conductivity cannot be explained by semi-conduction in dry olivine.
Therefore, water, melt or carbon may be present in the sub-lithospheric mantle.
8.1 Laboratory measurements under in-situ
Figure 1.5 summarises the electrical conductivities of some common Earth materials. MT studies around the world indicate that the deep crust and mantle are often more conductive and more electrically anisotropic than anticipated from laboratory measure- ments on uplifted lower-crustal rocks (e.g., Kariya and Shankland, 1983) or dry olivine (Constable et al., 1992), respectively. For example, the anisotropy of electrical conductivity of mylonites from the deep crust (Siegesmundet al., 1991) is too low to explain most observations of deep-crustal electrical anisotropy (Figure 8.2).
In situ crustal conductivities may be enhanced by saline fluids and/or graphite, metallic oxides and sulphides, or partial melt.
Mantle conductivities may be enhanced by fluids (including partial melts), graphite or hydrogen diffusivity (see Sections 9.2 and 9.3).
Macroscopic anisotropy is most easily explained if conductive phases are preferentially aligned (e.g., owing to crystal-preferred orientation) or exhibit a higher degree of interconnection in the more conductive direction.
Graphitic films have been observed in metamorphic rocks using Auger electron emission spectroscopy (Frost et al., 1989).
Amphibolite and gneiss core samples from depths of between
II To foliation II To foliation
x
y z
To lineation x
y z
To lineation
(b) (a)
(870) (1707)
1585
881 (1019)
(1365) 1359
858 648 (657) 1150 (1029) ρ(Ω m) Figure 8.2Macroscopic fabric
elements of mylonite rock samples from the Indian Ocean ridge and their direction-dependent electrical resistivities. The values in brackets are for low pressures, corresponding to shallow crustal depths, whilst the other values are for the high pressures expected at deeper crustal depths. The resistivities of these rock samples are too high, and their anisotropies too low to explain deep-crustal conductivity anomalies revealed by MT field
measurements. (Redrawn from Siegesmundet al., 1991.)
148 Conduction mechanisms
1.9–7 km in the KTB deep borehole that were saturated with NaCl solution exhibited increased conductivity in a direction at an angle to foliation when pressurised in the laboratory (Figure 8.3), despite the evacuation of fluid under pressure (Dubaet al., 1994; Shankland et al., 1997). The anisotropic conductivity increase, was explained in terms of reconnection of graphite under pressure. Ross and Bustin (1990) demonstrated that shear stress promotes graphitisation of rocks, whilst Robertset al.(1999) found evidence of deposition of carbon films on fracture surfaces during dilitancy of rocks under controlled laboratory conditions. The phase transition from con- ducting graphite to insulating diamond occurs at temperatures of 900–13008C and pressures of 45–60 kbar. Therefore, graphite is unlikely to be the cause of conductivity anomalies deeper than 150–200 km in the upper mantle.
Stesky and Brace (1973) demonstrated that serpentinised rocks from the Indian Ocean ridge are 3–4 orders of magnitude more conductive than serpentinite-free peridotite, gabbro and basalt
0 2 4 6 8 10
50 100 150 200 250
Pressure (MPa) Conductivity (×10–3 S m–1)
(a)
AC impedance
analyser 1 kHz–1 MHz
Shrink tubing Sample
Stainless-steel electrode Filter paper saturated with 1 M NaCl solution Hydrostatic pressure to 250 MPa (b)
Figure 8.3(a) Electrical conductivity of an amphibolite sample from a depth of 4.149 km in the KTB borehole.
Open symbols are measurements at 308to the normal of the foliation plane and closed symbols are for measurements in the plane of foliation. The conductivity in the plane of foliation decreases with increasing pressure, whereas the conductivity in the direction at an angle to the foliation increases with pressure, after decreasing with pressure in the first ~10MPa.
The conduction mechanism is interpreted to be the sum of highly conductive intergranular phases (such as graphite) and saline fluids.
(b) Schematic diagram of experimental setup used to measure electrical conductivities shown in (a).
(Redrawn from Dubaet al., 1994.)
8.1 Laboratory measurements underin-situconditions 149
from the same area. The enhanced conductivity may be associated with magnetite formed during serpentinisation. In addition to a conductivity anomaly, serpentinite should give rise to a magnetic anomaly.
