Polymers for Advanced Technologies Volume 8, pp. 371 377 Interlaminar Fracture Toughness of Carbon Fiber/Epoxy Composites using Short Kevlar Fiber and/or Nylon-6 Powder Reinforcement B. Y. Park, 1 S. C. Kim 2 and B. Jung 1 1 Agency for Defense Development, Yuseong P.O. Box 35-4, 305-600, 4-4-4, Daejon, Korea 2 Dept. of Chemical Engineering, Korea Advanced Institute of Science and Technology, Yuseong, Gusungdong 373-1, Daejon, Korea ABSTRACT Mode I (G IC ) and Mode II (G IIC ) interlaminar fracture toughness of carbon-fiber/epoxy composites have been investigated as a function of the amount of short Kevlar-29 fiber (SKF) and/or Nylon-6 powder (N6P) between continuous fiber layers. G IIC increased with increasing crack length as a consequence of the presence of SKFs bridging in the wake of propagating crack. G IIC of SKF alone could reach the maximum at an intermediate amount of SKF. G IIC of SKF and N6P was lower than that of SKF alone because N6P prevented the orientation of SKF to out-of-plane. The extent of SKF s bridging phenomenon may be influenced by the amount and orientation of SKF. G IC showed no significant effect with SKF and uniform irrespective of crack length. Scanning electron microscopy after G IIC test showed that new surfaces were created by extensive fiber bridging, pull-out and fracture of SKF in random direction without any fixed pattern. 1997 by John Wiley & Sons, Ltd. Polym. Adv. Technol. 8, 371 377 (1997) No. of Figures: 5 No. of Tables: 0 No. of Refs: 15 KEYWORDS: Mode I delamination toughness G IC, Mode II delamination toughness G IIC, short Kevlar fiber, Nylon- 6 powder, fiber bridging Correspondence to: B. Y. Park. INTRODUCTION The main advantages of carbon-fiber/epoxy composites are their superior fatigue resistance and strength- and stiffness-to-weight ratios. However, the primary deficiency of the current composites is their poor damage tolerance [1]. Conventional toughening approaches that involve dispersing a rubbery phase in the matrix resin have led to an improvement in damage tolerance at the sacrifice of hot/wet compressive performance [2]. Zeng et al. [3] suggested that toughening could be achieved by thin layers of matrix with modifier particles between plies in the carbon fiber/dicyanate composites. McGrail and Jenkins [4] used reactively terminated polysulphone as the particulate modifier in epoxy, bismaleimide and cyanate ester resin-based composites. Browning and Schwartz [5] ascribed the improved toughness to the incorporation of an adhesive in-lay either by itself or in combination with a Kevlar mat. Lin and Jang [6] suggested the incorporation of short Kevlar-29 fiber (SKF) in a thermosetting resin before being used for impregnating the continuous fibers or fabrics. Sohn and Hu [7] has reported that significant fiber bridging was observed for the Mode II (G IIC ) test while Mode I (G IC ) exhibited lower fracture energies with less apparent fiber bridging. Most of the fiber-bridging phenomena have been reported in Mode I fracture [8, 9], which arose from the misalignment of fibers across the crack plane or the growth of the crack in more than one plane, but is less understood for mode II delamination. CCC 1042 7147/97/060371 07 $17.50 Received 27 August 1996 1997 by John Wiley & Sons, Ltd. Revised 14 October 1996
372 / Park et al. The goal of this investigation was to study the effect of SKF bridging on G IC and G IIC of carbonfiber/epoxy composites using four different amounts of SKF and/or Nylon-6 powder (N6P). EXPERIMENTAL Preparation of Materials The carbon fabric (Fiberite, W-133) was eight-harness satin woven type. The epoxy resin (Ciba-Geigy Corp, 0510) and an aromatic amine curing agent (Ciba- Geigy Corp, HT 976) were used as received. Epoxy was mixed with curing agent at a weight ratio of 100 : 38 and the mixture was heated for the complete dissolution of curing agent in the epoxy for 20 min at 120 C. The prepreg was made by impregnating the resin mixture in the carbon fabric with a hand roller, heated during 2.0 hr at 120 C for B-stage condition and had fiber wt% fractions of 55.0% to 60.0%. Kevlar-29 was chopped to 6 mm in length and used as