Professional Documents
Culture Documents
PII: S1359-835X(98)00076-1
A technological solution to the testing and data reduction of single bre fragmentation tests
The single bre fragmentation test was rst introduced in 1965 for measuring interfacial adhesion in bre reinforced composites. However, this test is still far from a routine test and no suitable testing standard exists. Consequently, this method is hardly used by research laboratories of the reinforcing bre manufacturers, composite fabricators and other standardisation organisations. The main reasons for the failure of the fragmentation test to establish as a standard quality control tool are the time consuming manual testing procedures and the poor condence in the data reduction techniques used for calculating the interfacial adhesion parameter. An integrated machine, software and data reduction approach has been developed for the fragmentation test. The new approach allows the stoppage of the fragmentation test at pre-dened strains and digitisation of the whole embedded bre length using a microscope attached with a video-camera and Sun work-station. Subsequently, the fragment and associated debond lengths are measured using a Visual Basic program in a semi-automatic process. The problems associated with the huge amount of computer graphics data and compatibility of different graphics formats have been sorted out. This leads to substantial savings in the time required for the fragmentation test. Using this integrated machine and software approach, the single bre fragmentation test has been carried out on carbon bre samples with well-dened interfaces. The results were analysed using the conventional KellyTyson model and the cumulative stress transfer function (CSTF) technique, at different applied strains. It has been shown that the CSTF technique can be used for assessing the bre-matrix adhesion from the single bre fragmentation test at any applied strain before or after the saturation of the fragmentation process, thus eliminating the need for specially formulated resin for the fragmentation test. 1998 Published by Elsevier Science Ltd. All rights reserved
(Keywords: single bre fragmentation test; KellyTyson model; cumulative stress transfer function)
INTRODUCTION The fragmentation test is now widely used for measuring the effect of different surface treatments on the interfacial shear strength of glass and carbon bres because of its simplicity in specimen preparation, ease of testing and wealth of information obtained in terms of damage processes18. When an external stress is applied to a single bre embedded in a matrix, the tensile stress is transferred to the bre through an interfacial shear stress. As the tensile load increases, the tensile strain in the bre eventually exceeds the failure strain of the bre, and the bre fractures. The bre continues to fracture into shorter lengths as the load increases, until the fragment becomes too short to break. This situation is dened as the saturation in the bre fragmentation process. The shortest fragment length which can break on application of stress is dened as the critical bre length l c. Because of the statistical nature of bre strength, the single bre does not break into fragments of equal size and a wide variation in length is observed. The
* Corresponding author. Present address: RAPRA, Shawbury, Shrewsbury, UK.
fragment lengths in transparent matrix composites can be measured using a conventional optical microscope, and an average interfacial shear strength at the interface, t, can be estimated on the basis of the constant shear model proposed by Kelly and Tyson in 19659. It has been widely assumed that, at saturation, all the fragments are fully debonded or the matrix has completely yielded along the interface, providing the constant shear which the analysis requires: j d (1) t fu 2lc where d is the bre diameter and j fu is the bre strength at a length equal to the critical bre length l c. Early work by Ohsawa et al.10 provided the background for the semiempirical analysis of the test-data. The critical bre length is calculated by: 4 l (2) lc 3 where l is the average fragment length. The popularity of the fragmentation technique results from the fact that the data can be collected over a large area of the interface and the process itself replicates the in situ events in an actual composite material to a certain extent.
