Earth and Planetary Science Letters ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Contents lists available at SciVerse ScienceDirect

Earth and Planetary Science Letters

0012-82http://d

n CorrTel.: +1

E-m

Pleassequ

journal homepage: www.elsevier.com/locate/epsl

Frictional properties and slip stability of active faults withincarbonate–evaporite sequences: The role of dolomite and anhydrite

Marco M. Scuderi a,b,n, André R. Niemeijer c, Cristiano Collettini a,d, Chris Marone b

a Dipartemento di Scienze della Terra, Università degli Studi di Perugia, Piazza Università 1, 06100 Perugia, Italyb Department of Geosciences, The Pennsylvania State University, 16801 PA, USAc Utrecht University, Faculty of Geosciences, HPT Laboratory, Utrecht, The Netherlandsd Istituto Nazionale di Geofisica e Vulcanologia, Via di Vigna Murata 605, Rome, Italy

a r t i c l e i n f o

Article history:Received 14 September 2012Received in revised form28 January 2013Accepted 19 March 2013

Editor: P. Shearer

and anhydrite and differ from typical quartzofeldspathic rocks. Here, we report on laboratory experiments

Keywords:frictionearthquake mechanicsrock deformationcarbonate and evaporites rocks

1X/$ - see front matter & 2013 Elsevier B.V. Ax.doi.org/10.1016/j.epsl.2013.03.024

espondence to: 509 Deike Building, Universit814 753 2714.ail address: mms50@psu.edu (M.M. Scuderi).

e cite this article as: Scuderi, M.Mences: The role of dolomite and anhy

a b s t r a c t

Seismological observations show that many destructive earthquakes nucleate within, or propagate through,thick sequences of carbonates and evaporites. For example, along the Apennines range (Italy) carbonate andevaporite sequences are present at hypocentral depths for recent major earthquakes (5.0oMwo6.3).However, little is known about the frictional properties of these rocks, which are dominated by dolomite

designed to illuminate friction constitutive properties and associated microstructures of anhydrite anddolomite fault rocks. We performed frictional sliding experiments on powdered samples for a range ofnormal stresses (sn) from 10 to 150 MPa and sliding velocities from 1 to 300 μm/s. Experiments employedthe double direct shear configuration and were conducted both dry, at room temperature, and watersaturated at 75 1C under true triaxial stresses. At 25 1C, sliding friction is �0.65 for all samples and conditionsstudied. At 75 1C, samples develop cohesive strength, and sliding friction ranges from 0.4 to 0.5 for s′n from10 to 35 MPa. Velocity step tests were used to assess rate/state frictional behavior. At 25 1C the slip behavioris velocity strengthening/neutral for all samples. Microstructural observations indicate localized deformationon R1 Riedel and incipient Y-shear planes with marked grain-size reduction. At 75 1C and larger strain wefind that rate/state friction behavior correlates with velocity-dependent layer dilation and compaction andthis is associated with the development of through-going Y-shear planes. We conclude that elevatedtemperature, strain and the presence of pore fluids enhance shear localization into continuous Y-shearplanes, which could promote frictional instability at seismogenic depth within dolomite–anhydritesequences.

& 2013 Elsevier B.V. All rights reserved.

1. Introduction

Within many regions of the continental crust, moderate-to-large earthquakes nucleate at the base of the seismogenic layerwhere crystalline quartzo-feldspathic rocks are inferred to bepresent (Sibson, 1982; Scholz, 2002). To characterize the frictionaland poromechanical properties of quartzo-feldspathic fault rocks,a significant amount of experimental work has been carried out(e.g., Dieterich, 1972; Byerlee, 1978; Moore et al., 1994; Blanpiedet al., 1995; Beeler et al., 1996; Evans et al., 1997; Marone, 1998;Di Toro et al., 2004).

Recently, it has been shown that many earthquakes nucleatewithin, or propagate through, thick sequences of carbonates(both limestone and dolostone) that dominate the upper-crustal

ll rights reserved.

y Park, PA 16801, USA.

., et al., Frictional propertiedrite. Earth and Planetary Sc

sedimentary sequences in some regions. Examples of major earth-quakes within carbonate and carbonate-bearing terrains includethe 1995, Mw¼6.2, Aigion event in the Gulf of Corinth, Greece(Bernard et al., 2006), the 1997–1998, Mwmax¼6.0, Umbria–Marche seismic sequence in central Italy (Chiaraluce et al., 2003;Miller et al., 2004), the 2008, Mw¼7.9, Wenchuan earthquake inChina (Burchfiel et al., 2008; Xu et al., 2009), the 2009, Mw¼6.3,L'Aquila event in central Italy (Patacca et al., 2008; Chiarabba et al.,2009) and the 2012, Mw¼6.1, Emilia sequence (Ventura and DiGiovanbattista, in press).

For the 2009 L'Aquila earthquake (Fig. 1) it is difficult todetermine the exact rock types in the source region of the event(Chiaraluce, 2012), and for the Emilia sequence (Fig. 1) detailedstudies are still in progress. However in both sequences themainshocks at least propagated through the rocks of the sedi-mentary cover, which are made of carbonates and anhydrites, anda great number of aftershocks nucleated within these sedimentaryrocks (Chiaraluce, 2012). In the case of the Umbria–Marche 1997–1998 seismic sequence, the integration of seismological data with

s and slip stability of active faults within carbonate–evaporiteience Letters (2013), http://dx.doi.org/10.1016/j.epsl.2013.03.024i

Fig. 1. USGS Shake map showing recent earthquakes in carbonate and anhydrite-bearing terranes of the Appennines, Central Italy. Inset shows longitudinal cross sectionreconstructed using seismic reflection profiles for the 1997 Mw¼6.0 Colfiorito event. The mainshock (1) nucleated within the Trissic Evaporites (after Mirabella et al., 2008),followed by three other major events that nucleated within the same sedimentary sequence. The samples used in our study were collected from this region.

Coe

ffici

ent o

f Fric

tion

(µ)

Displacement, mm

(a-b) > 0Velocity Strengthening (a-b) < 0Velocity Weakening

Displacement, µm

= 75 MPa = 100 MPa = 150 MPa

m/sp2230

Fig. 2. (a) Biaxial loading frame with pressure vessel for fluid saturated and true-triaxial experiments. (b) Double direct shear configuration for room temperatureexperiments. Two layers of fault gouge are sandwiched between three loading platens. (c) Double direct shear configuration for hydrothermal experiments within thepressure vessel, showing rubber jacket, porous frits and pore fluids inlet and outlet. (d) Typical results for a complete experiment at three normal stresses. Insets showvelocity step test results from 1 to 300 μm/s and (lower inset) detailed data for a single velocity step, with raw data and a RSF model inversion. (e) Idealized RSF frictionresponse to a step increase in loading velocity. Black lines show velocity strengthening behavior and gray line shows velocity weakening behavior.

M.M. Scuderi et al. / Earth and Planetary Science Letters ∎ (∎∎∎∎) ∎∎∎–∎∎∎2

Please cite this article as: Scuderi, M.M., et al., Frictional properties and slip stability of active faults within carbonate–evaporitesequences: The role of dolomite and anhydrite. Earth and Planetary Science Letters (2013), http://dx.doi.org/10.1016/j.epsl.2013.03.024i

Table 1Experiment parameters. Experiments at elevated temperature and fluid saturated conditions are indicated by (v).

Experiment Simulatedgouge

T rn Pp Pc Shear strain (γ)

(1C) (MPa) (MPa) (MPa)

p2046 Anhydrite 25.5 10, 20, 35, 50 – – 9.2p2047 Anhydrite 25.9 75, 100, 150 – – 6.4p2097 Anhydrite (50%)+Dolomite (50%) 24.2 75, 100, 150 – – 9.0p2098 Anhydrite (50%)+Dolomite (50%) 23.7 10, 20, 35, 50 – – 13.3p2233 Dolomite from core sample 23.0 10, 20, 35, 50 – – 13.8p2230 Dolomite from core sample 22.7 75, 100, 150 – – 11.2p2231 Dolomite from outcrop 22.6 10, 20, 35, 50 – – 14.0p2232 Dolomite from outcrop 22.8 75, 100, 150 – – 10.5p2255v Anhydrite 75 10, 20, 35 2 3 10.9p2295v Anhydrite (50%)+Dolomite (50%) 75 10, 25, 35 2 3 23.5p2256v Dolomite from core sample 75 10, 20, 35 2 3 13.1p2257v Dolomite from outcrop 75 10, 20, 35 2 3 9.9

M.M. Scuderi et al. / Earth and Planetary Science Letters ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 3

seismic reflection profiles crossing the hypocentral area (Mirabellaet al., 2008), shows that the four major ruptures (5.6oMwo6.0)nucleated at about 6 km and within the sedimentary sequence ofthe Triassic Evaporites with the entire aftershock sequence con-fined within the sedimentary cover (Fig. 1).

