- Email: [email protected]
Fatigue properties of unidirectional carbon fibre composites at cryogenic temperatures
Fatigue properties of unidirectional carbon fibre composites at cryogenic temperatures* K. Pannkoke and H.-J. Wagner Kernforschungszentrum Karlsruhe, Institute of Material and Solid State Research IV, D-7500 Karlsruhe 1, Germany Design engineers working with composite materials are still confronted with uncertainties as to their fatigue behaviour, especially for cryogenic applications. In the course of cooling, different thermal contraction of the fibre and matrix gives rise to thermal stresses and strains which influence most of the mechanical properties. In this paper, the fatigue behaviour of unidirectional (UD) composites with different fibres and matrices will be described. A first step in understanding the failure mechanism under cyclic loading will be presented. In earlier tests excellent fatigue properties were found for carbon fibre UD composites made of T300 carbon fibres and an epoxy matrix 1'2. However, the applied epoxy resin was brittle, especially at low temperatures. Therefore the brittle resin was substituted by polycarbonate (PC), a tough thermoplastic polymer 3'4. Nevertheless, for a composite with that matrix the fatigue endurance limit, normalized to the static strength, was found to be much lower (43%). SEM studies illustrated a poor f i b r e - m a t r i x bond. To determine the bond's influence on fatigue properties, another tough matrix system was tested. The polymer PEEK is known to build a strong bond to carbon fibres, initiated by crystal growth onto the fibre surface 4'5. However, investigations on the fatigue behaviour of this composite at 77 K yielded the same low fatigue endurance limit as was found for the carbon f i b r e - P C system 4. At this point it can be concluded that the poor fatigue behaviour is not necessarily due to a strong or poor f i b r e - matrix bond. It is the purpose of this work to examine whether this different fatigue behaviour is due to matrix failure.
Keywords: carbon fibres; composites; fatigue
Nomenclature
a
N f E
Indices
Number of load cycles Frequency Initial Young's modulus
Greek letters
c~, ~ Shape parameters Strain
The fibres used in this work were high tensile carbon fibres. Their principal properties are summarized in Table 1. All fibre types exhibit a similar tensile strength. An epoxy resin (EP Rigidite ® R5212) and a modification (PAEK) of the thermoplastic, semicrystalline PEEK were chosen as matrix systems 6. The AS4/PAEK plates were manufactured in two different ways, by pressing the unidirectional (UD) layers (system 3a in Table 2)
*Paper presented at Non-Metallic Materials and Composites at Low Temperatures, 17 - 18 May 1990, Heidelberg, Germany O011 - 2 2 7 5 / 9 1 / 0 4 0 2 4 8 - 04 © 1991 B u t t e r w o r t h - H e i n e m a n n Ltd
:)48
Cryogenics 1991 Vol 31 April
U T F II D
Stress
Ultimate, upper Tensile Fibre In fibre direction After 107 load cycles
and by filament winding of a stretched hybrid yarn (system 3b in Table 2). The duroplastic matrix composite was produced by pressing thin sheets of a UD Table 1
Fibre properties
Fibres
aUT, (GPa)
CUTII (%)
EF If (GPa)
G30-500 AS-4 HTA7
3.9 3.8 3.9
1.7 1.6 1.6
228 210 236
Fatigue properties of UD carbon fibre composites: K. Pannkoke and H.-J. Wagner Table 2 Composite properties
Index
Matrix
Fibre
Quality
Bond
1 2 3a 3b 4
PC PEEK PAEK PAEK EP
HTA7 AS4 AS4 AS4 G30-500
+ + + +
+ + +
epoxy resin prepreg (fibre plane weight 145 45 g m -2, resin content 39%) in an autoclave. The quality of the UD plates (0~6) was determined by ultrasonic C-scanning which indicated satisfactory homogeneity and fibre alignment with the exception of the pressed AS4/PAEK composite (system 3a). The fibre content of all three composites was = 60 vol%. The specimens for experimental investigations were cut out of the plates.
OUT (MPa)
OD (MPa)
OD/OUT (%)
Reference
1980 2270 2304 2304 2349
851 1000 876 970 1386
43 44 38 42 59
4 4 -
l
1O0 ]
~
80 6(/
2~ 4(, o
2(,
Measurements The fatigue tests were performed using a high frequency resonance machine (50-100 Hz) 7. The magnitude of frequency depends on the stiffness of the machine and the tested panels. Energy dissipation causes only a small temperature rise inside the specimens 4. All devices were calibrated at low temperature and intercalibrated during the tests with devices outside the cryostat. After cooling down with LN 2 the experiments were performed under tensile threshold loading (R = 0.1) up to 107 cycles for run-outs. Parts of the fracture surfaces of several representative coupons, cut out in the load direction, were inspected by SEM.
