Frictional strength and healing behaviorof phyllosilicate-rich faults

Telemaco Tesei,1 Cristiano Collettini,1,2 Brett M. Carpenter,3,4 Cecilia Viti,5

and Chris Marone3

Received 5 February 2012; revised 18 July 2012; accepted 19 July 2012; published 6 September 2012.

[1] We study the mechanisms of frictional strength recovery for tectonic faults withparticular focus on fault gouge that contains phyllosilicate minerals. We report laboratoryand microstructural work from fault rocks associated with a regional, low-angle normalfault in Central Italy. Experiments were conducted in a biaxial deformation apparatus atroom temperature and humidity, nominally dry, under constant normal stresses of 20 and50 MPa, and at a sliding velocity of 10 mm/s. Our results for nominally dry conditionsshow good agreement with previous work conducted under controlled pore fluid pressure.The phyllosilicate contents of our samples, which include clay, talc and chlorite rangefrom 0 to 52 weight %. We study both intact rock samples, sheared in their in situgeometry, and powders made from the same rocks to address the role of fabric in faulthealing. We measured frictional healing, Dm, using slide-hold-slide tests with holdperiods ranging from 3 to 3000 s. Phyllosilicate-free materials show friction values of m ≈0.6 and healing rates that are larger in powdered samples, b ≈ 0.006 (Dm per decade intime, s) compared to intact wafers of fault rock, b ≈ 0.004. For phyllosilicate-bearingmaterials, healing rates are low, b < 0.002, and independent of fabric, phyllosilicate contentand normal stress. We observe that frictional strength decreases systematically withincreasing phyllosilicate content. Intact, phyllosilicate-bearing fault rock is consistentlyweaker than its powdered equivalent (0.2 < m < 0.3 versus 0.4 < m < 0.5, respectively). Wecompare our data to results from experiments conducted on a wide range of materialsand conditions. Deformation microstructures show localized slipping along sub-parallelshear planes. We suggest that low values of frictional strength and near zero healing rateswill combine to exacerbate the weakness of phyllosilicate-bearing faults and promotestable, aseismic creep.

Citation: Tesei, T., C. Collettini, B. M. Carpenter, C. Viti, and C. Marone (2012), Frictional strength and healing behavior ofphyllosilicate-rich faults, J. Geophys. Res., 117, B09402, doi:10.1029/2012JB009204.

1. Introduction

[2] Geologic observations, seismological data, and labo-ratory friction studies are consistent with the stick-slip modelof the seismic cycle [e.g., Byerlee, 1978; Scholz, 2002], inwhich slip instabilities (earthquakes) are associated withstress drops and interseismic strength recovery. Recent workhas documented a broad spectrum of slip behaviors, rangingfrom stable, aseismic creep to slow slip, tremor, and low

frequency earthquakes [e.g., Kanamori and Kikuchi, 1993;Outerbridge et al., 2010; Peng and Gomberg, 2010].[3] For all types of seismic and aseismic faulting, fault slip

is driven by a reduction in the ambient tectonic stress storedelastically in the surrounding rocks via frictional dissipation.However, the mode of this dissipation varies depending onthe rheology of the fault zone. Differences in frictionalbehavior can arise from the mineralogical and structuralevolution of the fault zone. Another fundamental require-ment of the seismic cycle of repeated failure is that faultsregain strength after failure, during the interseismic periodof stress re-accumulation. We refer to “fault healing” as theensemble of re-strengthening processes that occur afterslip, which likely involve cohesive- as well as frictional-strengthening [e.g., Dieterich, 1972; Muhuri et al., 2003].Fault healing is a central problem in earthquake physicsbecause it is likely to control the dynamics of rupture initi-ation and propagation [Das and Aki, 1977; Rice, 1980], aswell as the stress drop and recurrence time of earthquakes[Dieterich, 1972; Scholz, 2002].

1Dipartimento di Scienze della Terra, Università degli Studi di Perugia,Perugia, Italy.

2Istituto Nazionale di Geofisica e Vulcanologia, Rome, Italy.3Rock and Sediment Mechanics Laboratory, Pennsylvania State

University, University Park, Pennsylvania, USA.4Now at Istituto Nazionale di Geofisica e Vulcanologia, Rome, Italy.5Dipartimento di Scienze della Terra, Università di Siena, Siena, Italy.

Corresponding author: T. Tesei, Dipartimento di Scienze della Terra,Università degli Studi di Perugia, Piazza Università 1, IT-06123 Perugia,Italy. (telemaco.tesei@ingv.it)

©2012. American Geophysical Union. All Rights Reserved.0148-0227/12/2012JB009204

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 117, B09402, doi:10.1029/2012JB009204, 2012

B09402 1 of 13

1.1. Fault Gouge and Phyllosilicates[4] Published studies have documented the widespread

occurrence of phyllosilicate-bearing faults in different tec-tonic settings, including subduction zones [e.g., Vannucchiet al., 2008; Ikari et al., 2009a; Fagereng and Sibson,2010], crustal scale strike-slip structures [e.g., Wu et al.,1975; Vrolijk and van der Pluijm, 1999; Faulkner et al.,2003; Holdsworth 2004; Jefferies et al., 2006; Moore andRymer, 2007; Schleicher et al., 2010; Buatier et al., 2012;Holdsworth et al., 2011] and extensional detachments [e.g.,Wintsch et al., 1995; Manatschal, 1999; Collettini andHoldsworth, 2004; Numelin et al., 2007; Viti and Collettini,2009]. Phyllosilicate minerals commonly found in faultzones include micas, serpentines, talc, and different speciesof expandable and non-expandable clays such as chlorite,smectite, illite, kaolinite, saponite and other mixed-layerclays (see references above).[5] Phyllosilicate-bearing fault rocks have distinctly dif-

ferent mechanical properties than most other crustal materials.These properties include low friction (in general m < 0.4 for agreat number of phyllosilicates) and, for some mineral spe-cies, velocity strengthening frictional behavior under a vari-ety of crustal depth, temperature and pore pressure conditions[e.g.,Morrow et al., 1992; Saffer andMarone, 2003;Hong andMarone, 2005; Numelin et al., 2007; Moore and Lockner,2008; Carpenter et al., 2009; Haines et al., 2009; Ikari et al.,2009a; Ikari et al., 2009b; Smith and Faulkner, 2010;Collettini et al., 2011; Ikari et al., 2011a, 2011b; Moore andLockner, 2011; Sone et al., 2012]. The frictional properties ofphyllosilicates have been invoked to explain the weaknessand creeping behavior of faults such as the San Andreas[Brune et al., 1969; Wu et al., 1975; Carpenter et al., 2011;Lockner et al., 2011] and some low-angle normal faults[Chiaraluce et al., 2007; Collettini et al., 2009a;Hreinsdottirand Bennett, 2009]. Because of their common occurrenceand their properties, it is important to investigate the fric-tional and mechanical properties of phyllosilicate minerals.

1.2. Friction and Fault Healing[6] Fault healing is composed of a complex series of pro-

cesses. Published studies indicate that these processesinclude frictional healing [e.g., Dieterich, 1972; Beeler et al.,1994; Marone, 1998; Karner and Marone, 1998; Frye andMarone, 2002], pressure-solution welding [Bos and Spiers,2002; Niemeijer et al., 2008] and fault zone cementation[Hickman and Evans, 1992; Karner et al., 1997; Olsen et al.,1998;Muhuri et al., 2003; Tenthorey et al., 2003]. Frictionalstrength recovery has been most extensively studied in thelaboratory and most widely applied to fault healing [e.g.,Scholz et al., 1986; Kanamori and Allen, 1986; Vidale et al.,1994; Marone et al., 1995]. In laboratory experiments, fric-tional healing has been studied via slide-hold-slide experi-ments in which slip is imposed over a finite distance (usuallyat constant loading velocity) and then halted for a time, fol-lowed by reloading. In the lab, a combination of processesthat cause frictional strength to increase occur during holdperiods, in a manner that is similar to the interseismic intervalof the seismic cycle.[7] Frictional healing is a common effect and numerous

experimental studies show that friction increases roughlylinearly with the log of contact time [e.g., Coulomb, 1785;

Scholz, 2002]. Laboratory studies indicate that frictionalhealing arises from growth of the asperity contact junctionsthat determine friction [e.g., Rabinowicz, 1951; Dieterichand Kilgore, 1994] or from strengthening of these contacts[e.g., Beeler et al., 1994; Li et al., 2011]. For silicate rocks,existing works show that frictional healing involves hydro-lytic weakening and related effects [Dieterich and Conrad,1984; Frye and Marone, 2002].[8] Most of the geophysical experiments on frictional

healing have been carried out on simulated fault gouges(powders) or on relatively smooth, man-made rock surfaces,using mainly quartzo-feldspathic materials, at room tem-perature and humidity [e.g., Dieterich and Kilgore, 1994].Under these conditions, most materials have friction coeffi-cients close to m = 0.6–0.7 [Byerlee, 1978] and healing rates(b expressed as increase in friction per decade time) between0.005 and 0.01.[9] Despite the significant progress made in recent work,

the role of phyllosilicates in fault healing remains poorlyunderstood (see Bos and Spiers, 2000; Niemeijer et al.,2010; Carpenter et al., 2011]. The purpose of this paper isto describe a suite of laboratory experiments designed toaddress frictional healing of phyllosilicate-bearing rocks,including the role of fault zone shear fabric and fault rockmineralogy. Experiments have been performed at roomtemperature and humidity; nevertheless the role fluids, whichin general permeate tectonic faults, in influencing friction andhealing processes has been discussed in section 4.

