]

ILS

-Str

engt

h τ

[MP

a]

180°C

200°C190°C

8552/IM70

10

20

30

40

50

60

70

80

90

0 1 2 3 4 5

180°C

200°C190°C

M18-1/G939

Fig. 2. ILS strength of 8552/IM7 (left) and M18-1/G939 (right) versus mass loss at different aging temperatures.

0

200

400

600

800

1000

0 1 2 3 4 0 1 2 3 4 5

Mass loss Δm [%] Mass loss Δm [%]

Ten

sile

str

engt

h σ

[MP

a]

180°C

190°C

200°C8552/IM70

100

200

300

400

500

600

180°C190°C

200°CM18/G939M18-1/G939

Fig. 3. Tensile strength of 8552/IM7 (left) and M18-1/G939 (right) versus mass loss at different aging temperatures.

0

100

200

300

400

500

600

0 1 2 3 4 5 0 1 2 3 4 65

Com

pres

sive

str

engt

h σ

[MP

a]

180°C

200°C

190°C

8552/IM70

100

200

300

400

500

180°C

200°C

190°C

M18-1/G939

Mass loss Δm [%] Mass loss Δm [%]

Fig. 4. Compressive strength of 8552/IM7 (left) and M18-1/G939 (right) versus mass loss at different aging temperatures.

J. Wolfrum et al. / Composites Science and Technology 69 (2009) 523–530 525

epoxy band intensities can be observed compared to those of thepolyethersulfone as aging time increases. This means thermal deg-radation occurs predominantly within the epoxy resin of the ma-trix. While the polyethersulfone may still undergo some thermaldegradation, the rate of degradation of the epoxy is much higher.The IR spectra also show an intensive and broad band at about1650 cm�1. This band is characteristic of oxidation products repre-sented by carbonyle (C@O) species formed by reactions with oxy-gen [20,23]. Only a little increase in band intensity at 1650 cm�1

with aging time can be observed when bulk material of the CFRPis investigated.

To evaluate the effects of time and temperature more in detail,intensity ratios of the polyethersulfone (1486 cm�1) and the epoxyresin (1510 cm�1) bands for the surface and the bulk material areshown in Fig. 10. Other changes in the IR spectra such as the forma-tion of broad carbonyle bands (�1650 cm�1) and the decrease ofbroad bands characteristic for the degradation of aliphatic C–H-,N–H- and OH- groups of the epoxy resin with bands at 2900–3400 cm�1 [20] are determined less accurately.

The decrease in the epoxy band intensities is limited mainly tothe surface of the laminate. However, a slight degradation of theepoxy resin within the bulk of the laminate is observed. The band

0

100

200

300

400

500

600

700

800

900

1000

0 0.5 1 1.5 2 0 0.5 1 1.5 2

Strain ε [%] Strain ε [%]

Str

ess

σ [M

Pa]

Str

ess

σ [M

Pa]

8552/IM7

0 d/200°C

25 d/200°C

282 d/200°C

0

100

200

300

400

500

600M18/G939

0 d/200°C

14 d/200°C

158 d/200°C

M18-1/G939

Fig. 5. Stress–strain diagrams of 8552/IM7 (left) and M18-1/G939 (right) after different aging intervals at 200 �C.

Fig. 6. Representative broken tensile specimen of 8552/IM7: Unaged (top), aged for 25 days at 200 �C (middle) and 282 days at 200 �C (bottom).

526 J. Wolfrum et al. / Composites Science and Technology 69 (2009) 523–530

intensity ratios are used to calculate the relative content of theepoxy resin, when aged specimens are compared to the non-aged(see right axis in Fig. 10 giving the relative amount of residualepoxy resin).