the interlaminar reinforcement. The particles used were semicrystalline Nylon-6 (Atochem Corp, 1002D NAT) with a particle of 20 m in average diameter. SKF was spread manually after impregnating the resin in the carbon fabric. N6P was uniformly sieved onto the prepreg having SKF. The amount of SKF and N6P in equal weight ratio, SKF or N6P used was 4.0, 8.5, 17.0 and 25.5 g/m 2. The laminate consisted of eight plies of prepreg. The initial crack was formed by the folded sheet of 16 m thick aluminum foil embedded in the mid-plane of the laminate. Laminates were prepared by autoclave molding. Each specimen (150 24 3.2 mm 3 ) was cut from laminates by using a radial diamond saw. For G IC specimens, the hinge tab made of steel was bonded to the upper and lower side of the precracked specimen as a means of applying the load perpendicular to the interlaminar layer. The surfaces of the specimen and the hinge tab were prepared for bonding after carefully scraping with sandpaper and cleaning with acetone. The adhesive used for bonding the hinge tab to the specimen was adhesive film (Cyanamid Corp, FM73), which was cured at 120 C for 120 min. The hinge tab on each specimen was in alignment since the direction of the applied load for the Mode I crack opening within the specimens would be dictated by this alignment. The hinge must be free to rotate so that the minimal stiffening of G IC specimen is introduced when the hinge attached to the grip. The edges of each specimen before testing were polished manually with sandpaper down to 1000 mesh to produce a flat and smooth surface. To assist the observation of the crack propagation, the edges were then coated with a thin layer of diluted white correction fluid. Such a thin coating created a good contrast between the dark crack and the white intact area of the laminates. Marks 1 mm apart were made on the white background. Test Procedure The double cantilever beam (DCB) test was performed by the transverse tensile load at a constant rate of 2 mm/min. Several loading cycles were applied to the specimens. In each cycle, the load was increased from zero until the propagation of the delamination crack was about 10 mm, and was then reduced back to zero. During each loading cycle, the fracture load at the crack growth versus displacement was recorded with the crack length. The end-notched flexure (ENF) test was performed with the aid of a three-point bending apparatus with a fixed support distance of 100 mm. The test rate was 2.0 mm/min. The crack growth in the ENF tests has been shown to be unstable without SKF and stable with SKF. The specimens for stable crack growth were loaded until the crack propagated, the machine was stopped to measure the actual crack length and unloaded. This procedure was repeated until the crack reached near the center load point. Approximately two to seven crack length values were obtained. The specimens with unstable crack growth obtained only one value of G IC because the crack generally propagated to the central loading point in the unstable manner. The Instron testing machine (model 4204) was fitted with the transverse tensile and a three-point bending apparatus. For DCB and ENF tests, the first load application with test rate of 0.5 mm/min was applied on the specimen until the crack extended from the tip of the aluminum foil. The machine was then stopped and the specimen was unloaded to zero load. This precracking procedure was carried out in order to avoid the resin-rich areas in front of the started crack. A traveling optical microscope ( 25) was used to measure the length of the crack propagation. The Mode I delamination toughness G IC was calculated by eq. (1) based on the linear elastic beam theory: G IC = 3P 2 cc (1) 2Ba where P C is the fracture load required to extend crack length, a is the crack length, B is the specimen width and C is the compliance which is the inverse slope of P C - curve (i.e. C= /P C ), where is the loadline displacement. This method and other alternatives have been reviewed and analyzed in detail by Hashemi et al. [10]. From elastic beam theory, the Mode II delamination toughness G IIC is written as follows: 9P 2 cca 2 G IIC = (2) 2B(2L 3 +3a 3 ) where P C, C, a and B are as defined for eq. (1) and L is the span length between the central loading point and the support point. RESULTS AND DISCUSSION SEM Observation In the case of composites [1], the following fracture mechanisms are of special importance: (1) matrix deformation and fracture; (2) fiber and matrix debonding; (3) fiber pull-out; (4) fiber bridging and
