The Influence of Brines and Temperature on the Frictional Properties of Laboratory Fault Gouge. Cody W. Bomberger ABSTRACT

Transcription

1

2 The Influence of Brines and Temperature on the Frictional Properties of Laboratory Fault Gouge Cody W. Bomberger ABSTRACT The frictional properties of gouge material within tectonic faults zones dictate key aspects of earthquakes rupture dynamics and the seismic cycle. Modern friction laws are capable of describing the complete seismic cycle, however the underlying processes and thermodynamics of friction constitutive processes are poorly understood. This thesis describes a suite of experiments designed to investigate the processes and mechanisms of rate and state friction behavior. The work in this study, centers around laboratory experiments, in which quartz and gypsum powders are sheared while saturated in fluids composed of brines or deionized water. The effects of fluid chemistry on the frictional properties of gouge material are examined. Observations of quartz powders response to the change from deionized water to NaCl brine are unseen, if there is any effect, by the strain dependence of frictional healing. Prior investigations suggest that the strain dependence is related to the evolution of the gouge material and the number of grain contacts within the gouge. Gypsum powder experiments saturated in all fluids do not experience strain dependence but have a non log-linear healing rate indicating that the mechanism controlling healing in gypsum powder is different from quartz powder. A significant difference in frictional healing, frictional creep, and dilatancy occurs between the experiments being saturated in NaCl and deionized water. Na 2 SO 4 and CaCl 2 brines as saturating fluids were also compared with experiments in NaCl brine. Na 2 SO 4 exhibit the greatest healing rate, but the other the frictional properties appear similar for the three brine types. Experiments at elevated i

3 temperatures create little change in the frictional properties but cause a transition to unstable sliding. The process dominating the frictional properties of gypsum powder shows the characteristics of pressure solution. Healing controlled by pressure solution will be non loglinear and the rate that pressure solution occurs at increases in the presence of the brines used. For accurate interpretations of the strength of natural faults, the lithology of the faults and how the lithology is interacting with the fluids present is critical. ii

4 TABLE OF CONTENTS Acknowledgements... iv Introduction... 1 Experimental Methods... 3 Results and Observations... 5 Discussion Conclusion References Cited Appendix A iii

5 ACKNOWLEDGEMENTS I would like to thank Dan King for all of the hours he spent teaching me how to use the biax, how to process the data, and teaching me a great deal about rock mechanics. I would also like to thank Chris Marone for allowing me to complete this thesis with him and for all of the guidance he has given me. I would like to thank everyone at Penn State Rock Mechanics Laboratory who helped or discussed aspects of my research with me. iv

6 Introduction The strength and stability of a seismogenic fault depends upon the frictional properties of the fault gouge that shears during earthquakes. The environments in which these faults exist vary but nearly all are known to be saturated with groundwater. Water present in areas of active faulting is known to have physical and chemical effects on the gouge materials that can alter the material properties between seismic events. Groundwater contains many dissolved ions, some common ones being Na +, Mg 2+, Ca 2+,Cl -, and SO 2-4, that may cause further chemical or physical changes to gouge material (Feucht and Logan, 1990). The presences of fluids and the ions dissolved in these fluids are expected to influence the strength and stability of seismogenic faults and may play a significant role as some suggest (Blanpied et al.,1998; Bos and Spiers, 2002). Previous studies on quartz-rich, clay-bearing sandstones have shown that the ultimate strength, the greatest sliding friction achieved during an experiment, decreases in the presence of water and further decreases in the presence of dissolved ions (Feucht and Logan, 1990). Studies of quartz sandstones under hydrothermal conditions, however, show increased strength and cohesion due to pressure solution (Tenthory and Cox, 2006; Tenthory and Cox, 2003). Quartz is often used as a bulk representation of the continental lithosphere, and the effects of dissolved ions on shallow faults in the continental crust can be examined through the use of quartz powder. Gypsum at room temperature can be used as an analog for quartz at high temperature, likely representative of mid-crustal seismogenic depth (Muhuri et al., 2003). We hope to examine the effects of dissolved ions at greater depths within the lithosphere, than with the use of quartz powder alone, through the use of the gypsum analog. Through the use of both gypsum and quartz powder we hope to explore the contradictive studies summarized above. 1

