Page 156 - Fiber Fracture

STRENGTH OF GLASS FIBERS 141

foreign particles do not stick. Foreign particles stick to the fiber surface when the impact
occurs above the glass transition temperature. It is clear that fractography - if it can be
used - remains a powerful tool to identify the nature of the fracture initiating flaws.
In summary, much of the work in the past has used small-length samples and has
been focused on the extrinsic strength controlled by small flaws. Only recently, efforts
are being made to study the low strengths of long-length fibers. The following issues
about the extrinsic inert strength are not well understood:
(I) the identity and location of the fracture initiating flaws,
(2) the role of residual stresses around these flaws (especially inclusions and indents).
and
(3) the role of testing stress, humidity, and temperature in the nucleation of microcracks
around these flaws.

Extrinsic Fatigue Strength

In spite of an intuitive feel that flaws such as inclusions should not behave like cracks,
the fatigue behavior (dynamic fatigue, delayed failure, dependence on the moisture and
temperature of the test environment) of extrinsic flaws is surprisingly well modeled
by the assumption of pre-existing cracks and the theory of slow crack growth. One
of the major issues that has received a lot of attention (in the past and continues to
do so at present) is the precise nature of the equation for crack velocity during slow
crack growth, i.e., power function, the exponential function, or some other function
(Jakus et al., 1981; Gupta et al., 1994). This is important when the estimates for the
times for delayed failure need to be extrapolated to very low levels of stresses; the
different laws tend to diverge significantly (Kurkjian and Inniss, 1992). There is, at
present, strong experimental evidence (Michalske et al., 1991) and theoretical basis
(Wiederhorn et al., 1980) to support the exponential law. However, its consequences are
difficult to analyze mathematically. The power law, therefore, remains popular because
it is convenient to use. When using the power law, it is found that the stress corrosion
susceptibility, N, is not a constant for a composition. N has been found to increase with
decrease in temperature (Hibino et al., 1984). This is shown in Fig. 4 for silica fibers.
This is expected on the basis of Q. 11 and supports the validity of the exponential
equation. N is also found to increase with increasing strength (Kurkjian and Inniss,
1992). Armstrong et al. (1997) have reported a decrease in N with increase in the
humidity of the environment. However, Muraoka et al. (1996) measured crack growth
rates directly in silica fibers and did not observe a significant change in N with relative
humidity. Experiments can only provide empirical laws whose validities will necessarily
be limited. Atomistic simulations of slow crack growth are not mature enough at present
to be of use in applications (Fuller et al., 1980; Marder, 1996). This is an area where
much work remains to be done.
Artificial cracks generated by other techniques such as indentation or chemically
generated pits (Donaghy and Dabbs, 1988; Choi et al., 1990) or by abrasion have been
used to model the behavior of real cracks (Sakaguchi and Hibino, 1984; Inniss et al.,
1993). While this work is of much interest in itself, it has not yielded any new insights
as far as slow crack growth is concerned. This is probably because these artificial cracks