I IITIII[~ t iT?lff[l T N~T~HI~ I llIHIl( l 01
10 2
103
I TTItlTT~T {TTTflT I TFH11~ ~ I 05
I 0~ N
10 7 ,"
Figure 1 S - N c u r v e s o f UD A S 4 / P A E K c o m p o s i t e s at 77 K w i t h aUT = 2 3 0 4 MPa. O , Composite 3a, pressed UD layers; zx , corn posite 3b, f i l a m e n t w o u n d hybrid yarn
100
I
80
60
Results
I t)4
n
Tensile fatigue behaviour at 77 K The stress-life curves of the materials under tensile threshold loading were determined by using a non-linear regression analysis applying the following equation 8 ao (N) = aD +
out - OD exp(log N/o¢)~
40
(1)
where c~ and/3 are shape parameters. Table 2 shows the fatigue strengths, ao, after 107 cycles, together with values from other components, as mentioned above. The composite with duroplastic matrix shows the highest normalized fatigue endurance limit, aD/OUT, of all the tests. Table 2 confirms that the bond does not have an essential influence on the fatigue behaviour. The fibre-matrix combinations with the strongest bond (AS4/PAEK) yield the worst fatigue properties. As seen from Figure 1 for the AS4/PAEK system, inhomogeneity of the laminates and misalignment of the fibres do not affect the fatigue properties very much. The values at 10 7 cycles cannot be regarded as a fatigue endurance limit because the W6hler curves have not yet flattened out. On the other hand, the G30-500/epoxy system
0
"T
11i;II--T]ITn]II---FFFrqTrPI~rF~HTTII T ~ F T ~ F T I T r I ~ !
lip
102
103
104
105
7 TTTTTITl-
10ti N
107 ,~
Figure 2 S-N curve of UD G30-BOO/epoxycomposite at 77 K with
OUT
= 2349 MPa
yields a remarkably high normalized fatigue endurance limit of 59% (Figure 2). At this point it cannot be concluded that the fatigue behaviour is mainly determined by the matrix and its failure mechanism. The fatigue properties of T300 or G30-500 carbon fibres might also be superior to those of AS4 fibres.
Failure behaviour Photographs of representative fractured UD specimens are shown in Figure 3 which indicate that the appearance of the fracture is not uniform for both composites. In the
C r y o g e n i c s 1991 Vol 31 April
249
Fatigue properties of UD carbon fibre composites: K. Pannkoke and H.-J. Wagner
';,
tI
i
(a)
(b)
Figure 3 Fractured UD specimens of: (a) AS4/PAEK; (b) G30500/epoxy
case of the AS4/PAEK system, the failure is characterized by a splitting and fraying of the fibres which looks like a brush. By contrast, the G30500/epoxy specimens generally exhibit a fracture surface with two different parts, namely a brush-like part and a transverse fracture line. Figures 4 and 5, which are taken from the brush-like part, show that the duroplastic matrix composite has fractured in a brittle manner. The fibres are bare as a consequence of the weak fibre- matrix bond (Figure 5).
Figure 4 SEM observation of fracture surface for G30500/epoxy composite, showing the brittle manner of fracture (see text for details)
250
Cryogenics 1991 Vol 31 April
Figure 5 SEM observation of fracture surface for G30500/epoxy composite (at higher magnification than Figure 4), showing the brittle nature of fracture (see text for details)
The resin surface fracture profile consists almost entirely of rows of cusps. As evident from Figure 5, their formation commences by development of transverse microcracks in the matrix along the fibre. As loads are transmitted from fibre to fibre by matrix shear, the increasing fibre fracture rate leads to high interlaminar shear stresses; thus a line of transverse cracks originates along the interface area. On account of the low ductility of the epoxy resin these cracks open very easily and so form the typical fracture surface profile of cusps.
Figure 6 SEM observation of fracture surface for G30500/epoxy composite, showing coalescence of cracks to form a longitudinal crack along the f i b r e - m a t r i x interface (see text for details)
Fatigue properties of UD carbon fibre composites: K. Pannkoke and H.-J. Wagner
lOpm ..... ~w~,ii~ ¸
Figure 7
SEM observation of fracture surface for AS4/PAEK composite. In contrast to Figure 6, no characteristic damage zone can be identified (see text for details)
Failure occurs when these cracks simultaneously bend over and merge to form a longitudinal crack. In the case of poor bonding, this coalescence follows the fibre-matrix interface, as illustrated in Figure 6 for the duroplastic matrix composite. In contrast to this, no characteristic damage zone could be identified in the PAEK composite (Figure 7). Even under high SEM magnification no microcracks are visible. The fibre-matrix bond is very strong, hence failure has taken place within the matrix, leaving a thin film of polymer on the fibre surface.