7 A contributing factor to what makes gypsum a room temperature analog for quartz at higher temperatures is that the chemical activity of gypsum, at room temperature, is comparable to the chemical activity of quartz at higher temperatures. The elevated chemical reaction rate of gypsum compared to quartz also proves useful when studying chemical effects of dissolved ions on fault gouge (Rimstidt and Barnes, 1980; Meer et al., 2000). Pressure solution of gypsum is known to happen at room temperature under low normal stresses, less than 5MPa (Bos and Spiers, 2002). Pressure solution is controlled by three processes, dissolution, diffusion, and precipitation. All three of these processes are temperature dependent by an Arrhenius relationship (Yasuhara et al., 2003). The presence of ions may decrease the dissolution rate at the grain boundaries, and the rate of precipitation and cementation in the pore space may increase (De Meer and Spiers, 1995; Meer and Spiers, 1999). An increase in precipitation and cementation will cause an increase in frictional strength after a hold period, meaning the healing rate of some faults should increase in the presence of brines, sometimes this may be referred to as an increase in cohesion or lithification (Yasuhara et al., 2005; Ikari and Kopf, 2011; Ikari et al., 2011). 2

8 Experimental Methods The granular quartz was a high-purity ( > 99% quartz), fine-grained material purchased from the U.S. Silica Company as F110 foundry sand; 95% of the grains are in the range µm and the mean grain size is 127 µm. F110 is a common laboratory standard for granular studies, and its frictional properties are well established over a wide range of conditions (e.g. Mair and Marone, 1999; Anthony and Marone, 2005; Marone et al., 2008). The Gypsum powder was made in a ball mill from solid samples of Alabaster from Pomaie Italy from Ward s. Layers of quartz or gypsum powder were created with a thickness of 5-mm in a leveling jig and sheared between rough steel forcing blocks with a surface area of 5 x 5 cm 2. For all experiments a double direct shear configuration was used in a servo-controlled hydraulic testing apparatus. The double direct shear blocks were submerged in either deionized water or brine with the use of a plastic membrane open at the top. Elevated temperatures were achieved and held constant with an immersion heater. Brines used in this study were made to specific concentrations by weight with deionized water. The normal force and shear force applied to the double direct shear configuration, seen in Figure 1, were measured via BeCu load cells and the displacement of these load cells were measured via Direct Current Displacement Transducers (DCDTs). More information about the experimental apparatus can be found in (Mair and Marone, 1999; Karner and Marone, 2001; Frye and Marone, 2002). The normal stress for the quartz and gypsum experiments were held constant at 25 MPa and 10 MPa respectively. Samples were sheared at a constant velocity of 10 µm/s for 5 mm or 8 mm to achieve stable sliding friction. This initial run-in was followed by a series of velocity steps and slide-hold-slide sequences or immediately by a series of slide-holdslides sequences. An example of a common experiment form can be seen in Figure 1 with the 3

9 velocity pattern used during each set of velocity steps and the hold duration for all hold sequences. Table 1, Appendix 1, lists the experiment numbers, saturating fluid, the number of hold sequences, and number velocity step sequences. Figure 1. Shear stress versus shear displacement for a 5-mm gouge layers sheared in the double direct shear experiment setup shown in the upper left corner and submerged in either deionized water or one of the three brines. Shown are the initial loading to achieve stable sliding friction, velocity steps, and a single sequence of slide-hold-slides. The inset shows the values of the sliding velocity during the velocity steps and the duration of the holds during a single slide-hold-slide sequence. 4

10 Results and Observations Quartz Powder Experiments and Strain Dependence A comparison of the experimental data of friction as a function of shear displacement can be a useful tool to visually compare velocity steps and slide-hold-slide sequences among experiments. A comparison of the slide-hold-slide sequences at similar shear displacement for quartz powder experiments saturated in three different fluids (deionized water, 1M NaCl, and 3M NaCl) shows no obvious difference with the changes in fluid chemistry (Figure 2). When comparing the frictional healing and the frictional creep measurements taken from the experiments saturated in these three different fluids, no obvious relationship between salinity and frictional healing or frictional creep is seen (Figure 3). Frictional creep is the amount the friction decreases throughout the hold period and frictional healing is the peak of friction upon re-shear greater than the sliding friction before the hold period. Salinity not having a measurable effect on the frictional properties of quartz powder gouge indicates that the frictional properties are dominantly controlled by a different factor at the conditions of these experiments. The evolution of the gouge through shear displacement can cause strain to be a dominate influence on the frictional properties and is known to occur in quartz powder gouge from previous studies (Richardson and Marone, 1999). The dependence of frictional healing on shear displacement can be seen in Figure 4, which includes all quartz powder experiments and all fluids. Fictional healing shows a strong linear trend with shear displacement while the frictional creep has no clear relationship or dependence on shear displacement. The linear trend seen for all 100 second holds is also seen for all 1000 second holds. Using this linear trend of healing with strain, the frictional healing was normalized for 5

