Frictional stability of Longmaxi shale gouges and its implication for deep seismic potential in the southeastern Sichuan Basin

Microearthquakes accompanying shale gas recovery highlight the importance of exploring the frictional and stability properties of shale gouges. Aiming to reveal the in ﬂ uencing factors on fault stability, this paper explores the impact of mineral compositions, effective stress and temperature on the frictional stability of Longmaxi shale gouges in deep reservoirs located in the Luzhou area, southeastern Sichuan Basin. Eleven shear experiments were conducted to de ﬁ ne the frictional strength and stability of ﬁ ve shale gouges. The speci ﬁ c experimental conditions were as follows: temperatures: 90 – 270°C; a con ﬁ ning stress: 95 MPa; and pore ﬂ uid pressures: 25 – 55 MPa. The results show that all ﬁ ve shale gouges generally display high frictional strength with friction coef ﬁ cients ranging from 0.60 to 0.70 at the aforementioned experiment condition of pressures, and temperatures. Frictional stability is signi ﬁ cantly affected by temperature and mineral compositions, but is insensitive to variation in pore ﬂ uid pressures. Fault instability is enhanced at higher temperatures (especially at >200°C) and with higher tectosilicate/carbonate contents. The results demonstrate that the combined effect of mineral composition and temperature is particularly important for induced seismicity during hydraulic fracturing in deep shale reservoirs.


| INTRODUCTION
In recent years, induced earthquakes resulting from unconventional resource extraction activities have attracted growing public attention.Such activities include hydraulic fracturing for shale gas exploitation, wastewater injection, and CO 2 geological sequestration (Atkinson et al., 2020;Bao & Eaton, 2016;Cappa & Rutqvist, 2011;Clarke et al., 2019;Elsworth et al., 2016;Juanes et al., 2016).Several moderate to large earthquakes have been shown to be a threat to society.For example, the M L 5.7 earthquake on December 16, 2018 in southwest China related to shale gas exploitation by hydraulic fracturing has resulted in significant losses (Lei et al., 2019b).The Mw 5.5 earthquake that occurred on November 15, 2017 in Pohang of South Korea was known as the largest induced earthquake related to enhanced geothermal systems (EGS) in the world (Grigoli et al., 2018).In addition, induced earthquakes mainly associated with EGS and deep injection of wastewater occurred in Basel (Grigoli et al., 2017), Oklahoma and Arkansas (Ellsworth, 2013), and Texas (Schultz et al., 2020) as well.All of these show a clear correlation between induced seismicity with fluid injection and highlight the importance of understanding the underlying mechanisms of the reactivation of preexisting faults/fractures (Candela et al., 2018;Lei et al., 2019a;Verdon & Bommer, 2020).
The Sichuan Basin (Figure 1a) of China is rich in shale gas resources and has optimistic prospects for development (C.Zou et al., 2015).However, the number of earthquakes in this area has increased sharply since the initiation of hydraulic fracturing for shale gas exploitation and recovery (An, Zhang, Elsworth, et al., 2020) in 2011, principally due to hydraulic fracturing, especially in the Changning and Weiyuan areas (Ma, 2018).Meanwhile, multiple earthquakes with Mw > 4 have been reported, including the Mw 4.7 earthquake in the Changning block on January 27, 2017, 9 days after hydraulic fracturing (Lei et al., 2017), the Mw 4.8 earthquake in the Changning block on January 3, 2019, and the Mw 4.9 earthquake in the Weiyuan block on September 7, 2019 (Lei et al., 2019b).These induced seismic events have resulted in important socioeconomic impacts.All these events were demonstrated to be highly correlated to hydraulic fracturing.At present, the shale gas recovery in the Sichuan Basin is still in the developing stages and the main target shale reservoirs are generally located at depths <3.5 km.With the increase in reservoir depth in the southeastern Sichuan Basin, both the shale compositions and the geologic environments have significantly changed.In this sense, exploring the frictional and stability properties of the deep Longmaxi shale faults under hydrothermal conditions is of great significance in providing an insight into understanding the mechanisms of induced seismicity during hydraulic fracturing and subsequent production of the deep shale reservoirs.This research offers a basis for better choices in production safety and risk mitigation.
Frictional and stability properties of shale faults at shallow depth have been extensively studied in recent years (Fang et al., 2017;Guglielmi et al., 2015;Jia et al., 2019;Scuderi & Collettini, 2018;F. Zhang et al., 2017F. Zhang et al., , 2019)).Evidently, as is shown by the results of previous related studies, the frictional properties of the preexisting fault or fracture exert some significant controls on its reactivation, especially the mineral compositions, temperatures, and applied stresses (Fang et al., 2016;Scuderi & Collettini, 2016;Segall & Lu, 2015;Yasuhara et al., 2004).In terms of mineralogy, the contents of phyllosilicate minerals play an important role in fault stability (An, Zhang, Elsworth, et al., 2020;Fang et al., 2018;L. Zhang & He, 2016), and faults with less phyllosilicates are prone to exhibit unstable slips (Kohli & Zoback, 2013).The high temperature not only changes the gouge composition through chemical reactions but also accelerates fluid-assisted stages including pressure dissolution (Yasuhara et al., 2005).In addition, the change in effective stress was generally associated with the change in pore pressure caused by fluid injection.A higher pore fluid pressure often leads to reduced effective stress on pre-existing faults, thus resulting in fault instability (Candela et al., 2015;Elsworth et al., 2016).Considering the frequent fluid injection in the southern Sichuan Basin and the ambiguous mechanisms of induced earthquakes in Longmaxi shales, it is particularly important to explore the parameters controlling the instability of deep shale faults (López-Comino & Cesca, 2018).
The first-order trends were defined in the mineral composition of the Longmaxi shale in the southeastern Sichuan Basin.Based on the data on these compositions, tri-axial shear experiments were conducted at high temperatures (90-270°C) and high fluid pressures (25-55 MPa) to study the evolution of friction and stability in the simulated Longmaxi shale gouges.The geological background of the southeastern Sichuan Basin and the existing record of induced earthquakes (Chang et al., 2020) were combined to define the coupled influences of the mineral composition, effective stress and temperature on the clusters of earthquakes in the Longmaxi formation due to hydraulic fracturing.

