matrix interface strength using the microbond test

Composites Science and Technology 57 (1997) 991-994 0 1597 Elsevier Science Limited Printed in Northern Ireland. All rights reserved PII:

SO266-3538(96)00146-7

0266-3538/97/$17.00

A STUDY OF THE EFFECTS OF TIME AND TEMPERATURE ON THE FIBER/MATRIX INTERFACE STRENGTH USING THE MICROBOND TEST

Alexander Fiber Science Program,

Department

Straub, Michael Slivka & Peter Schwartz* of Textiles and Apparel,

Cornell

University,

Ithaca, New York 14853-4401,

USA

(Received 17 June 1996; revised 4 September 1996; accepted 1 October 1996) proportions ranging from 100 % DER331 (E = 1620 MPa) to 50 % DER331/50 % DER732 (E = 538 MPa). SFC tests were conducted at two very low strain rates-O.004 and 0.0007 min-‘. At the higher strain rate, the IFSS increased from 48 to 66MPa, as the DER732 weight fraction was increased from 0 to 0*4 (E = 678 MPa), and dropped to 42 MPa upon increase of the DER732 content to 50 %. Also, the IFSS was relatively strain rate insensitive (at the two low strain rates used) when using neat DER331, but decreased with decreasing strain rate as the flexibilizer was added. Rao and Drza12 also studied the effect of matrix stiffness on IFSS of AS-4-graphite-fiber/epoxy composites by means of the SFC test. Two types of epoxies were used for the matrix, a DGEBA-based epoxy and a trifunctional epoxy (MY720), and the stiffness was varied by using different types of hardeners. They found that as E for the DGEBAbased epoxy increased from 669 MPa to 3300 MPa (Tg from 18 to 130“(Z), the IFSS increased from 39 to 75 MPa), and as E for the trifunctional epoxy increased from 451 to 1080 MPa (Tg from 22 to 4O”C), the IFSS increased from 41 to 54 MPa; the strain rate used was not reported. Their results are the opposite of those reported earlier by Netravali et al.’ Piggot and Wang, using a single-fiber pull-out technique at a testing rate of 0.5 mm/min, studied the effects of temperature on IFSS using single AS-4 fibers embedded in DGEBA-based epoxy (EPON815). As the test temperature increased from 20 to 8O”C, the IFSS decreased from 56 to 44 MPa, but at llO”C, the IFSS jumped back up to 60MPa. Approximating a Tg of 120°C (curing temperature +20”(Z), as the absolute difference between the testing temperature and Tg increased, the IFSS decreased for the lower range of testing temperatures, consistent with the observations of Rao and Drzal.2 Piggot and Wang3 also measured the ultimate shear strength of the unreinforced DGEBA epoxy by the Iopescu method at a rate of 1 mm min-’ . They found that the shear strength varied inversely with ambient temperature, consistent with the results of Netravali et al.’ A comparison of these

Abstract With samples of p-aramid fibers embedded in a DGEBA-based epoxy (T, =llO’C), a series of experiments were conducted by the microbond method to determine the effect of testing rate and temperature on the fiber/matrix interfacial shear strength (IFSS). Microbond tests were conducted ouer three orders of magnitude of testing rate (0.1-100 mm min -‘) and a range of temperatures (21-130°C). While the general trend observed is that the IFSS decreased with increasing testing rate, the effect was more pronounced below Tg than near or above Tg, where the IFSS seemed to be insensitive to testing rate. We have analyzed these data by using an exponential breakdown model proposed by Zhurkov, enabling the determination of an activation energy for the failure process. 0 1997 Elsevier Science Limited Keywords:

