Water Weakening of Mantle Materials, a, b and g Olivine

Jiuhua Chen, Toru Inoue, Donald J. Weidner, Yujun Wu and Michael T. Vaughan

Center for High Pressure Research, SUNY at Stony Brook
       

Rheological properties of different phases of olivine, both dry and hydrous, are studied by monitoring the diffraction peak broadening as a function of pressure, temperature and time. The measure-ments are carried out up to a pressure of 10 gigaPascals for the a- and b-phases, 20 gigaPascals for the g-phase; and up to a temperature of 600oC for the a- and b-phases and 1000oC for the g-phase.

FWHM (full width half maximum) analysis of the energy dispersive diffraction patterns shows that water

weakens the a-phase dramatically while the hydrous samples of the b- and g-phases behave slightly weaker than their dry forms.

The effective strain relaxation in the hydrous a-phase above 400oC is hardly observed, but the dry a-phase still follows an effective power law strain relaxation with an exponent of 10 at 600oC. The exponents of strain relaxation for the dry and hydrous b-phases are 14 and 12 at 600oC, and those for the hydrous g-phase remains as high as 20 at 800oC.

     
       
       

Rheological properties of the different phases of olivine are important properties that constrain mantle convection and have implications regarding the origin of deep focus earthquakes. All the three phases of olivine, a, b and g, can take different amounts of water, from hundreds ppm to few percent by weigh at mantle conditions. The upper mantle is estimated containing 100 to 500 ppm water at average. Many mantle materials, including different phases of olivine may, therefore, exist in hydrous forms in the earth. Hydrogen ions introduced into minerals in the presence of water may significantly effect physical properties of the minerals. Karato reported the electrical conductivity of olivine is considerably enhanced by the presence of water, and later he claimed that water may also greatly effect the seismic wave velocity and attenuation. Melting experiment in the system Mg2SiO4-MgSiO3 show that the presence of water drastically changes the melting temperature, liquidus phases and the coexisting melt composition. Apparently, the effect of water on physical properties of minerals must be taken into account in constraining geophysical phenomena. It is widely recognized that water considerably weakens olivine at confining pressures up to one GPa. Extrapolation of this result to higher pressure is necessary to understand the water effect on the mineral at the upper mantle condition. On the other hand, in the high pressure phases, b and g, hydrogen is structurally bound, with uncertain implications on its ability to mechanically weaken these minerals.

 

Here we report results of the rheology measurements on the both dry and hydrous phases of a-, b- and g- olivine. Strain

in the sample at high pressure is derived from broadening of x-ray diffraction peaks. The methodology is described in at length.

 

The anhydrous sample of the a-phase is San Carlos olivine. The hydrous a-phase is prepared by annealing the anhydrous olivine in a water environment at high pressure and temperature using a uniaxial split sphere anvil apparatus, USSA2000. Powdered San Carlos olivine with saturated amount of water is sealed in a Pt capsule with an outer diameter of 3 mm, length of 3 mm and wall thickness of 0.1 mm. The sample is annealed at 12 GPa and 1200oC for 1 hour. Presence of water after the experiment is confirmed by observing water bubbles in a recovered sample chunk. The recovered sample is examined by scanning electron microprobe (SEM) and x-ray diffraction (XRD) to confirm the composition and structure. The water content in the sample is estimated referring published experimental results at a similar P-T condition to be few hundreds weight parts per million. The b-and g-phases are synthesized from starting materials of a mixture of MgO and SiO2 for anhydrous forms and a mixture of MgO, SiO2 and Mg(OH)2 for hydrous forms. Saturated amount of water (11.3 wt %) is added in form of Mg(OH)2 in the starting materials for hydrous phase syntheses. Both the anhydrous and hydrous b-phases are synthesized at 13.5 GPa and 1200oC, and the g-phase at 19.0 gigaPascals and 1300 oC. Annealing duration for the b- and g-phase is 40 min and 60 min respectively. Recovered samples are investigated by SEM and micro-focused XRD. Water content is measured to be 3.8wt% for the hydrous b-phase and 2.2wt% for the g-phase by secondary ion mass spectrometry (SIMS).

     
       
             
           
             
Most of the strain measurements are performed with a DIA-type cubic-anvil apparatus, SAM85(10) at the superconductor wiggler beamline (X17B1) at the National Synchrotron Light Source (NSLS). The measurement for the hydrous g-phase is conducted with a newly developed T-cup high pressure cell(11) at the same beamline, to achieve a higher pressure.
             
         
In the experiment of hydrous g-phase, the sample is loaded in one half of the of the T-cup sample cell (below). The other half of the cell is filled by the mixture of NaCl and BN.
       
     
Samples are ground into powders with an average grain size of several micrometers. In the measurements of the a- and b-phases, anhydrous and hydrous samples are loaded in pairs into a sample chamber about 1 mm in diameter and 2 mm in length. The two samples are separated by a layer of a mixture of powdered NaCl and BN, which also serves as a internal pressure calibrant. Cell assembly is shown above.  
     
