Page 238 - Fiber Fracture

STRENGTH AND FRACTURE OF METALLIC FILAMENTS 223

had on both sides an additional layer of 0.05 pm Ti. Micro-lithography was used to
pattern the tensile samples with a gauge section of 200 pm width and 600 wm length.
The silicon waver was removed by etching for 4 h at 100°C. This also determined
the final annealing state. The average grain size was then found to be 0.98 pm. His
tensile specimens have therefore only one or a few grains through the thickness. The
extremely delicate fatigue tests were performed with specially designed equipment in
the load-controlled tension-tension mode with a frequency of 1/15 Hz Read (1998b).
The dispersion on the fatigue life is much larger than for the foils of Hong and Weil. but
of the same order (1 decade) as for measurements on wires and the other foils shown in
Fig. 39 (see also Table 6). The fatigue resistance for these micro-specimens is quite high
and comparable to as-drawn wires and eletrodeposited films. Previous measurements on
similar AI foils (Read and Dally, 1995), however, gave fatigue lives that were 2 to 3
decades below those of AI sheets (Forrest, 1966). From his TEM studies Read confirms
the observation of Hong and Weil that cyclic deformation does not generate cell walls in
small grains and concludes that dislocations move individually and escape through the
surface.
By returning once more to Fig. 39 we conclude that most curves have features
similar to another, but taken all together there is no clearly visible trend. These curves
have been obtained on very different grain sizes, sample thicknesses and grain size
to thickness ratios, but do not show a distinct trend whether one of these parameters
decisively governs fatigue life in microscopic samples. Of course Cu can be prepared in
a broad range of tensile strengths. From over 500 MPa down to values that are difficult
to specify, especially when necking, as is usual in fatigue tests on microscopic samples.
is not taken into account (Le. engineering rather than effective stress is used). Obviously
the initial tensile strength is a decisive parameter for the duration of the fatigue life,
particularly in the low-cycle regime, but in the high-cycle range many crossovers can be
observed. It finally appears that many other factors, such as texture, crystalline defect
density and surface defects, play an equally important role.
Nevertheless, the question of size effects in fatigue that are related to the proximity
of the surface, where dislocations can escape, is of scientific intcrcst. In view of the
diversity of the results presented on small-grained Cu, such studies have to be camed
out with coarse-grained samples prepared from the same base material. In the following
section we will present results that were obtained from studies on thin Cu wires and
ribbons.

Size Efect in the Fatigue Behavior of Thin Cu Wires and Foils

Results of fatigue tests on micro-Cu wires with bamboo structures which is the
coarsest grain size possible in wires are given in Fig. 40. The S-N curves obtained
on 30, SO and 95 km diameter wires show that the fatigue life at stress amplitudes
below half the tensile strength of thin wire is 2 to 3 decades longer than for the
thick ones. These wires have been annealed for 2 h at 600°C and are therefore highly
ductile (for their mechanical properties see Table 5). Nevertheless, the thick wire
that fails first, shows a brittle transgranular fracture above IO4 cycles (Fig. 35 right),
whereas the thin wire always shows severe necking, even at the highest cycles measured