| Gouge preparations
Five samples of the deep Longmaxi shales were collected from rock core cuttings in the southeastern Sichuan Basin.All samples (Figure 1c) were recovered from boreholes with depths > 3700 m in the Luzhou block (Figure 1b).After removing the surface impurities, all shales were crushed through a pestle and mortar and then sieved to a grain diameter < 74 μm to simulate comminuted fault gouge.The mineral compositions of the five samples were tested by X-ray diffraction method and the results are shown in Table 1.Abundant minerals in Longmaxi shales include quartz, plagioclase, calcite, dolomite, and clay minerals with a trace of pyrite.With an increase in depth, the quartz content first increases and then decreases, while the total clay content displays the opposite trend.The highest quartz content in the studied shales reaches 68 wt.%, and the minimum is only 26 wt.%.The total clay content ranges from 5 to 38 wt.% (Table 1).The mineral composition of the Longmaxi shale samples is characterized by an increase in tectosilicate contents.

| Apparatus and testing procedure
The shear experiments on simulated gouges were completed in an argon-gas-confined high-temperature and high-pressure tri-axial shear apparatus (Figure 2) at the Institute of Geology, China Earthquake Administration, Beijing, China.The maximum temperature for the apparatus is 600°C, the maximum confining pressure is 420 MPa, and the maximum pore fluid pressure is 200 MPa.A 1-mm-thick gouge layer was emplaced between the upper and lower forcing blocks of saw-cut cylinders (20 mm in diameter; 40 mm in height) and at an inclination of 35°.Two perpendicular holes were drilled in the upper forcing block as a channel for pore fluid entry.The porous brass filter was inserted into the two boreholes to prevent gouge extrusion while maintaining a high permeability of pore fluid.The gouge thickness was guaranteed by a specially made cylindrical leveling jig.
Once filled with a layer of gouge, the gabbro block was inserted into a copper jacket with a thickness of 0.35 mm, followed by placing the cylindrical tungsten-carbide and corundum blocks on both sides.The boron nitride powder was adopted to fill the space between the copper tube and the heating furnace to prevent gas convection.Two  O-rings were placed on the steel end blocks on either side of the copper tube to isolate samples from the argon gas.
The high-pressure seal on the upper piston could prevent the gas from escaping from the pressure vessel and thus seal the whole experimental system.The experimental procedures are as follows: First, the confining and pore fluid pressures were raised to 70%-80% of the target values.Then, the temperature was raised to the target value at a heating rate of 5°C/min.A thermocouple was installed at the pore fluid inlet to measure the temperature.The temperature was maintained constant within ±2°C by an independent controller during the experiments.When the temperature was raised to the desired value, the confining and pore fluid pressures were elevated to the target values and remained stable within ±0.5 and ±0.3 MPa by the servo control system.At the beginning of each experiment, the sample was loaded at a constant axial shear velocity of 1.0 μm/s.Once the steady state of friction was achieved, the axial shear velocity first declined and then increased, following the sequence of 1-0.2-0.04-0.2-1-0.2-0.04 μm/s, equivalent to the velocities along the shear direction switching between 1.22, 0.244, 0.0488, 0.244, 1.22, 0.244, and 0.0488 μm/s successively.The final shear displacements were generally within the range of 3-4 mm.