1099
EXPERIMENTAL The use of single bre multifragmentation or, in short, the fragmentation test to measure the quality of the interface requires two individual experiments. First, the strength of a number of single bres at a xed gauge length and the Weibull modulus calculated so that the bre strength at the critical bre length can be estimated. Second, a single bre is embedded in a block of polymer matrix and subjected to a high extensional force to obtain multiple bre fractures. The material Carbon bre. HTA7 carbon bres were supplied by Toho Carbon Fibre Inc. HTA7 is a high performance carbon bre manufactured from a polyacrylonitrile precursor with high strength and standard modulus. Surface-treated epoxy compatible (EP03) sized and surface-treated unsized bres were used for the fragmentation test. The bre diameter was measured using scanning electron microscopy. Epoxy resin. MTM60 resin from Advanced Composites Group was used for the fragmentation test. It is a medium temperature curing resin whose prepregs are mainly used for the manufacture of low cost tooling. The following curing schedule was used: 60C/30 min; 80120C at 20C/hr; 90 min at 120C, followed by natural cooling to room temperature. The cured resin had a distinctive light yellow colour and is semi-transparent. This resin was found to be suitable for the fragmentation of both bres. However, it was not possible to see the bre breaks or fragment ends, even in the presence of polarised light, for bres with an extremely poor interface prepared from plasma polymerisation of hexane directly onto the carbon bre surface. Therefore, this example of an extremely poor interface could not be used for the validation of the data reduction methodologies.
1100
1101
Figure 1
A typical force vs extension curve of single carbon bre specimen obtained from the mini-tensile testing machine
Figure 2 Modied stress/strain curve, obtained from the modied force/extension curve, of the single carbon bre obtained from the mini-tensile testing machine
procedure is explained elsewhere3,20. Care was taken to ensure that the central part of the bre was not touched, since it forms part of the gauge length. The bre was secured in place by folding and gluing the opposite card with a strong adhesive. Filaments which were not properly aligned at the centre of the holes were discarded. Each specimen was examined using an optical microscope to make sure that only one bre lament was present. The specimens with a gauge length of 6.35 mm were tested under uniaxial tension in the mini-tensile testing machine at a displacement rate of 0.52 mm per min. The supporting paper for the bre lament was cut before tensile testing. The bre was carefully aligned, parallel to the loading direction and carefully monitored to ensure that the bre failed in tension; otherwise the data was discarded. Single bre fragmentation test. Single laments were extracted at random from the tow, mounted on a PTFE mould and embedded in a hot cured epoxy resin. The laments for testing were not touched and were fully aligned. The cured specimens were then ground and polished to
dimensions of 80 mm 9.8 mm 1.65 mm. Test specimens were subjected to an uniaxial tension at a displacement rate of 0.13 mm/min. The fragmentation of a single bre was continuously monitored across the specimen length. Above a chosen value of 4% strain, the test was interrupted at each percentage increment of applied strain and all the fragments in the specimen were digitised into pictures. The initial value of applied strain was chosen to be just below the known value of saturation for this system. The digitisation of the whole sample length normally took 5 min. The bre fragments, which could not be accommodated in the video screen for the digitisation of the picture (typically greater than 1.5 mm), were directly measured using the vernier calliper tted to the microscope. The test was stopped after the tensile failure of the specimen.
The calculation procedure Single bre strength measurement. The maximum likelihood method was used to obtain the Weibull modulus (m)
1102
where j 1 is the failure strength of the lament at a gauge length of l f. Single bre fragmentation test. The quality of the brematrix interfacial adhesion was estimated from two data reduction methodologies as given below: The KellyTyson approach. The critical bre length and interfacial shear strength are calculated using eqns (1) and (2). A computer programme discussed elsewhere3 was used for the calculation. The CSTF approach. According to the CSTF technique for assessing bre-matrix adhesion from a single bre fragmentation test, a better interface leads to a higher stress transfer across the interface and, consequently, a higher CSTF value. CSTF values can be calculated as follows:
iN i1 Li 0
jf (x)dx (4) Li
CSTF
iN i1
where N is the number of fragments and L i is the length of ith fragment. The tensile stress prole across the bre fragment jf (x) is a function of the distance from the bre-end, x. As it has been dened, the CSTF technique uses the plasticity effect model to calculate the tensile stress prole in the fragment21. For the elastic-elastic component of the stress transfer function Nairns variational model has been employed. The area of inuence of the bre (R/r) in the matrix was assumed to be 40, where R/r is the ratio of the matrix and bre radii15,21. Thus the calculated, tensile stress prole is a function of the bre and matrix properties, the stress state of the specimen and associated yielding and/or debonding features. The CSTF technique offers several major advantages over the KellyTyson approach and are discussed elsewhere12,17. The CSTF value is also presented as a reinforcing efciency of a continuous lament composite [jf (max) c Ef ] which is a measure of the maximum possible CSTF value. It has already been reported that the shear stress prole at the interface of a single short bre reinforced composite is very complex and the above assumption is not completely accurate15,18,21. Hence, the percentage reinforcing efciency gures are relative, but are useful for the comparison of the bre/matrix interface quality. We would like to stress that the CSTF values are much more reliable and accurate measures of the quality of the interface.