Although seismic rupture nucleation and/or propagation withincarbonates seems to be quite common, little attention has beengiven to characterizing the frictional properties of carbonate-bearing faults under conditions within the shallow crust. Highvelocity friction experiments on calcite (Han et al., 2007), dolomite(De Paola et al., 2011b) and anhydrite (Di Toro et al., 2011) havedocumented a significant decrease in sliding friction for seismicsliding velocities of the order of 1 m/s. These data show that athigh slip rates, friction decreases to very low values, but they donot shed light on the question of earthquake nucleation withinevaporites or carbonate-bearing faults.

The purpose of this paper is to report on an experimentalinvestigation of the frictional properties and slip stability ofdolomite and anhydrite samples. We study the two mineralogiesof the Triassic Evaporites (Ciarapica and Passeri, 1976) at tempera-tures of 25 and 75 1C and compare the results with the micro-structures of an ancient and exhumed evaporite-bearing normalfault (e.g., De Paola et al., 2008).

2. Experimental methods

We performed double-direct shear experiments on layers ofpowdered samples at room temperature, under nominally dryconditions and under hydrothermal conditions, using a biaxialtesting machine and a true-triaxial pressure vessel (Fig. 2a). Rocksamples were collected from outcrops in Tuscany (Fig. 1) and deepboreholes in the Umbria–Marche Apennines (Trippetta et al.,2010). Intact solid samples were crushed and sieved to obtain agrain size fraction ≤150 μm. XRD and ICP-MS analysis were used todetermine the mineralogical composition of the cores, whichshowed that the core samples were relatively pure (495 wt%dolomite and 495 wt% anhydrite, cf. De Paola et al., 2011a),however the dolostone samples recovered from outcrop containedover 15 wt% calcite and gypsum (Trippetta et al., 2010). Experi-ments were conducted on dolomite (both from core and outcrop),anhydrite, and a 50–50% mixture (by weight) of anhydrite anddolomite from outcrop (Table 1).

2.1. Dry and room temperature experiments

Layers of simulated fault gouge were sandwiched in a three-blocks sample configuration consisting of a central forcing block

Please cite this article as: Scuderi, M.M., et al., Frictional propertiesequences: The role of dolomite and anhydrite. Earth and Planetary Sc

and two side blocks (Fig. 2b). The forcing block surfaces, in contactwith gouge material, were grooved by machining teeth of 0.8 mmheight and 1 mm spacing (Anthony and Marone, 2005). The gougelayers were prepared using leveling jigs in order to obtain a uniforminitial layer thickness of 3 or 5 mm and 5 cm�5 cm nominalcontact area. Subsequently, a rubber membrane was attached tothe bottom of the blocks to avoid loss of material during experi-ments and steel guide plates were fixed on each side of theassembly to reduce lateral extrusion of the gouges (Hong andMarone, 2005). The prepared sample assembly was then placedinside the biaxial loading frame and a normal force was applied viaa horizontal piston, initially controlled in a displacement feedbackmode until a small load was reached, at which point the piston wasswitched to a load feedback mode and load was increased until thetarget normal stress was reached. The applied normal load wascontrolled and measured, using custom-built load cells with anamplified output of 75 V and an accuracy of 70.1 kN.

Shear and normal displacements were measured using DCDTsattached to the vertical and horizontal pistons, with an accuracy of70.1 μm. Load point displacement measurements were corrected forstiffness of the testing apparatus, with nominal values of 0.5 kN/μmfor the vertical frame and 0.37 kN/μm for the horizontal load frame.After normal stress was applied, and the sample did not compact anyfurther, the vertical piston was moved down in displacement feed-back mode, applying force on the central block and consequentlyinducing shear within the two gouge layers. Data were acquired witha 24-bit710 V analog-to-digital converter at a rate of 10,000 sam-ples/s and averaged to obtain recording rates from 1 Hz to 10 kHzdepending on the sliding velocity. At the end of each experiment thesample assembly was removed and the blocks were carefullyseparated in order to recover the sheared gouges samples intact.Layers were subsequently impregnated with epoxy resin and stan-dard thin sections were cut for microstructural analysis (opticalmicroscope and Scanning Electron Microscope, SEM).

2.2. Wet and hot experiments

We conducted experiments at elevated temperature and underfluid-saturated conditions using the biaxial testing machine fittedwith a pressure vessel to allow true triaxial loading (Fig. 2a). Thesample assembly consisted of the same double-direct shear con-figuration used for room temperature experiments, but here theforcing blocks were different (Fig. 2c). The central and side blocksare equipped with internal channels, which provide access andflow paths for water to enter the gouge layers (Ikari et al. 2009;Samuelson et al., 2009). Sintered stainless steel porous frits(k�10−14 m2), positioned in depressions within the blocks, wereused to uniformly distribute fluids over the sample. The grooves

s and slip stability of active faults within carbonate–evaporiteience Letters (2013), http://dx.doi.org/10.1016/j.epsl.2013.03.024i

M.M. Scuderi et al. / Earth and Planetary Science Letters ∎ (∎∎∎∎) ∎∎∎–∎∎∎4

on the frits are 0.5 mm in height and spaced 1 mm apart. Gougelayers were prepared in a similar way as for the room temperatureexperiments, with a layer thickness of 5 mm and 5.7 cm�5.4 cmnominal contact area. In order to separate the gouge sample fromthe confining fluid, the assembly (gouge layers+forcing blocks)was jacketed following a three step procedure: (1) a flexiblerubber sheet of �4 mm thickness was taped around the blocksto provide support for the layers during the successive jacketingsteps; (2) two layers of latex rubber sheet were used to cover thefrits exposed on the central forcing block; and (3) two custom-made latex boots were used to cover the entire sample assembly.The assembly was then sealed using steel wires around the finallatex boot at the position of rubber O-rings on the blocks (Fig. 2c).At this point, fittings and tubing internal to the pressure vesselwere connected, and the sample was placed inside the pressurevessel. The pressure vessel is equipped with two dynamic seals,which provide access for the horizontal and vertical pistons.Additional fluid and confining oil ports allow connection of theinternal tubing for pore fluid pressure through the pressure vesselto external upstream and downstream pore pressure intensifiers.Pore pressures and confining pressure were servo-controlled usingfast-acting hydraulic servocontrollers. Fluid pressures were mon-itored with in-line diaphragm pressure transducers accurate to77 kPa. Confining pressure was applied using a non-toxic, hydro-genated, paraffinic white oil (XCELTHERM 600, Radco Industries).

The pressure vessel is accessed via removable front and rearsdoors, which are sealed with O-rings. Heaters are inserted into thepressure vessel via access holes in the doors. Three cartridgeheaters were used to increase the temperature of the oil duringexperiments, two of those (600 W heaters) were positioned on theback door, and one (1000 W heater) on the front door in order tominimize temperature variation within the pressure vessel. Twothermocouples were introduced in the pressure vessel throughaccess ports. One was placed �5 cm above the sample and used asfeedback to the temperature controller. The other was placed nextto the downstream pore pressure line to monitor fluid tempera-ture. The fluid pressure lines were coiled within the oil and fluidflow rates were kept sufficiently low to ensure that the pore fluidwas at temperature before it entered the sample. The inside of thepressure vessel was insulated using 1 cm thick ceramic plates toreduce internal temperature gradients. Initial heating to the targettemperature, T¼75 1C, was accomplished within 1 h. For our runconditions, the temperature controller maintained oil temperature towithin 72 1C and pore fluid temperature remained constant to ≤2 1C.