11 shear displacement by removing the increase in healing gained at greater strain, these results are shown in Figure 5. After the normalization was completed, the frictional healing data continued to show no dependence on brine salinity. This result is likely related to the low chemical reactivity of quartz at room temperature (Rowe and Survey, 1960; Rimstidt and Barnes, 1980). 6

12 Figure 2. Hold sequence, as seen in Figure 1, comparison of difference concentrations of NaCl brine saturating quartz powder fault gouge. Load point displacement verses friction with the concentration increasing from top to bottom. The observed pattern is nearly identical between the three different concentrations. 7

13 Figure 3. Frictional healing (left) and frictional creep (right) as a function of hold time for quartz powder of the experiments in Figure 2. Blue circles represent experiments with deionized water, red triangles represent 3M NaCl, and green squares represent 1M NaCl. Frictional healing and frictional creep show a log linear relationship hold time as expected, but there is no clear relationship between the salt concentration and frictional properties. Figure 4. Frictional healing for 100 second and 1000 second holds as a function of shear displacement (left panels) and frictional creep for 100 second and 1000 second holds as a function of shear displacement. All four panels include the same experiments saturated in varying fluid composition. Healing increases with increasing strain regardless of the salinity, while creep shows no obvious relationship with strain. 8

14 Figure 5. Fictional healing as a function of hold time for quartz experiments after a linear correction for strain dependence. There is no obvious relationship between fluid salinity and the healing rate after the correction. Gypsum Powder Experiments In another set of experiments, we explored the frictional behavior of gypsum powder saturated in a variety of brines. A comparison of the slide-hold-slide sequences of two gypsum powder experiments saturated in deionized water and 2M NaCl (Figure 6) shows noticeable differences, between the two brines. 9

15 Figure 6. Friction as a function of shear displacement for gypsum experiments saturated in deionized water (top) and 2M NaCl (bottom). A difference in friction during relaxation and upon reshear can be seen directly from experimental run plots. The peak of friction upon reshear is consistently greater during the experiment saturated NaCl brine. A comparison of frictional healing, frictional creep, and dilatancy as a function of hold time show greater healing, creep, and dilatancy when gypsum gouge is saturated in NaCl brine than when saturated in deionized water (Figure 7). Diltancy is the change layer thickness during the hold time. 10

16 Figure 7. Frictional healing, frictional creep, and dilatancy as a function of hold time during experiments with gypsum powder saturated in deionized water, blue circles, and saturated in 2M NaCl brine, red squares. The experiment saturated in brine shows greater healing, creep, and dilatancy, while the relationship between healing and hold time is not log linear for either as expected by rate- and- state friction laws. 11

17 The analysis of the dependence of frictional healing and creep on shear strain did not show the same dependence as the experiments with quartz powder. The healing rate for both experiments is a non-log linear dependence on hold time which does not follow conventional rate- and statefriction expectation. The mean and standard deviation of all gypsum experiments saturated in NaCl brine at room temperature is plotted in Figure 8 and shows that the different between healing in deionized water and NaCl brine is significant and that the healing rate is significantly non-log linear in NaCl brine. 12

18 Figure 8. The mean and standard deviation for all experiment of gypsum saturated with 2M NaCl at room temperature. The Non log-linear aspect of the healing rate is outside of one standard deviation from the mean. The range of the standard deviation is also outside of the experiments saturated in deionized water. To further investigate the ways that common dissolved ions in groundwater could affect the frictional properties of the simulated gouge, brines of varying cation and anion contents were used to compare to the previous NaCl brine saturated experiments. The additional brines used were 2M Na 2 SO 4 and 2M CaCl 2. By stacking the hold sequences of three experiments each, saturated in a different brine (Figure 9), it can be seen that the frictional creep during the holds 13

19 and the peak of friction upon reshear for the experiments saturated in 2M Na 2 SO 4 and 2M CaCl 2 brines appear similar. With the help of the quantitative measurements in Figure 10 it is shown that they differ from the experiment saturated in NaCl. Figure 9. Hold sequences as a function of load point displacement of three gypsum powder experiments saturated in three different brines, 2M NaCl (top), 2M Na 2 SO 4 (middle), and 2M CaCl 2 (bottom). The overall change in friction throughout each slide-hold-slide appear greater in experiments saturated in Na 2 SO 4 and CaCl 2. 14