The duroplastic matrix composite shows multiple transverse crack propagation but good fatigue behaviour. The thermoplastic matrix composite by contrast shows little crack formation and a lower fatigue endurance. Thus, it can be concluded that multiple crack formation is good for energy absorption and better fatigue behaviour will result. It leads to multiple splitting of fracture energies, stress concentrations within the duromer are reduced and, furthermore, catastrophic crack extension is inhibited. On the other hand, SEM gives no information about crack development in the AS4/PAEK composite. This may be due to the toughness of the polymer. Hence, an efficient fracture energy absorbing mechanism is missing, some cracks lead to high stress concentrations and catastrophic failure occurs within the matrix. The AS4/PAEK material exhibits no longitudinal cracking but complete brush-like fractured specimens. The duroplastic matrix composite exhibits longitudinal cracking within the brush-like fractured area, indicating an interrupted shear load transfer by the matrix. It is astonishing that no longitudinal cracks have been found within the area of the clear cut fracture line. This feature is not yet understood, Further investigations on the fatigue life of composites in general and their shear fatigue behaviour in particular are necessary to give an estimation of the influence of matrix and fibre properties on the fatigue endurance. Furthermore, it is important to understand whether differences exist between duroplastic and thermoplastic matrix failure mechanisms. Further fatigue tests will concentrate on torsional loading of hoop-wound, thin-walled tubes. Comparison will be made between the fatigue behaviour under tensile and pure shear loading.
References I 2 3 4
Discussion The G30-500/epoxy system exhibits good fatigue resistance but poor fibre-matrix bonding, whereas AS4/PAEK has poor fatigue life but excellent fibrematrix bonding. The fibre-matrix bond probably has little influence on the fatigue resistance of the composite.
5 6 7 8
Hartwig, G. and Knaak, S. Co,ogenics (1984) 24 639-647 Hartwig, G. Adv Cryog Eng Mat (1981) 28 179-189 Ahlborn, K. Cryogenics (1988) 28 267-272 Ahlborn, K. Mechanische Eigenschaften yon Kohlenstoffaserverst~irken Thermoplasten ffir die Anwendung in der Tieftemperaturtechnologie PhD Dissertation University of Karlsruhe, Germany (1989) Peacock, J.A., Fife, B., Nield, E. and Barlow, C.Y. Proc ICCI-1 Cm~f North Holland, Amsterdam, The Netherlands (1986) 143-149 Miinstedt, H. and Zeiner, H. Kunststoffe (1989) 79 (10) 993-996 Hartwig, G. En~wclopediaof Polymer Science and Engineering Vol 4, Wiley and Sons, New York, USA (1986) Gecks, M. and Och, F. Ermittlung dynamischer Festigkeitskennlinien durch nichtlineare Regressionsanalyse, Sonderheft DFVLR, Strukturmechaniktagung, Ottobrunn (1977)
Cryogenics 1991 Vol 31 April
251
Keywords: carbon fibres; composites; fatigue
Nomenclature
a
N f E
Indices
Number of load cycles Frequency Initial Young's modulus
Greek letters
c~, ~ Shape parameters Strain
The fibres used in this work were high tensile carbon fibres. Their principal properties are summarized in Table 1. All fibre types exhibit a similar tensile strength. An epoxy resin (EP Rigidite ® R5212) and a modification (PAEK) of the thermoplastic, semicrystalline PEEK were chosen as matrix systems 6. The AS4/PAEK plates were manufactured in two different ways, by pressing the unidirectional (UD) layers (system 3a in Table 2)
*Paper presented at Non-Metallic Materials and Composites at Low Temperatures, 17 - 18 May 1990, Heidelberg, Germany O011 - 2 2 7 5 / 9 1 / 0 4 0 2 4 8 - 04 © 1991 B u t t e r w o r t h - H e i n e m a n n Ltd
:)48
Cryogenics 1991 Vol 31 April
U T F II D
Stress
Ultimate, upper Tensile Fibre In fibre direction After 107 load cycles
and by filament winding of a stretched hybrid yarn (system 3b in Table 2). The duroplastic matrix composite was produced by pressing thin sheets of a UD Table 1
Fibre properties
Fibres
aUT, (GPa)
CUTII (%)
EF If (GPa)
G30-500 AS-4 HTA7
3.9 3.8 3.9
1.7 1.6 1.6
228 210 236
Fatigue properties of UD carbon fibre composites: K. Pannkoke and H.-J. Wagner Table 2 Composite properties
Index
Matrix
Fibre
Quality
Bond
1 2 3a 3b 4
PC PEEK PAEK PAEK EP
HTA7 AS4 AS4 AS4 G30-500
+ + + +
+ + +
epoxy resin prepreg (fibre plane weight 145 45 g m -2, resin content 39%) in an autoclave. The quality of the UD plates (0~6) was determined by ultrasonic C-scanning which indicated satisfactory homogeneity and fibre alignment with the exception of the pressed AS4/PAEK composite (system 3a). The fibre content of all three composites was = 60 vol%. The specimens for experimental investigations were cut out of the plates.