1103
Figure 3
Weibull strength and strain distributions for the treated and sized bres
Table 1 Fibre strength test data for different carbon bres used in the fragmentation test Property No. of bre diameter measurements Average bre diameter (mm) No. of bre strength measurements Average bre strength at 6.25 mm length (GPa) Average bre failure strain (%) Elastic modulus of the bre (GPa) Welbull modulus for strength/strain statistics Treated and unsized carbon bre 18 6.79 0.50 85 4.58 1.00 2.53 0.73 187 130 6.11/3.82 Treated and sized carbon bre 17 7.45 0.30 86 4.32 1.19 2.35 0.58 191 65 3.69/4.72
Table 2 Mechanical properties of the epoxy resin used for the fragmentation test Mechanical properties Tensile modulus (GPa) Tensile yield strength (MPa) Tensile cold draw strength (MPa) Tensile yield strain (%) Failure strain in tension (%) MTM60 Epoxy resin 3.21 84.1 84.1 7 7.7
Table 3 Material properties used for the calculation of CSTF value Material property Transverse modulus of bre (GPa) Longitudinal Poissons ratio of bre Transverse Poissons ratio of bre Longitudinal shear modulus of bre (GPa) Friction coefcient Fibre volume fraction Longitudinal thermal expansion coefcient of bre (K 1) Transverse thermal expansion coefcient of bre (K 1) Thermal expansion coefcient of matrix (K 1) value 14 0.20 0.25 20 0.2 0.001 0.1 10 6 18 10 6 4O 10 6
stress curve resumed its original trend (Figure 5). The other bre and matrix properties used for the calculations are given in Table 3. Treated sized bre. The single bre fragmentation test results for treated sized bre are shown in Table 4. During fragmentation of treated sized carbon bres, it was observed that the rst bre failures occurred at an applied strain greater than 2.5%. There was very few bre fragments smaller than 1.5 mm at applied strains of less than 3%. The rst detailed measurements of bre fragment and debond lengths were carried out at an applied strain of 4%, because there were few fragments of length 1.5 mm. Additionally, above this strain the fragment length distribution became narrower, rapidly. Therefore, as the applied strain increased, the number of bre fragments increased and the average fragment length decreased (Figure 6). A true measure of the fragment saturation is the number of fragments per unit length of the test specimen. We have assumed a unit length of 10 mm for
this purpose. It can be seen from Table 4 and Figure 7 that the number of fragments per 10 mm reaches a plateau at 5% applied strain. Hence, it can be concluded that saturation in the fragmentation process occurs at 5% applied strain. Interfacial debonding was observed during the fragmentation test at applied strains of 5%. Figure 8 shows that the percentage of debonded length increases with applied strain. However, complete debonding at higher applied strains could not be established because the failure strain of the resin was insufcient. Treated unsized bre. A fragmentation prole, similar to treated sized bres, was observed for the treated unsized bre (Table 5). However, in the case of the treated unsized bre, the onset of fragmentation began at a lower applied
1104
Figure 4
Weibull strength and strain distributions for the treated and unsized bres
Figure 5
Tensile stress/strain curve of the epoxy resin (a) during mechanical testing of the resin and (b) during the fragmentation test
strain. Consequently more fragments were observed at an applied strain of 3% in comparison to treated sized bres. In addition, as with the fragmentation test on treated sized bres, the distribution in fragment lengths became narrower as the applied strain increased. However, at a particular applied strain, the distribution was broader for the treated unsized bres. This is very difcult to explain, since the Weibull modulus for the treated unsized bres is higher than that for the treated sized bres and was expected to have caused a narrower distribution. Two possible explanations for this may be: (1) the presence of sizing causes the bres in a tow to adhere together. When these tows are used for the selection of a bre for the fragmentation test, bres of low strength cannot be removed for testing. This would result in an apparently narrower strength distribution in comparison to that for the treated unsized bres, which are loosely bound and readily removed from the tow for testing; (2) the presence of sizing leads to the modication and/or the creation of an interphase. This interphase provides a smoother stress transfer across the interface.
