J O U R N A L OF M A T E R I A L S S C I E N C E L E T T E R S 14 (1995) 1 4 5 1 - 1 4 5 4

Fractographic analysis of fatigue crack propagation in an amorphous copolyester E. J. M O S K A L A , T. J. PECORINI, L. T. G E R M I N A R I O Eastman Chemical Company RO. Box 1972 Kingsport, TN 37662 USA

The fatigue crack propagation (FCP) behaviour of amorphous glassy polymers has been the topic of numerous investigations [1-8]. These studies have shown that fatigue crack propagation typically occurs by the formation of striations and/or discontinuous growth (DG) bands. A striation is a fracture surface marking which corresponds to the position of the crack tip as a result of an individual cycle or load excursion. A DG band, on the other hand, is a fracture surface marking which corresponds to an increment of crack growth caused by multiple cycles. Striations and DG bands may be found on the same fatigue fracture surface; DG bands typically occur at locations associated with slow crack growth rates, whereas striations occur at locations associated with rapid crack growth [1]. Fatigue fracture surfaces are commonly examined by scanning electron microscopy (SEM), despite the fact that this technique suffers from some limitations. In conventional scanning electron microscopy, specimens must be coated with either carbon or a conductive metal in order to eliminate surface charging and to reduce radiation damage. Although the new generation of field emission-based scanning electron microscopes obviate the need for suface coating, accurate measurements of topographical features in three dimensions cannot be obtained. A relatively new type of microscopy, scanning force microscopy (SFM), on the other hand, is capable of providing three-dimensional images of surfaces in air or under a fluid, without the need for a vacuum or special surface treatment [9]. A primary objective of this study was to determine the applicability of SFM for fractographic analysis of polymeric materials. Fatigue crack propagation tests were performed on an amorphous, glassy thermoplastic copolyester of dimethyl terephthalate with ethylene glycol and 1,4cyclohexanedimethanol (EASTAR~'ETG, Eastman Chemical Company). Compact tension specimens were machined from 6.3-ram thick injection-moulded plaques of the copolyester. FCP tests were performed at room temperature on a MTS closed-loop servohydraulic testing machine using a sinusoidal waveform with a frequency of 10Hz. The tests were conducted in load-control using a minimum-tomaximum load ratio of 0.1. Crack lengths were measured by computer using an elastic compliance technique. FCP rates, da/dN, were computed by an incremental polynomial method and plotted versus stress intensity factor range, AK, according to the Paris equation da/dN = A A K ~

0261-8028 © 1995 Chapman & Hall

where A and m are material parameters [10]. Further experimental details of the testing procedures can be found in our previous work [2, 3]. SEM studies were performed on a Cambridge Instruments Stereoscan 200 scanning electron microscope. Fracture surfaces were coated with a thin layer of gold in a sputtering chamber before the SEM observation. SFM studies were performed with a Digital Instruments Nanoscope III scanning probe microscope under ambient condition, using the large area (J-head) piezoelectric element. Sample surfaces were scanned with triangular, 200 gm Nanoprobe cantilevers, supplied with integrated pyramidal Si3N4 tips. Topographic, 3-D profiles were obtained in constant force mode. Crack growth occurs from left to right in all SEM and SFM micrographs shown in this paper. A log-log plot of da/dN versus AK for the copolyester is shown in Fig. 1. The data can be described by the following Paris equation da/dN = 5.47 X 10-4AK389

Crack length and AK were measured in ram/cycle and MPam 1/2, respectively. A cursory examination of the fracture surface by optical microscopy revealed that fatigue crack propagation occurs by DG band formation at values of AK below approximately -2

x

E E

v

×x

-3-

"O

Str'ations

v O

DGBs ~

-4-

-0.2

o:2

0.4

Iog(AK) (MPam 1/2) Figure 1 Log-log plot of da/dN versus AK.

1451

1 MPam 1/2 and by striation formation at values of A K above 1 MPam ~/2, as is indicated in Figure 1. A scanning electron micrograph of striations at &K = 1.2 MPam v2 is shown in Fig. 2. The striations are approximately 1 gm wide, which agrees well with the FCP rate of 1.1 gin/cycle predicted by the above Paris equation. It should be noted that it was not possible to obtain clear micrographs of the striations at higher magnifications due to the material's susceptibility to beam damage. A scanning electron micrograph of DG bands at AK= 0.8 MPam 1/2 is shown in Fig. 3. The DG bands in this region are approximately 30gin wide and correspond to approximately 130 cycles. The topography of these bands is generally similar to that reported for discontinuous growth bands in other amorphous glassy polymers [1,5-7]; relatively smooth dark regions alternate with fibrillated, light regions. However, the detailed structure of these DG bands is somewhat unusual and will be discussed in the SFM studies. A three-dimensional scanning force micrograph of the striations is shown in Fig. 4. DSll and KSnczS1 have shown that a growing fatigue crack is always preceded by a crazed region [7]. The fragments of the broken craze fibrils are clearly evident as small nodules on the fracture surface in Fig. 4 with the

Figure 3 Scanning electron micrograph of discontinuous growth bands at &K = 0.8 MPam 1/2.

o.

15 .................

/JM

Figure 4 Three-dimensional atomic force micrograph of striations at AK = 1.2 MPam ~/2. (X: 5 grrddivision, Z: 0.5 ~tm/division)

Figure2 Scanning electron micrograph of striations at &K = 1.2 MPa m 1/2.