OUT (MPa)
OD (MPa)
OD/OUT (%)
Reference
1980 2270 2304 2304 2349
851 1000 876 970 1386
43 44 38 42 59
4 4 -
l
1O0 ]
~
80 6(/
2~ 4(, o
2(,
Measurements The fatigue tests were performed using a high frequency resonance machine (50-100 Hz) 7. The magnitude of frequency depends on the stiffness of the machine and the tested panels. Energy dissipation causes only a small temperature rise inside the specimens 4. All devices were calibrated at low temperature and intercalibrated during the tests with devices outside the cryostat. After cooling down with LN 2 the experiments were performed under tensile threshold loading (R = 0.1) up to 107 cycles for run-outs. Parts of the fracture surfaces of several representative coupons, cut out in the load direction, were inspected by SEM.
I IITIII[~ t iT?lff[l T N~T~HI~ I llIHIl( l 01
10 2
103
I TTItlTT~T {TTTflT I TFH11~ ~ I 05
I 0~ N
10 7 ,"
Figure 1 S - N c u r v e s o f UD A S 4 / P A E K c o m p o s i t e s at 77 K w i t h aUT = 2 3 0 4 MPa. O , Composite 3a, pressed UD layers; zx , corn posite 3b, f i l a m e n t w o u n d hybrid yarn
100
I
80
60
Results
I t)4
n
Tensile fatigue behaviour at 77 K The stress-life curves of the materials under tensile threshold loading were determined by using a non-linear regression analysis applying the following equation 8 ao (N) = aD +
out - OD exp(log N/o¢)~
40
(1)
where c~ and/3 are shape parameters. Table 2 shows the fatigue strengths, ao, after 107 cycles, together with values from other components, as mentioned above. The composite with duroplastic matrix shows the highest normalized fatigue endurance limit, aD/OUT, of all the tests. Table 2 confirms that the bond does not have an essential influence on the fatigue behaviour. The fibre-matrix combinations with the strongest bond (AS4/PAEK) yield the worst fatigue properties. As seen from Figure 1 for the AS4/PAEK system, inhomogeneity of the laminates and misalignment of the fibres do not affect the fatigue properties very much. The values at 10 7 cycles cannot be regarded as a fatigue endurance limit because the W6hler curves have not yet flattened out. On the other hand, the G30-500/epoxy system
0
"T
11i;II--T]ITn]II---FFFrqTrPI~rF~HTTII T ~ F T ~ F T I T r I ~ !
lip
102
103
104
105
7 TTTTTITl-
10ti N
107 ,~
Figure 2 S-N curve of UD G30-BOO/epoxycomposite at 77 K with
OUT
= 2349 MPa
yields a remarkably high normalized fatigue endurance limit of 59% (Figure 2). At this point it cannot be concluded that the fatigue behaviour is mainly determined by the matrix and its failure mechanism. The fatigue properties of T300 or G30-500 carbon fibres might also be superior to those of AS4 fibres.
Failure behaviour Photographs of representative fractured UD specimens are shown in Figure 3 which indicate that the appearance of the fracture is not uniform for both composites. In the
C r y o g e n i c s 1991 Vol 31 April
249
Fatigue properties of UD carbon fibre composites: K. Pannkoke and H.-J. Wagner
';,
tI
i
(a)
(b)
Figure 3 Fractured UD specimens of: (a) AS4/PAEK; (b) G30500/epoxy
case of the AS4/PAEK system, the failure is characterized by a splitting and fraying of the fibres which looks like a brush. By contrast, the G30500/epoxy specimens generally exhibit a fracture surface with two different parts, namely a brush-like part and a transverse fracture line. Figures 4 and 5, which are taken from the brush-like part, show that the duroplastic matrix composite has fractured in a brittle manner. The fibres are bare as a consequence of the weak fibre- matrix bond (Figure 5).