An IR spectroscopic analysis of M18-1/G939 shows bands at1720 and 1780 cm�1 characteristic of polyetherimide as tough-ening agent [24]. The band at 1510 cm�1 originates from theepoxy resin. A detailed assignment of the occurring bands is gi-ven in references [22,24]. Fig. 11 is a plot of the matrix spectrarecorded after different durations of aging at 180 �C. It can beobserved that the epoxy resin content decreases over time withrespect to the polyetherimide. This is indicated by a decreasingband intensity at 1510 cm�1 characteristic of the epoxy resincompared to the polyetherimide band at 1780 cm�1 (Fig. 12),similar to the behaviour of the 8552 matrix (Fig. 10). However,the degradation of the epoxy resin in the bulk material occursfaster and to a larger extent compared to 8552/IM7. A severeoxidative damage of the matrix is indicated by a broad andintensive band at 1650 cm�1 on the laminate surface, whereasin the bulk only a slight increase in band intensity at1650 cm�1 is observed.

When samples of aged M18-1/G939 and 8552/IM7 specimensare thermally desorbed at 200 �C and analysed by GC/MS, princi-pally two groups of emitted components are formed for bothcomposites:

Firstly aliphatic alcohols, aldehydes, ketones, esters and carbonacids with a maximum number of six carbon atoms; such as: 1-propanole, 2-propanole, 2-propenole, acetic aldehyde, acroleine,propionic aldehyde, 2-butanone, 1-hydroxy-2-propanone, aceticacid etc.

Secondly aromatic amines such as aniline, N-methylaniline,N,N-dimethylaniline, pyridine etc.

These components can be attributed to a thermal degradationand oxidation of the epoxy resin. Fig. 13 shows a characteristic seg-ment of the epoxy resin based on a detailed analysis of the epoxyresin components [18,22]. When the epoxy resin degrades, theweakest bonds are cleaved predominately. This leads to the forma-tion of aliphatic oxygen containing species and aromatic amines(see above). With increasing durations of aging, the amount ofhighly oxidised species such as carbonic acids increases, whereascomponents which are lower oxidised (alcohols and aldehydes)are desorbed in lower amounts. Therefore the proceeding oxida-

Fig. 7. SEM pictures of cross sections of unaged (top) and aged (69 days at 200 �C, bottom) 8552/IM7 specimen used for ILS strength tests.

Fig. 8. SEM pictures of fracture surfaces of an unaged (left) and aged (34 days at 200 �C, right) M18-1/G939 specimen.

J. Wolfrum et al. / Composites Science and Technology 69 (2009) 523–530 527

tion of the epoxy resin with increasing durations of thermal agingis evident. No products can be identified which originate exclu-sively from the thermoplastic tougheners.

5. Discussion

5.1. Interpretation of mechanical performance by the degradation ofthe matrix

A significant reduction in mechanical strength and an increasein mass loss are observed when samples are aged at 180 to200 �C. As ILS strength is dominated by the properties of the matrixand the fibre-matrix interphase, it is clear that the extensive deg-radation of both the matrix and the interphase is responsible forthe significantly higher decrease in ILS strength of M18-1/G939compared to 8552/IM7. The matrix content in M18-1/G939 is�40 wt% (�34 wt% epoxy resin, �6 wt% polyetherimide) and

�35 wt% (�29 wt% epoxy, �6 wt% polyethersulfone) in 8552/IM7, respectively [22]. The epoxy resins of both composites showa similar thermal stability when analysed by thermogravimetrydue to their similar composition [18,22,25,26]. Slight differencesin thermal stability of the epoxy resins can be explained by the dif-ferent curing agents. 8552/IM7 exclusively contains diaminodi-phenylsulfone (DDS) as curing agent, whereas M18-1 additionallycontains 4,40-Methylenbis-(2,6-diethyaniline) (MBDA) and 4,40-Methylenbis-(2-isopropyl-6-methyaniline) (MBIMA) [22]. Thesecomponents show aliphatic side groups, which are more suscepti-ble to thermo-oxidative degradation than the purely aromatic DDSin the cured 8552-resin [26,27]. Epoxy components do not appearto be the crucial factor for a significant difference in thermalbehaviour as both resins contain Tetraglycidylmethylendianiline(TGMDA) and 8552 additionally contains Triglycidyl-p-aminophe-nole (TGPAP) [22]. However, the epoxy resins are less thermallystable compared to the toughening thermoplastics. In a thermo-

1200130014001500160017001800

Wave numbers [cm-1]

AT

R a

.u.