IFSS,

viscoelastic,

microbond,

glass

transition 1 INTRODUCTION The strength of the interface between the reinforcing fibers and the matrix in a composite is a function of the surface chemistry and topography of the fiber and the chemistry of the matrix. For viscoelastic matrices, it is also dependent upon the rate of loading or, equivalently, the ambient temperature. Furthermore, because the Young’s modulus changes considerably as the temperature varies through the glass transition temperature, Tg, it is expected that the strength of the interface will change with the Tg, and likewise, with the stiffness of the matrix. Netravali et al.’ studied the effect of the matrix stiffness and strain rate on the interfacial shear strength (IFSS) of IM-6-graphite-fiber/epoxy composites by means of the single-fiber composite (SFC) test. The Young’s modulus, E, of the epoxy was varied by adding a polyglycol diepoxide flexibilizer (DER 732) to a stiff DGEBA-based epoxy (DER331) in * To whom correspondence

should be addressed. 991

992

A. Straub et al.

results with those found in our experiments will be discussed later. In this paper, we present the results of experiments to study both rate and temperature effects on IFSS of Kevlar-49/DGEBA-based-epoxy composites. The microbond test4 was chosen over the SFC method because the IFSS could be measured directly and high testing rates could be used. and over traditional pull-o$ tests so meniscus effects were reduced. While in theory the interface shear stress is independent of the bead size, in reality a size effect has been noted: very small beads yield higher shear stresses. 2 EXPERIMENTAL 2.1 Materials Para-aramid fibers (Kevlar 49) were used because they are relatively strain rate insensitive over a wide range of loading rates,” and their mechanical properties have been widely reported. The epoxy was DER331 (DGEBA) cured with a tetraethylene pentamine (TEPA) hardener, DEH26, both supplied by Dow Chemical Co. DGEBA-based epoxies are extensively used in high-performance, fiber-reinforced composites, and were a natural choice for this work. 2.2 Sample preparation The resin and hardener were mixed in stoichiometric proportions recommended by the manufacturer, and degassed under a vacuum. Epoxy droplets (80200 pm) were placed on single fibers by attaching a short carbon fiber to a wooden stick, then dipping the tip of the carbon fiber into the epoxy, leaving a small bead on the end, then touching the bead to the Kevlar fiber (past work has indicated that, in the range given above, the IFSS is relatively insensitive to bead size,” and our results showed no bead size effects within the experimental error). At the same time, unreinforced resin from the batch was used to make samples for thermal analysis. Samples were then cured at 100°C for 1 h. 2.3 Thermal analysis A Perkin-Elmer DSC4 was used to measure the Tg of the cured epoxy. Five, 5 mg epoxy samples were scanned at a rate of 40”C/min. 2.4 Mechanical testing Mechanical testing was carried out in an Instron Model 1122 tensile tester fitted with an Instron temperature chamber, Model A74. The ambient temperature in the chamber was measured with a therniocouple connected to an Omega Model 199 digital temperature recorder, and mounted inside the chamber near the bottom jaw. The configuration used was typical of microbond testing procedures reported throughout the literature.6 Testing was carried out at

21, 50, 90, and 13o”C, over three magnitudes of cross-head speeds, 0.1, 2, 10, and lOOmm/min. For tests conducted at elevated temperatures, the samples were allowed to equilibrate for at least 1 min prior to starting the test. 3 RESULTS

AND DISCUSSION

The thermal analysis yielded an average Tg of 110°C for the five samples of cured resin. The Tg for each was determined by the instrument software, and was the midpoint of the range from onset to completion of the transition. The results of the microbond tests are given in Table 1, and illustrated in Fig. 1. Assuming that the load is uniformly distributed along the fiber embedded length, the IFSS is calculated as the maximum average interfacial shear stress at failure (7) using ,=L

ndl

(1)

where F is the maximum test load, d is the fiber diameter, and 1 the embedded length, both measured using an optical microscope. As seen from Table 1 and Fig. 1, at slow and moderate testing rates, the IFSS clearly decreased as the ambient temperature increased. Piggot and Wang’ described this effect as being consistent with the increased molecular mobility in the resin as one approaches the Tg. Our data followed a trend essentially similar to that for carbon fiber/DGEBA systems reported by Rao and Drzal,’ as illustrated in Fig. 2. The results of Netravali et al.’ and Piggot and Wang3 are also shown in Fig. 2. On looking at the same data with the testing rate as the independent variable, two different regimes were apparent. For T < T,, the IFSS decreased with increasing testing rate; for T = Tg and T > Tg the IFSS was relatively insensitive to the testing rate. Our results are opposite to those reported by Netravali et al.,’ and this may be due to the fact that they used much slower testing rates and a different test configuration. For the regime T < Tg, Piggot’ argued that because the measured epoxy modulus increased with increas-