20 Figure 10. Frictional healing, frictional creep, and dilatancy all as a function of hold time for gypsum powder saturated in 2M NaCl,2M Na 2 SO 4, and 2M CaCl 2 shown respectively as blue circles, red squares, and green triangles. The greatest healing rate occurs during the experiment saturated in Na 2 SO 4, but no distinguishable patterns are shown in the frictional creep or dilatancy. Gypsum experiment saturated in Na 2 SO 4 brine shows the greatest healing while experiments saturated in CaCl 2 and NaCl brines have similar frictional healing. The frictional creep and dilatancy of experiments saturated in Na 2 SO 4 and CaCl 2 are similar and greater than the experiments in NaCl brine. Temperature Variations during Gypsum Powder Experiments The final group of experiments were conducted at elevated temperature for all fluids for the duration of the experiment, where the fluid was heated to either 4 or 60 C. Experiments saturated in NaCl show a decrease in sliding stability with an increase in temperature. Figure 11 shows a few stick-slip occurrences at 40 C and periodic stick-slip during the slide-hold-slide sequences. The NaCl experiment also shows a drastic decrease in the 15

21 healing for the 1000 second holds which may be related to the period doubling occurring during the stick-slips. Figure 11. Hold sequences as a function of load point displacement of gypsum powder experiments with temperature variations saturated in 2M NaCl. Temperature increases from top to bottom, top 20 C, middle 40 C, and bottom 60 C. Unstable sliding occurs at the higher temperature with stick-slip behavior can be seen before the 3 second hold at 40 C and throughout the entire slide-hold-slide at 60 C. A comparison of the frictional measurements for the three sequences in Figure 11 are shown in Figure 12 and reveal no clear temperature dependence but do show the importance of the stick- 16

22 slip instabilities in making the measurements at 60 C inconsistent. Figure 12. Frictional healing, frictional creep, and dilatancy as a function of hold time for gypsum powder saturated in NaCl brine with temperature variations, 60 C red circles, 40 C green squares, and 20 C blue triangles. The frictional measurements show no clear dependence upon temperature, but are still non-log linear with hold time. Some aspects of the unstable sliding appear in the healing with the drastic decrease in healing during the longer holds at higher temperatures. As with NaCl, experiments saturated in Na 2 SO 4 exhibit some stick-slip instabilities at 40 C and periodic stick-slip instabilities at 60 C (Figure 13). The hold sequences for experiments at room temperature and 40 C appear similar but the experiment at 60 C has a higher stable sliding friction along with audible periodic stick-slip instabilities. As the hold time increases the displacement required to return to the periodic instabilities increases. The longer hold times allow the gouge to temporarily move away from unstable sliding. Period doubling occurred during the 60 C experiments, as in the NaCl experiments at 60 C, but were not as pronounced. 17

23 Figure 13. Hold sequences as a function of load point displacement of gypsum powder experiments with temperature variations saturated in Na 2 SO 4. Temperature increases from top to bottom, top 20 C, middle 40 C, and bottom 60 C. A few stick-slip instabilities can be seen after the 3 second hold at 40 C. Periodic stick-slip instabilities can be seen throughout the hold sequence at 60 C and following the longer holds a greater sliding distance is required to return to the periodic behavior. 18

24 A comparison of the frictional measurements with temperature variations for experiments saturated in Na 2 SO 4 (Figure 14) show no clear dependence on temperature as expected. A slight increase in healing and creep may occur at the change to 60 C but is small, and no difference between room temperature and 40 C appears. Figure 14. Frictional healing, frictional creep, and dilatancy as a function of hold time for temperature variations of gypsum powder saturated in Na 2 SO 4 brine. Temperatures are 60 C red circles, 40 C green squares, and 20 C blue triangles. A slight increase in healing and creep with temperature is observed with overlap occurring between the temperatures. The last group of experiments with variations in temperature were saturated in CaCl 2 (Figure 15). The holds sequence shown occurred at similar sliding friction with only a few stickslip instabilities seen at 60 C. Comparison of frictional measurements from these hold sequences (Figure 16) show a small increase in healing rate with temperature, but the variation between experiments overlaps. The healing rate and frictional creep at 40 C is lower than both room temperature and 60 C, which may be due to variations between experiments. 19

25 Figure 15. Hold sequences as a function of load point displacement of gypsum powder experiments with temperature variations saturated in CaCl 2. Temperature increases from top to bottom, top 20 C, middle 40 C, and bottom 60 C. Stick-slip instabilities are seen during the experiment at 60 C, but these instabilities are not as periodic during the slide-hold-slides as previous experiments at 60 C. 20