1510

1486

0 days

1 day

5 days

9 days

70 days

Fig. 9. IR spectra recorded on the surface of a 8552/IM7 specimen after differentdurations of aging at 180 �C.

00 100 200 300 400

Time t [d]

Inte

nsity

ratio

Ir [I

( 15

10

cm-1) /

I (1

48

6 c

m-1)]

100%

75%

50%

25%

0%

180°C/surface

200°C/surface

200°C/bulk

180°C/bulk

8552/IM7

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Fig. 10. Intensity ratios of bands characteristic of the epoxy resin (1510 cm�1) andthe polyethersulfone (1486 cm�1) within the matrix 8552; the right axis indicatesthe relative amount of residual epoxy resin.

1200130014001500160017001800

AT

R a

.u.

1510

1780

0 days

1 day

5 days

9 days

40 days

Wave numbers [cm-1]

Fig. 11. IR spectra recorded on the surface of a M18-1/G939 specimen afterdifferent durations of aging at 180 �C.

0

1

2

3

4

5

6

7

8

9

10

0 100 200 300 400

Time t [d]

Inte

nsity

ratio

Ir [I

( 15

10

cm-1

) /I (

17

80

cm

-1)]

100%

75%

50%

25%

0%

180°C/surface

200°C/surface

200°C/bulk

180°C/bulk

M18/G939M18-1/G939

Fig. 12. Intensity ratios of bands characteristic of the epoxy resin (1510 cm�1) andthe polyetherimide (1780 cm�1) within the matrix M18-1, the right axis indicatesthe relative amount of residual epoxy resin.

NS

O

O

CH2

CH

CH2

OH

N

H R

curing agent

aromatic

epoxy component

aromatic aliphatic

Fig. 13. Simplified characteristic segment of the chemical formula of the curedepoxy resins of the systems M18-1 and 8552 based on a detailed analysis of thecomponents [15,17].

528 J. Wolfrum et al. / Composites Science and Technology 69 (2009) 523–530

gravimetric analysis both thermoplastics – separated from the ma-trix – showed 200 �C higher degradation temperatures comparedto the epoxy resins [27]. Polyetherimide and polyethersulfone arewell known tougheners which improve mechanical performanceand thermal aging characteristics of composites [28].

Therefore the higher mass loss and the greater loss of ILSstrength for M18-1/G939 is explained by the predominant degra-dation of the epoxy resin, which is present in a higher amountand less thermally stable compared to 8552/IM7. The decrease inultimate tensile strength and strain (Fig. 5) is also explained bythe degradation of the matrix and the degradation of the fibre-ma-trix interphase, as well as the formation of cracks and delamina-tions observed by scanning electron microscopy and thebrooming of aged specimens after tensile test (Fig. 6).

A chemical analysis confirms the predominant degradation ofthe epoxy resin compared to the thermoplastic tougheners. Espe-cially on the specimen surface, the presence of oxygen leads to se-vere material damage due to the formation of oxidation products.A detailed analysis by TD-GC/MS attributes carbonyle group con-taining species, which were identified by IR spectroscopy, to theformation of highly oxidised species such as carbonic acids. Thesecomponents originate from the epoxy resin. IR band intensities thecharacteristic of aging products increase with aging time and cor-relate with the amount of desorbed oxidation products. In the bulkmaterial, only a slight increase in the concentration of oxidationproducts with increasing aging time can be observed by IR spec-troscopy. Based on an IR spectroscopic analysis, a depth profilefrom the surface into the bulk depicts a severe thermal and oxida-

J. Wolfrum et al. / Composites Science and Technology 69 (2009) 523–530 529

tive damage limited to about 0.1 mm below the surface for a 8552/IM7 specimen aged for 9 days at 200 �C [22].