Carbon Fiber/Epoxy Composites / 373 FIGURE 1. Mode II fracture surfaces of carbon-fiber/epoxy composite: (a) with no modifier; (b) having SKF of 17.0 g/m 2 (fractured SKF); (c) having SKF of 17.0 g/m 2 (pull-out of interlocking SKF); (d) having SKF of 17.0 g/m 2 (fracture of SKF oriented to out-of-plane); (e) having N6P of 17.0 g/m 2 (debonding between N6P and the matrix with hackle). fracture. All of these mechanisms consume energy and contribute to the toughness of the composite. Which of them actually occur, and to what extent, during the failure of the material depends largely on the partial properties of the microstructural elements of the composite: matrix, fiber and interface, and on the geometrical arrangement and form of the reinforcement. No apparent matrix plastic deformation in this study was observed because the deformation zone was limited to the thin geometrical spacing between SKF or N6P in the matrix resin. Owing to the poor interfacial bond strength between SKF or N6P and matrix resin, it was not expected that the fracture toughness would be increased by debonding. To obtain a high crack resistance, the fracture mechanism should occur by fiber bridging and fiber
374 / Park et al. fracture that can absorb large energy than the other [8, 9]. The fracture surfaces for ENF test specimens were also examined by SEM (JEOL, model JXA-840A). Figure 1(a) shows the fracture surface of the specimen without SKF or N6P in the interlaminar region. A clean fracture surface with a few hackle structures around the continuous fibers can be seen on the micrograph of the control sample. During the G IIC test, the composite with the matrix only in the interlaminar layer causes the fracture plane to intersects the plane of the fibers [11]. The fiber plane generally restricts any further growth of the crack, resulting in a series of parallel hackles in the matrix between the fiber plane. A large hackle structure always occurs in the brittle composite system subjected to the shear loading [12]. The SEM fractrographs of the composite having SKF, Fig. 1(b) (d), show that new surfaces were created in random direction without any fixed pattern. These include the fracture of matrix, pull-out of interlocked SKF and fracture of SKF oriented to out-of-plane. SKF oriented to out-of-plane. SKF orientation would be expected to be placed at small angle or parallel to in-plane orientation. SKF that was initially placed parallel to in-plane moved by resin flow during curing, oriented to other direction. In addition, as shown in Fig. 1(c), many SKF were pulled out by another interlocking SKF underneath and debonded from the matrix, which may have acted as a bridge crossing the fracture surface during delamination. There existed no hackle structure around pulled-out and fractured SKF. SKF constrained the plastic deformation of the matrix under shear loading during G IIC test. The fibrillation and fracture of SKF oriented to out-of-plane, Fig. 1(d), were evidence of SKF s bridging. The fracture surface with N6P, Fig.1(e), showed debonding between N6P and the matrix with the slight hackle in the matrix around N6P. Effect of SKF and/or N6P on G IC The only difference between the G IC and G IIC tests is the orientation of maximum tensile stress under which microscopic failure occurs [11]. The maximum tensile stress during G IC and G IIC tests is oriented at 90 and 45 to in-plane orientation. In general, the fracture plane in any loading condition of the carbon fiber/epoxy composite is perpendicular to the maxi- FIGURE 2. Moide I crack growth resistance of carbon/epoxy having SKF in varying amounts. The different symbols represent the repeated tests: (a) 4.0 g/m 2 ; (b) 8.5 g/m 2 ; (c) 17.0 g/m 2 ; (d) 25.5 g/m 2.