2. Sampling and Experimental Procedure

[10] We collected foliated samples from the basalphyllosilicate-rich horizon of the Zuccale low-angle normalfault, Central Italy [Collettini and Holdsworth, 2004; Smithet al., 2007]. Samples from this regional, low-angle normalfault have been the subject of previous experimental workfocused on frictional properties [Smith and Faulkner, 2010],the role of phyllosilicate-rich fabrics in fault weakness[Collettini et al., 2009b], and the relationship between faultzone structure, frictional properties, and mixed-mode slipbehavior [Collettini et al., 2011].[11] We determined the mineralogical composition of our

samples using X-ray powder diffraction (XRPD) coupledwith thermal analyses and scanning electron microscopy(SEM) equipped with energy dispersive spectrometry(EDS). Thermo-gravimetry (TG) provided reliable estimatesof calcite content, via decarbonation weight loss. Emittedgas analysis (EGA) mass spectrometry was coupled with TGin order to determine the decarbonation temperature range.XRPD provided qualitative mineralogical determinations,revealing calcite, amphiboles, talc, and clay minerals (pre-dominantly smectite and mixed-layer clays with minorchlorite). XRPD patterns were recorded on oriented samples,both untreated and glycolated, allowing expandable clays tobe distinguished from chlorite (Figure 1). Semiquantitativemineralogical determinations for phyllosilicates and amphi-boles were done on the basis of XRPD peak intensities in theglycolated patterns. Mineral contents are summarized inTable 1, and show that samples T1, T7 and T8 have similarmineralogical compositions and probably belong to the samefault rock analyzed by Viti and Collettini [2009]. Sample T4differs slightly from previous samples due to the absence of

TESEI ET AL.: FRICTION AND FAULT HEALING B09402B09402

2 of 13

talc and to the occurrence of trioctahedral mixed-layer clays,which resemble corrensite based on glycolated XRPD pat-terns and EDS chemical compositions. Sample T2 is exclu-sively composed of amphibole (in particular, actinolite) andrepresents a completely different protolith.[12] During field collection, we documented fabric orien-

tation and in situ shear direction using traditional kinematicindicators, i.e., sigmoid shape and orientation and grooves.To produce the laboratory samples, fault rocks were first cutusing a precision ring saw (Gemini Saw Co.) to producewafers !0.9–1.2 cm thick and 5x5 cm nominal area. Theseintact wafers were then hand-machined, using a rotarygrinder, on the loading platens for double-direct shear.Samples were oriented such that they could be subject tofrictional shear parallel to the in situ shear plane. We crushedthe fragments left after cutting wafers and powdered these ina disk mill to attain an equivalent composition fault powderwith a grain size <150 mm. Experiments on powders werecarried out on layers 0.5 cm thick and 5x5 cm in area.[13] Experiments were conducted in a servo-controlled,

biaxial testing apparatus at the Pennsylvania State Univer-sity Rock and Sediments Mechanics lab using a double-direct shear configuration [e.g., Mair and Marone, 1999;Ikari et al., 2011a] in which two layers (or intact wafers) aresheared simultaneously between three steel forcing blocks(inset in Figure 2a) in a simple shear strain mode. Ourexperiments were conducted at room temperature andhumidity, under nominally dry conditions. Normal stress

(sn) was applied and held constant via a hydraulic ram act-ing on one of the steel side blocks, with the other side blocksupported by the opposite side of the load frame. The blockswere positioned so that the central steel block was directlybelow the vertical loading ram after layer compaction. Sheardeformation was accomplished by driving the vertical cen-tral block down between the two stationary side blocks (insetin Figure 2a). Stresses were measured continuously usingBe-Cu strain gauge load cells, accurate to "0.01 MPa.Friction (m) was calculated by dividing the shear stress by thenormal stresses assuming zero cohesion. Displacement nor-mal to the layer represents compaction or dilation and wasmeasured continuously, along with shear displacement, viadirect-current displacement transducers accurate to"0.1 mm.We derived a continuous measure of average shear strain

Table 1. Mineralogical Composition of Tested Samplesa

T1 T2 T4 T7 T8

Calcite 54 0 33 55 51Tremolite 20 0 15 9 5Actinolite 0 100 0 0 0Chlorite 5 0 0 3 1Smectite 17 0 25 9 16Mixed-layer clays 0 0 27 0 0Talc 4 0 0 24 27Total 100 100 100 100 100Total Phyllosilicates 26 0 52 36 44

aUnit is wt. %.

Figure 1. XRPD patterns of phyllosilicate-rich samples in the 4–15# 2-theta range, showing basal reflec-tions of smectite (sm), mixed-layered clays (m.l.), talc (tlc) and chlorite (chl), together with a sharp peak oftremolite (tr). The low angle peaks at !14.8 Å shift to 15.9–16.6 Å in glycolated samples T1g, T7g andT8g, indicating the occurrence of expandable smectite instead of chlorite. The XRPD pattern of glycolatedT4g is consistent with the occurrence of trioctahedral mixed-layered clays (chlorite/smectite interlayering).Int = diffraction intensity (arbitrary units, a.u.).

TESEI ET AL.: FRICTION AND FAULT HEALING B09402B09402

3 of 13

within the layers by dividing displacement increments by themeasured layer thickness and summing. Digital data wererecorded with a"10V, 24-bit system at 10 kHz and averagedfor recording from 1 to 100 Hz, depending on the shearingrate. Experiments were performed at constant applied normalstress (20 and 50 MPa) and a constant sliding velocity of10 mm/s. The two normal stresses were chosen as represen-tative of very shallow crustal depths, representative of thefault exhumation (<6 km) [see Collettini and Holdsworth,2004], and so that we could assess normal stress depen-dence of frictional healing. At our experimental conditions

the precision of friction coefficient (and thus frictionalhealing) measurements is "0.001 or less.[14] All experiments began with the application of normal

stress until sample compaction ceased, followed by a sheardisplacement “run-in” of 5–7 mm (Figure 2). The run-inensures a consistent state of sample compaction, and isparticularly important for experiment-to-experiment repro-ducibility in powdered experiments. We typically observe asteady state dynamic friction md after a small amount ofshearing (Figure 2a).[15] Following steady shear, we performed slide-hold-slide

tests by sliding at 10 mm/s for 100 s followed by holds(driving velocity v = 0 mm/s) ranging from 3 to 3000 s(Figure 2a) following the same procedure as in Carpenteret al. [2011]. During holds, shear stress decays due toinelastic creep within the sample (Figure 2b). After eachhold, shearing is resumed and another cycle is started. Staticfriction, ms, measured upon re-shear is the yield strength ofthe simulated fault, whereas the frictional healing,Dm, is thedifference between the static friction and the pre-hold steadydynamic friction. The dynamic friction is generally identicalbefore and after the hold. We vary the duration of the holdtime to assess the healing rate b =Dm/Dlog10(th). At the endof each experiment the sheared layers, which experiencedshear displacements of 12 to 15 mm, were carefully removedfrom the sample assembly, preserved in epoxy resin, andsliced parallel to the sense of shear for petrographic analysis,optical microscopy, and scanning electron microscopy(SEM). Table 2 summarizes parameters for our suite ofexperiments. In general, experiments on powders showhigher values of mean shear strain because they compactmore in comparison to intact wafers.[16] Because our experiments were conducted on natural

rock samples, which vary slightly from one location toanother, it was not possible to maintain the exact samesample mineralogy for tests at each normal stress. However,all samples, except T2, have phyllosilicate contents in therange of 21–52 weight%.