The observed fracture behaviour (Figs. 6 and 8) is typical of a re-duced matrix to fibre adhesion after thermal load. The carbon fi-bres are impregnated with a finish consisting of an epoxy resin,which is known to be less thermally stable than the toughened ma-trix. Because of this characteristic it is the fibre to matrix inter-phase which is degraded first, when the bulk material is underthermal load. This may open a preferred pathway for oxygen topenetrate the matrix. It is well known, that when the oxidativedegradation reaction takes place at the exposed surface of thespecimen, the reaction zone progresses into the unreacted corezone dependent on thermal treatment [28–32].

5.2. Empirical correlation of mechanical properties and IR bandintensity ratio

The intensity ratio of IR bands characteristic of the degradingepoxy resin and the non degrading thermoplastic component ofthe matrix systems of 8552 and M18-1 can be expressed as a func-tion of time (Figs. 10 and 12). The mechanical properties over themass losses give almost linear correlations for the residual tensile,compressive and ILS strengths (see Figs. 2–4). Linear regressionswere calculated averaging the mechanical data in the temperaturerange from 180 to 200 �C. For the following calculations, the IR datawere averaged in the temperature range from 180 �C to 200 �C byfitting an exponential decay. This data basis was used to calculatecurves expressing the dependence of tensile, compressive and ILSstrength on the intensity ratio of characteristic IR bands.

Fig. 14 shows the result of the curve fit for 8552/IM7. The curvesof the mechanical properties dependent on IR band intensity ratioare almost identical, since the relative decrease in tensile, com-pression and ILS strength is almost identical over the mass loss.Nevertheless, there are differences in the location where the IRspectra were recorded. For each specimen, the strength data werecorrelated to both the bulk and surface IR spectra. For the surface,which experiences greater degradation, lower values of the bandintensity ratio are observed. This result in curves with a lower de-cay compared to the bulk data. For example, a measured IR bandintensity ratio of about 0.7 results in a residual strength of about70%, if spectra were recorded for the bulk material. The same ratiocorrelates to a loss of mechanical strength below 3%, if spectrawere recorded on the specimen surface.

Fig. 15 shows the result of the curve fit for M18-1/G939. For thismaterial, the relative decrease in different mechanical properties

70

75

80

85

90

95

100

00.20.40.60.81

8552/IM7

Fig. 14. Mechanical properties (ILS strength: s; tensile strength: rt, compressionstrength: rc) versus the intensity ratio of characteristic IR bands (epoxy resin:1510 cm�1 and polyethersulfone: 1486 cm�1) for 8552/IM7 normalised to 100 % forthe unaged specimen.

versus mass loss was not identical as the slopes of the linearregressions vary significantly (Figs. 2–4). These effects are ex-plained by the severe degradation of the matrix and the fabricstructure of the M18-1 prepreg. As a consequence individualcurves for each property are drawn. Again due to greater degrada-tion at the surface, lower IR band intensity ratios are observed forsurface data.

In summary, the empirical correlation provides a tool to esti-mate the decrease in mechanical strength from intensity ratios ofIR spectra. The spectra can be taken from either the surface orthe bulk to estimate the decrease in overall strength. As the surfacedegrades to a larger extent compared to the bulk material, it is rec-ommended to choose bulk IR spectra to predict long-term thermaldegradation effects. However, for a non-destructive evaluation of areal component only surface IR spectra can be recorded. As surfacespectra represent only a small fraction of the overall material, thismay be a source of inaccuracy. Moreover influences by the speci-men dimensions and the fibre lay-up have to be taken into account.

6. Conclusion

The initial goal was to determine the effects of thermal degrada-tion on two commonly used composites. Mass loss, loss of mechan-ical strength, and changes in fracture behaviour after thermal agingwere interpreted by the initial chemical composition and the deg-radation of the matrix. Differences in thermal behaviour betweenthe investigated resin systems are mainly due to a better ther-mo-oxidative stability of exclusively aromatic curing agents in8552 when compared to M18-1. Thermal and oxidative degrada-tion of the epoxy resin was observed by IR spectroscopy and TD-GC/MS, whereas the thermoplastic tougheners do not degradesignificantly.