In each experiment, normal force was applied following theprocedure described above, and then the pressure vessel wasclosed, filled with oil and a confining pressure of 3 MPa wasapplied. Due to the sample geometry, part of the confiningpressure contributes to the normal stress (Ikari et al., 2009). Anupstream pore pressure (usually 0.2 MPa) was then applied, usingdeionized water (DI) as the pore fluid, until flow-through wasestablished. The downstream pore fluid line was then connectedand a pore pressure of 2 MPawas applied and maintained constantthroughout the experiment (i.e. drained conditions). Once flow inthe system was equilibrated, we began warming the oil. After atemperature of 75 1C was reached, the sample was left to equili-brate for 1 h. We then followed the same procedure as describedfor room temperature experiments. A correction was applied toexperiments in the pressure vessel to account for elastic compres-sion of the rubber jacket and jacket strength. The values of shearstress and shear displacement reported below have been socorrected. Once the experiment was finished, heaters were turnedoff, pore pressure was removed and the sample was left to cooldown for �30 min. When the oil temperature was low enough(�50 1C), the confining pressure was removed, the pressure vesselemptied and the front door removed. The assembly was left to cool

Please cite this article as: Scuderi, M.M., et al., Frictional propertiesequences: The role of dolomite and anhydrite. Earth and Planetary Sc

for an additional hour, after which pore pressure lines and ther-mocouples were disconnected, normal stress was removed, andthe sample assembly removed from the pressure vessel. Carefullyattention was given during jacket removal in order to collect thesheared gouge samples without damaging them. Samples werethen impregnated with epoxy resin and standard thin sectionswere cut for microstructural analyses. The complexity of thesample assembly and jacketing procedure made it difficult torecover large, intact portions of the sheared samples.

2.3. Friction measurements

We investigated friction constitutive properties as a function ofapplied normal stress in the range 10–150 MPa for all gouge types(Table 1). For each normal stress shear began at constant velocity of10 μm/s for �5 mm, until a steady state shear strength was achieved(Fig. 2d). Velocity was then increased stepwise in roughly 3-foldincrements from 1 to 300 μm/s with a constant shear displacementof 500 μm for each step. Shear stress was then removed, normalstress was increased and the procedure was repeated, but with ashorter run-in at 10 μm/s (�3 mm). Total shear displacement variedfrom �36 mm (for the experiments with four normal stresses) and�20 mm (for experiments with three normal stresses), correspond-ing to engineering shear strains between 9 and 23.5. The overallshear strength of the gouges was measured at 10 μm/s for eachnormal stress and fitted with the Coulomb–Mohr failure criterion:

τ¼ C þ msss′n ð1Þwhere τ is the shear stress, μss is the coefficient of stable sliding friction,C is cohesion (Handin, 1969), and s′n represents the effective normalstress, calculated as the difference between the applied normal stressand the pore pressure (Hubbert and Rubey, 1959). For experiments inthe pressure vessel, the effective normal stress is given by

s′n ¼ sn þ 0:506Pc−Pp ð2Þwhere sn is the normal stress, Pp is the average pore pressure, Pc is theconfining pressure and the prefactor (0.506) accounts for the con-tribution of Pc to normal stress (Samuelson et al., 2009; Ikari et al.,2009). We assume that powdered samples are cohesionless in ourrange of normal stresses, which is consistent with previous works (e.g.,Byerlee, 1978; Saffer and Marone, 2003). However, at elevatedtemperature and under fluid-saturated conditions, chemical reactionsmay be rapid enough to cause cementation and thus we evaluate thepossible role of cohesion as described below.

To investigate friction constitutive behavior and stability offault slip, we performed velocity step tests (Fig. 2d) and evaluatedthe rate- and state-dependent frictional behavior (Dieterich, 1979):

μ¼ μ0 þ alnðV=V0Þ þ blnðθV0=DcÞ ð3Þ

dθ=dt ¼ 1−Vθ=Dc ð4Þwhere μ0 represents a reference value of friction at a reference velocityV0, V is the slip velocity, θ is a state variable which may be thoughtof as the average life time of contact junctions or the porosity of agranular aggregate, a is the friction direct effect, b is the frictionevolution effect, and Dc is the critical slip distance, which representsthe distance required to renew the asperity contact population after aperturbation (Fig. 2e). The friction rate parameter a−b represents thechange in steady state friction with slip velocity:

a−b¼Δμss=ΔlnðVÞ ð5ÞPositive values of a−b denote velocity strengthening behavior,

which indicates an increase in shear strength with increasing strainrate and results in stable, aseismic slip. On the other hand, a negativevalue of a−b denotes velocity weakening behavior, which is char-acteristic of unstable or conditionally unstable behavior and is aprerequisite for dynamic slip instability.

s and slip stability of active faults within carbonate–evaporiteience Letters (2013), http://dx.doi.org/10.1016/j.epsl.2013.03.024i

0

0.01

1720

1735

18.1

1 μm/s 3 μm/s

0

0.01

0

2

4

1 μm/s 3 μm/s

ΔH

Laye

r Thi

ckne

ss(μ

m)

Dila

tion

(μm

)

Displacement (mm) Displacement (mm)

Laye

r Thi

ckne

ss (μ

m)

Coe

ffcie

nt o

f Fric

tion

(μ)

Displacement (mm)

1400

1600

1800

2000

2200

2400

2600

2800

0

p2295vp2098

Layer Thickness(p2098)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0

Layer Thickness(p2098)

18.4 18.7 18.1 18.4 18.7

5 10 15 20 255 10 15 20 25Shear Strain (γ)

μ μ

Fig. 3. (a) Data showing the evolution of frictional strength and layer thickness as a function of shear displacement (left side panel) and shear strain (right side panel). Blacklines show friction for the 50/50 mixture of anhydrite and dolomite at room temperature/dry. Red lines show data for experiment at T¼75 1C/wet. Gray lines show layerthickness for the room temperature/dry experiment (p2098). (b) Zoom on a velocity step (p2098), showing the evolution of frictional strength in response to a velocityperturbation (top curve). Friction data are corrected for strain weakening linear trend. Bottom curve shows the raw layer thickness evolution for the same velocity step.Dashed line represents the linear trend that was removed in order to analyze dilation. (c) Top curve shows the same velocity step of (b). Bottom curve shows the equivalentdilation (ΔH) measured after the linear trend due to geometrical thinning was removed. (For interpretation of the references to color in this figure legend, the reader isreferred to the web version of this article.)

M.M. Scuderi et al. / Earth and Planetary Science Letters ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 5

In order to determine the friction parameters a, b and Dc fromexperimental data, the finite elastic stiffness of the testingmachine must be taken into account. We do so using a singledegree of freedom elastic relation for the testing machine, sampleassembly and layer:

dμ=dt ¼ kðVlp−V Þ ð6Þwhere k is the elastic stiffness normalized by effective normalstress (for our apparatus, k¼3�10−3 μm−1 at 25 MPa normalstress), Vlp is the load point velocity and V is the slip velocity.Eqs. (3)–(6) were solved using a fifth-order Runge–Kutta numer-ical integration. To obtain best-fit constitutive parameters a, b andDc, we inverted experimental data using an iterative least squaremethod (Fig. 2d) (Blainped et al., 1998).

2.4. Dilatancy measurements

Previous works show that shear strength of gouge depends onboth frictional properties and volumetric strain. Dilatancy occursduring shear under some conditions and the evolution of volumetricstrain exhibits rate and state dependent properties (e.g., Marone et al.,

Please cite this article as: Scuderi, M.M., et al., Frictional propertiesequences: The role of dolomite and anhydrite. Earth and Planetary Sc

1990; Marone and Kilgore, 1993). In order to investigate the degree ofshear localization in response to step changes in the imposed slipvelocity, we analyzed changes in gouge layer thickness by monitoringthe displacement of the horizontal ram throughout the experiment.Dilation associated with a velocity perturbation was measured afterremoving monotonic layer thinning due to simple shear (Fig. 3aand b). We report dilation as the change in layer thickness normalizedby the change in slip velocity:

α¼ΔH=ΔlogðVÞ ð7Þ

where ΔH is the change in steady state layer thickness in response to avelocity step (Fig. 3c) (Marone and Kilgore, 1993). For granular faultgouges, the gouge layer volume typically increases upon an upstep insliding velocity, leading to a positive value of α.

Layer dilation with increasing velocity could cause dilatancyhardening if pore fluid pressure drops, potentially inhibiting seismicrupture nucleation or propagation (Samuelson et al., 2009). On theother hand, a decrease in layer thickness with increasing velocitycould cause compaction and/or shear localization, which would favornucleation of dynamic slip.

s and slip stability of active faults within carbonate–evaporiteience Letters (2013), http://dx.doi.org/10.1016/j.epsl.2013.03.024i

Fig. 4. Shear strength and coefficient of sliding friction as a function of normal stress for all samples investigated. (a) Room temperature data. Dashed line representsCoulomb failure envelope with cohesion of 0.03 MPa; (b) data for T¼75 1C fluid-saturated conditions. Dashed lines show Coulomb failure envelope for each gouge type;(c) coefficient of friction as a function of normal stress for dry experiments at room temperature (T¼25 1C) assuming cohesion of zero. (d) Water saturated experiments atT¼75 1C. Open symbols represent coefficient of friction calculated assuming negligible cohesion (μ¼τ/sn), and black symbols show the corresponding coefficient of frictionat s′n¼20 MPa resulting from the values of cohesion derived from the failure envelopes shown in panel b for each gouge type.

