The enstatite of (Mg, Fe) SiO₃ is the second most important phase in the Earth's upper mantle. Experimental study of the kinetic behavior of enstatite at upper mantle conductions has showed that the enstatite transformed to β (or γ) phase plus stishovite at 20 GPa and transformed to ilmenite structure only at pressure greater than 21 GPa^[1]. This implies that the metastable field of (Mg, Fe) SiO₃ enstatite could approach greater depths than 410 km discontinuity in transition zones. Thus, knowledge of enstatite stability-fields and phase transitions are essential for the determination of the stable mineral assemblages in the upper mantle.

It is well known that the electrical conductivity is a very useful physical property to provide much information about mineral composition, temperature, iron content and oxygen fugacity of the Earth’s deep interior^[2~7]. On the other hand, it is also useful for us to explain the available results of geo-electromagnetic sounding, as well as the partial melting and dehydration. Therefore, laboratory measurements of the electrical conductivity of minerals and rocks are significant in studying the electrical conductivity distribution through the upper mantle even trarusition zone because the electrical conductivity is sensitive to composition and temperature. In the past decades, numerous studies have been conducted for the electrical conductivity of (Mg, Fe) SiO₃ enstatite and its polymorph^[8~11]. Duba^[8] and Dvorak^[9] measured the electrical conductivity of pyroxene and enstatite at ambient pressure, and reported the electrical conductivity dependence on oxygen fugacity. Xu et al.^[8] measured the electrical conductivity of orthopyroxene and its high pressure phase at pressure up to 21 GPa and temperature of 1000~1400℃, showed quite good consistency in comparison their results with previous experimental data. Dai et al.^[11] measured the electrical conductivity of pyroxenite at pressure up to 4 GPa and 800~1150℃ under different oxygen fugacities, and reported a positive activation volume and targe activation energy. However, there is no reported in the electrical conductivity of (Mg, Fe) SiO₃ enstatite simultaneously at high pressure and high temperature of equivalent transition zone conditions, furthermore most previous investigators never considered the influence of water on the electrical conductivity of enstatite.

In the present study, we measured the electrical conductivity of (Mg_0.9Fe_0.1) SiO₃ enstatite at pressures of 10~20 GPa and temperatures of 750~1600 K using a KAWAI-type multianvil high-pressure apparatus, at the same time, the water content was determined after the electrical conductivity measurement. In light of the new experimental results, we also discuss the conduction mechanism of enstatite.

The starting material was enstatite with a (Mg_0.9Fe_0.1) SiO₃ composition, which was the same as that used in our previous measurement of the electrical conductivity of silicate perovskite^[12]. The oxygen partial pressure was controlled by a H₂ and CO₂ mixed gas with a ratio of 1: 1, and the oxygen fugacity lies between the IW buffer and the WM buffers.

The high-pressure and high-temperature experiments were performed using a KAWAI type multianvil high-pressure apparatus USSA-5000 at the Institute of the Earth's Interior (ISEI), Okayama University. Fig. 1 shows a schematic cross-section of the sample assembly for the electrical conductivity measurements. A MgO + 5% Cr₂O₃ octahedral pressure medium of 10 mm edge length was used and a 4.0 mm diameter hole was drilled in it to accommodate a heater of Re two-foil with 4.20×10.02 mm width and 0.025 mm thickness. The tungsten carbide cubes with edge length of 31 mm and truncation of 4.0 mm, and a pyrophyllite gasket of 1.40 mm in thickness were used. Temperature of sample was monitored using two pairs of W₉₇ Re₃-W₇₅ Re₂₅ thermocouple with 0.125 mm diameter, which were mechanically connected to each Fe electrode in contact with the sample. The thermocouples were also used as lead wires for conductivity measurement. The uncertainty of temperature values caused by temperature gradients was estimated to be less than 50 K in the sample chamber. The generated pressure was calibrated at ambient temperature using phase transition of Bi (2.55 and 7.7 GPa), GaAs (18.3 GPa) and GaP (22.6 GPa). The relative uncertainty in pressure was estimated to be±lGPa. Before measurements, the sample and all of other parts were cleaned out in acetone using ultrasonic, and then heated at 1000℃ about 12 h to remove the influence of adsorbed water in the sample assembly.