Because of this effect the thermal degradation can be quantifiedby IR spectroscopy and correlated to the mechanical properties.The developed relationship can be used to predict the mechanicalproperties by the IR spectra. As many modern epoxy systems aremodified with high temperature resistant thermoplastics, thisprinciple can be applied to many commercial composites.

The described method provides the opportunity to evaluate athermal preload of these composites, for example by estimatingthe incurred temperature during a mishap and the material perfor-mance without the necessity of performing mechanical tests. Nev-ertheless, further tests should be performed to establish thereported correlations for a wider range of temperatures. The accu-

40

50

60

70

80

90

100

0123456789

M18/G939

IR band intensity ratio: I(1510 cm-1) / I(1780 cm-1)

Fig. 15. Mechanical properties (ILS strength: s; tensile strength: rt, compressionstrength: rc) versus the intensity ratio of characteristic IR bands (epoxy resin:1510 cm�1 and polyetherimide: 1780 cm�1) for M18-1/G939 normalised to 100 %for the unaged specimen.

530 J. Wolfrum et al. / Composites Science and Technology 69 (2009) 523–530

racy of predicted properties could be further improved by using awider range of test samples considering for example variousdimensions and lay-ups and by more component representativespecimen. The tool can then be used to design or modify futurecomposites and predict their performance under thermal load.

The IR spectroscopic technique can be applied very rapidly andit can be performed in a non-destructive manner, when CFRP with-out coatings are investigated. This opens a way for a non-destruc-tive in-service control of composites for aircraft maintenance.

Even though this investigation focuses on the heat damage ofcomposites, the deep insight into the composition and degradationpathway of the polymer matrix may be used to assess material per-formance and aging at lower temperatures. However, it is wellknown that degradation mechanisms may change at lowertemperatures.

References

[1] Bank LC, Gentry TR, Barkatt A. Accelerated test methods to determine the long-term behavior of frp composite structures: environmental effects. J Reinf PlastCompos 1995;14:559–87.

[2] Tsotsis TK. Thermo-oxidative aging of composite materials. J Compos Mater1995;29(3):410–22.

[3] Fang J, Hu X, Yue CY. Thermal degradation study of interpenetrating polymernetwork based on modified bismaleimide resin and cyanate ester. Polym Int2003;52:15–22.

[4] Bowles KJ, Nowak G. Thermo-oxidative stability studies of Celion 6000/PMR-15 unidirectional composites, PMR-15, and Celion 6000 fibre. J Compos Mater1988;22:966–85.

[5] Bowles KJ, Roberts GD, Kamvouris JE. Long-term isothermal aging of carbonfabric-reinforced PMR-15 composites: compression strength. ASTM STP1997;1302:175–90.

[6] Hancox NL. Thermal effects on polymer matrix composites: part 2. Thermaldegradation. Mater Des 1998;19:93–7.

[7] Pering GA, Farrell PV, Springer GS. Degradation of tensile and shear properties ofcomposite exposed to fire or high temperature. In: Springer: environmental effects oncomposite materials. Berlin: Technomic Publishing Company Inc.; 1981.

[8] Milke JA, Vizzini AJ. Thermal response of fibre-exposed composites. J ComposTechnol Res JCTRER 1991;13(3):145–51.

[9] Guetta B, Bunsell AR, Belinski C, Keraly FL. Effect of thermal spiking on carbonToray 300B reinforced polystyrylpyridine composites: a kinetic, spectroscopicand mechanical study. Composites 1989;20(5):461–5.

[10] Pritt HL. Thermal degradation of graphite–epoxy composites. Final report No.N62669-79-C-0270, Naval Air Development Center, 1980.

[11] Burcham LJ, Eduljee RF, Gillespie JW. Investigation on the microcrackingbehavior of bismaleimide composites: during thermal aging. Polym Compos1995;16(6):507–17.

[12] Mascia L, Zhang J. Mechanical properties and thermal aging of aperfluoroether–modified epoxy resin in castings and glass fibre composites.Composites 1995;26(5):379–85.

