| The brittle-to-ductile transition (BDT) in silicon has been studied extensively over the last two decades, leading to the development of several theories to explain the observed characteristics of the transition. However to date, most of the experiments have been conducted on notched or pre-cracked samples because it is widely believed that the BDT results from some crack-tip phenomenon. We believe that the BDT occurs when a critical combination of stress and temperature results in the generation of an avalanche of dislocations in the material, permitting general yielding. In this view, a crack-tip is not necessary for the transition but it may serve as the stress concentrator that precipitates the avalanche of dislocations. To test this hypothesis, a new method of sample preparation using photolithography was developed. This process facilitated the production of a large number of uniform, dislocation-free, notch-free silicon specimens. These silicon beams were tested in bending at various elevated temperatures and strain-rates. These experiments reveal that a strain-rate dependent brittle-to-ductile transition does occur in these defect-free samples. In addition, the large number of uniform test specimens allowed the characteristics of the transition region to be probed in greater detail than previously possible. These tests reveal a deviation from the expected upper yield point behavior that coincides with the brittle-to-ductile transition temperature (BDTT) and may be related to the sudden generation of dislocations. This prompted an investigation of dislocation sources using dislocation etch-pit analysis. The etched sample surfaces show that massive dislocation activity occurs prior to yielding in the highest stressed portions of the beam. In addition, the scanning electron microscope images reveal a qualitative increase in dislocation activity above the BDTT, in agreement with the observed upper yield point behavior. Collectively, these observations necessitate a reevaluation of the theoretical treatment of the BDT. In particular, the observed massive dislocation activity associated with the yielding of the initially dislocation-free samples, must be taken into account. The Khantha-Pope-Vitek model of the BDT is proposed to address these new observations. |