Table 1. Microbond test results for interface strength, in MPa, for Kevlar-49/DGEBA-based epoxy samples over a range of testing rates and ambient temperatures”

21°C 50°C 90°C 130°C ~ 0.1 mm min-’ 18.5 (7.9) 21.3 (10.5) 13.5 (6.2) 9.0 (5-4) 2mmmin-’ 20.0 (8.1) 18.8 (7.8) 10.0 (4.7) 4.8 (2.3) 10 mm min.-’ 15.8 (6.3) 17.3 (7.5) 10.8 (7.0) 4.8 (2.9) 100mm min-’ 9.0 (3.6) 9.8 (4.2) 7.1 (2.8) 5.1 (3.4) “ Values in parentheses represent standard deviations in MPa.

Fiber/matrix

i‘nterface strengths

993

regardless of rate, no more energy could be absorbed. Thus the interface strength should be relatively insensitive to rate in this (T = Tg) regime, as seen in our experimental data. In Fig. 3, using semi-logarithmic coordinates, IFSS was plotted versus testing rate. Taking into account experimental error, the approximately linear trends in the data suggested that the strength of the interface, r followed a logarithmic rule as a function of testing rate (u) r = Aln(u) + B (2) with A and B constants. Considering the droplet embedded length, 1, v may be replaced by a pseudo-failure time t = llv, so eqn (2) may be recast as a function of t z = &n(t) + p (3)

5

0 0:1

lb

i

Noting this relationship, and motivated by Zhurkov’s work,’ the time to failure is assumed to follow an exponential process of the form

100

Testing rate [mm/min] Fig. 1. Interface strength as a function of temperature: indicate the standard error.

bars

ing rate, at high rates, the interface could absorb very little energy prior to failure, and the interface strength was depressed. Near and above Tg, the energy absorbing capacity was near its maximum and,

where @ is a constant, U. is the thermodynamic interaction energy, V, is an activation volume, related to but not exactly the free volume required by segmental mobility, that depends on the absolute temperature, T, and k is Boltzmann’s constant. Solving for r as a function of In (t) yields r = $ln(t)

+

T

U. - kTln(@)

(5)

VT

20-

T-UK1 +-.

K4artDGEBA.

-t-

Mii

IME GnphiiESA.

SFC Test (Netmvali ei al. Ill) SFC Tesl (Rae and Drzal PI)

- --c

-

AS4 GrsphiiGEBA,

- -e

-

AS4 GlaphitaMY720,

-----c--

Test (2mmlmin)

AS4 GRphiiESA.

SFC Tea Pulcwt

(Rm and Drzal PI) Test (Pii

and Wang 131) J

Fig. 2. Composite plot illustrating the experimental results along with those of Netravali et al.,’ Rao and Drzal,* and Piggot and Wang.3 (Note: for Refs 1 and 3 Tg was inferred from the curing conditions and the authors’ observations.)