[13] Skourlis TP, McCullough RL. The effect of temperature on the behavior of theinterphase in polymeric composites. Compos Sci Technol 1993;49:363–8.

[14] Ng SJ, Cramer RJ, Mehrkam PA. Characterization of IM7/8552 polymercomposites subjected to heat damage, Composites for the real world. In:Proc. of the 29th international SAMPE technical conference, Orlando FL, 1997.p. 776–86.

[15] Kerr JR, Haskins JF. Effects of 50,000 hours of thermal aging on graphite/epoxyand graphite/polyimide composites. In: Structures, structural dynamics andmaterials conference, 23rd, 1982, collection of technical papers Part. 1,American Institute of Aeronautics and Astronautics, p. 101–8.

[16] Tsotsis TK, Lee SM. Long-Term thermo-oxidative aging of composite materials:failure mechanisms. Compos Sci Technol 1998;58:355–68.

[17] Gates TS, Grayson MA. On the use of accelerated aging methods for screeninghigh temperature polymeric composite materials. In: Structures, structuraldynamics and materials conference, 40th, 1999, collection of technical papersPart 2, American Institute of Aeronautics and Astronautics, p. 925–35.

[18] Dao B, Hodkin J, Krstina J, Mardel J, Tian W. Accelerated aging versus realisticaging in aerospace composite materials, II chemistry of thermal aging in astructural composite. J Appl Polym Sci 2006;102:3221–32.

[19] Bondzic S, Hodkin J, Krstina J, Mardel J. Chemistry of thermal aging inaerospace epoxy composites. J Appl Polym Sci 2005;100:2210–9.

[20] Dara IH, Ankara A, Akovali G, Suzer S. Heat damage assessment of carbon-fiber-reinforced polymer composites by diffuse reflectance infraredspectroscopy. J Appl Polym Sci 2005;96:1222–30.

[21] Product data sheets, Hexcel Inc, www.hexcel.com.[22] Eibl S. Observing inhomogeneity of plastic components in carbon fibre

reinforced polymer materials by ATR–FTIR spectroscopy in the micrometerscale. J Compos Mater 2008;42:1231–46.

[23] Musto P, Ragosta G, Russo P, Mascia L. Thermal-oxidative degradation of epoxyand epoxy-bismaleimide networks: kinetics and mechanism. Macromol ChemPhys 2001;202:3445–58.

[24] Hummel O, Scholl F. Infrared analysis of polymers resins and additives anatlas, 2. Winheim: Verlag Chemie; 1998.

[25] Mortaigne B, Regnier N. Study of epoxy and epoxy–cyanate networks thermaldegradation to predict materials lifetime in use conditions. J Appl Polym Sci2000;77:3142–53.

[26] May CA. Epoxy resins – chemistry and technology. 2nd Ed. New York,Basel: Marcel Dekker; 1988.

[27] Eibl S. Moderne chemische Analytik zur Bewertung mechanischer undthermischer Eigenschaften von Kohlefaser verstärkten Kunststoffen (CFK) immilitärischen Luftfahrzeugbau. 6te Würzburger Tage in der instrumentellenAnalytik in der Polymertechnik. In: Proceedings, 2003.

[28] Hayes BS, Seferis JC. Modification of thermosetting resins and compositesthrough preformed polymer particles: a review. Polym Compos2001;22:451–67.

[29] Nam JD, Seferis JC. Anisotropic thermo-oxidative stability of carbon fiberreinforced polymeric composites. SAMPE Quart 1992;24:10–8.

[30] Bowles KJ, Nowak G. Thermo-oxidative stability studies of Celion 6000/PMR-15 unidirectional composites, PMR-15, and Celion 6000 fiber. J Compos Mater1988;22:966–85.

[31] Potter BD, Yuan FG, Pater RH. The effect of molecular weight on transversemicrocracking in high-temperature LaRC-RP46T polyimide composites. J AdvMater 1993;25(1):30–4.

[32] Tandon GP, Pochiraju KV, Schoeppner GA. Modeling of oxidative developmentin PMR-15. Polym Degrad Stabil 2006;91:1861–9.