26 Figure 16. Frictional healing, frictional creep, and dilatancy as a function of hold time for temperature variations of gypsum powder saturated in CaCl 2 brine. Temperatures shown are 60 C as red circles, 40 C as green squares, and 20 C as blue triangles. These measurements show a slight increase with temperature with a decrease at 40 C. An analysis of stick-slip behavior for the experiments at 60 C saturated in NaCl and Na 2 SO 4 shows that during the velocity steps increases in velocity coincided with increase in the stress drop frequency, as seen in Figure 17. At the velocity change seen in Figure 17, the break in slope of displacement verse time, the magnitude of the stress drops decreases. 21

27 Figure 17. Stick-slips occurring during the velocity stepping segment of a gypsum experiment saturated in Na 2 SO 4 at 60 C. The left axis is shear stress as a function of time and the right axis is shear displacement as a function of time. The velocity change (the break in slope of shear displacement) coincides with the change in frequency of the stress drops and a decrease in the magnitude. The stress drops have a strong negative log linear dependence on the sliding velocity as expected by previous studies (Karner and Marone, 2000). The stress drop dependence on sliding velocity for the experiment saturated in Na 2 SO 4 is shown in Figure

28 Figure 18. Stress drop, of stick slips, as a function of the shear velocity. The log linear relation shows that as the velocity increases the stress drops approach zero. During portions of the experiment the stress drops occurred with multiple modes of periodicity which contributed to the variability in the magnitude of the stress drops. The variation among the stress drops for a single velocity can be attributed to the occurrence of period doubling seen in the inset. The experiment saturated in NaCl had a greater range of velocities used but shows greater variance among the individual sliding velocities (Figure 19). 23

29 The variation in stress drops is greatest at a shear velocity of 10 µm/s, which coincides with the most prominent occurrence of period doubling during this experiment (shown in the inset). 24

30 Figure 19. Stress drops of stick slips as a function of shear velocity. The log linear relation is shown between velocity and stress drop, but with large scattering occurring for some velocities. The scatter was caused by period doubling of the stick slips shown in the inset. The stick slip behavior occurring after the holds of the slide-hold-slides in the experiment saturated in Na 2 SO 4 at 6 required a greater sliding distance following the longer holds to return to periodic stick slip behavior. After the sliding behavior returns to a periodic and measureable stick-slip behavior, the magnitude of the stress drops decrease as a function of the duration of the preceding hold (Figure 20). This process shows a negative log-linear relationship with the hold time preceding the stick-slip instabilities. 25

31 Figure 20. The top panel shows frictional healing as a function of hold time for an experiment with stick slip occurring during the sliding portions of the slide-hold-slides. The bottom panel shows the average stress during unstable sliding as a function of the hold time that preceded the sliding during the slide-hold-slide. 26

32 Discussion Strain Dependence of Healing and Gouge Evolution The frictional properties of the quartz powder experiments showed no dependence on the fluid chemistry. Quartz is known to have a low chemical activity at room temperature (Rowe and Survey, 1962; Rimstidt and Barnes, 1980), meaning that reaction rates with the fluid chemistry and the quartz powder will act on time scales longer than observed in our experiments. This leads to physical and mechanical processes dominating the frictional properties instead of a chemical process. A 50% increase in healing in less than 10 millimeters of displacement for the 100 second holds shows that strain in all of our experiments with quartz powder is a controlling factor in the healing rate. An increase in healing can be accommodated by an increase in contact area between grains in the gouge material (Marone and Scholz, 1989). Two ways to accommodate grain contact growth is through a reduction in the grain size or through the welding of the grains. The welding of grains could increase contact area through the activity of pressure solution during a holding period, but this contact area would be reset upon re-shear and would be hold time dependent. A reduction in grain size would be a continuous process that would increase throughout all portions of an experiment and increase with shear displacement through the breaking and crushing of grains. The crushing and breaking of grains during sliding will create a wider distribution of grain sizes. A wider grain size distribution allows a greater packing density and increases the contact area of grain by decreasing pore space seen as in Figure

33 Figure 21. Two schematic cross sections of laboratory fault gouge cut though an equal area of grains, the area of grains on the left is the same as on the right. The panel on the left represents before grain crushing occurs and the panel on the right is after grain crushing. The number of contacts between grains nearly doubles between panels with the shift in grain size, creating a greater area for the normal force to act upon. The increase in contact area causes an increase in the strength of the fault gouge by increase the area over which the normal force is acting for the duration of the hold. Through schematic diagrams of simulated fault gouge (Figure 21) it can be shown how the contact area can be increased through a reduction in grain size. The smaller grains fill the pore space previously unoccupied by gouge material, creating new contacts. Non-log Linear Healing Rates As expected by rate-and-state frictional laws, healing as a function of hold time has a linear relationship with the base ten logarithm of hold time (Marone, 1998). However, during all experiments of gypsum powder the healing rate shows a powder law dependence on hold time. This indicates that the process that controls the healing rate during the gypsum experiments is 28