Carbon Fiber/Epoxy Composites / 375 FIGURE 3. Mode II crack growth resistance of carbon/epoxy composite having SKF in varying amounts. The different symbols represent the repeated tests: (a) 4.0 g/m 2 ; (b) 8.5 g/m 2 ; (c) 17.0 g/m 2 ; (d) 25.5 g/m 2. mum tensile direction. When the fibers lay perpendicular to the maximum tensile direction, a single cleavage plane parallel to the fibers results. The river pattern and feather pattern features found on the Mode I failure surface fall into this category [12]. In this study, most of SKF in interlaminar layer were oriented between 45 and 90 to the maximum stress plane of Mode I. The effect of fiber pull-out, fiber bridging and fiber fracture was not expected. In addition, the carbon fabric used did not show any fiber bridging due to fabrication-induced fiber misalignment, since the specimen had uniform G IC values irrespective of the crack length. G IC values are low for all samples irrespective of the presence of SKF and/or N6P. The control had G IC of 0.5 kj/m 2. For composites having SKF, both the increase in fracture surface area through SKF and the more tortuous path of crack growth would suggest greater resistance to crack growth for increasing Mode I loading. A fairly large experimental scatter was noticeable on composites with SKF (Fig. 2), probably because of the difficulty encountered in fixing the random orientation of SKF during resin curing. Thus, within the experimental scatter, very little effect with the amount of SKF was observed. SKF and N6P showed lower G IC values compared with SKF alone, but had smaller experimental scatter. N6P showed no improvement in G IC owing to the poor bonding between N6P and matrix. Effect of SKF and/or N6P on G IIC For fracture mechanics except fiber bridging [8, 9], G IC and G IIC are independent of the crack length. G IIC of either SKF or SKF and N6P in this case increased with increasing the crack length. Thus, the fiber bridging is responsible for the increased G IIC. The orientation of some SKF shown in the Fig. 1(d) was placed to be parallel to the principal stresss plane of Mode II. SKF s bridging means the bridged zone of unbroken fibers inclining and interconnecting the opposite fracture surfaces behind a crack tip. Therefore, SKF s bridging requires a higher load of more energy to produce the crack propagation than the other fracture mechanisms [8, 9]. Figures 3 and 4 show that G IIC values exhibited a pronounced dependence for the specimens containing either 8.0 25.5 g/m 2 of SKF, or 8.0 17.0 g/m 2 of SKF and N6P. Typically, G IIC values were in the range of 1.0 4.0 kj/m 2. It was observed that the values commenced well around 1.0 kj/m 2 and rose sharply to the 3.0 4.0 kj/m 2 range with increasing crack propagation. It is possible that G IIC of the composite with SKF could be enhanced further if the tests were
376 / Park et al. FIGURE 4. Mode II crack growth resistance of carbon/epoxy composite having SKF and N6P in equal weight ratio in varying amounts. The different symbols represent the repeated tests: (a) 4.0 g/m 2 ; (b) 8.5 g/m 2 ; (c) 17.0 g/m 2 ; (d) 25.5 g/m 2. FIGURE 5. Mode II interlaminar fracture toughness as a function of the amount of N6P modifier. The different symbols represent the repeated tests. continued. G IIC tests could not be continued owing to the flexural mode fracture of the specimen when the crack tip passed beyond the central loading point. Slepetz and Carlson [13] found that the G IC value measured with fiber-bridged specimens showed an initial increase with crack length and then levelled off at a value more than double the value without fiber bridging. This stabilized energy level is thought to correspond to the full development of the fiberbridged zone. Thus, for this system, the leveled-up G IIC with crack length will be expected. The results with 8.5 or 17.0 g/m 2 of SKF did not indicate any difference in G IIC. G IIC of composites having 25.5 g/m 2 of SKF was lower than that with 8.5 or 17.0 g/m 2 of SKF. The out-of-plane orientation of SKF by increased amount of SKF decreased owing to the interference between SKF during resin curing. G IIC was constant with crack length for the specimens containing 4.0 g/m 2 of SKF, 4.0 or 25.5 g/m 2 of SKF and N6P. In the case of 4.0 g/m 2 of either SKF or SKF and N6P, the absolute amount of SKF oriented towards out-of-plane was too small to have significant fiber bridging. The value of 25.5 g/m 2 of SKF and N6P in equal weight ratio had relatively large volume of N6P owing to the low density of N6P (0.52 g/cm 3 ) compared with SKF (1.45 g/cm 3 ). Thus, N6P prevented the orientation of SKF to out-of-plane during resin curing. G IIC of composites having SKF and N6P was lower than that of SKF alone because N6P had a negative effect on the orientation of SKF to out-of-plane. With N6P (Fig. 5), G IIC increased