M.M. Scuderi et al. / Earth and Planetary Science Letters ∎ (∎∎∎∎) ∎∎∎–∎∎∎6

3. Results

3.1. Frictional strength

We measured the overall shear strength τ and coefficient ofstable sliding friction μss at steady state for each gouge (Fig. 4). Forexperiments conducted at room temperature and humidity,all samples show a linear relationship between shear stress andnormal stress with zero cohesion (Fig. 4a). At elevated temperatureand under fluid-saturated conditions, the values of frictional strengthare similar and consistent with brittle, linearly pressure-dependentfailure (Fig. 4b). However, our best-fit Coulomb failure parameters(Eq. (1)) for T¼75 1C indicate non-zero cohesion (Fig. 4b). Cohesionvalues range from 0.8 MPa to 5.2 MPa, with the highest values for the50–50% mixture of anhydrite and dolomite. Our data set for 75 1C islimited, thus we cannot rigorously distinguish between cohesivestrength and variation of internal friction with normal stress. Weconsider both possibilities.

For room temperature and humidity, our measured values ofsliding friction range from �0.6 to 0.7 and do not vary systematically

Please cite this article as: Scuderi, M.M., et al., Frictional propertiesequences: The role of dolomite and anhydrite. Earth and Planetary Sc

with normal stress (Fig. 4c). At elevated temperature and under fluid-saturated conditions, friction values vary from 0.45 to 0.75, if cohesionis assumed to be zero (Fig. 4d). These values decrease with increasingnormal stress. The solid symbols in Fig. 4d show internal frictionvalues for Coulomb failure envelopes that account for cohesion, shownfor our intermediate normal stress. These values range from �0.3 to0.6. Gouges composed of 100% dolomite from outcrop show cohesionof 0.8 MPa, comparable with the dry case. However, gouges composedof 100% dolomite from core samples (C¼2.7 MPa), 100% anhydrite(C¼4.4 MPa) and the 50–50% anhydrite and dolomite mixture(C¼5.2 MPa), show higher cohesion in the presence of water andelevated temperature.

3.2. Velocity dependence of friction

Experiments performed at room temperature and humidityare characterized by an overall velocity strengthening behavior(Fig. 5). The frictional constitutive parameter a−b shows values inthe range of −0.0020 and 0.0039 for all gouges, with the highestpositive values corresponding to the highest applied normal

s and slip stability of active faults within carbonate–evaporiteience Letters (2013), http://dx.doi.org/10.1016/j.epsl.2013.03.024i

Fig. 5. Friction parameter a−b for dry experiments at room temperature and different applied normal stresses. (a) Anhydrite; (b) anhydrite/dolomite mixture; (c) dolomitefrom outcrop samples; and (d) dolomite from drill core.

Fig. 6. Friction parameter a−b for water saturated experiments at T¼75 1C and a range of effective normal stresseses. (a) 100% Anhydrite; (b) anhydrite/dolomite mixture;(c) dolomite from outcrop samples; and (d) dolomite from drill core.

M.M. Scuderi et al. / Earth and Planetary Science Letters ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 7

Please cite this article as: Scuderi, M.M., et al., Frictional properties and slip stability of active faults within carbonate–evaporitesequences: The role of dolomite and anhydrite. Earth and Planetary Science Letters (2013), http://dx.doi.org/10.1016/j.epsl.2013.03.024i

Fig. 7. Dilatancy parameter (α) as a function of sliding velocity for all normal stresses under dry and hydrothermal conditions. (a) 100% Anhydrite simulated fault gouge;(b) anhydrite/dolomite mixture; (c) T¼75 1C, water saturated dolomite from drill core samples. (d) T¼75 1C, water saturated anhydrite simulated gouge; (e) T¼75 1C, watersaturated anhydrite/dolomite mixture; and (f) T¼75 1C, water saturated dolomite from drill core samples.

M.M. Scuderi et al. / Earth and Planetary Science Letters ∎ (∎∎∎∎) ∎∎∎–∎∎∎8

stresses. Gouges composed of 100% anhydrite, dolomite from core,and dolomite from outcrop do not show any particular evolutionof a−b with increasing velocity, with values in the range of−0.0006oa−bo0.003 (Fig. 5a, c and d). On the other hand, themixture of 50–50% anhydrite and dolomite shows an evolutionwith increasing velocity from strengthening toward velocity neu-tral behavior (Fig. 5b). Values of a−b range from −0.002oa−bo0.0039 at s′n¼10 MPa, to ≈0oa−bo0.002 at s′n¼150 MPa.

At elevated temperature and under fluid-saturated conditions,simulated gouges composed of 100% anhydrite and 100% dolomitefrom core show velocity strengthening behavior (Fig. 6a and d),with the frictional constitutive parameter a−b in the range of≈0oa−bo0.01. At effective normal stress of 20 MPa, the gougecomposed of 100% dolomite from outcrop evolves from velocitystrengthening to velocity weakening behavior with increasingvelocity, with a−b going from 0.01 to −0.0004 (Fig. 6c). The50–50% mixture of anhydrite and dolomite is characterized byan evolution from velocity strengthening to velocity weakeningbehavior with increasing velocity for both effective normal stres-ses of 10 MPa and 20 MPa (Fig. 6b). Values of a−b drop to −0.0005at s′n¼10 MPa and 300 μm/s. At s′n¼35 MPa no trend is observedand the behavior is velocity strengthening.

By comparing the two sets of experiments, we find that theelevated temperature experiments show a wider range of a−bvalues than the corresponding room temperature experiments.Moreover, the a−b values at elevated temperature are in mostcases higher than the corresponding values at room temperature.No particular trend is observed for the evolution of a−b withincreasing normal stress for experiments at elevated temperature.The gouge composed of 100% dolomite from outcrop and the50–50% mixture of anhydrite and dolomite show in general anevolution from velocity strengthening to velocity weakeningbehavior with increasing velocity, which is not observed at roomtemperature and dry conditions.

Please cite this article as: Scuderi, M.M., et al., Frictional propertiesequences: The role of dolomite and anhydrite. Earth and Planetary Sc

3.3. Dilation

Layer thickness was measured continuously throughout ourexperiments. Here, we focus on perturbations in thickness duringvelocity stepping, as quantified by the parameter α (Eq. (7)). Experi-ments conducted at room temperature and dry conditions show thatα increases with increasing velocity (Fig. 7a–c). Simulated gougescomposed of 100% anhydrite and the anhydrite/dolomite mixtureshow the same range of α values, 0.05 μmoαo0.9 μm, with thehighest values corresponding to the lowest normal stress (Fig. 7aand b). The gouges composed of 100% dolomite show lower α values,ranging from �0 to 0.4 μm without any particular evolution withnormal stress (Fig. 7c).

With increasing temperature and under saturated conditionsthree different behaviors are observed for the three gouge litholo-gies (Fig. 7d–f). Gouge composed of 100% dolomite shows anincrease in α with increasing velocity for the entire range of normalstresses (Fig. 7f). 100% anhydrite gouges are characterized by anincrease in α with increasing sliding velocity up to 30 μm/s abovewhich α remains constant at values between 1 and 1.5 μm (Fig. 7d).The 50–50% mixture of anhydrite and dolomite is characterized by alog-linear decrease of α with increasing velocity, from 2.5 μm to�1 μm, for the entire range of normal stresses applied (Fig. 7e).Generally at elevated temperature and in the presence of water, thevalues of α are larger than at room temperature and dry conditionsfor the same range of velocities.

