HIGH-TEMPERATURE CREEP OF OLIVINE SINGLE-CRYSTALS .3. MECHANICAL RESULTS FOR UNBUFFERED SAMPLES AND CREEP MECHANISMS

The dependence of steady-state creep rate on oxygen fugacity f(O2), temperature T, applied stress sigma, and loading orientation has been measured experimentally for unbuffered single crystals of San Carlos olivine. The stress exponent, approximately 3.5, is the same for all three 45-degrees loading orientations and is also identical with the stress exponent for buffered samples. When compressed along the [101]c or [011]c axis, the mechanical results for unbuffered samples are similar to those for orthopyroxene (opx) buffered samples. However, when compressed along the [110]c orientation, the creep behaviour of unbuffered samples is substantially different from that of opx-buffered samples. For unbuffered samples compressed along the [110], axis three power-law relations are required to describe the dependence of strain rate on T and f(O2) over the full range of experimental conditions. It is proposed that each power-law relation describes the creep behaviour of a different rate-controlling creep mechanism. For the three power-law relations, a deconvolution procedure yields values for the activation energy Q of 320,810 and 200 kJ mol-1 and values for the oxygen fugacity exponent m of 0.44, 0.09 and 0.04. At an oxygen fugacity of about 10(-8) atm, the values of both Q and m obtained for unbuffered samples are smaller for T > 1370-degrees-C but are larger for T < 1300-degrees-C than the values obtained for opx-buffered samples. For 1300-degrees-C < T < 1370-degrees-C, the values of these parameters for unbuffered samples are smaller but close to those for opx-buffered samples. It is proposed that the same creep mechanisms operate during the deformation of both unbuffered and opx-buffered samples at similar experimental conditions. The differences between the creep behaviour of the unbuffered and the opx-buffered [110]c samples are attributed to the dependence of creep rate on opx activity. From the creep data together with published results for the non-stoichiometry of olivine, the dependence of creep rate on opx activity has been determined. For [110]c samples and at f(O2) values in the middle of the f(O2) stability field of olivine, the opx-activity exponents are q = -1 +/- 1 and +0.8 +/- 0.7 for the rate-controlling creep mechanism dominating creep rate at high and low temperatures respectively; q less-than-or-equal-to 0 for the creep mechanism dominating the creep rate at intermediate temperatures. Likewise, the similarity between the mechanical results for the unbuffered and the opx-buffered [101]c or [011]c samples is due to the very weak opx-activity dependence of creep rate. For [101]c, q = 0 +/- 1 and 0 +/- 1 for the creep mechanisms controlling creep rate at low and high f(O2) values respectively for temperatures around 1400-degrees-C. For [011]c,q = 0 +/- 1 for the two creep mechanisms operating in the present study. For deformation experiments under unbuffered conditions, the creep behaviour of samples annealed before deformation (pre-annealed) in opx powders is different from the creep behaviour of samples pre-annealed in magnesiowustite powders. Thus the creep behaviour of unbuffered olivine is influenced by the ratio of (Mg, Fe) to silicon. This difference in creep behaviour is probably due to a change in charge neutrality conditions. Finally, by combining with the mechanical results and dislocation structure observations for opx-buffered samples, the obtained opx activity exponents and oxygen fugacity exponents were used to model the physical mechanisms which determined the rate of deformation of olivine under various experimental conditions