Boreholes drilled to 12.3 km depth on Russia’s Kola peninsula and to 8.9 km in Bavaria, southern Germany revealed that the rock at these mid-crustal depths was saturated with highly saline fluids – a most unexpected result. At mid-crustal temperatures and pres- sures, highly saline fluids can be expected to have conductivities of the order 25–50 S (Nesbitt, 1993).
Although mantle minerals are nominally anhydrous, they are able to incorporate dissociated H+ and OHions (Mackwell and Kohlstedt, 1990; Bell and Rossman, 1992). Measured concentra- tions of hydrogen content in natural samples indicate that pyroxene minerals are the most important reservoirs for hydrogen in the upper mantle, followed by garnet and olivine (Bell and Rossman, 1992). Hydrogen diffusion in the mantle could enhance the electrical conductivity of the mantle according to the Nernst–Einstein equation (Section 9.3, Equation (9.1)). Laboratory measurements indicate that hydrogen diffusivities vary depending on mineral type, and that they may be strongly anisotropic in single crystals.
Tyburczy and Waff (1983) showed that partial-melt conductiv- ities are far less dependent on pressure than on temperature.
Laboratory conductivity measurements on partially molten samples and mixing-law calculations are compared in Roberts and Tyburczy (1999). A compilation of results of laboratory electrical conductivity measurements on dry olivines and basaltic melts at a range of temperatures is shown in Figure 1.4. At upper-mantle temperatures, partial melt is approximately 1–2 orders of magnitude more conduct- ive than dry olivine. However, many conductivity anomalies occur in stable regions at depths where the temperature is expected to be less than the solidus temperature of mantle silicates, and would require unrealistic quantities of partial melt to explain them. Based on a comparison of laboratory measurements made under controlled oxygen fugacities and electrical conductivity models of the Western Cordillera in northern Chile derived from MT data (Echternacht et al., 1997), at least 14% partial melt would be required below the magmatic arc in the central Andes to explain the observed deep-crustal conductance, which exceeds 40 000 S (Schilling et al., 1997).
Upper-mantle temperatures and pressures can be simulated in a multi-anvil press, in which 1–20 mm3samples are sealed from interaction with the external atmosphere, pressurised, and heated
150 Conduction mechanisms
with a laser beam. A typical experimental setup for performing mantle conductivity measurements is shown in Figure 8.4. The molyb- denum shield and electrodes help to minimise leakage currents and temperature gradients, and to control oxygen fugacity (fO2). The importance of controllingfO2 when determining electrical conduct- ivities in the laboratory has been discussed by Duba and Nicholls (1973), Dubaet al.(1974), Duba and von der Go¨ nna (1994) and Duba et al.(1997).
The upper mantle is composed predominantly of olivine and pyroxene. Laboratory measurements indicate an2 orders of mag- nitude increase in electrical conductivity arising from a phase change from olivine to wadsleyite, which is expected to occur at depths of 410 km (Figure 8.5; Xu et al., 1998). Less dramatic increases in conductivity are expected at the wadsleyite to ring- woodite to prevoskite+magnesiowu¨stite transitions that are expected to occur a depths of520 km and660 km respectively.
At upper-mantle temperatures and pressures olivine and pyroxene have similar electrical conductivities, whereas in the transition zone, the electrical conductivities of pyroxene and ilmenite–garnet com- positional systems are 1–2 orders of magnitude lower than those of wadsleyite and ringwoodite (Xu and Shankland, 1999).
Electrode wire
Shield (Mo)
Thermocouple (Electrode wire)
ZrO2
MgO Al O2 3
Al O cement2 3 Mo
Furnace (LaCrO )3 Sample
Electrode (Mo)
Figure 8.4Experimental setup for complex impedance measurements under mantle conditions of pressure, temperature and oxygen fugacity. (Redrawn from Xu et al., 1998.)
1 0 –1 –2 –3
–40 200 400 600 800 1000
OWR SKC J93
BOS93–1 SPP93
BOS93–2
B 69
log10[σ (S m–1)]
Depth (km)
Figure 8.5Conductivity distribution in and around the transition zone between the upper and lower mantle.
Laboratory data: OWR (Xu et al., 1998) and SPP93 (Shanklandet al., 1993).
Magnetotelluric and geomagnetic models: B69 (Banks, 1969), BOS93-1, -2 (Bahret al., 1993) and SKCJ93 (Schultzet al., 1993).
(Redrawn from Xuet al., (1998.)
8.1 Laboratory measurements underin-situconditions 151