图 1

Fig. 1

Fig. 1 Schemical cross-section of the cell assembly for electrical conductivity measurement in the multi-anvil press at simultaneous high pressure and high temperature

The electrical conductivity measurements were performed at pressure of 10, 15 and 20 GPa, and temperature increasing from 750 to 1600 K in 50~100 K steps, respectively. The powdered enstatite sample was embedded in an Al₂O₃ sleeve and two Fe electrodes with a diameter of 1 mm were placed in contact with the sample, so that the oxygen fugacity could be controlled to the iron-wlistite buffer during the experiment. At the same time, current leakage could be minimized using an Al₂O₃ sleeve between the sample and furnace. The further details of experimental processes and the electrical circuits for electrical conductivity measurement have been described earlier by Katsura and cO-workers^[12~14]. A reference resistance of 10²~10⁶ Ω was connected to the sample in series, and a sinusoidal signal with amplitude of 1.0 V and frequency of 0.1 Hz was applied to the circuit. Sample resistance was obtained by dividing the voltage applied on the sample by the circuit current. The circuit current was obtained from the voltage applied on the reference resistance connected to the sample in a series. To obtain high quality and stable data, background noise was always monitored during the data acquisitions and, to avoid bias in the data, several heating and cooling cycles were repeated at low temperature range. Insulation test was conducted by means of substituting blank MgO disk for enstatite powder in the same size of cell assembly before electrical conductivity measurements. The insulation resistance was exceeded 10⁸ Ω at ambient temperature, and decreased to around 10⁵ Ω at temperature up to 1600 K. However, the insulation resistance was more than 2 orders of magnitude higher than the sample resistance at high temperature. In the runs for electrical conductivity measurements, the sample was firstly compressed to desired pressure, then pre-heating at 500 K in order to eliminate the adhesive water. After that, several heating and cooling were repeated in temperature range from 750 to 1600 K.

The electrical conductivity was determined from sample resistance by the equation:

(1)

where Lis the thickness of the sample, R is the direct current resistance, S is the cross-section area of electrode. The thickness and diameter of the sample were measured using an electron microscope after decompression, it is found that the changes of L and S under pressure are negligible in Eq.(1) for the calculation of the electrical conductivity, because L and S change not more than 10% under high pressure in this study, on the other hand, the conductivity changes by orders of magnitude.

According to formula (1), the temperature and pressure dependence of electrical conductivity can be expressed as Arrhenius law:

(2)

where σ is electrical conductivity, σ₀ is a preexponential factor, the activation enthalpy ΔH=ΔE + PΔV with activation energy ΔE and activation volume ΔV, P is pressure, T is the absolute temperature, and k is the Boltzmann constant. The logarithmic conductivity of (Mg_0.9Fe_0.1) SiO₃ enstatite at different pressure is plotted against the reciprocal temperature in Fig. 2. The electrical conductivity of enstatite increases slightly with increasing pressure from 10 to 15 GPa. Activation enthalpy (ΔH) and pre-exponential terms (ο₀) resulting from fitting Eq.(2) to the data are summarized in Table 1.

图 2

Fig. 2

Fig. 2 Logarithm of electrical conductivity vs. reciprocal temperature for (Mg0.9Fe0.1) SiO3 enstatite at pressure of 10, 15 and 20 GPa

表 1(Table 1)

Table 1 Activation enthalpies and preexponential terms from fitting Eq. (2) to the experimental data

Table 1 Activation enthalpies and preexponential terms from fitting Eq. (2) to the experimental data

In present study, the conductivities of enstatite were measured at pressure of 10, 15 and 20 GPa, and temperature increasing up to 1600 K. Previous works^{[1, 15]} have suggested that the transformation of enstatite to high-pressure phases (such as β-(Mg, Fe)₂SiO₄/γ-(Mg, Fe)₂SiO₄ + stishovite) occurs preferentially at high temperature and high pressure. After the conductivity measurement was completed, the recovered sample was characterized by X-ray diffraction, electron microprobe analysis and electron microscopic observation. The phases of final product determined by X-ray diffraction are assemblages of ringwoodite and stishovite. As a consequence, the last runs for conductivity measurements at 20 GPa was not the electrical conductivity of enstatite, but the assemblages of ringwoodite and stishovite. On the other hand, doubly polished sample with a thickness of less than 100 μm was prepared for the IR analysis? the water content of the sample was 0.23 wt% determined by non-polarized Fourier-transform infrared spectroscopy. The Paterson calibration was used to determine the water content from the total integration according to each infrared absorption peak1-16-1.Although no water was added to the sample on purpose before measurements in this study, we cannot exclude the possibility which may come from the surrounding pressure medium at high temperature. We did not measure the water content of sample before measured conductivity > as our starting materials are powder.