Carbon Fiber/Epoxy Composites / 377 gradually with the amount added and showed lower values compared with similar case due to the uneven distribution of N6P in the interleaf layer [14]. One possible explanation for the uneven distribution of N6P was the manual sieving method adopted in this study. The composite with SKF produced large improvement in G IIC, but small increase in G IC. Several researchers have shown G IIC to have approximately linear correlation with the residual compressive strength after impact for composites [14, 15]. Thus, the presence of SKF between carbon fiber/epoxy prepreg should correspond to a large improvement in composite impact properties. CONCLUSIONS G IC and G IIC tests conducted on composites having SKF and/or N6P between carbon fiber/epoxy prepreg led to the following conclusions: (1) The orientation and amount of SKF were important factors in determining the extent of SKF s bridging. The major mechanism for increasing G IIC with crack length was SKF s bridging. G IC showed no SKF bridging and a lower value because SKF was oriented perpendicular to the maximum tensile direction. (2) For the G IC test, a fairly large experimental scatter was noticed on composites with SKF, probably because of the random orientation of SKF during resin curing. Thus, within the experimental scatter, little effect with the amount of SKF was observed. Composites having SKF and N6P in equal weight ratio showed lower G IC values compared with those with SKF alone, but had smaller experimental scatter. N6P showed no improvement in G IC owing to the poor bonding between N6P and the matrix. (3) Within the range of SKF addition of 4.0 25.5 g/ m 2, G IIC reached as high as 3.7 kj/m 2 from the initial value of about 1.0 kj/m 2. The G IIC of composites having SKF and N6P of 8.5 or 17.0 g/m 2 was 3.1 kj/m 2 with SKF s bridging. G IIC of composites having SKF and N6P of 4.0 or 25.5 g/m 2 was 1.0 and 1.7 kj/m 2 respectively without SKF s bridging. The former was too small to have significant fiber bridging of the absolute amount of SKF oriented out-ofplane. The latter had a relatively large volume of N6P which prevented the orientation of SKF to out-of-plane. With N6P, G IIC increased slowly with the amount added. SYMBOLS AND ABBREVIATIONS a crack length B specimen width C compliance DCB double cantilever beam ENF end notched flexure G IC Mode I interlaminar fracture toughness G IIC Mode II interlaminar fracture toughness L span length between the central loading point and the support point P c the fracture load required to extend the crack length SEM scanning electron microscopy loadline displacement REFERENCES 1. K. Schulte and W. W. Stinchcomb, in Application of Fracture Mechanics to Composite Materials, Vol. 6, Elsevier Science Publishers, Amsterdam (1989). 2. R. E. Evans and J. E. Masters, in Toughened Composites, ASTM STP 937, American Society for Testing and Materials, Philadelphia, PA (1987). 3. S. Zeng, M. Hoisington and J. C. Seferis, Polym. Comp., 14, 458 (1993). 4. P. T. McGrail and S. D. Jenkins, Polymer, 34, 677 (1993). 5. C. E. Browning and H. S. Schwartz, in Composite Materials. Testing and Design (Seventh Conf.), ASTM STP 893, American Society for Testing and Materials, Philadelphia (1986). 6. T. L. Lin and B. Z. Jang, Ann. Techn. Conf. (ANTEC), 35, 1552 (1989). 7. M. S. Sohn and X. Z. Hu, Comp. Sci. Technol., 52, 439 (1994). 8. X. N. Hung and D. Hull, Comp. Sci. Technol., 35, 283 (1989). 9. X. Z. Hu and Y. W. Mai, Comp. Sci. Technol., 46, 147 (1993). 10. S. Hashemi, A. J. Kinloch and J. G. Williams, Proc. R. Soc. Lond., A427, 173 (1990). 11. J. E. Masters, in Engineering Materials Handbook, Vol. 1, Composites, ASM International, Metals Park, OH (1988). 12. H. J. Sue, R. E. Jones and E. I. Garcia-Meitin, J. Mater. Sci., 28, 6381 (1993). 13. J. M. Slepetz and L. Carlson, in Fracture of Composite, ASTM STP 593, American Society for Testing and Materials (1975). 14. N. Odagiri, H. Kishi and T. Nakae, 6th American Society for Composites, New York (1991). 15. J. E. Masters, Key Engineering Materials, 37, 317 (1989).