| Data analysis
In the experiments, the raw data including the axial load, pore fluid pressure, temperature, and axial displacement at a sampling frequency of 1 Hz were recorded.The raw data were corrected according to the method proposed by He et al. (2006) to incorporate the influences caused by the change in contact area with increasing displacement and the shear resistance of the copper tube.The frictional strength of the studied gouges was determined by the frictional coefficient μ, which could be calculated from shear stress τ divided by effective normal stress σ neff (ignoring the cohesion), shown as follows: where σ n represents the applied normal stress and p f denotes the applied pore fluid pressure.
The velocity dependence of friction (a − b) can be explained according to the rate-and state-friction (RSF) constitutive theory (Dieterich, 1978(Dieterich, , 1979;;Marone, 1998;Ruina, 1983).In the category of RSF theory, the frictional coefficient μ is calculated as where μ is the coefficient of friction at the instantaneous shear velocity V, μ 0 denotes the coefficient of friction at the reference velocity V 0 , θ represents a state variable, a and b refer to the frictional parameters representing the direct and evolutional effects caused by the step change The positive value of (a − b) indicates velocitystrengthening behavior under a velocity step (Figure 3a) and generally promotes stable fault sliding.Conversely, a negative value of (a − b) represents velocity-weakening behavior, and unstable sliding occurs whenever the critical stiffness is satisfied (Figure 3a).When the gouges show stick-slip behavior, the method for calculation of (a − b) is as shown in Figure 3b.In addition, the slope was also detrended in calculating the (a − b) values following the methods in Figure 3c.

| Impact of mineral composition on shale gouge friction
By using the method in Section 2.3, five shear experiments on different shale gouges were conducted under experimental conditions of σ c = 95 MPa, p = 40 MPa, and T = 130°C, with the coefficient of friction-shear displacement curves shown in Figures 4 and 6b.In the first ~0.5 mm of shear displacement, each gouge displayed a linear increase in friction until approaching a steady state.After a shear displacement of ~1.0 mm, all gouges showed a strain-hardening response.The steady-state frictional coefficients were evaluated at a shear displacement of approximately ~2.5 mm and a shear velocity of 1.22 μm/s and the results are shown in Figure 5a.The frictional coefficient μ of the five shale gouges was in the range of ~0.55-0.7,consistent with those of previous studies on shale gouges with similar phyllosilicate contents (Kohli & Zoback, 2013).A higher frictional coefficient in shale gouge SSC3740-2 was possibly due to a higher carbonate content (~25 wt.%) and a lower phyllosilicate content (~5 wt.%) (Figure 5a and Table 1).Conversely, the shale gouge SSC3750 showed the lowest coefficient of friction (~0.57) and this possibly resulted from its highest phyllosilicate content (38 wt.%).By using the methods in Section 2.3, frictional stability (a − b) was calculated and the results are presented in Figure 5b.Values of (a − b) for the five shale gouges were in the range of ~0.001-0.008,suggesting that all shales were prone to contribute to a stable fault slip (Figure 5b and Table 3).Meanwhile, there existed a clear correlation between the frictional stability (a − b) and the phyllosilicate content and this phenomenon is in accord with the results of Ikari et al. (2011).