3. Results

3.1. Frictional Strength[17] In our tests, the coefficient of friction is measured

continuously and commonly evolves with shear displacement(Figure 2). To evaluate reproducibility between tests and

Table 2. Slide-Hold-Slide Experiments and Main ExperimentalCharacteristics

Experiment Sample FabricNormalStress

Max. ShearStrain

P3260 T7 Wafer 50 2.16P3261 T2 Wafer 50 2.17P3262 T1 Wafer 50 2.55P3263 T7 Powdered 50 4.37P3264 T2 Powdered 50 4.31P3265 T1 Powdered 50 3.73P3281 T7 Wafer 20 1.88P3282 T8 Wafer 20 2.23P3283 T4 Wafer 20 2.08P3291 T7 Powdered 20 4.39P3292 T8 Powdered 20 2.17P3293 T4 Powdered 20 4.21

Figure 2. (a) Representative slide-hold-slide frictionexperiment (p3260). During the “run-in,” friction reachesa maximum and then stabilizes around a md value duringsteady sliding. After the run-in, the slide-hold-slide sequenceis started. This experiment shows zero healing and evenslight strain weakening after hold periods. Inset shows thedouble-direct experimental configuration where 3 steelblocks shear two identical layers of a sample. (b) Detail offriction versus time for a 100 s hold from another experiment(p3264). Friction decays from the steady dynamic value, md,during the hold. The static friction, ms, is the maximumfriction measured after the hold. Frictional healing, Dm, isthe difference between peak static friction and the pre-holddynamic friction.

TESEI ET AL.: FRICTION AND FAULT HEALING B09402B09402

4 of 13

compare results among samples, we picked dynamic frictionvalues at the end of both the run-in phase and just before eachhold step (Figures 2a and 2b). Phyllosilicate mineralogy andcontent has a striking influence on the frictional strength ofour samples. In our samples and at our experimental condi-tions (room dry), with the exception of chlorite that is presentin limited amount (5% in T1, 3% in T7 and 1% in T8), all theother phyllosilicates (smectite, mixed-layer clays and talc)are inferred to be weak. Subsequently, when we compare thefrictional properties of the fault rocks with the percent of thephyllosilicates, we always refer only to weak phyllosilicateminerals (smectite, mixed-layer clays and talc).[18] Friction of powdered fault gouge (Figure 3) decreases

systematically with increasing weak, phyllosilicate content.For example, at a normal stress of 50 MPa, sample T2,which has 0% phyllosilicates, has dynamic friction md ≈0.62. All phyllosilicate-rich samples are weaker (md < 0.5).That is: sample T1 (21% phyllosilicates) has md ≈ 0.48 andsample T7 (33% phyllosilicates) md ≈ 0.45 (Figure 3a).We find that phyllosilicate-rich samples are weak at bothnormal stresses studied. For experiments at 20 MPa normalstress, sample T7 (33% phyllosilicates) has dynamic friction,

md ≈ 0.49, sample T8 (43% phyllosilicates) md ≈ 0.44 andsample T4 (52% phyllosilicates) md ≈ 0.39 (Figure 3b).[19] Figure 4 shows shear strain versus dynamic fiction for

all the tested samples. To a first approximation, friction isindependent of net strain, with differences in dynamic fric-tion less than 0.03; the only exception being the solid waferT2 sample, which exhibits a marked strain weakening. Thefrictional strength of intact phyllosilicate-rich rocks is sig-nificantly lower (md < 0.3) than the frictional strength of theirpowdered equivalents (0.4 < md < 0.5). In contrast,amphibolite solid wafer T2 exhibits friction 0.56 < md < 0.62with little difference compared to its own powder. The mostphyllosilicate-rich sample, T4, exhibits the lowest frictionvalues for both intact- and powdered-fault rock.

3.2. Healing[20] We measured healing as the difference between static

friction at reshear and the pre-hold steady state dynamicfriction (Dm) for each hold (as in previous works byMarone[1998], Richardson and Marone [1999], and Frye andMarone [2002]). For some samples, we find a slight strain-related trend in friction vs shear strain curves. We removedthese slopes to pick a more precise reference dynamic fric-tion value after the hold prior to calculating Dm. We deter-mined the healing rate b = Dm/Dlog10(th) where th is thehold time. Figure 5 shows healing results for experimentsconducted at each normal stress. In Figure 5a data forphyllosilicate-poor samples (T2), which have high friction,show higher rates of frictional healing respect to samplesT1 and T7, which are frictionally weak. For T2 samples,healing rate is markedly different for the solid and the

Figure 3. Average friction of powders as function ofshear strain and weak phyllosilicates content (in brackets).(a) Experiments carried out at 50 MPa normal stress and(b) experiments carried out at 20 MPa normal stress. Frictiondecreases with increasing phyllosilicate content.

Figure 4. Evolution of dynamic friction, md, with shearstrain during all the experiments. md values have been col-lected at the end of the run-in procedure and before eachhold. Phyllosilicate-poor sample T2 shows friction withinthe Byerlee’s range. Samples with phyllosilicates aremechanically weak and weaken slightly with increasingshear strain. Intact samples of foliated and phyllosilicate-richrocks exhibit lower friction than their powdered equivalents.Data points with same color have the same mineralogicalcomposition.

TESEI ET AL.: FRICTION AND FAULT HEALING B09402B09402

5 of 13

powdered specimens, b = 0.005 and b = 0.007 respectively.Phyllosilicate-rich samples have b ≤ 0.0015 and small dif-ferences between intact wafers and powders. Sample T7’sintact wafer showed a slight weakening over all holding times.Experiments carried out at a lower normal stress (Figure 5b)on phyllosilicate-rich materials highlight that healing isreduced in all samples, ranging from 0.0002 < b < 0.002,with rock fabric having again little influence.

3.3. Microstructural Observations[21] SEM investigations on powdered and foliated, intact

samples show that deformation-induced microstructures are

controlled by a combination of mineralogical compositionand rock fabric. In powdered samples (Figure 6), deforma-tion occurs through cataclastic processes, resulting in a faultrock characterized by localized Riedel (R1), Y and boundary(B) shear planes [e.g., Logan et al., 1992;Mair and Marone,1999]. Quite often, R1 shear planes bound fault regionscharacterized by grain size reduction (GSR in Figure 6). Insamples T1, T4 and T7 (21, 52 and 33 wt.% phyllosilicates,respectively), strong calcite crystals and tremolite aggregatesare embedded in a fine to ultrafine phyllosilicate-rich matrix.This matrix is homogeneously distributed around grainboundaries, lining both large and small fractured grains(Figure 6c) suggesting adhesion of ultrafine flakes and fibersto the strong, coarse clasts. Powders of sample T2 arecomposed of amphibole polycrystalline aggregates, up to150 mm in size (bright BSE contrast in Figure 6b), that are inpart the result of the initial crushing and sieving procedure.The aggregates consist of randomly oriented submicronfibers of actinolite (Figure 6d) and are embedded in a fine toultrafine-grained amphibole-rich matrix. Minor epidotegrains are also observed.[22] Intact rock wafers of samples T1 and T7 show a

marked foliation where much of the deformation is accom-modated through localized slipping along sub-parallel shearplanes (Figure 7a). Foliation is represented by continuousand interconnected horizons, 10–200 mm thick, boundingstrong calcite and tremolite sigmoids. The foliation horizonsare formed by ultrafine oriented lamellae of talc and smec-tite, with minor tremolite fibers. Tremolite fibers also showpreferred orientation, with fiber axes parallel to the phyllo-silicate foliation (see Viti and Collettini [2009] for nanos-tructural details). Deformation is accommodated by slipparallel to the phyllosilicate lamellae, leaving strong calcitecrystals fractured but rather undeformed (Figure 7b).[23] Sample T4 shows microstructural features similar to

samples T1 and T7, consisting of strong calcite and tremolitesigmoids surrounded by smectite- and tremolite-rich ultrafinefoliation (Figure 7c, upper side). Shear occurs in foliation-parallel surfaces, as in samples T1 and T7. However, sampleT4 also revealed the occurrence of a sub-millimetric horizoncharacterized by different mineralogical composition, micro-structure and deformation style (Figure 7c, lower side). Thishorizon consists of relatively coarse lamellae of mixed-layeredclays (chlorite/smectite), up to 10 mm in thickness, that enve-lope coarse calcite grains (bright in Figure 7c) and ultrafine-grained sigmoids (Figure 7d). The same material of the mainfoliation forms the ultrafine sigmoids, i.e., they consist of anultrafine assemblage of smectite lamellae and tremolite fibers.Here, deformation is accommodated mainly along sigmoidedges (Figure 7d), through delamination and foliation-parallelslip along (001) clay lamellae, but also through irregularmicrofolds (inset in Figure 7d).[24] T2 fault rock hand specimens are characterized by a

macroscopic foliation that, in contrast with other samples,does not contains phyllosilicates. In agreement with X-raydiffraction data, the sample is predominantly composed ofamphibole (actinolite with variable Fe contents), occurring assubmicron-sized fibers in preferred orientation. The anisot-ropy of this sample is therefore due to a preferential align-ment of amphibole fibers, but it is also related to chemicalheterogeneities. In particular, the Fe content of amphibole islocally variable, resulting in an evident dark/bright layering