It can be seen from Tables 4 and 5 that, at a 7% applied strain, although the average fragment length is the same in the two cases, 20% more interfacial debonding was observed with the treated unsized bres. This point is further strengthened when the average fragment aspect ratio is compared in the two cases. The average fragment aspect ratio of treated sized bres is lower than that of the treated unsized bres. It should be pointed out that the two bres nominally come from same supplier, but will have been produced on differing lines, which accounts for the reported variation in bre diameter. It is therefore coincidental that the average fragment lengths are similar and it can be concluded that the bre-matrix interface in the case of the treated sized bre is better than that in the case of the treated unsized bre.
DISCUSSION The single bre fragmentation test results for treated sized
1105
Figure 6
Average fragment length at different applied strains obtained during the fragmentation of single bre composites with different interfaces
Figure 7 Number of fragments per 10 mm at different applied strains obtained during the fragmentation of single bre composites with different interfaces
Figure 8
Percentage debonding at different applied strains obtained during the fragmentation of single bre composites with different interfaces
and treated unsized bres were used to calculate a brematrix interface parameter, using KellyTyson and CSTF methodologies, at different applied strains. The results are shown in Tables 4 and 5, respectively. The details of the material properties used for the CSTF calculation are given in Tables 13. Fragments longer than 1.5 mm (aspect ratio 200) were
not included in the calculation of CSTF values because they will behave essentially as continuous bres. It can be seen from Table 4 that the critical bre length decreases with applied strain, while the bre strength at the critical bre length increases. The interfacial shear strength and CSTF values for the treated sized bres are given in Table 4. It can be seen from
1106
Calculated results Critical bre length (mm) Fibre strength at critical bre length (GPa) KellyTyson ISS (MPa) SD in KellyTyson ISS (MPa) CSTF value (MPa) Reinforcing efciency (%)
Table 5 Fragmentation test results for treated and unsized carbon bre at different applied strains Applied strain (%) Shortest bre (mm) Longest bre (mm) No. of bres longer than 1.25 mm No. of specimens No. of fragments Average fragment length (mm) SD in average fragment length (mm) Average fragment aspect ratio Total length of fragments (mm) No. of fragments per 10 mm Average debonding length (mm) SD in average debonding length (mm) % debonding Calculated results Critical bre length (mm) Fibre strength at critical bre length (GPa) KellyTyson ISS (MPa) SD in KellyTyson ISS (MPa) CSTF value (MPa) Reinforcing efciency (%) 4 0.12 1.45 19 5 386 0.46 0.24 67.75 177.37 21.76 2.44 l0 2 4.79 l0 2 5.31 5 0.09 1.42 6 4 421 0.42 0.18 61.86 174.84 24.08 9.40 10 2 9.87 10 2 22.64 6 0.09 0.72 1 4 508 0.37 0.13 54.49 187.02 27.16 0.13 0.10 35.12 7 0.10 0.73 0 3 362 0.38 0.13 55.96 137.50 26.33 0.19 0.12 49.42
Figure 9 that the value of interfacial shear strength increases with applied strain. This is expected because the major assumption in the calculation of interfacial shear strength from the KellyTyson model is that a better interface leads to shorter fragment lengths. The effect of partial interfacial debonding is not included in the calculations. In addition, the interfacial shear strength calculated from the data occurred at applied strains of 5%, so have little value because saturation in the bre fragmentation has not been achieved. On the other hand, the CSTF value decreases with an increase in applied strain (Figure 10). This is in accordance with the assumptions of CSTF, because it assumes that short bres are subject to lower levels of stress than long bres. Interfacial debonding causes a further