34 different from the grain crushing that is suggested to occur during the quartz experiments. A common process which increases strength in gypsum at room temperature is pressure solution (De Meer and Spiers 1995; De Meer and Spiers, 1999; Bos and Spiers, 2002; Niemeijer et al., 2008). During the holds of the slide-hold-slide sequences, the gypsum gouge is under uniaxial compression allowing pressure solution to occur in a simple stress state. Models of pressure solution from previous studies show that pressure can act on the time scale of the holds and should be time dependent (Yasuhara et al., 2005). The three processes controlling pressure solution are dissolution, precipitation, and diffusion. Dissolution occurs at the grain contacts due to the normal force applied and as the gypsum dissolves, it is transported from high pressure to low pressure along the grain boundary to the pore space. The dissolved gypsum then precipitates in the pores space while adjoining the initial grain to the surrounding grains. These three processes are all time dependent and limited by a rate coefficients (Yasuhara et al., 2003;Yasuhara et al., 2005). The coefficients depend on the gouge material and the fluid in contact with the gouge. The presence of fluid may increase the ability of gypsum to dissolve under pressure, transport along the grain boundaries, and precipitate in the pore space. A schematic diagram of the processes controlling pressure solution is shown in Figure

35 Figure 22. A schematic diagram of pressure solution occurring between four grains that would be in a greater assemblage of gouge material. Dissolution occurs at the grain contacts perpendicular to the normal force, diffuse into the pore space, and precipitate in the area of low stress within the pore space. The experiments on quartz powder did not indicate the occurrence of pressure solution, but the rates controlling pressure solution follow an Arrhenius relationship with temperature. The occurrence of pressure solution is known to occur at temperatures higher than those achieved on quartz powder in this study (Tenthory and Cox, 2006). This shows that at depth within the crust pressure solution is an important mechanism for fault healing. The degree to which healing will be controlled by pressure solution in crustal material depends on the amount of time the fault is stationary and temperature of the fault. Healing Increase in the Presence of Brines The healing rate of gypsum powder saturated in the three brines was greater than when saturated in deionized water. The increased healing rate due to the presence of brine is directly related to how the ions introduced to the fault gouge affect the chemical kinetics of the pressure solution process. The addition of ions should lower the ability for the gypsum gouge to dissolve into solution. The diffusion rate along the grain boundary will also decrease due to the change in 30

36 the concentration gradient along the grain boundary. The addition of the alkali ions to the saturating fluid will increase the rate of precipitation (Meer and Spiers, 1999). The expectation will be that the precipitation rate of pressure solution will control the healing rate of the gouge material. In nature the formation of gypsum occurs in the presence of brines, and any residual brine left in the pore space upon faulting of gypsum layers will cause greater fault healing than expected under the assumptions using deionized water. Temperature Variations and Unstable Sliding The most prominent result from the elevated temperature experiments was the occurrences of unstable slide shown as stick-slip instabilities. The rate coefficients controlling pressure solution all follow an Arrhenius function and will increase with increasing temperature (Yasuhara et al., 2003). This means the rate of pressure solution should increase with increasing temperature, and increasing the rate of pressure solution will increase the healing. A clear increase in healing rate was not observed as expected. The lack of clear temperature dependence may be partially attributed to the unstable sliding with the variations in temperature. The occurrence of stick-slip behavior causes a fundamental change in the frictional behavior of the gouge material. A second factor that may damp the expected temperature dependence of healing is that the presence of the ions in solution alone may dominate the healing and other fictional properties, over temperature. The occurrence of period doubling was observed in experiments saturated in NaCl and Na 2 SO 4. The clarity of doubling was greatest when saturated in NaCl. From classical mechanics period doubling occurs in many natural systems of oscillatory motion, commonly shown as coupled pendulums or a double pendulum. The period doubling phenomena is thought to be an 31

37 intermediate step between simple harmonic motion and chaotic motion with unpredictably periodicity. The implication of this phenomena in a fictional sliding surface is that the failure of the gouge material is becoming chaotic and will not follow a predictable pattern. This also indicates that the force required for failure at a given slide velocity (or shear stress) is no longer predictable. 32