01 0.01

0.1

1

10

100

II

0

Testing rate [mm/min]

Fig. 3. Semi-logarithmic

plot of IFSS as function rate.

of testing

A. Straub

994

Table 2. Calculated values for the constants in the exponential breakdown model desaibed by eqn (3) (intrinsic activation energy U, = 8% x lo-” J; time constant @ = 8.30 x lo-' s)

Ambient temperature (“C)

Activation volume, V, ( X lbz’ m3)

R2

21 50 90 130

3.09 3.17 5.78 10-9

0.79 0.73 0.65 0.69

Transforming of t using the

the experimental data into functions average I of the data set used to

determine each, linear least-squares (LLS) regression techniques were used to determine, for each of the four testing temperatures, the best linear fit of eqn (3) to the data. The regression lines for 50 and 90°C were then slightly adjusted (within the error in the experimental data) to force them to intersect at the natural intersection point of the 21 and 130°C regression lines. Using eqn (5) with the LLS regression equations, the unknowns 4, UC,,and V, were calculated, and these are presented in Table 2 along with the value of R2 for each of the LLS regression equations. The intrinsic energy for debonding was determined to be 8.98 X 10p21J, and was the same for any choice of regression lines used. The activation volume, as increased with increasing temperature, expected, consistent with an increase in free volume necessary

et al. for segmental motion as the temperature is increased. In Fig. 4, using the values given in Table 2, the fit of the actual data to the exponential model is presented. 4 CONCLUSIONS In the above the results of microbond experiments at four different temperatures-21, 50, 90 and 13O”Cand testing rates covering four orders of magnitude0.1, 2, 10 and 100 mm min-‘-were presented. Below T,, the trends compared favorably to other reported experiments in the literature, the interface strength decreased with increased temperature. For T 1 T,, there was very little noticeable rate effect, consistent with the maximum energy absorption at this temperature. The effect of testing rate observed in our tests was opposite to that reported by others Using an exponential energy argument, an intrinsic activation energy of 8.98 X lo-*’ J was calculated for

interface failure between the p-aramid fiber and the DGEBA-based epoxy used in these experiments. ACKNOWLEDGEMENTS The authors wish to acknowledge Matthew Fred, a student in Materials Science and Engineering, for his tireless efforts in producing many of the microbond samples. We also wish to acknowledge support from the Cornell University Agricultural Experiment Station Grant NYS329-424 and the College of Human Ecology. Partial tuition relief for Alexander Straub was provided by the College of Engineering. REFERENCES

25

20

7

-9.

21 ‘C

-

50 ‘C

-e-

QO ‘C

-*

130%

I

z

$

I5

b

1. Netravali,

A. N.. Henstenburg, R. B., Phoenix, S. L. and Schwartz, P. Interfacial shear strength studies using the single-filament-composite test. I: Experiments on graphite fibers in epoxy. Polym. Compos., 1989, 10,226. 2. Rao, V. and Drzal, L. T. The dependence of interfacial shear strength on matrix and interphase properties. Polym. Compos., 1991,X2,48. 3. Piggot, M. R. and Wang, Z. N., Relations between _ polymer and fibre-polymer interface properties. Proc. American

8 f

lo

4.

5

.5.

r

6. 0 o.oooo1

0.0001

0.001

0.01

0.1

10

1

7.

1. Failure time [set]

Fig. 4. Plot of IFSS versus failure time illustrating

the

exponential

breakdown model, experimental data.

eqn

the fit of (4), to the

8.

Society for

Composites

6th Technol.

ConJ:

Technomic Publishing, Lancaster, PA, 1991. Miller, B., Muri, P. and Rebenfeld, L. A microbond method for determination of the shear strength of a fiber/resin interface. Conlpos. Sci. Technol., 1987, 28, 17. Wu, H. F., Phoenix, S. L. and Schwartz, P. Temperature dependence of lifetime statistics for single Kevlar 49 filaments in creep-rupture. J. Mater. Sci., 1988, 23, 1851. Ktipper, K. and Schwartz, P. Modification of fiber-matrix interface of p-aramid fibers using gas plasmas. J. Adhesion Sci. Technol., 1991, 5, 165. Piggot, M. R. Failure processes in the fibre-polymer interphase. Compos. Sci. Technol., 1991, 42, 57. Zhurkov, S. N. Kinetic concept of the strength of solids. ht. J. Fract. Mech., 1965, 1, 311.