Figure 5. Frictional healing as function of hold time for(a) experiments carried out at 50 MPa normal stress and(b) experiments carried out at 20 MPa normal stress. Solidlines are log linear best fit relations. In Figure 5a healing ratesb are remarkably different between strong T2 (T2 wafer b =0.0046, T2 powder b = 0.0066) and weak T1 and T7 samples(b ≤ 0.0015). In Figure 5b the fit line highlights the highesthealing rate (T8 wafer b = 0.0022).

TESEI ET AL.: FRICTION AND FAULT HEALING B09402B09402

6 of 13

in BSE images (e.g., Figure 7e, where brighter layers corre-spond to higher Fe contents). The sample also hosts euhedralcrystals of epidote, 50–200 mm in size that are located withinsigmoidal-like bodies (Figure 7e). Figure 7f shows details ofthe ultrafine-grained amphibole matrix, indicating that mostof the deformation occurs within thin bands (less than 50 mmthick), parallel to rock anisotropy, and characterized by per-vasive micro cracking. In these bands frictional slidingamong the small fibers or fibers aggregates predominantlyaccommodates the deformation.

4. Discussion

4.1. Comparison With Previous Friction Data[25] Our experiments conducted on dry powders composed

of differing amounts of strong minerals (calcite, amphiboles,epidote) and phyllosilicates (talc, smectite, mixed layer claysand minor chlorite) confirm the role of weak phyllosilicatesin fault weakening. In particular, the frictional strength ofsamples composed only of amphibole is about 0.6, consistentwith existing studies. For phyllosilicate-rich samples (cf.Table 1), deformation is accommodated by frictional slidingalong phyllosilicate lamellae and grain-to-grain sliding or

crushing for the strong mineral phases. The result of thismixed-mode of deformation is that the higher the phyllosi-licate content, the lower the friction coefficient (Figure 3). Inorder to achieve a significant weakening effect in powders,md < 0.4, the percent of phyllosilicates has to be >50%. Thismicromechanical interpretation is in agreement with themodel proposed by Tembe et al. [2010] in which the tran-sition from a stress supporting framework of strong roundedgrains to a clay matrix embedded with disperse strong grainsoccurs for clay content around 50%. These results are also inagreement with previous data on simulated [Shimamotoand Logan, 1981; Logan and Rauenzahn, 1987; Brown et al.,2003; Saffer and Marone, 2003; Carpenter et al., 2009;Ikari et al., 2011a] and natural phyllosilicate-rich fault rocks[Numelin et al., 2007; Smith and Faulkner, 2010; Ikari et al.,2011b]. On the other hand, experiments on mixtures ofstrong (lizardite or antigorite) and weak (talc) mineral phases,all with platy habits, show that limited amounts of talcweaken fault gouges [Moore and Lockner, 2011].[26] All the experiments conducted on solid foliated

samples, except T2, are characterized by dynamic frictionmd < 0.3 (Figure 4). In solid phyllosilicate-rich samples,sliding occurs by slipping along foliated phyllosilicate-rich

Figure 6. SEM-BSE images of powdered samples. (a) T1 sample showing cataclastic texture associatedwith grain size reduction (GSR) zone that is bounded by R1 (oblique). Y (shear parallel) and B (gouge-forcing block boundary) planes. (b) Cataclastic processes in sample T2: larger grains are actinolite aggre-gates that are disaggregated along GSR zones associated with Riedel shear bands. (c) Detail of sample T1showing calcite (Cal) and tremolite (Tr) crystals embedded in a fine to ultrafine phyllosilicate-rich matrix(dark gray contrast). (d) Detail of an actinolite aggregate in sample T2 consisting of submicrometric fiberswith random orientation.

TESEI ET AL.: FRICTION AND FAULT HEALING B09402B09402

7 of 13

Figure 7. SEM-BSE images of solid samples. (a) Sample T1 foliated microstructure with thickphyllosilicate-rich horizons (Sm + Tlc + Tr, dark gray) oriented with the sense of shear; light gray portionscorrespond to calcite sigmoids (Cal). Deformation is accommodated through slip parallel to phyllosilicatelamellae (Slip Parallel Foliation SPF). (b) Sample T7; detail of the boundary between calcite andphyllosilicate-rich ultrafine foliation with deformation occurring by slip along sub-parallel planes not affect-ing calcite sigmoids. (c) Complex T4 foliation with two different deformation styles separated by a principalSPF slip surface. Upper side: typical Sm + Tr ultrafine foliation, in which deformation occurs throughfoliation-parallel slipping, analogously to T1 and T7 samples. Bottom: mixed layered clays (Mlc)-rich foli-ation with anastomosing slip surfaces. (d) Detail of sample T4, showing relatively coarse lamellae of mixedlayered clays, surrounding ultrafine-grained Sm + Tr sigmoids. Inset: deformation in Mlc, accommodatedalong curvilinear slip surfaces and by microfolding. (e) Foliated T2 sample composed of ultrafine actinolitewith variable Fe contents that surrounds epidote-bearing sigmoids (Ep). (f) Microcracking bands accommo-dating the deformation in sample T2. Microcracking does not produce an effective grain-size reduction, butonly disaggregation, because of the extremely fine-grained size of the amphibole fibers.

TESEI ET AL.: FRICTION AND FAULT HEALING B09402B09402

8 of 13

horizons without or with limited friction between hardminerals (e.g., sample T1 and T7 in Figures 7a and 7b).These results confirm that the foliated fabric that commonlyoccurs in weak-phyllosilicate-rich fault rocks significantlyreduces the rock’s yield strength when deformation isaccommodated along interconnected and weak phyllosili-cate horizons [Collettini et al., 2009b; Niemeijer et al.,2010]. Moreover, the frictional properties can be affectedby local heterogeneities in phyllosilicate mineralogy anddistribution as demonstrated by variability of friction ofwafers (bottom of Figure 4) from the same fault zone.[27] T2 solid wafers are characterized by a foliation that is

evident in hand samples. SEM investigations show that thisfoliation is the result of both oriented amphibole fibers andchemical heterogeneities of the fault rock (Figures 7e and 7f).Here the deformation is accommodated by frictional slidingof the amphibole aggregates without grain-size reduction,resulting in a high friction coefficient. The comparison ofsolid foliated samples with their powdered equivalentshighlights that, in the absence of weak minerals like phyllo-silicates, foliation and/or mineral orientation cannot produceweakening of the fault rock.[28] Due to the fact that our experiments were conducted

under dry conditions, it is important to note that furtherweakening in phyllosilicates can be induced by the presenceof water, as observed in serpentine [Morrow et al., 2000],smectite [Moore and Lockner, 2007] and chlorite [Brownet al., 2003; Ikari et al., 2009b].[29] In order to test the applicability of our results to nat-

ural phyllosilicate-rich faults that are likely dominated by thepresence of fluids, we compared our experiments conductedunder room-dry conditions to other studies in which thefriction of phyllosilicates has been studied under water sat-uration or internal fluid pressure conditions. We used fric-tional data from synthetic powder mixtures of weakphyllosilicates and quartz [Logan and Rauenzahn, 1987;Brown et al., 2003; Ikari et al., 2007; Takahashi et al., 2007;Crawford et al., 2008; Ikari et al., 2009a, 2009b; Tembeet al., 2010; Moore and Lockner, 2011] and samples ofboth intact and powdered rock of the Zuccale Fault[Collettini et al., 2009b; Smith and Faulkner, 2010] withphyllosilicate amounts similar to our samples. In our com-parison, mixtures of hard, equant minerals (quartz) and weakphyllosilicates have been selected as proxies for our naturalsamples made of calcite-tremolite and phyllosilicates.[30] Differences in frictional strength between wet and dry

powders of gouge are negligible until phyllosilicate contentreaches !50% (Figure 8). This reflects the fact that forpowdered material, the stress-supporting framework ofstrong grains like quartz or calcite is not influenced by the

Figure 8. Comparison between dry (this work) and wet[Logan and Rauenzahn, 1987; Brown et al., 2003;Takahashi et al., 2007; Crawford et al., 2008; Tembe et al.,2010; Smith and Faulkner, 2010; Moore and Lockner,2011; Collettini et al., 2009b] friction for phyllosilicate-richmixtures: (a) ≈ 20%, (b) ≈ 30% and (c) ≈ 50% of weak phyl-losilicates. For phyllosilicate amounts <50%, friction of dryand wet powders is similar. Friction of dry wafers is alwayssignificantly lower than either dry or wet powders andslightly higher than similar wet wafers. Red and blue sym-bols are for dry and wet experiments respectively.