4. Microstructures

At room temperature, cataclasis, grain size reduction, grainboundary sliding, and rotation are the dominant deformationmechanisms. We performed microstructural studies of thin sec-tions of our deformed samples, and find that R1 shear planes are

s and slip stability of active faults within carbonate–evaporiteience Letters (2013), http://dx.doi.org/10.1016/j.epsl.2013.03.024i

Fig. 8. SEM images of gouge samples from experiment at room temperature and dry conditions. (a) Simulated gouge composed of dolomite from outcrop samples subject toa final normal stress of 150 MPa. Note prevalence of R1 shear bands and overall grain size reduction. (b) Zoom of R1 shear band from panel (a). (c) Simulated gouge composedof dolomite from core samples, subjected to a final normal stress of 50 MPa. Note R1 shear bands and general grain size reduction. (d) Simulated gouge composed of 100%anhydrite, subjected to a final normal stress of 50 MPa. Note R1 shear bands and in some cases Y-shear planes. (e) Zoom on R1 shear band from panel (d). (f) Simulated gougecomposed of anhydrite/dolomite mixture, subjected to a final normal stress of 50 MPa. Note R1 planes associated with Y-shear planes. (g) Zoom on R1 plane from panel (f).

M.M. Scuderi et al. / Earth and Planetary Science Letters ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 9

the dominant features (e.g. Logan et al., 1979; Beeler et al., 1996;Marone, 1998; Niemeijer et al., 2010). The dolomite gouges testedup to s′n¼150 MPa (Fig. 8a) are characterized by pervasive R1

shear planes linked by sub-horizontal fractures. Grain size reduc-tion occurs along the entire fault zone where the larger clasts havedimensions of about 20 μm (note that the starting grain size waso150 μm). Along R1 shear planes, grain size reduction is morepronounced and here the fault gouge consists of micron and sub-micron grains of angular dolomite (Fig. 8b). For samples deformedat s′n¼50 MPa (Fig. 8c), there are fewer R1 shear planes with amore pronounced grain size reduction along these shear planes.The comparison of the microstructure for experiments p2232(s′n¼150 MPa, shear strain 10.5, Fig. 8a) and p2233 (s′n¼50 MPa,shear strain 13.8, Fig. 8c) suggests that the maximum appliednormal stress is a key parameter for developing abundant R1 shearplanes and pervasive grain size reduction. R1 shear planes withstaircase trajectories and incipient Y-planes with irregular geome-try characterize the microstructure of the anhydrite gouge(Fig. 8d). Grain-size reduction occurs along and around theseshear planes, with remnants of rounded anhydrite clasts withinthe fine-grained matrix (Fig. 8d and e). For mixtures of anhydriteand dolomite, tested up to normal stress of 150 MPa, deformation

Please cite this article as: Scuderi, M.M., et al., Frictional propertiesequences: The role of dolomite and anhydrite. Earth and Planetary Sc

is concentrated along R1 and Y-shear planes where grain sizereduction is significant (Fig. 8f and g). The comparison betweenthe microstructure of experiments p2232 (Fig. 8a) and p2098(Fig. 8f) suggests that the presence of anhydrite favors thelocalization of deformation along R1 and Y-shear planes wheregrain-size reduction concentrates.

At 75 1C, the dolomite-bearing fault gouge is characterized by azone where isolated large clasts, up to 150 μm, are surrounded by azone of finer ground mass (Fig. 9a, upper part). In general, the gougemicrostructure does not appear to be dominated by localization alongR1 and Y-shear planes nor by pervasive grain size reduction (Fig. 9aand b). Due to the low effective stress and elevated fluid pressure(s′n¼35MPa, confining pressure¼3MPa, fluid pressure¼2MPa andshear strain of 13.1) it is likely that much of the deformation wasaccommodated by grain sliding, rotation and translation. At 75 1C, theanhydrite fault gouge shows several R1 shear planes with an en-echelon geometry (Fig. 9c). The experiment at 75 1C on the 50–50%mixture of dolomite–anhydrite shows a distinct R-Y-P fabric (Fig. 9dand e). Here the Y-shear planes consist of through-going structuralfeatures with dolomite and minor anhydrite grains with micron andsub-micron dimensions (Fig. 9e). The P-foliation is defined by foldedbands of very fine-grained dolomite and anhydrite grains (Fig. 9f).

s and slip stability of active faults within carbonate–evaporiteience Letters (2013), http://dx.doi.org/10.1016/j.epsl.2013.03.024i

Fig. 9. SEM images of simulated gouges from water saturated experiments at elevated temperatures. (a) Simulated gouge composed of dolomite from core samples. Notedevelopment of B shear planes along the boundaries, parallel to the shear direction, and R1 oblique plane throughout the gouge. (b) Zoom of panel (a), showing grain sizereduction above B plane. (c) Simulated gouge composed of 100% anhydrite. Note R1 planes with en-echelon geometry. (d) Simulated gouge composed of anhydrite/dolomitemixture. Note development of R1 and Y-shear planes. (e) Zoom on R1 shear plane from panel (d). (f) Particular of P foliation from panel (d).

M.M. Scuderi et al. / Earth and Planetary Science Letters ∎ (∎∎∎∎) ∎∎∎–∎∎∎10

5. Discussion

5.1. Comparison with previous data

There have been relatively few previous studies of the frictionalproperties of anhydrites and dolostones fault rocks. Weeks andTullis (1985) conducted rotary shear experiments on solid dolo-mite marble surfaces (cut and roughened with 80 grit alumina),at a normal stress of 75 MPa and sliding velocities from 0.01 to10 μm/s. They found friction values in the range of 0.56–0.58. DePaola et al. (2011b) performed experiments at high slip rates(V41 m/s) and normal stress of 1.2 MPa on the same dolomite weused (our Dolomite Outcrop samples). They found that frictiondecays exponentially from peak values of about 0.8 to extremelylow steady-state values of about 0.1 (De Paola et al., 2011b). Thesmall difference between our data, 0.6oμo0.7 and those ondolomite marble (Weeks and Tullis, 1985) might be attributed todifferences in lithology, humidity and/or differences betweenextended solid surfaces and aggregates. The higher values of peakfriction documented in the high-velocity experiments of De Paolaet al. (2011b) are typical of high-speed rotary shear experimentsand may represent the coupled effect of friction, inertia and/or lowapplied normal stress.

Please cite this article as: Scuderi, M.M., et al., Frictional propertiesequences: The role of dolomite and anhydrite. Earth and Planetary Sc

For the velocity dependence of friction, our data confirm thevelocity strengthening behavior of dolomite gouge documented byWeeks and Tullis (1985). However, our data show a pronouncedevolution effect upon a velocity-step (large b-value), not docu-mented in Weeks and Tullis, (1985). This is an important differ-ence because it is the large value of b that can potentially lead to avelocity-neutral or velocity weakening behavior. We do not have adetailed understanding of this difference, however we suspect thatit is related to differences between solid surfaces and granularlayers. We note that previous works document a clear increase inthe friction parameter b with shear strain in granular layers (e.g.,Richardson and Marone, 1999).

Saw-cut type sliding experiments on fine-grained anhydritefault gouge show a decrease in friction from 0.47 to 0.39 uponaddition of water and water plus CO2 (Pluymakers, 2010). Anhy-drite gouge without water exhibited unstable stick-slip behavior(Pluymakers, 2010). High velocity friction experiments at lownormal stress (o5 MPa) show that anhydrite gouge evolves froma friction coefficient of about 0.8 at 0.01 m/s to 0.35 at 1 m/s(Di Toro et al., 2011). For experiments run under dry and roomtemperature conditions, our data confirm a frictional strengthwithin Byerlee's range, 0.6oμo0.7. At elevated temperature(75 1C) and in the presence of water, we document a strain

s and slip stability of active faults within carbonate–evaporiteience Letters (2013), http://dx.doi.org/10.1016/j.epsl.2013.03.024i

M.M. Scuderi et al. / Earth and Planetary Science Letters ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 11

weakening behavior resulting in a decrease in friction from 0.75 to0.44 (Fig. 3a). The friction behavior is predominantly velocitystrengthening with some occurrences of velocity-neutral or velo-city weakening behavior. As far as we know, there are no experi-mental data on the frictional properties of mixtures of anhydritesand dolostones.