Figure 4 SEM observation of fracture surface for G30500/epoxy composite, showing the brittle manner of fracture (see text for details)
250
Cryogenics 1991 Vol 31 April
Figure 5 SEM observation of fracture surface for G30500/epoxy composite (at higher magnification than Figure 4), showing the brittle nature of fracture (see text for details)
The resin surface fracture profile consists almost entirely of rows of cusps. As evident from Figure 5, their formation commences by development of transverse microcracks in the matrix along the fibre. As loads are transmitted from fibre to fibre by matrix shear, the increasing fibre fracture rate leads to high interlaminar shear stresses; thus a line of transverse cracks originates along the interface area. On account of the low ductility of the epoxy resin these cracks open very easily and so form the typical fracture surface profile of cusps.
Figure 6 SEM observation of fracture surface for G30500/epoxy composite, showing coalescence of cracks to form a longitudinal crack along the f i b r e - m a t r i x interface (see text for details)
Fatigue properties of UD carbon fibre composites: K. Pannkoke and H.-J. Wagner
lOpm ..... ~w~,ii~ ¸
Figure 7
SEM observation of fracture surface for AS4/PAEK composite. In contrast to Figure 6, no characteristic damage zone can be identified (see text for details)
Failure occurs when these cracks simultaneously bend over and merge to form a longitudinal crack. In the case of poor bonding, this coalescence follows the fibre-matrix interface, as illustrated in Figure 6 for the duroplastic matrix composite. In contrast to this, no characteristic damage zone could be identified in the PAEK composite (Figure 7). Even under high SEM magnification no microcracks are visible. The fibre-matrix bond is very strong, hence failure has taken place within the matrix, leaving a thin film of polymer on the fibre surface.
The duroplastic matrix composite shows multiple transverse crack propagation but good fatigue behaviour. The thermoplastic matrix composite by contrast shows little crack formation and a lower fatigue endurance. Thus, it can be concluded that multiple crack formation is good for energy absorption and better fatigue behaviour will result. It leads to multiple splitting of fracture energies, stress concentrations within the duromer are reduced and, furthermore, catastrophic crack extension is inhibited. On the other hand, SEM gives no information about crack development in the AS4/PAEK composite. This may be due to the toughness of the polymer. Hence, an efficient fracture energy absorbing mechanism is missing, some cracks lead to high stress concentrations and catastrophic failure occurs within the matrix. The AS4/PAEK material exhibits no longitudinal cracking but complete brush-like fractured specimens. The duroplastic matrix composite exhibits longitudinal cracking within the brush-like fractured area, indicating an interrupted shear load transfer by the matrix. It is astonishing that no longitudinal cracks have been found within the area of the clear cut fracture line. This feature is not yet understood, Further investigations on the fatigue life of composites in general and their shear fatigue behaviour in particular are necessary to give an estimation of the influence of matrix and fibre properties on the fatigue endurance. Furthermore, it is important to understand whether differences exist between duroplastic and thermoplastic matrix failure mechanisms. Further fatigue tests will concentrate on torsional loading of hoop-wound, thin-walled tubes. Comparison will be made between the fatigue behaviour under tensile and pure shear loading.
References I 2 3 4
Discussion The G30-500/epoxy system exhibits good fatigue resistance but poor fibre-matrix bonding, whereas AS4/PAEK has poor fatigue life but excellent fibrematrix bonding. The fibre-matrix bond probably has little influence on the fatigue resistance of the composite.
5 6 7 8
Hartwig, G. and Knaak, S. Co,ogenics (1984) 24 639-647 Hartwig, G. Adv Cryog Eng Mat (1981) 28 179-189 Ahlborn, K. Cryogenics (1988) 28 267-272 Ahlborn, K. Mechanische Eigenschaften yon Kohlenstoffaserverst~irken Thermoplasten ffir die Anwendung in der Tieftemperaturtechnologie PhD Dissertation University of Karlsruhe, Germany (1989) Peacock, J.A., Fife, B., Nield, E. and Barlow, C.Y. Proc ICCI-1 Cm~f North Holland, Amsterdam, The Netherlands (1986) 143-149 Miinstedt, H. and Zeiner, H. Kunststoffe (1989) 79 (10) 993-996 Hartwig, G. En~wclopediaof Polymer Science and Engineering Vol 4, Wiley and Sons, New York, USA (1986) Gecks, M. and Och, F. Ermittlung dynamischer Festigkeitskennlinien durch nichtlineare Regressionsanalyse, Sonderheft DFVLR, Strukturmechaniktagung, Ottobrunn (1977)
Cryogenics 1991 Vol 31 April
251