reduction in the CSTF value. The fragmentation test results for the treated unsized bres show a similar trend. The treated sized bres hava a higher interfacial shear strength than the treated unsized bres at similar applied strains. This prediction of interfacial quality by the Kelly Tyson model is acceptable, since the treated sized bres showed lower degrees of interfacial debonding, although the fragment length remained approximately the same. However, the interfacial shear strength for the treated sized bres, which is in the region of 60 MPa is considerably higher than the shear yield strength of the matrix (48.55 MPa) estimated from von Mises yield criterion, ignoring the effect of hydrostatic pressure on the epoxy resin. Since interfacial debonding was observed during the
1107
Figure 9 Interfacial shear strength values at different applied strains obtained during the fragmentation of single bre composites with different interfaces
Figure 10 Cumulative stress transfer function values at different applied strains obtained during the fragmentation of single bre composites with different interfaces
fragmentation test, the interfacial shear strength should have been considerably lower than the shear yield strength of the matrix. This is one of the major limitations of interfacial shear strength calculated from the KellyTyson model and has been highlighted in several other studies12,15,18,21. In addition, a difference of 15 MPa between the values of interfacial shear strengths for the treated unsized and treated sized bres is rather large, especially when both the samples showed considerable interfacial debonding, differing by only 8%. Additionally, it can be seen from Figures 9 and 10 that the interfacial shear strengths and CSTF values for the treated sized bres are higher than those for the treated unsized bres. These results agree well with the fracture information obtained from the fragmentation test. The differences between the two CSTF values at the particular applied strain for the two cases is relatively small, since both of the interfaces exhibited similar degrees of interfacial debonding. Hence, it can be concluded that the CSTF
technique provides a more reliable prediction of interfacial adhesion, showing the same trends irrespective of the applied strain and whether saturation had been achieved. This particular observation paves the way for the prediction of the quality of the interface at an arbitrary applied strain, provided a sufcient number of bre fragments are obtained during the fragmentation test at that particular applied strain. The CSTF approach attempts to differentiate between good interfaces, where matrix yielding dominates and poor interfaces where debonding and frictional stress transfer is dominant, thereby allowing the true effect of surface modication to be studied.
CONCLUSIONS An attempt has been made to overcome the major limitations of the fragmentation test by addressing relevant material, machine, software, experimental and data analysis
1108
5. 6. 7.
8. 9. 10.
11.
12. 13.
ACKNOWLEDGEMENTS This work was carried out under the framework of an Engineering and Physical Sciences Research Council (EPSRC) project on the Quantication of Fibre-Matrix Adhesion for Correlation of Polymer Composite Performance, via grant no. GR/K/81881. We gratefully acknowledge the nancial assistance from the Department of Trade and Industry Link project for the purchase of the machine, and Advanced Composites Group Ltd for the supply of material. Thanks to Dr Helen Atkinson for the use of Sun work-station and Dr S. Hayes for advice.
14. 15.
16. 17.
18.
REFERENCES
19. 1. 2. 3. 4. Drzal, L.T., Adhesion of graphite bres to epoxy matrices: II. The effect of bre nish. Journal of Adhesion, 1983, 16, 133152. Cheng, T.H., Jones, F.R. and Wang, D., Effect of bre conditioning on the interfacial shear strength of glass-bre composites. Composites Science and Technology, 1993, 48, 8996. Cheng, T.H., Zhang, J., Yumitori, S., Jones, F.R. and Anderson, C.W., Sizing resin structure and interphase formation in carbon bre. Composites, 1994, 25, 661670. Yumitori, S., Wang, D. and Jones, F.R., The role of sizing resins in 20.
21.
1109