5.2. Fault zone microstructure and mechanical data

In the following, we merge our laboratory data and microstructuralobservations to discuss the mechanical behavior of experimental faultsin dolomite and anhydrite. The experimental procedure allows us toinvestigate relatively low strains (γo14) and only in one experiment(T¼75 1C, 50–50% mixture of dolomite–anhydrite) we achievedγ¼23.5. Therefore, we emphasize that our data are most relevant tomicrostructural evolution in the initial stages of tectonic fault devel-opment. Our laboratory data show that fault gouges composed ofdolomite, anhydrite andmixtures of the two are frictionally strong andpredominantly velocity strengthening for low temperature conditionsand γo14. For experiments at low temperature and high stresses,deformation is accommodated by both inhomogeneously distributedcataclasis with grain fracturing (Fig. 8) and cataclasis with grain-sizereduction along R1 and rare, incipient Y-shear planes. We suggest thatthe low strain achieved is not sufficient to localize the deformationalong a continuous Y-shear plane, resulting in a velocity strengthen-ing/neutral behavior and an increase in dilation with sliding velocity(Figs. 5 and 7). For the same rocks, high-velocity friction experimentsshow that at higher strains, slip localization occurs on a continuousslip zone made of ultra fine-grained material that favors slip weaken-ing (De Paola et al., 2011b). For the experiments conducted at 75 1C onanhydrite (100%) and dolomite (100%) gouges, distributed cataclasiswith grain boundary sliding and grain rotation is facilitated both bythe low effective normal stresses (≤35MPa) and the presence of fluids.The absence of Y-shear planes with localized grain size reduction(Fig. 9a and c), the predominant velocity strengthening behavior forthese rocks and the increase of dilationwith increasing sliding velocity(Fig. 7d and f) indicate that deformation was distributed (Marone andKilgore, 1993; Niemeijer et al., 2009; Samuelson et al., 2009; Niemeijeret al., 2010).

The only experiment at larger strain (γ¼23.5) is p2295v (T¼75 1C,50–50% mixture of dolomite-anhydrite) where we observe a markedstrain weakening for γ415 (Fig. 3a). At elevated temperature,approaching hydrothermal conditions, dolomite is still strong, whereasanhydrite and anhydrite–dolomite mixtures exhibited moderate strainweakening. A possible explanation for the strain weakening weobserve with mixtures of anhydrite and dolomite at 75 1C, could bethat the increasing rehydration of anhydrite to form gypsumweakensthe fault. However, we consider this unlikely because (1) the increasein temperature allows the gypsum–anhydrite transformation and notvice-versa; and (2) experiments on identical anhydrites conducted forthe same experimental time (several hours) under wet conditions donot show the development of gypsum (De Paola et al., 2009). There-fore, the observed weakening is probably not related to the formationof gypsum, but perhaps arises from microstructural evolution in thegouge itself. Indeed, in this experiment, we observe a microstructurethat is characterized by a through-going Y-shear plane, made of ultrafine-grained material (Fig. 9d), suggesting efficient slip localizationonce the Y-shear plane achieved a continuous structure. This observa-tion is consistent with both the evolution from velocity strengtheningto velocity weakening and the decrease in dilation with increasingsliding velocity observed in the mechanical data. Other frictionexperiments and microstructural analysis on gouge samples showthat the transition from more distributed deformation, i.e., distributedcataclasis and Riedel and anti-Riedel type structures, to localizeddeformation, i.e., Y and B shears type structures (Logan et al., 1979;Rutter, 1986; Logan et al., 1992), is associated with the switch from

Please cite this article as: Scuderi, M.M., et al., Frictional propertiesequences: The role of dolomite and anhydrite. Earth and Planetary Sc

velocity strengthening to velocity weakening behavior (Beeler et al.,1996; Collettini, 2011).

5.3. Application to natural tectonic faults

For the 1997–1998 Umbria–Marche sequence (central Italy),seismic reflection profiles crossing the epicentral area show thatthe four largest events, 5.5oMwo6.0, nucleated within theTriassic Evaporites (Fig. 1) (Mirabella et al., 2008). This lithologyconsists of anhydrites and dolostones (Ciarapica and Passeri, 1976).Evaporitic rocks are thought to act as detachment horizons (i.e. bypromoting plastic behavior (Rutter, 1986; Green and Marone,2002) even at low pressure and temperature) in many fold andthrust belts, including the Northern Appennines (Bally et al., 1986;Barchi et al., 1998) and the French–Swiss Jura (Muller and Briegel,1980; Jordan and Nüesch, 1989). However, recent field studies(Collettini et al., 2008, 2009; De Paola et al., 2008) on ancientfaults that represent exhumed analogs of the active structuresresponsible of the seismicity within Triassic Evaporites ofthe Appennines show the typical characteristics of active faultswithin the shallow crust (e.g. Chester and Chester, 1998; Shiptonand Cowie, 2001; Faulkner et al., 2003; Wibberley and Shimamoto,2003; Agosta and Kirschner, 2003). In particular, large displace-ment structures (exhumation 42 km, displacement 4100 m)within the Triassic Evaporites are characterized by an inner faultcore affected by cataclastic processes with localization alongPrincipal Slip Surfaces (PSS), and a foliated outer fault core(Fig. 10a, and details in De Paola et al. (2008) and Collettini et al.(2009)). Microstructural detail (Fig. 10b) from the PSS shows that itis made of a fine-grained dolomite and Ca-sulphate rich cataclasite(in most of the outcrops the Ca-sulphate is secondary gypsumresulting from re-hydration of anhydrite at shallow crustal depth,i.e. To60 1C). Along the PSS, deformation is accommodated bycataclasis with grain size reduction and localization along theY-shear planes.

Although we do not have unambiguous evidence that the PSSformed during seismic faulting, the localization of the deformationalong continuous and thin (o1 mm) PSS (Fig. 9a and b) indicatesan extremely localized fault structure (e.g. Chester and Chester,1998; Wibberley and Shimamoto, 2005) that during slip is capableof facilitating thermally activated processes that may play a key-role in triggering earthquake slip instability (Rice, 2006; Di Toroet al., 2011). Finding the signature of thermally activated processesalong exhumed natural faults is an unsolved problem in faultmechanics (e.g. Cowan, 1999).

If we consider slip localization along the PSS documented inthe exhumed faults within the Triassic Evaporites (Fig. 10b) asevidence of a seismic fault, the microstructures documented alongthe natural fault are very similar to those obtained in ourexperiments on mixtures of anhydrite and dolomite at 75 1C(Fig. 10c). Here, we observe grain size reduction along R1 andcontinuous, through-going Y-shear planes. Clasts of dolomite arefragmented and smeared out along these shear planes resulting incontinuous boundary-parallel slip surfaces made of micrometricgrains of dolomite within an even finer matrix composed ofintermingled anhydrite–dolomite. The localization of deformationalong Y-shear planes is consistent with an evolution from velocitystrengthening to velocity weakening and from high to low dilationwith increasing sliding velocity (Fig. 10d).

Our results are dominated by velocity strengthening/neutralfriction behavior and these data do not provide a simple explanationfor frictional instabilities in anhydrites and dolomite. However,earthquakes occur as reactivation of mature faults (Scholz, 1998)and the mainshocks of the Umbria–Marche sequence nucleated at6 km depth, on evaporite-bearing faults with ∼500 m of displace-ment (Mirabella et al., 2008). If we look at our experimental data

s and slip stability of active faults within carbonate–evaporiteience Letters (2013), http://dx.doi.org/10.1016/j.epsl.2013.03.024i

IFC

OFC

OFC

PSSY Shear

(a-b

)

Velocity, µm/s Velocity, µm/s

Dila

tanc

y, µ

m

Fig. 10. (a) Exhumed Triassic Evaporite bearing normal fault. The fault is characterized by a cataclastic Inner Fault Core where slip is localized along Principal Slip Surfaces(PSS) and a foliated Outer Fault Core, OFC (for details see De Paola et al. (2008)). (b) Microstructural detail from the natural Principal Slip Surface consisting of a dolomite andCa-sulphate rich cataclasite: here the deformation is accommodated by cataclasis with grain size reduction and localization along a Y-shear plane.(c) Microstructural detail ofa fault rock formed at 75 1C by shearing (γ¼23.5) a mixture of anhydrite and dolomite (experiment p2295). Here the deformation is accommodated by cataclasis with grain-size reduction along R1 and Y shear planes. (d) Mechanical data for experiment p2295. With increasing sliding velocity we observe an evolution toward a velocity weakeningbehavior associated to a reduction in dilation.

M.M. Scuderi et al. / Earth and Planetary Science Letters ∎ (∎∎∎∎) ∎∎∎–∎∎∎12

collected under hydrothermal conditions (T¼75 1C, 50–50% mixtureof dolomite–anhydrite, γ¼23.5), they show a transition from velocitystrengthening to velocity weakening and decreasing dilation, con-sistent with frictional instabilities. Thus, we propose that withincreasing shear strain, the presence of strong dolomite and weakanhydrite within the fault core favors slip localization alongboundary-parallel Y or B shear planes. Slip localization, accompaniedby decreasing dilation with increasing slip velocity is our preferredmechanism for velocity weakening and thus the potential develop-ment of frictional instabilities in these rocks.