TESEI ET AL.: FRICTION AND FAULT HEALING B09402B09402

9 of 13

presence of water. Intact, dry rock wafers show significantlylower friction with respect to wet powders and they appearto be slightly weakened by the presence of water (e.g.,Figure 8a). We thus suggest that experiments conducted onfoliated and phyllosilicate-rich fault rocks are more repre-sentative than powders in order to study the frictionalstrength of phyllosilicate-rich faults.[31] The presence of a published data set [Smith and

Faulkner, 2010] on the frictional properties of thephyllosilicate-rich horizon of the Zuccale fault, gives usthe possibility to make a detailed comparison (Figure 9).Talc-rich (70%) horizons are characterized by friction in therange of 0.2–0.3, and are the weakest rocks of the Zuccale fault(sample L3.2 of Smith and Faulkner [2010]). In a suite ofexperiments conducted on the same fault rock (C. Collettini,unpublished data on L3.2, 2008), we document the samefriction with only minor differences between dry and wetsamples (Figure 9), in agreement with previous observationson the reduced effect of water in lowering the shear strengthof talc [e.g.,Morrow et al., 2000]. Experiments conducted onwet and powdered samples of fault rocks with a loweramount of phyllosilicates (L3.3, 12% chlorite and 16% talc)show friction in the range of 0.4–0.45 [Smith and Faulkner,2010]. For foliated and phyllosilicate-rich samples tested inthis manuscript, we note that the strength of intact waferswith 21–52% of phyllosilicate amounts approaches thestrength of fault rock powders with 70% of talc. However, itis worth noting that in a foliated rock dominated by thepresence of expandable clays, the frictional strength can besignificantly lowered by the effects of the hydration state ofclay interlayers [e.g., Moore and Lockner, 2004; Ikari et al.,2007].

4.2. The Mechanisms of Frictional Healing[32] The frictional healing data of our experiments shows

very low b values for all phyllosilicate-rich samples, T1,T4 T7 and T8 and higher b values for the strong sample T2.Sample T2 shows lower healing rates for the intact foliated

rocks compared to its powdered equivalent. These healingdifferences can be explained as follows. In powdered sam-ples, actinolite fibers cross each other in a random orienta-tion playing the role of asperities (Figure 6d): during holds,the real contact area (and thus friction) grows. In T2 solidrock wafers, the preferred orientation of amphibole fibers(Figure 7e) results in larger contact area and thus lessasperity expansion during holds, resulting in lowered heal-ing rates.[33] Samples T1, T4, T7 and T8 all show b < 0.002 with

limited differences that are weakly controlled by fabric (i.e.,wafers versus powders), phyllosilicate content and normalstress. These healing rates are extremely low compared topreviously reported healing results on quartz gouges(Figure 10) studied to infer healing properties of seismo-genic faults [e.g., Marone, 1998], but are consistent withrecent tests on clay-rich cuttings (random fabric and wetconditions) obtained from deep drilling of the San AndreasFault [Carpenter et al., 2011]. From a micromechanicalpoint of view, this negligible yield strength gain duringquasi-stationary contact time implies no growth of realcontact areas via creeping expansion of asperities. Previousworks have proposed that the platy structure of phyllosili-cates allow real contact areas to saturate quickly, perhapsduring shear, within the fault rock [e.g., Saffer and Marone,2003; Ikari et al., 2009a].[34] In our experiments, the comparison between friction

and healing values between solid, foliated samples andpowdered equivalents suggests that the phyllosilicate-fabrichas an important role in fault strength but minor or no effecton healing rates. We propose the following as a mechanicalexplanation for this behavior. In powdered material,

Figure 9. Comparison of frictional data from the Zuccalefault phyllosilicate-rich horizon. Room-dry and wet experi-ments are shown in red and blue respectively.

Figure 10. Comparison between frictional healing datafrom this study and previous work on quartz gouge[Marone, 1998] and clay-rich cuttings from the San AndreasFault [Carpenter et al., 2011]. Fit lines highlight the healingrate of quartz (b = 0.0086), T2 powdered amphibolite (b =0.0066), the highest and the lowest rate among phyllosili-cate-rich samples of this work, intact T8 wafer (b =0.0022) and T7 wafer respectively (b = 0), and healing rateof San Andreas Fault clay-rich cuttings (b = 0.0006).

TESEI ET AL.: FRICTION AND FAULT HEALING B09402B09402

10 of 13

irrespective of the grain-size, the microstructure is repre-sented by strong calcite and tremolite grains, homoge-neously surrounded by thin, phyllosilicate-rich horizons(Figure 6c). During hold periods, the rotation and adhesionof phyllosilicate lamellae to calcite and tremolite grainsprevents the adhesive growth of contact area that is typicalof strong granular materials, leading to limited healing asobserved. However, during sliding periods, this thin matrixcannot completely sustain shear, allowing for grain-to-grainsliding and/or crushing, resulting in higher friction (0.4 < md< 0.5). In solid wafers, during holds, the phyllosilicate foli-ation allows quick saturation of the contact area and thus nore-strengthening. Moreover, phyllosilicate horizons are thick(Figure 7a) and well-oriented to sustain much of the shear-ing, preventing strong grain-to-grain contact and crushingand resulting in lower friction (md < 0.3). A phyllosilicate-rich matrix with the ability to saturate contact area, thusresulting in near-zero healing, is likely to occur above acritical phyllosilicates content that needs further investiga-tion. Our results are in agreement with recent detailed stud-ies of the role of shear fabric on friction [e.g., Ikari et al.,2011b].[35] In discussing the applicability of our results to natu-

ral faults, it is important to emphasize that healing datapresented in this work does not consider other healingprocesses such as pressure solution [e.g., Bos and Spiers,2002; Niemeijer et al., 2010] or cementation [e.g., Olsenet al., 1998; Tenthorey et al., 2003]. Nevertheless pressure-solution experiments on clay-rich simulated fault gougesshow that healing is limited by the presence of phyllosilicatesin the fault rock [Niemeijer et al., 2008]. Healing in phyllo-silicate-rich faults is thus likely to be extremely low at leastfor short quasi-stationary contact times before hydrothermalcementation takes place. Hydrothermal cementation takesplace over longer timescales with respect to our experiments,hours to years depending on reaction rates [e.g., Karneret al., 1997; Olsen et al., 1998; Muhuri et al., 2003] andleads to a significant restrengthening. Widespread calciteand/or quartz-rich veins with crack-and-seal texture indicatethe presence of fluid assisted fracturing and cementationprocesses in several phyllosilicate-rich faults [Imber et al.,2001; Collettini et al., 2006; Vannucchi et al., 2008; Fagerenget al., 2010]. However, for each weak and phyllosilicate-rich horizon, there are thousands of weak potential slidingsurfaces associated with phyllosilicate lamellae (001 planes)and we observe many sub-parallel slipping surfaces distrib-uted within these horizons (Figure 7a). Distributed defor-mation implies that the energy needed to create the fractureand propagate slip is similar for each sliding surface. Whenthe formerly active sliding surface is cemented, slip is ener-getically favored to proceed along another weak phyllosili-cate layers without re-shearing the cemented materials andthus, for thick, and weak phyllosilicate-rich faults, the pro-cesses would have a limited effect on the global frictionalstrength (e.g., Figure 7c). This is confirmed by the observa-tion that in the talc-smectite rich foliation of the Zuccalefault, foliation-parallel calcite veins are widespread andcharacterized by single or multiple crack-and-seal events.Phyllosilicate-rich faults can therefore remain weak afterseveral mineralizing events.[36] The combination of low friction and low rates of