| Impact of temperature on shale gouge friction
To explore the temperature dependence of frictional coefficient and stability, four shear experiments were conducted on shale gouge SSC3730-1 under the  3)., the frictional coefficient slightly rose from 0.67 to 0.69 when the pore fluid pressure was raised from 25 to 55 MPa, showing that the effective stress variation exerts only a minor impact on gouge frictional strength (Figure 9a).As pore fluid pressure rose, the gouge frictional stability also increased slightly, especially at 0.244 -0.0488 μm/s, indicating that varying the pore fluid pressure also has a minor impact on shale gouge stability (Figure 9b).7b).Meanwhile, this stability is accompanied by a slight increase in frictional coefficient.This temperature dependence of gouge friction and stability is generally consistent with previous results on phyllosilicate-bearing gouge (Den Hartog et al., 2012) (Figure 10).The variation of the gouge frictional stability with temperature can be partially clarified by a microphysical model for phyllosilicate-quartz-dominated gouge (Den Hartog & Spiers, 2014;Niemeijer & Spiers, 2007).In this model, the changes in frictional stability with temperature are caused by the competition between thermally activated gouge compaction and shear-induced dilation.At higher temperatures (T = 270°C), the gouge compaction is strong due to the pressure solution and this leads to the compaction dominating the gouge deformation at an upward velocity step.As a consequence, the coefficient of friction would be reduced (Samuelson et al., 2009) and contributes to the observed velocityweakening behavior.On the contrary, at lower temperatures (T = 90-200°C), the gouge dilation dominated the gouge deformation upon an upward velocity step.This leads to an increased coefficient of friction and the observed velocity-strengthening behavior.
The mineral compositions in shales can mainly be divided into three groups, namely, tectosilicates, carbonates, and phyllosilicates.Tectosilicate or carbonate-rich gouges are prone to promote fault instability.Compared with the gouge SSC3730-1, the gouge SSC3740-2 at increasing temperatures exhibits a higher coefficient of friction (μ) and an earlier emergence of negative frictional stability (a − b).This phenomenon is closely related to the mineral composition of two shale gouges.The gouge SSC3740-2 shows a higher tectosilicate content but a lower phyllosilicate content than the gouge SSC3730-1.Compared with many previous research works (An, Zhang, Chen, et al., 2020;Chen et al., 2015;Fang et al., 2018;Kohli & Zoback, 2013), the mineral compositions are shown to be an important factor for shale fault stability.

| Implications for shale gas recovery
According to previous studies, seismicity associated with hydraulic fracturing during shale gas recovery is closely related to the reactivation of pre-existing faults (An, Zhang, Chen, et al., 2020;Igonin et al., 2021).The results obtained in this study are of great significance for exploring the frictional response of deep faults in shale reservoirs of the southeastern Sichuan Basin.The experimental results show that the fault frictional properties are influenced by the temperature and mineral compositions during hydraulic fracturing.Since the depth of current hydraulic fracturing is generally shallower than 3.5 km, the in situ temperature in these reservoirs should be less than 150°C (Ma & Xie, 2018).According to our experimental results, most shale fault slips in these shallow reservoirs should be aseismic.However, three earthquakes with Mw ≥ 4.7 have been detected in this area (2017-2019), which have been proven to be closely associated with hydraulic fracturing (Wang et al., 2022).This manifests the possibility of moderate earthquakes in shallow shale reservoirs.With the shale gas exploitation toward a deeper depth in the southeastern Sichuan Basin (Xu et al., 2019;C. Zou et al., 2015), the temperature would be much higher than the shallow reservoirs and thus the seismic hazard would also be increased in deeper reservoirs.
The magnitude and number of seismic events are closely related to the reservoir temperature.From the experimental results, the frictional instability of shale gouges is enhanced with an increase in reservoir temperature, indicating that the seismic hazard also increases at higher temperatures.The burial depths of the Longmaxi formation are generally in the range of 1.5-5 km in the southern Sichuan Basin, with corresponding in situ temperatures in the range of 50-180°C.As the depth of the target reservoir increases, the potential for fault instability and hazard of induced seismicity will increase dramatically.Although temperature is not the main factor controlling seismicity in the shallow shale reservoirs (An, Zhang, Chen, et al., 2020), the high temperature can impact the fault stability in deep shale reservoirs.Additionally, the mineral composition of the deep Longmaxi shale reservoir varies significantly and may also favor velocity-weakening behavior at increased depths.From Figure 5b, it can be seen that the mineral composition changes significantly with the change in burial depth and this change exerts a strong control on frictional stability.Therefore, the combined effect of temperature and mineral composition is particularly important for the occurrence of seismic activities at deep shale reservoirs.