6. Conclusions

We performed laboratory experiments on natural fault rocks andsimulated fault gouges composed of 100% dolomite (from core andoutcrop samples), 100% anhydrite and a 50/50 mixture of anhydriteand dolomite, at room temperature and 75 1C. Within the range ofnormal stresses (10–150MPa) and velocities (1–300 μm/s) tested,simulated gouges show primarily velocity strengthening behaviorat room temperature, accompanied by increasing dilation and thedevelopment of abundant R1 shear bands. At elevated temperatureand under fluid-saturated conditions, shear localization occurred andwas accompanied by an evolution from velocity strengthening fric-tional behavior to velocity weakening behavior, a decrease in dilation,and the development of Y-shear planes. These results highlight thatshear localization is favored by shear strain and the concomitantpresence of anhydrite (weak) and dolomite (strong) within faultgouge. In natural faults within evaporitic and dolomitic rocks, it hasbeen shown that deformation is highly localized in Principal SlipSurfaces (PSS) of mixtures of anhydrite and dolomite. Our studysuggests that at seismogenic depth (i.e. �6 km, T�130 1C, fluidsaturated) shear localization within evaporitic rocks can promotefrictional instabilities resulting in earthquake nucleation. However,further studies under hydrothermal conditions are needed in order tobetter constrain the frictional behavior, stability and physicochemicalinteractions that might affect the seismogenic potential.

Please cite this article as: Scuderi, M.M., et al., Frictional propertiesequences: The role of dolomite and anhydrite. Earth and Planetary Sc

Acknowledgments

This research has been carried out within the ERC Starting GrantGLASS (No. 259256) awarded to Cristiano Collettini. Laboratory workwas supported by the NSF grants OCE-0648331, EAR-0746192, andEAR-0950517 to C. Marone. AN is supported by the VENI-grant No.863.09.013 awarded by the Netherlands Organisation for ScientificResearch (NWO) We thank S. Swavely for technical help andB. Carpenter and C. Viti for insightful discussions. We also wish tothank Nicola De Paola and Hiroyuky Noda for their constructivecomments.

References

Agosta, F., Kirschner, D.L., 2003. Fluid conduits in carbonate-hosted seismogenicnormal faults of Central Italy. J. Geophys. Res. 108 (B4), 2221, http://dx.doi.org/10.1029/2002JB002013.

Anthony, J.L., Marone, C., 2005. Influence of particle characteristics on granularfriction. J. Geophys. Res. 110, B08409, http://dx.doi.org/10.1029/2004JB003399.

Bally, A., Burbi, L., Cooper, C., Ghelardoni, R., 1986. Balanced sections and seismicreflection profiles across the Central Apennines. Mem. Soc. Geol. Ital. 35,257–310.

Barchi, M.R., Minelli, G., Pialli, G., 1998. The crop 03 profile: a synthesis of results ondeep structures of the Northern Apennines. Mem. Soc. Geol. Ital. 52, 383–400.

Beeler, N.M., Tullis, T.E., Blanpied, M.L., Weeks, J.D., 1996. Frictional behavior oflarge displacement experimental faults. J. Geophys. Res. 101 (B4), 8697–8715,http://dx.doi.org/10.1029/96JB00411.

Bernard, J.E., Hubert, W., Diraison, M., 2006. Seismicity wedge beneath the UpperRhine Graben due to backwards Alpine push? Tectonophysics 428, 49–64, http://dx.doi.org/10.1016/j.tecto.2006.08.009.

Blanpied, M.L, Lockner, D.A., Byerlee, D.J., 1995. Frictional slip of granite at hydrothermalconditions. J. Geophys. Res. 100 (B7), 13,045–13,064, http://dx.doi.org/10.1029/95JB00862.

Blainped, M.L., Tullis, T.E., Weeks, J.D., 1998. Effects of slip, slip rate, and shearheating on the friction of granite. J. Geophys. Res. 103 (B1), 489–511.

Byerlee, J., 1978. Friction of rocks. Pure Appl. Geophys. 116, 615–626.Burchfiel, B.C., Royden, L.H., van der Hilst, R.D., Heger, B.H., 2008. A geological and

geophysical context for the Wenchuan earthquake of 12 May 2008, Sichuan,People's Republic of China. GSA Today 18, 4–11.

Chester, F.M., Chester, J.S., 1998. Ultracataclasite structure and friction processes ofthe Punchbowl fault, Punchbowl fault, San Andreas system, California. Tecto-nophysics 295, 199–221.

s and slip stability of active faults within carbonate–evaporiteience Letters (2013), http://dx.doi.org/10.1016/j.epsl.2013.03.024i

M.M. Scuderi et al. / Earth and Planetary Science Letters ∎ (∎∎∎∎) ∎∎∎–∎∎∎ 13

Chiarabba, C., Amato, A., Anselmi, M., Baccheschi, P., Bianchi, I., Cattaneo, M., Cecere,G., 2009. The 2009 L'Aquila (central Italy) MW 6.3 earthquake: main shock andaftershocks. J. Geophys. Res. Lett. 36 (18), 1–6, http://dx.doi.org/10.1029/2009GL039627.

Chiaraluce, L., Ellsworth, W.L., Chiarabba, C., Cocco, M., 2003. Imaging the complex-ity of an active normal fault system: the 1997 Colfiorito (central Italy) casestudy. J. Geophys. Res. 108 (B6), 1–19, http://dx.doi.org/10.1029/2002JB002166.

Chiaraluce, L., 2012. Unravelling the complexity of Apenninic extensional faultsystems: a review of the 2009 L'Aquila earthquake (Central Apennines, Italy).J. Struct. Geol. 42, 2–18.

Ciarapica, G., Passeri, L., 1976. Deformazione da fluidificazione ed evoluzionediagenetica della formazione evaporitica di Burano. Boll. Soc. Geol. Ital. 95,1175–1199.

Collettini, C., Cardellini, C., Chiodini, G., De Paola, N., Holdsworth, R.E., Smith, S.A.F.,2008. Fault weakening due to CO2 degassing in the northern appennines:short- and long-term processes. Geol. Soc., London, Spec. Publ. 299, 175–194,http://dx.doi.org/10.1144/SP299.11.

Collettini, C., De Paola, N.D., Faulkner, D.R., 2009. Insights on the geometry andmechanics of the Umbria-Marche earthquakes (Central Italy) from the integra-tion of field and laboratory data. Tectonophysics 476, 99–109, http://dx.doi.org/10.1016/j.tecto.2008.08.013.

Collettini, C., 2011. The mechanical paradox of low-angle normal faults: currentunderstanding and open questions. Tectonophysics 510, 253–268.

Cowan, D.S., 1999. Do faults preserve a record of seismic slip? A field geologist'sopinion. J. Struct. Geol. 21, 995–1001.

De Paola, N., Collettini, C., Faulkner, D.R., Trippetta, F., 2008. Fault zone architectureand deformation processes within evaporitic rocks in the upper crust. Tectonics27 (TC4017), 1–21, http://dx.doi.org/10.1029/2007TC002230.

De Paola, N., Faulkner, D.R., Collettini, C., 2009. Brittle versus ductile deformation asthe main control on the transport properties of low-porosity anhydrite rocks.J. Geophys. Res. 114, B06211, http://dx.doi.org/10.1029/2008JB005967.

De Paola, N., Chiodini, G., Hirose, T., Cardellini, C., Caliro, S., Shimamoto, T., 2011a.The geochemical signature caused by earthquakes propagation in carbonate-hosted faults. Earth Planet. Sci. Lett. 310 (3–4), 225–232.

De Paola, N., Hirose, T., Mitchell, T., Di Toro, G., Viti, C., Shimamoto, T., 2011b. Faultlubrication and earthquake propagation in thermally unstable rocks. Geology39 (1), 35–38, http://dx.doi.org/10.1130/G31398.1.

Dieterich, J.H., 1972. Time-dependent friction in rocks. J. Geophys. Res. 77 (20),3690–3697.

Dieterich, J.H., 1979. Modeling of rock friction experimental results and constitutiveequations. J. Geophys. Res. 84 (B5), 2161–2168.

Di Toro, G., Goldsby, D.L., Tullis, T.E., 2004. Friction falls towards zero in quartzrock as slip velocity approaches seismic rates. Nature 427 (January), 774–777,http://dx.doi.org/10.1038/nature02249.