frictional healing observed in our experiments suggests that

fault behavior and the mode of tectonic slip should vary withphyllosilicate content of the fault zone. Fault zones that aredevoid of phyllosilicates are expected to have ordinary ratesof frictional healing, and thus they may host repeated stick-slip frictional sliding. Moreover, the combination of ourobservations and previous data on friction velocity depen-dence for both powdered [e.g., Saffer and Marone, 2003;Moore and Lockner, 2007; Ikari et al., 2007; Carpenteret al., 2011] and intact, foliated [Ikari et al., 2011a, 2011b;Collettini et al., 2011] samples suggests that phyllosilicate-rich faults should creep aseismically and remain weak overlong time scales, at least at shallow crustal depth conditionsat which most weak phyllosilicate minerals are stable. Onepossible implication is that mechanical segmentation offaults, into regions of velocity strengthening and velocityweakening frictional behavior, is controlled by the mineral-ogy and distribution of weak phyllosilicate phases withinfault rocks.[37] In the last two decades, the complexity of global-scale

faulting through mixed modes of failure combining seismicslip, slow slip earthquakes, tremors and creeping along faultpatches with different rheology has been documented [e.g.,Ben-Zion et al., 1993; Kanamori and Kikuchi, 1993; Shellyet al., 2007; Ito et al., 2007; Hreinsdottir and Bennett,2009; Outerbridge et al., 2010; Rubinstein et al., 2010;Perfettini et al., 2010]. The mechanical data presented heresupport the thesis of long-term stability of these creeping/slow-slipping fault portions and future work will be neededto clarify how they can load/interact with seismically activefault patches.

5. Conclusion

[38] We present results from a suite of 12 friction experi-ments focused on phyllosilicate-rich fault materials. Wecompare the behavior of intact and powdered samples of faultrock from the Zuccale Fault, central Italy. Experiments wereperformed under constant normal stress (20 or 50 MPa), atroom temperature and dry conditions. Our results, and thecomparison with existing studies, indicate that intact, natural,weak phyllosilicate-bearing rocks are mechanically weak(md < 0.3), much weaker than powdered samples of the samefault rock (0.4 < md < 0.5), even in the presence of pore fluids.[39] Our microstructural studies show that fabric-induced

weakness varies with phyllosilicate composition and internalfault rock heterogeneities. The frictional strength of foliatedfault rock is determined by the weakest phase and the mostcritically oriented fabric elements. We demonstrated thatfoliation itself cannot induce weakness in the fault rockwithout the presence of weak phyllosilicate minerals. Oursamples show low frictional healing rates (b < 0.002) in bothpowdered samples and intact rocks, and this result holds fora range of normal stresses and over a large range of phyllo-silicate content, regardless of the weak phyllosilicates speciesinvolved in the process. We interpret this as the result ofsaturation of real contact area due to the presence of phyllo-silicate layers that experience little or no time-dependentgrowth within the timescales studied experimentally (3–3000 s). During sliding episodes in solid wafers, deformationoccurs by sliding entirely within thick phyllosilicate folia-tions while in powdered equivalents shear occurs by fric-tional sliding of hard grains only damped by the presence of

TESEI ET AL.: FRICTION AND FAULT HEALING B09402B09402

11 of 13

phyllosilicates. The reduced re-strengthening may also beinsensitive to other healing processes such as pressure-solution and hydrothermal cementation therefore leading toa prolonged weakness of phyllosilicate-rich faults.[40] Phyllosilicates are commonly associated with mature

fault zones and their weak frictional properties and reducedhealing rates could have an important role in crustal faultingin a variety of geodynamic environments. Our data suggestsprolonged apparent weakness of fault segments with aseis-mic and other non-seismic fault behaviors.

[41] Acknowledgments. This research was supported by the ERCSt. G. Nr. 259256, GLASS project to C. Collettini, and NSF grantsOCE-0648331, EAR-0746192, and EAR-0950517 to C. Marone. We thankSteve Swavely for technical help, and Marco M. Scuderi for assistance withthe biaxial apparatus and insightful discussions. We thank R. E. Holdsworthand D. E. Moore for critical reviews that helped to significantly improve thequality of this work.

ReferencesBeeler, N. M., T. E. Tullis, and J. D. Weeks (1994), The roles of time anddisplacement in the evolution effect on rock friction, Geophys. Res. Lett.,21(18), 1987–1990, doi:10.1029/94GL01599.

Ben-Zion, Y., J. R. Rice, and R. Dmowska (1993), Interaction of the SanAndreas creeping segment with adjacent great rupture zones and earth-quake recurrence at Parkfield, J. Geophys. Res., 98(B2), 2135–2144,doi:10.1029/92JB02154.

Bos, B., and C. J. Spiers (2000), Effect of phyllosilicates on fluid-assistedhealing of gouge-bearing faults, Earth Planet. Sci. Lett., 184, 199–210,doi:10.1016/S0012-821X(00)00304-6.

Bos, B., and C. J. Spiers (2002), Fluid-assisted healing processes in gouge-bearing faults: Insights from experiments on a rock analogue system,Pure Appl. Geophys., 159, 2537–2566, doi:10.1007/s00024-002-8747-2.

Brown, K. M., A. Kopf, M. B. Underwood, and J. L. Weinberger (2003),Compositional and fluid pressure controls on the state of stress on theNankai subduction thrust: A weak plate boundary, Earth Planet. Sci.Lett., 214, 589–603, doi:10.1016/S0012-821X(03)00388-1.

Brune, J. N., T. L. Henyey, and R. F. Roy (1969), Heat flow, stress, and rateof slip along the San Andreas Fault, California, J. Geophys. Res., 74(15),3821–3827, doi:10.1029/JB074i015p03821.

Buatier, M. D., A. Chauvet, W. Kanitpanyacharoen, H. R. Wenk, J. F. Ritz,and M. Jolivet (2012), Origin and behaviour of clay minerals in the Bogdfault gouge, Mongolia, J. Struct. Geol., 34, 77–90, doi:10.1016/j.jsg.2011.10.006.

Byerlee, J. (1978), Friction of rocks, Pure Appl. Geophys., 116, 615–626,doi:10.1007/BF00876528.

Carpenter, B. M., C. Marone, and D. M. Saffer (2009), Frictional behaviorof materials in the 3D SAFOD volume, Geophys. Res. Lett., 36, L05302,doi:10.1029/2008GL036660.

Carpenter, B. M., C. Marone, and D. M. Saffer (2011), Weakness of the SanAndreas Fault revealed by samples from the active fault zone, Nat.Geosci., 4, 251–254, doi:10.1038/ngeo1089.

Chiaraluce, L., C. Chiarabba, C. Collettini, D. Piccinini, and M. Cocco(2007), Architecture and mechanics of an active low-angle normal fault:Alto Tiberina Fault, northern Apennines, Italy, J. Geophys. Res., 112,B10310, doi:10.1029/2007JB005015.

Collettini, C., and R. E. Holdsworth (2004), Fault zone weakening andcharacter of slip along low angle normal faults: Insights from the Zuccalefault, Elba, Italy, J. Geol. Soc., 161, 1039–1051.

Collettini, C., N. De Paola, and N. R. Goulty (2006), Switches in theminimum compression direction induced by overpressure beneath a lowpermeability fault zone, Terra Nova, 18, 224–231, doi:10.1111/j.1365-3121.2006.00683.x.

Collettini, C., C. Viti, S. A. F. Smith, and R. E. Holdsworth (2009a), Thedevelopment of interconnected talc networks and weakening of conti-nental low-angle normal faults, Geology, 37, 567–570, doi:10.1130/G25645A.1.

Collettini, C., A. Niemejer, C. Viti, and C. Marone (2009b), Fault zone fab-ric and fault weakness, Nature, 462, 907–910, doi:10.1038/nature08585.

Collettini, C., A. Niemeijer, C. Viti, S. A. F. Smith, and C. Marone (2011),Fault structure, frictional properties and mixed-mode fault slip behavior,Earth Planet. Sci. Lett., 311, 316–327, doi:10.1016/j.epsl.2011.09.020.

Coulomb, C. A. (1785), Théorie des machines simples, en ayant égard aufrotement de leurs parties, et a la roideur des cordages,Mem. Math. Phys.,X, 161–342.

Crawford, B. R., D. R. Faulkner, and E. H. Rutter (2008), Strength, poros-ity, and permeability development during hydrostatic and shear loadingof synthetic quartz-clay fault gouge, J. Geophys. Res., 113, B03207,doi:10.1029/2006JB004634.

