The strength and fatigue behavior of bulk fused silica is well understood in terms of the growth of microcracks under the combined influence of stress and environmental attack. The behavior of high strength, flaw free silica optical fiber shows significant differences from the bulk material for poorly understood reasons making long term predictions unreliable. It is known that silica fiber strength and fatigue are sensitive to such environmental parameters as temperature, humidity and pH. However, this paper presents results which also show a sensitivity to ionic species in the environment. These results are interpreted in terms of possible models for the fiber behavior.
INFRODUCTIONIn most ceramic materials and, in particular, in oxide glasses, failure can occur after prolonged application of a constant stress which is significantly lower than the stress required to cause catastrophic failure on a short time scale. This phenomenon, called fatigue or delayed failure, can determine the long term reliability of ceramic components. A complete understanding of the mechanisms that produce delayed failure would allow one to make reliable lifetime predictions for components (in our case, optical fiber) under operating conditions.The mechanism that causes fatigue in most ceramic materials consists of slow growth of pre-existing cracks up to the critical size for fast crack growth. The slow growth is due to stress-enhanced environmental attack of the strained bonds at the crack tip. The empirical relationship between the rate of growth of a macroscopic crack, dc/dt, and the applied stress intensity factor, Kj, may be described by a power 1 = AK1.( 1) where n is the stress corrosion susceptibility parameter. By combining it with the Griffith relation, K1 = aYc1'2 , (2) and integrating, two equations can be obtained:relates the time to failure, tj, to a constant applied stress, oa, (static fatigue) while on+1 = 2&[a]fl2. (4) relates the strength, a, to the loading rate, do/dt (dynamic fatigue). It is noted that dynamic fatigue experiments can be used to determine the unknown parameters in Eq. (3) and hence predict time to failure under static conditions.The slow crack growth model provides a good explanation of fatigue behavior for crack containing bulk glass. By considering the published data for high strength silica fiber, however, some discrepancies are noted. The value of n derived from Eq. (3) and (4) for weak, bulk silica is around 40, in good agreement with crack velocity measurements made