38 Conclusion The frictional properties of quartz powder at room temperature show no clear dependence on brine composition. In quartz powders experiencing grain crushing, the history of the fault material is an important aspect to the frictional properties. Other processes occurring concurrently may be masked by effects of shear displacement. Gypsum powder has a strong non log-linear healing rate representative of pressure solution and increases in cohesion. The addition of ions, from common salts, to the pore fluid of gypsum gouge causes a significant increase in the strength recovery of the gouge material after sliding. As the temperature of the gouge material and saturating solution increases a transition to unstable sliding occurs regardless of the composition of the brine. Geologic implications of these effects of common groundwater ions on simulated gouge material lie in the way the strength of natural faults are estimated. Predictions of fault strength based on previous studies in dry conditions or in distilled or deionized water may be an under estimation. Due to gypsums lower strength relative to other natural materials slip occurs within gypsum layers before different materials that may surround it. Gypsum layers form in alkali ions present and the strength of natural gypsum layers need to take into account these ions being present in the pore fluid. The strength recovery of faults after rupture should be evaluated not only on the basis of lithology but should incorporate the pore fluid chemistry. 33

39 References Cited Anthony, J. L., & Marone, C. (2005). Influence of particle characteristics on granular friction. Journal of Geophysical Research, 110(B8), B doi: /2004jb Blanpied, M. L., Marone, C., Lockner, D. A., Byerlee, J. D., & King, D. P. (1998). Quantitative measure of the variation in fault rheology due to fluid-rock interactions. Journal of Geophysical Research, 103(B5), Bos, B., & Spiers, C. J. (2002). Fluid-assisted Healing Processes in Gouge-bearing Faults: Insights from Experiments on a Rock Analogue System. Pure and Applied Geophysics, 159(11-12), doi: /s De Meer, S., & Spiers, C. J. (1999). Influence of pore-fluid salinity on pressure solution creep in gypsum. Tectonophysics, 308(3), doi: /s (99) De Meer, S., & Spiers, C. J. (1995). Creep of wet gypsum aggregates under hydrostatic loading conditions. Tectonophysics, 245(3-4), doi: / (94)00233-y Feucht, L. J., & Logan, J. M. (1990). Effects of chemically active solutions on shearing behavior of a sandstone. Tectonophysics, 175(1-3), doi: / (90)90136-v Frye, K. M., & Marone, C. (2002). Effect of humidity on granular friction at room temperature. Journal of Geophysical Research, 107(B11), doi: /2001jb Ikari, M. J., Marone, C., & Saffer, D. M. (2010). On the relation between fault strength and frictional stability. Geology, 39(1), doi: /g Ikari, M. J., Niemeijer, A. R., & Marone, C. (2011). The role of fault zone fabric and lithification state on frictional strength, constitutive behavior, and deformation microstructure. Journal of Geophysical Research, 116(B8), B doi: /2011jb Ikari, M. J., & Kopf, A. J. (2011). Cohesive strength of clay-rich sediment. Geophysical Research Letters, 38(16), L doi: /2011gl Karner, S. L., & Marone, C. (2000). Effects of Loading Rate and Normal Stress on Stress Drop and Stick-slip Recurrence Interval, Karner, S. L., & Marone, C. (2001). Frictional restrengthening in simulated fault gouge: Effect of shear load perturbations. Journal of Geophysical Research, 106(B9), 19,319 19,337. King, D. S. H., & Marone, C. (2012). Frictional properties of olivine at high temperature with applications to the strength and dynamics of the oceanic lithosphere. Journal of Geophysical Research, 117(B12), B doi: /2012jb Mair, K., & Marone, C. (1999). Friction of simulated fault gouge for a wide range of velocities and normal stresses. of Geophysical Research: Solid Earth (1978, 104(B12), 28,899 28,914. Retrieved from Mair, K., Frye, K. M., & Marone, C. (2002). Influence of grain characteristics on the friction of granular shear zones. Journal of Geophysical Research, 107(B10), 1 9. doi: /2001jb Marone, C., Carpenter, B., & Schiffer, P. (2008). Transition from Rolling to Jamming in Thin Granular Layers. Physical Review Letters, 101(24), doi: /physrevlett Marone, C. (1998). Laboratory-Derived Friction Laws and Their Application To Seismic Faulting. Annual Review of Earth and Planetary Sciences, 26(1), doi: /annurev.earth