Das, S., and K. Aki (1977), Fault plane with barriers: A versatile earth-quake model, J. Geophys. Res., 82(36), 5658–5670, doi:10.1029/JB082i036p05658.

Dieterich, J. H. (1972), Time-dependent friction in rocks, J. Geophys. Res.,77, 3690–3697, doi:10.1029/JB077i020p03690.

Dieterich, J. H., and G. Conrad (1984), Effect of humidity on time- andvelocity-dependent friction in rocks, J. Geophys. Res., 89(B6), 4196–4202,doi:10.1029/JB089iB06p04196.

Dieterich, J. H., and B. D. Kilgore (1994), Direct observation of frictionalcontacts: New Insights for state-dependent properties, Pure Appl. Geophys.,143, 283–302, doi:10.1007/BF00874332.

Fagereng, A., and R. H. Sibson (2010), Melange rheology and seismicstyle, Geology, 38, 751–754, doi:10.1130/G30868.1.

Fagereng, A., F. Remitti, and R. H. Sibson (2010), Shear veins observedwithin anisotropic fabric at high angles to the maximum compressivestress, Nat. Geosci., 3, 482–485, doi:10.1038/ngeo898.

Faulkner, D. R., A. C. Lewis, and E. H. Rutter (2003), On the internal struc-ture and mechanics of large strike-slip faults: Field observations from theCarboneras fault, southeastern Spain, Tectonophysics, 367, 235–251,doi:10.1016/S0040-1951(03)00134-3.

Frye, K. M., and C. Marone (2002), Effect of humidity on granular frictionat room temperature, J. Geophys. Res., 107(B11), 2309, doi:10.1029/2001JB000654.

Haines, S. H., B. A. van der Pluijm, M. Ikari, D. Saffer, and C. Marone(2009), Clay fabrics in natural and artificial fault gouge, J. Geophys.Res., 114, B05406, doi:10.1029/2008JB005866.

Hickman, S. H., and B. Evans (1992), Growth of grain contacts in halite bysolution transfer: Implications for diagenesis, lithification and strengthrecovery, in Fault Mechanics and Transport Properties of Rocks,pp. 253–260, Academic, London, doi:10.1016/S0074-6142(08)62825-9.

Holdsworth, R. E. (2004), Weak faults–rotten cores, Science, 303, 181–182,doi:10.1126/science.1092491.

Holdsworth, R. E., E. W. E. van Diggelen, C. J. Spiers, J. H. P. de Bresser,R. J. Walker, and L. Bowen (2011), Fault rocks from the SAFOD coresamples: Implications for weakening at shallow depths along the SanAndreas Fault California, J. Struct. Geol., 33, 132–144, doi:10.1016/j.jsg.2010.11.010.

Hong, T., and C. Marone (2005), Effects of normal stress perturbations onthe frictional properties of simulated faults, Geochem. Geophys. Geosyst.,6, Q03012, doi:10.1029/2004GC000821.

Hreinsdottir, S., and R. A. Bennett (2009), Active aseismic creep on theAlto Tiberina low-angle normal fault, Italy, Geology, 37, 683–686,doi:10.1130/G30194A.1.

Ikari, M. J., D. M. Saffer, and C. Marone (2007), Effect of hydrationstate on the frictional properties of montmorillonite-based fault gouge,J. Geophys. Res., 112, B06423, doi:10.1029/2006JB004748.

Ikari, M. J., D. M. Saffer, and C. Marone (2009a), Frictional and hydrologicproperties of a major splay fault system, Nankai subduction zone,Geophys. Res. Lett., 36, L20313, doi:10.1029/2009GL040009.

Ikari, M. J., C. Marone, and D. M. Saffer (2009b), Frictional and hydrologicproperties of clay-rich fault gouge, J. Geophys. Res., 114, B05409,doi:10.1029/2008JB006089.

Ikari, M. J., C. Marone, and D. M. Saffer (2011a), On the relation betweenfault strength and frictional stability, Geology, 39, 83–86, doi:10.1130/G31416.1.

Ikari, M., A. Niemeijer, and C. Marone (2011b), The role of fault zone fab-ric and lithification state on frictional strength, constitutive behavior, anddeformation microstructure, J. Geophys. Res., 116, B08404, doi:10.1029/2011JB008264.

Imber, J., R. H. Holdsworth, C. A. Butler, and R. A. Strachan (2001),A reappraisal of the Sibson-Scholz fault zone model: The nature of thefrictional to viscous (“brittle-ductile”) transition along a long-lived,crustal-scale fault, Outer Hebrides, Scotland, Tectonics, 20(5), 601–624,doi:10.1029/2000TC001250.

Ito, Y., K. Obara, K. Shiomi, S. Sekine, and H. Hirose (2007), Slow earth-quakes coincident with episodic tremors and slow slip events, Science,315, 503–506, doi:10.1126/science.1134454.

Jefferies, S. P., R. E. Holdsworth, C. A. J. Wibberley, T. Shimamoto,C. J. Spiers, A. R. Niemeijer, and G. E. Lloyd (2006), The nature andimportance of phyllonite development in crustal-scale fault cores: Anexample from the Median Tectonic Line, Japan, J. Struct. Geol., 28(2),220–235, doi:10.1016/j.jsg.2005.10.008.

Kanamori, H., and C. R. Allen (1986), Earthquake repeat time and averagestress drop, in Earthquake Source Mechanics, Geophys. Monogr. Ser.,

TESEI ET AL.: FRICTION AND FAULT HEALING B09402B09402

12 of 13

vol. 37, edited by S. Das, J. Boatwright, and C. H. Scholz, pp. 227–235,AGU, Washington, D. C., doi:10.1029/GM037p0227.

Kanamori, H., and M. Kikuchi (1993), The 1992 Nicaragua earthquake: Aslow Tsunami earthquake associated with subducted sediments, Nature,361, 714–716, doi:10.1038/361714a0.

Karner, S. L., and C. Marone (1998), The effect of shear load on frictionalhealing in simulated fault gouge, Geophys. Res. Lett., 25(24), 4561–4564,doi:10.1029/1998GL900182.

Karner, S. L., C. Marone, and B. Evans (1997), Laboratory study of fault heal-ing and lithification in simulated fault gouge under hydrothermal conditions,Tectonophysics, 277, 41–55, doi:10.1016/S0040-1951(97)00077-2.

Li, Q., T. E. Tullis, D. Goldsby, and R. W. Carpwick (2011), Frictionalageing from interfacial bonding and the origins of rate and state friction,Nature, 480, 233–236, doi:10.1038/nature10589.

Lockner, D. M., C. Morow, D. Moore, and S. Hickman (2011), Lowstrength of deep San Andreas fault gouge from SAFOD core, Nature,472, 82–85, doi:10.1038/nature09927.

Logan, J. M., and K. A. Rauenzahn (1987), Frictional dependence of gougemixtures of quartz and montmorillonite on velocity, composition and fab-ric, Tectonophysics, 144, 87–108, doi:10.1016/0040-1951(87)90010-2.

Logan, J. M., C. A. Dengo, N. G. Higgs, and Z. Z. Wang (1992), Fabrics ofexperimental fault zones: Their development and relationship to mechan-ical behaviour, in Fault Mechanics and Transport Properties of Rocks,edited by B. Evans and T. Wong, pp. 33–67, Academic, San Diego,Calif., doi:10.1016/S0074-6142(08)62814-4.

Mair, K., and C. Marone (1999), Friction of simulated fault gouge for a widerange of velocities and normal stresses, J. Geophys. Res., 104(B12),28,899–28,914, doi:10.1029/1999JB900279.

Manatschal, G. (1999), Fluid-and reaction-assisted low-angle normal fault-ing: Evidence from rift-related brittle fault rocks in the Alps (Err nappe,eastern Switzerland), J. Struct. Geol., 21, 777–793, doi:10.1016/S0191-8141(99)00069-3.

Marone, C. (1998), The effect of loading rate on static friction and the rateof fault healing during the earthquake cycle, Nature, 391, 69–72,doi:10.1038/34157.

Marone, C., J. E. Vidale, and W. L. Ellsworth (1995), Fault Healinginferred from time dependent variations in source properties of repeatingearthquakes, Geophys. Res. Lett., 22, 3095–3098, doi:10.1029/95GL03076.

Moore, D. E., and D. A. Lockner (2004), Crystallographic control of thefrictional behavior of dry and water-saturated sheet structure minerals,J. Geophys. Res., 109, B03401, doi:10.1029/2003JB002582.