| CONCLUSIONS
We explored the effects of the variations in mineral composition, temperature, and effective stress on the frictional and stability properties of Longmaxi shale gouges at depths that are representative of the southeastern Sichuan Basin, southwest China.The mineral compositions of the collected shales were analyzed and the shear experiments were performed under hydrothermal conditions.The mineral compositions in five Longmaxi shales from the Luzhou block mainly include quartz, plagioclase, calcite, dolomite, and clay minerals (mainly illite and chlorite) and vary with increasing depth.Both the mineral composition and temperature can impact the frictional coefficient and stability of Longmaxi shale gouges, while the effective stress variation only exerts a minor control.The frictional strengths increase with increasing tectosilicate and carbonate contents and the gouge instability is also enhanced at higher tectosilicate and carbonate contents.With the increasing temperature, the frictional strength of shale gouges increases slightly.However, a transition from velocity-strengthening to velocity-weakening behavior is identified at a threshold of temperature of 200°C.The combined effect of temperature and mineral composition is important for controlling the shale fault frictional response, especially for the deeper shale reservoirs in the southeastern Sichuan Basin.
The location of the research region and the shale samples.(a) The research region of Sichuan Basin.(b) Luzhou area and Weiyuan area in the southeastern Sichuan Basin (Lei et al., 2020).(c) The sketch plots of the collected five shale samples.

|
Experimental curves showing the gouge SSC3730-1 at a confining pressure of 95 MPa, a pore fluid pressure of 40 MPa, and temperatures of (a) 90°C, (b) 130°C, (c) 200°C, and (d) 270°C, respectively.Impact of effective stress on shale gouge friction Three shear experiments were conducted on the shale gouge SSC3740-2 at σ c = 95 MPa, T = 200°C, and pore fluid pressures of 25, 40, and 55 MPa to examine the stress dependence of friction and stability.The coefficient of friction-shear displacement curves are shown in Figure 8.At T = 200°C

F
Experimental results of μ and (a − b) of the shale gouge SSC3730-1 at a confining pressure of 95 MPa, a pore fluid pressure of 40 MPa, and different temperatures.The legend shows the shear velocities.(a) coefficient of friction, (b) frictional stability.I G U R E 8 Experimental curves showing the gouge SSC3740-2 at a confining pressure of 95 MPa, a temperature of 200°C, and pore fluid pressures of (a) 25 MPa, (b) 40 MPa, and (c) 55 MPa.4.1 | Dependence of shale gouge friction on temperature and mineral composition The depth of the Longmaxi formation shales varies widely in the southern Sichuan Basin and the Longmaxi formation, therefore, covers a wide range of temperature conditions.From our experimental results, the temperature can affect both the frictional coefficient (μ) and stability (a − b) of shale gouges.Four experiments on the shale gouge SSC3730-1 conducted at σ c = 95 MPa, P f = 40 MPa, and T = 90-270°C exhibit two kinds of velocity dependence.It is characterized by velocity-strengthening behavior (a − b > 0) at T = 90-200°C and a velocity-weakening behavior (a − b < 0) at T = 270°C (Figure Experimental results of μ and (a − b) of the shale gouge SSC3740-2 at a confining pressure of 95 MPa, a temperature of 200°C, and pore fluid pressures of 25, 40, and 55 MPa.The legend shows the shear velocities.(a) coefficient of friction, (b) frictional stability.U R E 10 Experimental results of (a) coefficient of friction μ and (b) frictional stability (a − b) of the shale gouge SSC3740-2 at a confining pressure of 95 MPa, a pore fluid pressure of 40 MPa, and temperatures of 130°C and 200°C.The legend shows the shear velocities.
shear velocity, respectively, and D c denotes the critical slip distance from the past steady state to a new steady state.In a steady state of friction, the velocity dependence (a − b) is calculated as Abbreviations: σ c , confining pressure; l final , final shear displacement; P f , pore fluid pressure; ss, stick-slip; T, temperature; vs, velocity-strengthening; vw, velocity-weakening. of Results of μ and (a − b) values for all shear experiments T A B L E 3