Di Toro, G., Han, R., Hirose, T., De Paola, N., Nielsen, S., Mizoguchi, K., Ferri, F., Cocco,M., Shimamoto, T., 2011. Fault lubrication during earthquakes. Nature 471,494–498, http://dx.doi.org/10.1038/nature09838.

Evans, J.P., Forster, C.B., Goddard, J.V., 1997. Permeability of fault-related rocks, andimplications for hydraulic structure of fault zones. J. Struct. Geol. 19 (11),1393–1404.

Faulkner, D., Lewis, A., Rutter, E., 2003. On the internal structure and mechanics oflarge strike–slip faults: field observations from the Carboneras fault, south-eastern Spain. Tectonophysics 367, 235–251.

Green, H.W., II, Marone, C., 2002. Instability of deformation, plastic deformation ofrocks. In: Karato, S., Wenk, H.R. (Eds.), Reviews in Mineralogy and Geochem-istry. ISBN:0-939950-63-4.

Han, R., Shimamoto, T., Ando, J., Ree, J.H., 2007. Seismic slip record in carbonate-bearing fault zones: an insight from high-velocity friction experiments onsiderite gouge. Geology 35, 1131–1133, http://dx.doi.org/10.1130/G24106A.1.

Handin, J., 1969. On the Coulumb–Mohor failure criterion. J. Geophys. Res. 74 (22),5343–5348.

Hong, T., Marone, C., 2005. Effect of normal stress perturbations on the frictionalproperties of simulated faults. Geochem. Geophys. Geosyst. 6, Q03012, http://dx.doi.org/10.1029/2004GC000821.

Hubbert, K.M., Rubey, W.W., 1959. Role of fluid pressure in mechanics of over-trusting faulting. Bull. Geol. Soc. Am. 70, 115–166.

Ikari, M.J., Saffer, D.M., Marone, C., 2009. Frictional and hydrologic properties ofclay-rich fault gouge. J. Geophys. Res. 114, B05409, http://dx.doi.org/10.1029/2008JB006089.

Jordan, P., Nüesch, R., 1989. Deformation structures in the Muschelkalk anhydritesof the Schafisheim well (Jura overthrust, northern Switzerland). Eclogae Geol.Helv. 82, 429–454.

Logan, J.M., Friedman, M., Higgs, N., Dengo, C., Shimamoto, T., 1979. Experimentalstudies of simulated fault gouge and their application to studies of natural fault

Please cite this article as: Scuderi, M.M., et al., Frictional propertiesequences: The role of dolomite and anhydrite. Earth and Planetary Sc

zones, In: Proceedings of Conference VIII—Analysis of Actual Fault Zones inBedrock Ed. US Geological Survey Open-file Report 79-1239101–120.

Logan, J.M., Dengo, C.A., Higgs, N.G., Wang, Z.Z., 1992. Fabric of experimental faultzones: their development and relationship to mechanical behavior. In: Evans,B., Wong, T.F. (Eds.), Fault Mechanics and Transport Properties of Rocks.International Geophysics Series. Academic Press, London, pp. 33–67.

Marone, C., Raleigh, B.C., Scholz, C.H., 1990. Frictional behavior and constitutivemodeling of simulated fault gouge. J. Geophys. Res. 95 (B5), 7007–7025.

Marone, C., Kilgore, B., 1993. Scaling of the critical slip distance for seismic faultingwith shear strain in fault zones. Nature 362, 618–620.

Marone, C., 1998. Laboratory-derived friction laws and their application to seismicfaulting. Annu. Rev. Earth Planet. Sci. 26, 643–696.

Miller, S.A., Collettini, C., Chiaraluce, L., Cocco, M., Barchi, M.R., Kaus, B., 2004.Aftershocks driven by a high pressure CO2 source at depth. Nature 427,724–727.

Mirabella, F., Barchi, M., Lupattelli, A., Stucchi, E., Ciaccio, M.G., 2008. Insights on theseismogenic layer thickness from the upper crust structure of the Umbria–Marche Apennines (central Italy). Tectonics 27 (1), 1–15, http://dx.doi.org/10.1029/2007TC002134.

Moore, D.E., Lockner, D.A., Byerlee, J.D., 1994. Reduction of permeability in graniteat elevated-temperatures. Science 265, 1558–1561, http://dx.doi.org/10.1126/science.265.5178.1558.

Muller, W.H., Briegel, U., 1980. Mechanical aspects of the Jura overthrust. EclogaeGeol. Helv. 73, 239–250.

Niemeijer, A., Elsworth, D., Marone, C., 2009. Significant effect of grain sizedistribution on compaction rates in granular aggregates. Earth Planet. Sci. Lett.284, 386–391, http://dx.doi.org/10.1016/j.epsl.2009.04.041.

Niemeijer, A., Marone, C., Elsworth, D., 2010. Fabric induced weakness of tectonicfaults. Geophys. Res. Lett. 37 (3), 1–5, http://dx.doi.org/10.1029/2009GL041689.

Patacca, E., Scandone, P., Di Luzio, E., Cavinato, G.P., Parotto, M., 2008. Structuralarchitecture of the central Apennines: interpretation of the CROP 11 seismicprofile from the Adriatic coast to the orographic divide. Tectonics 27 (3), 1–36,http://dx.doi.org/10.1029/2005TC001917.

Pluymakers, 2010. MS.c Thesis. Utrecht University, HPT Laboratory, unpublished.Rice, J.R., 2006. Heating and weakening of faults during earthquake slip. J. Geophys.

Res. 111, B05311.Richardson, E., Marone, C., 1999. Effects of normal stress vibration on frictional

healing. J. Geophys. Res. 104 (B12), 28859–28878.Rutter, E.H., 1986. On the nomenclature of mode of failure transitions in rocks.

Tectonophysics 122, 381–387.Saffer, D.M., Marone, C., 2003. Comparison of smectite- and illite-rich gouge

frictional properties: application to the updip limit of the seismogenic zonealong subduction megathrusts. Earth Planet. Sci. Lett. 215, 219–235, http://dx.doi.org/10.1016/S0012-821X(03)00424-2.

Samuelson, J., Elsworth, D., Marone, C., 2009. Shear-induced dilatancy of fluid-saturated faults: Experiment and theory. J. Geophys. Res. 114 (B12404), 1–15,http://dx.doi.org/10.1029/2008JB006273.

Scholz, C.H., 1998. Earthquakes and friction laws. Nature 391, 37–42.Scholz, C.H., 2002. The mechanics of Earthquakes and Faulting, 2nd edition

Cambridge University Press. (508) pp.Shipton, Z.K., Cowie, P.A., 2001. Analysis of three-dimensional damage zone

development over a micron to km scale range in high-porosity Navajosandstone, Utah. J. Struct. Geol. 23, 1825–1844.

Sibson, R.H., 1982. Fault zone models, heat flow, and the depth distribution ofearthquakes in the continental crust of the United States. Bull. Seismol. Soc. Am.72 (1), 151–163.

Trippetta, F., Collettini, C., Vinciguerra, S., Meredith, P.G., 2010. Laboratory measure-ments of the physical properties of Triassic Evaporites from Central Italyand correlation with geophysical data. Tectonophysics 492 (1–4), 121–132,http://dx.doi.org/10.1016/j.tecto.2010.06.001.

Ventura, G., Di Giovanbattista, R. Fluid pressure, stress field and propagation styleof coalescing thrust from the analysis of the 20 May 2012 ML 5.9 Emiliaearthquake (Northern Appennines, Italy). Terra Nova, in press.

Weeks, J.D., Tullis, T.E., 1985. Frictional sliding of dolomite: a variation inconstitutive behavior. J. Geophys. Res. 90 (B9), 7821–7826.

Wibberley, C.A.J., Shimamoto, T., 2003. Internal structure and permeability of majorstrike slip fault zones: the median tectonic line in Mie Prefecture, SouthwestJapan. J. Struct. Geol. 25, 59–78.

Wibberley, C.A.J., Shimamoto, T., 2005. Earthquake slip weakening and asperitiesexplained by thermal pressurization. Nature 436, 689–692.

Xu, X., Wen, X., Yu, G., Chen, G., Klinger, Y., Hubbard, J., Shaw, J., 2009. Coseismicreverse-and oblique-slip surface faulting generated by the 2008Mw 7.9 Wenchuanearthquake, China. Geology 37, 515–518.

s and slip stability of active faults within carbonate–evaporiteience Letters (2013), http://dx.doi.org/10.1016/j.epsl.2013.03.024i