Fig. 2 displays the conductivity of (Mg_0.9Fe_0.1) SiO₃ enstatite under 10, 15 and 20 GPa as a function of reciprocal temperature. It is clear that the Arrhenius relations can be divided into high temperature and low temperature regimes. The activation enthalpies are larger in the high temperature than these in the low temperature regimes (Table 1), in agreement with the electrical conductivity in hydrous olivine ^[3]. Similarly, we suggest the dominant conduction mechanisms change from protons to small polaron as water is determined in the recovered sample in this study. The low-T activation enthalpies for proton conduction increased from 0.75 to 0.79 eV because of a tiny loss of water during conductivity meas-urements. This behavior is also comparable with those for proton conduction in hydrous olivine^[3].The high-T activation enthalpies of enstatite for hopping conduction are 1.06 and 1.39 eV corresponding to the pressure of 10 and 15 GPa, respectively, which are lower than 1.6~1.8 eV in orthopyroxene^{[10, 11]}, consequently, our measurements of conductivity are much higher than previous results^{[10, 11]}. The discrepancy may be due to the oxygen buffers^{[6, 7]} and composition. Dai et al.^{[6, 7]} adopted a novel method to control oxygen fugacity for measurements of grain boundary conductivities of rocks at high pressure and temperature, and they found that the changes of grain boundary electrical conductivity were close to one order of magnitude by using different oxygen buffers under the same conditions of pressure and temperature. A weak pressure effect also supports the small polaron model as the dominant conduction mechanism in the high-T region^[17], as the electrical conductivity of enstatite increases lightly with increasing pressure, and if all of the reproducible and reversible conductivity data in the high-T regions at pressure of 10 and 15 GPa are fitted to Arrhenius equation (2) using a Levenberg-Marqu-ardt nonlinear procedure^[18], the preexponential factor, activation energy and activation volume are calculated to be σ₀=736±1.8 S/m, △E=1.21±0.05 eV and ΔV=-0.14±0.20 cm³/mol, respectively. The small and negative activation volume is in consistency with conduction mechanism of small polaron.

The conductivity at pressure of 20 GPa was the electrical conductivity of ringwoodite. For ringwoodite, the activation energy (1.46 eV) for small polaron is consistent with previous deter-mination^{[4, 19]}, while the activation energy (0.8 eV) for proton conduction is a little lower than previous study^[4]. Fig. 3 indicates a comparison of the electrical conductivity for ringwoodite with the existing results^{[4, 19, 20]}, water are in good agreement with the available measurements of ringwoodite with 0.22 wt% water^[4].

图 3

Fig. 3

Fig. 3 Comparison of the electrical conductivity for ringwoodite with previous results. The thick dashed line from Xu (1998) [19]; The dot-dashed lines from Huang (2005)[20]; The open circles from Yoshino (2008)[4] The numbers represent water content in wt%

In conclusion, due to pressure induced phase transition in enstatite, the present results indicate the conductivity of ringwoodite is compatible with other studies. The higher conductivity of enstatite in this study might arise from higher hydrogen concentration or different oxygen buffers^{[6, 7]}, in the sample. Thus the water content influence on the electrical conductivity of enstatite should be further investigated in future.

This work was supported by funds from the Natural Science Foundation of China (No. 40874034, 40537033, ) and Program for New Century Excellent Talents in University (NCET-05-0553), and also partially supported by funds of the graduate innovation of USTC (No. KD2007059) and funds of Heifei University-Technology (No. 2009HGXJ002). We gratefully acknowledge financial support from COE-21 Collaborative Research Program at ISEI, Misasa, Japan.