         
       

Stress in a-phase upon loading and then heating: dry (upper) and hydrous (lower)
     

Energy-dispersive x-ray diffraction mode with a solid state detector at a fixed scattering angle of 7.5o is used to collect the diffraction data. Data gathering times are less than 1 minute with an x-ray beam of 100´200 mm (v´h).

The experiments are performed by first compressing the powdered sample to 10 gigaPascals (20 gigaPascals for the hydrous g sample) at room temperature and then heating the sample stepwise. At each step the temperature is held for several tens of minutes during which diffraction data are recorded as a function of time. Figure 1 shows strains in anhydrous and hydrous phases of both a- and b- olivine.

 

In a-olivine, stress increases with pressure for both the anhydrous and hydrous phases. Stress in the hydrous phase, however, begins to fall beneath the elastic line at a lower pressure than in the anhydrous phase, which shows that the hydrous phase reaches the yield point earlier. When the samples are heated to 400oC, stress relaxation in the hydrous sample is almost completed while the anhydrous sample still holds half amount of the stress. Upon further heating to 600oC, there is no significant further stress relaxation observed in the hydrous phase confirming that the sample does not hold stress any more. On the other hand, a further stress drop in the anhydrous sample is obvious when the temperature is increased to 600oC. Furthermore, decrease of the stress as a function of time can be observed at 600oC in the anhydrous sample, which indicates that the stress is not yet totally relaxed at this temperature.

     
       
       

Stress in b-phase upon loading and then heating: dry (upper) and hydrous (lower) .
     
In b-olivine, the pressure of the stress yield point is, in general, higher than for the a-phase, and the difference of the yielding pressure between the anhydrous and hydrous phases is less obvious compared to the a-olivine. The hydrous phase remains as strong as the anhydrous phase at 400oC. No significant strain drop is observed at temperatures up to 400oC. Upon heating to 600oC, the hydrous phase shows a greater strain drop than the anhydrous phase, and both phases show further strain relaxation as a function of time at this temperature.
     
       
       

Stress in g-phase upon loading and then heating: dry (upper) and hydrous (lower) .
     
Time resolved stress relaxation of the hydrous g sample at 20 GPa with some single temperature stress data of the anhydrous g sample measured at 10 GPa. As with the b-phases, the hydrous g-phase becomes weaker than the anhydrous phase when temperature is increased to 600oC. Stress drop can still be observed when the temperature jumps from 600oC to 800oC and from 800oC to 1000oC, which indicates that the hydrous g-phase is also much stronger than hydrous a-phase at high temperature. Up to 1000oC, the sample shows further strain relaxation as a function of time at a constant temperature.
     
       
       

The conventional power-law expression relates the strain rate with the stress to a power:

 

¶e /¶t = ksn ,

 

where e is the elastic strain, t is the time, s is the stress, and k is a material-dependent constants. If s is expressed as an elastic modulus times the strain, this becomes:

 

¶e /¶t = k'en,

 

where k' includes the elastic modulus. Integrating and taking the logs,

 

(1-n) log(e) = log(t) + k''.

 

The figure shows log of stress as a function of log of time for the different phases. The data define a straight line yielding value for n at each temperature. A smaller n value indicates a higher strain rate, that is a weaker material. Therefore in general, the lower temperature data have larger n values.

The indeterminate values for the hydrous a-phase indicate that the amplitude of further strain relaxation is smaller than the instrument resolution.

 

The b-phases have n value of 31 and 28 at 400oC, and 14 and 12 at 600oC (anhydrous and hydrous, respectively), which are greater than the corresponding values for the a-phases at each temperature. In other words, the b-phase is stronger than a-phase. The hydrous b-phase has smaller n values than the anhydrous phase at each temperature.

 

The g-phase is the strongest one of the three phases. The n values of hydrous g-phase at 400oC, 600oC and 800oC are 46, 23 and 20 respectively. The data at 1000oC become scattered due to instrumentation resolution.

     
       
         
   
       
Stress relaxation at various temperatures, on a log-log plot. A smaller n represents a higher strain rate, or a weaker material. Lower temperatures have larger values. For the strongest samples, there is too much scatter to give a good determination of n.    
       

All the a-, b-, and g-phases of olivine are weakened by the presence of small amounts of water. The considerable weakening effect on a-olivine is confirmed at high pressure up to 10 GPa. The average water content of hundreds ppm in the upper mantle can probably make the a-olivine release all the strains at 400 - 600oC.

The higher pressure phases, b and g, show much stronger behavior under the water weakening although they can take more water than the a-phase. A comparison of the three hydrous forms of the a-, b- and g-phases at 400oC and 600oC in terms of stress is shown above.

Assuming the three phases of olivine are the main minerals of the earth mantle and the subduction zone contains at least several hundreds ppm of water, the rheological properties of a-, b- and g-phases may be the confining factor regarding the origin of deep focus earthquakes.