O o

most highly extended fibrils occurring at the front (i.e. at the back of the craze). A dimensional SFM image of the striations and a profile of this region are shown in Fig. 5. The

am

VV O

O O

"5 0

(b) I

5.0

I

10.0 gm

Figure 5 (a) Two-dimensional atomic force micrograph and (b) depth profile for striations at &K = 1.2 MPam 1/2,

1452

I

15.0

crack twodepth depth

profile, Fig. 5b, corresponds to the region enclosed in a rectangular box in Fig. 5a. It was recorded over the length of the box and averaged over the width of the box. The arrowheads in the SFM image, Fig. 5a, correspond with the arrowheads in the depth profile and delineate an individual striation. The strong periodic nature of the striations is clearly evident in Fig. 5b. The average distance between striation peaks is approximately 1.4 gm. The average depth of a striation is approximately 20 nm (0.020/zm), giving a width-to-depth ratio of 70:1. Kitagawa used interference microscopy to show that the width-todepth ratio for striations in poly(methyl methacrylate) is about 100:1 [8]. A three-dimensional scanning force micrograph of the DG bands is shown in Fig. 6. According to a model for discontinuous crack growth proposed by D611 and K6ncz61 [7], a large rounded peak attributed to melted craze fibrils identifies the beginning of a DG band (points A and E in Fig. 6). Several sharp spikes from fibrils which fail in a pseudo-brittle fashion follow (from points B to D). Finally, a relatively flat region marks the end of the DG band (point D). Enlargements of regions D and E are shown in Figs 7 and 8, respectively. In Fig. 7, the remnants of very short craze fibrils, reminiscent of the striation topography in Fig. 4, are clearly evident. In Fig. 8, the topography is distinctly different, highly elongated, melted fibrils precede the fibrils which fail in a pseudo-brittle fashion. A two-dimensional SFM image of the DG bands and a depth profile of this region are shown in Fig. 9. Once again, the depth profile, Fig. 9b, corresponds to the region enclosed in a rectangular box in Fig. 9a. It was recorded over the length of the box and averaged over the width of the box. The arrowheads in the SFM image correspond with the arrowheads in the depth profile and delineate an individual DG band. The depth profile, Fig. 9b, shows that each DG band begins with a tall, broad peak, followed by a series o f sharper peaks with progressively smaller heights. However, approximately midway across the DG band (point C), the height of the peaks increases slightly and then continues to decrease once again. The overall decrease in height within one DG band is approximately 0.5gm. Closer examination of the region between points C and D in Fig. 6 reveals that the broken fibrils are aligned in bands approximately 0.8 gm wide. (It should be noted that the above Paris equation predicts that the amount of crack extension per cycle should be about 0.3 gm.) The fibrils in the region between points B and C, however, appear to be arranged randomly. These observations suggest a modification to the model proposed by D611 and K6ncz61 [7] for DG band formation, as is outlined in Fig. 10. The craze preceding the crack tip lengthens after N cycles. Then, the most highly extended fibrils fail thermally on a per cycle basis creating increments of craze extension, Ac. Finally, the bulk of the craze fibrils fail in a pseudo-brittle fashion creating the profile shown in Fig. 9b. The incremental craze extension that occurs during thermal failure of the

2.0

35

........

-

............. 40

"""

UM

Figure 6 Three-dimensional atomic force micrograph of discontinuous growth bands at AK= 0.8 MPam v2. (X: 10 [,tm/division, Z: 1.5 gm/ division)

Figure 7 Three-dimensional atomic force micrograph of region D in Fig. 6. (X: 1 gm/division, Z: 1 gm/division)

Figure 8 Three-dimensional atomic force micrograph of region E in Fig. 6. (X: 1 l,tm/division, Z: 1 ].tm/division)

1453

~m o o

T O

o (b) O

0

I

10.0

I

20.0

I

lam

30.0

I

40.0

Figure 9 (a) Two-dimensional atomic force micrograph and (b) depth profile for discontinuous growth bands at 2~K= 0.8 MPam 1/2.

Cyole

most highly extended craze fibrils gives rise to the bands of broken fibrils in the region between points C and D in Fig. 6.

I'-"

No

>

q

I',

Acknowledgement The authors wish to thank Eastman Chemical Company for permission to publish this work. This paper was originally presented at the Society of Plastics Engineers' ANTEC '94 in San Francisco, California, and has been reprinted with the permission of the Society.

y!

No+ N

EASTAR®is a registerd trademark of Eastman Chemical Company. No+N+

1

References Ac2

I1-No+N+2

No+N+3

Figure 10 Model for the formation of discontinuous growth bands.

1454

1. R. W. HERTZBERG and J. A. MANSON, "Fatigue of engineering plastics" (Academic, London, 1980). 2. E. J. MOSKALA, J. Appl. Polym. Sci. 49 (1993) 53.

3. Idem., J. Mater. Sci. in press. 4.

H. KIM, R. W. TRUSS, Y. MAI, and B. COTTERELL,

Polymer 29 (1988) 268. 5. M. E. MACKAY, T.-G. TENG, J. M. SCHULTZ, J. Mater. Sci 14 (1979) 221. 6. M. T. TAKEMORI and R. P. KAMBOUR, J. Mater. Sci. Lett. 16 (1981) 1108. 7. W. DI~LL and KONCZOL, Adv. Polym. Sci. 91/92 (1990) 137. 8. M. KITAGAWA, Bull. J. Soc. Mech. Eng. 18 (1975) 240. 9. G. BINNIG, C. F. QUATE, and C. GERBER, Phys. Rev. Lett. 56 (1986) 930. 10. P. C. PARIS and F. ERDOGAN, J. Bas. Eng. Trans. ASME Ser. D 85 (1963) 528.

Received 14 October 1994 and accepted 27 April 1995