40 Marone, C., Hobbs, B. E., & Ord, a. (1992). Coulomb constitutive laws for friction: Contrasts in frictional behavior for distributed and localized shear. Pure and Applied Geophysics PAGEOPH, 139(2), doi: /bf Marone, C., & Scholz, C. (1989). Particle-size distribution and microstructures within simulated fault gouge. Journal of Structural Geology, 11(7), Retrieved from Meer, S. D., Spiers, C. J., & Peach, C. J. (2000). Kinetics of precipitation of gypsum and implications for pressure-solution creep. Journal of the Geological Society, 157(2), doi: /jgs Muhuri, S. K., Dewers, T. a., Scott, T. E., & Reches, Z. (2003). Interseismic fault strengthening and earthquake-slip instability: Friction or cohesion? Geology, 31(10), 881. doi: /g Nielsen, A. E. (1987). Mechanisms and Rate Laws in Electrolyte Crystal Growth from Aqueous Solution, Niemeijer, a. R., & Spiers, C. J. (2007). A microphysical model for strong velocity weakening in phyllosilicate-bearing fault gouges. Journal of Geophysical Research, 112(B10), B doi: /2007jb Niemeijer, A., Marone, C., & Elsworth, D. (2008). Healing of simulated fault gouges aided by pressure solution: Results from rock analogue experiments. Journal of Geophysical Research, 113(B4), B doi: /2007jb Richardson, E., & Marone, C. (1999). Effects of normal stress vibrations on frictional healing. Journal of Geophysical Research:, 104(B12), 28,859 28,878. Retrieved from Rimstidt, J., & Barnes, H. (1980). The kinetics of silica-water reactions. Geochimica et Cosmochimica Acta, 44, Retrieved from Rowe, J. J., & Survey, U. S. G. (1962). The solubility of quartz in water in the temperature interval from 25 to 3OO C *, 26(1960), Stephen L. Karner, & Marone, C. (2000). Effects of Loading Rate and Normal Stress on Stress Drop and Stick-slip Recurrence Interval, Tenthorey, E., & Cox, S. F. (2006). Cohesive strengthening of fault zones during the interseismic period: An experimental study. Journal of Geophysical Research, 111(B9), B doi: /2005jb Tenthorey, E., & Cox, S. F. (2006). Cohesive strengthening of fault zones during the interseismic period: An experimental study. Journal of Geophysical Research, 111(B9), B doi: /2005jb Yasuhara, H., Elsworth, D., & Polak, A. (2003). A mechanistic model for compaction of granular aggregates moderated by pressure solution. Journal of Geophysical Research, 108(B11), doi: /2003jb Yasuhara, H., Marone, C., & Elsworth, D. (2005). Fault zone restrengthening and frictional healing: The role of pressure solution. Journal of Geophysical Research, 110(B6), B doi: /2004jb

41 Appendix A: Table of Experiment Experiment Powder Saturating Solution Procedure p3598 Quartz DI Water VS, SHS p3599 Quartz 3M NaCl VS, SHS p3614 Quartz DI Water VS, SHS p3615 Quartz 1M NaCl VS, SHS p3616 Quartz DI Water SHS p3617 Quartz DI Water VS, SHS p3651 Quartz 2M NaCl SHS p3652 Quartz 2M NaCl SHS p3653 Quartz 3M NaCl VS, SHS p3665 Gypsum 2M NaCl VS, SHS p3666 Gypsum DI Water SHS p3667 Gypsum 2M NaCl SHS p3680 Gypsum 2M NaCl VS, SHS p3681 Gypsum 2M NaCl VS, SHS p3752 Gypsum 2M Na2SO4 VS, SHS, SHS p3753 Gypsum 2M CaCl2 VS, SHS, SHS p3754 Gypsum 2M CaCl2 VS, SHS, SHS p3755 Gypsum 2M Na2SO4 VS, SHS, SHS, VS p3756 Gypsum 2M NaCl VS, SHS, SHS p3759 Gypsum 2M Na2SO4 VS, SHS p3760 Quartz 2M Na2SO4 VS, SHS p3862 Quartz DI Water SHS p3863 Gypsum 2M CaCl2 VS, SHS p3864 Quartz 2M CaCl2 VS, SHS p3865 Quartz 2M CaCl2 VS, SHS p3866 Quartz 2M NaCl VS, SHS p3867 Gypsum DI Water VS, SHS p3868 Gypsum DI Water VS, SHS p3869 Gypsum DI Water VS, SHS Table 1. List the experiments starting with the experiment number, the gouge material, saturating fluid, and the procedure after the run-in to achieve stable sliding friction. VS means velocity steps sequences and SHS means slide-hold-slide sequence. 36