Moore, D. E., and D. A. Lockner (2007), Friction of the smectite clay mont-morillonite, in The Seismogenic Zone of Subduction Thrust Faults, editedby T. Dixon and C. Moore, pp. 317–345, Columbia Univ. Press, NewYork.

Moore, D. E., and D. A. Lockner (2008), Talc friction in the temperaturerange 25#–400#C: Relevance for fault-zone weakening, Tectonophysics,449, 120–132, doi:10.1016/j.tecto.2007.11.039.

Moore, D. E., and D. A. Lockner (2011), Frictional strengths of talc‐serpentine and talc‐quartz mixtures, J. Geophys. Res., 116, B01403,doi:10.1029/2010JB007881.

Moore D. E., and M. Rymer (2007), Talc-bearing serpentinites and thecreeping section of the San Andreas fault, Nature, 488, 795–797,doi:10.1038/nature06064.

Morrow, C. A., B. Radney, and J. Byerlee (1992), Frictional strength andthe effective pressure law of montmorillonite and illite clays, in FaultMechanics And Transport Properties of Rocks, pp. 69–88, Academic,San Diego, Calif.

Morrow, C. A., D. E. Moore, and D. A. Lockner (2000), The effect of min-eral bond strength and adsorbed water on fault gouge frictional strength,Geophys. Res. Lett., 27(6), 815–818, doi:10.1029/1999GL008401.

Muhuri, S. K., T. A. Dewers, T. E. Scott Jr., and Z. Reches (2003), Inter-seismic fault strengthening and earthquake-slip instability: Friction orcohesion?, Geology, 31, 881–884, doi:10.1130/G19601.1.

Niemeijer, A., C. Marone, and D. Elsworth (2008), Healing of simulatedfault gouges aided by pressure solution: Results from rock analogueexperiments, J. Geophys. Res., 113, B04204, doi:10.1029/2007JB005376.

Niemeijer, A. R., C. Marone, and D. Elsworth (2010), Fabric induced weak-ness of tectonic faults, Geophys. Res. Lett., 37, L03304, doi:10.1029/2009GL041689.

Numelin, T., C. Marone, and E. Kirby (2007), Frictional properties ofnatural fault gouge from a low-angle normal fault, Panamit Valley,California, Tectonics, 26, TC2004, doi:10.1029/2005TC001916.

Olsen, M., C. H. Scholz, and A. Lèger (1998), Healing and sealing of a sim-ulated fault gouge under hydrothermal conditions: Implications for faulthealing, J. Geophys. Res., 103(B4), 7421–7430, doi:10.1029/97JB03402.

Outerbridge, K. C., T. H. Dixon, S. Y. Schwartz, J. I. Walter, M. Protti,V. Gonzalez, J. Biggs, M. Thorwart, and W. Rabbel (2010), A tremor and

slip event on the Cocos-Caribbean subduction zone as measured by aglobal positioning system (GPS) and seismic network on the NicoyaPeninsula, Costa Rica, J. Geophys. Res., 115, B10408, doi:10.1029/2009JB006845.

Peng, Z., and J. Gomberg (2010), An integrated perspective of the contin-uum between earthquakes and slow-slip phenomena, Nat. Geosci., 3,599–607, doi:10.1038/ngeo940.

Perfettini, H., et al. (2010), Seismic and aseismic slip on the central Perumegathrust, Nature, 465, 78–81, doi:10.1038/nature09062.

Rabinowicz, E. (1951), The nature of static and kinetic coefficients offriction, J. Appl. Phys., 22, 1373–1379, doi:10.1063/1.1699869.

Rice, J. R. (1980), The mechanics of Earthquake rupture, in Physics of theEarth Interior, edited by A. M. Dziewonski and E. Boschi, pp. 555–649,North Holland, Amsterdam.

Richardson, E., and C. Marone (1999), Effects of normal stress vibrationson frictional healing, J. Geophys. Res., 104(B12), 28,859–28,878,doi:10.1029/1999JB900320.

Rubinstein, J. L., D. R. Shelly, and W. L. Ellsworth (2010), Non-volcanictremor: A window into the roots of fault zones, in New Frontiers inIntegrated Solid Earth Sciences, pp. 287–314, Springer, Dordrecht,Netherlands, doi:10.1007/978-90-481-2737-5_8.

Saffer, D. M., and C. Marone (2003), Comparison of smectite- and illiterich gouge frictional properties: Application to the up-did limit of theseismogenic zone along subduction mega-thrusts, Earth Planet. Sci. Lett.,215, 219–235, doi:10.1016/S0012-821X(03)00424-2.

Schleicher, A. M., B. A. van der Pluijm, and L. N. Warr (2010), Nanocoat-ings of clay and creep of the San Andreas fault at Parkfield, Calif. Geol.,38, 667–670, doi:10.1130/G31091.1.

Scholz, C. H. (2002), The Mechanics of Earthquakes and Faulting,2nd ed., Cambridge Univ. Press, Cambridge, U. K., doi:10.1017/CBO9780511818516.

Scholz, C. H., C. A. Aviles, and S. G. Wesnouki (1986), Scaling differencesbetween large interplate and intraplate earthquakes, Bull. Geol. Soc. Am.,76(1), 65–70.

Shelly, D. R., G. C. Beroza, and S. Ide (2007), Non-volcanic tremors andlow-frequency earthquakes swarms, Nature, 446, 305–307, doi:10.1038/nature05666.

Shimamoto, K., and J. M. Logan (1981), Effects of simulated clay gougeson the sliding behavior of Tennessee sandstone, Tectonophysics, 75,243–255, doi:10.1016/0040-1951(81)90276-6.

Smith, S. A. F., and D. R. Faulkner (2010), Laboratory measurements of thefrictional strength of a natural low-angle normal fault, J. Geophys. Res.,115, B02407, doi:10.1029/2008JB006274.

Smith, S. A. F., R. E. Holdsworth, C. Collettini, and J. Imber (2007), Usingfootwall structures to constrain the evolution of low angle normal faults,J. Geol. Soc., 164, 1187–1191, doi:10.1144/0016-76492007-009.

Sone, H., T. Shimamoto, and D. E. Moore (2012), Frictional properties ofsaponite-rich gouge from a serpentinite-bearing fault zone along theGokasho-Arashima Tectonic Line, central Japan, J. Struct. Geol., 38,172–182, doi:10.1016/j.jsg.2011.09.007.

Takahashi, M., K. Mizoguchi, K. Kitamura, and K. Masuda (2007), Effectsof clay content on the frictional strength and fluid transport property offaults, J. Geophys. Res., 112, B08206, doi:10.1029/2006JB004678.

Tembe, S., D. A. Lockner, and T.-F. Wong (2010), Effect of clay contentand mineralogy on frictional sliding behavior of simulated gouges:Binary and ternary mixtures of quartz, illite, and montmorillonite,J. Geophys. Res., 115, B03416, doi:10.1029/2009JB006383.

Tenthorey, E., S. F. Cox, and H. F. Todd (2003), Evolution of strengthrecovery and permeability during fluid-rock reaction in experimental faultzones, Earth Planet. Sci. Lett., 206(1–2), 161–172, doi:10.1016/S0012-821X(02)01082-8.

Vannucchi, P., F. Remitti, and G. Bettelli (2008), Geological record of fluidflow and seismogenesis along an erosive subducting plate boundary,Nature, 451, 699–703, doi:10.1038/nature06486.

Vidale, J. E., W. L. Ellsworth, A. Cole, and C. Marone (1994), Variations inrupture process with recurrence interval in a repeated small earthquake,Nature, 368, 624–626, doi:10.1038/368624a0.

Viti, C., and C. Collettini (2009), Growth and deformation mechanisms oftalc along a natural fault: A micro/nanostructural investigation, Contrib.Mineral. Petrol., 158, 529–542, doi:10.1007/s00410-009-0395-4.

Vrolijk, P., and B. A. van der Pluijm (1999), Clay gouge, J. Struct. Geol.,21, 1039–1048, doi:10.1016/S0191-8141(99)00103-0.

Wintsch, R. P., R. Christoffersen, and A. K. Kronenberg (1995), Fluid-rock reaction weakening of fault zones, J. Geophys. Res., 100(B7),13,021–13,032, doi:10.1029/94JB02622.

Wu, T., L. Blatter, and H. Roberson (1975), Clay gouges in the San AndreasFault System and their possible implications, Pure Appl. Geophys., 113,87–95, doi:10.1007/BF01592901.

TESEI ET AL.: FRICTION AND FAULT HEALING B09402B09402

13 of 13