Earth and Planetary Science Letters 413 (2015) 2536

Contents lists available at ScienceDirect

y

m

S icin

Sa Db Rc Id D

a

ArReReAcAvEd

Kelocagodyeaex

1.

by(Ealsem

*

ht00epartment of Geology, University of Otago, Dunedin 9054, New Zealandock Mechanics Laboratory, Department of Earth Sciences, Durham University, Durham, England, United Kingdomstituto Nazionale di Geosica e Vulcanologia (INGV), Rome 00143, Italyipartimento di Geoscienze, Universit degli Studi di Padova, Via G. Gradenigo 6, 35131 Padova, Italy

r t i c l e i n f o a b s t r a c t

ticle history:ceived 14 October 2014ceived in revised form 13 December 2014cepted 21 December 2014ailable online 14 January 2015itor: P. Shearer

ywords:calizationlciteugenamic weakeningrthquakesperiments

To determine the role of strain localization during dynamic weakening of calcite gouge at seismic slip rates, single-slide and slideholdslide experiments were conducted on 23-mm thick layers of calcite gouge at normal stresses up to 26 MPa and slip rates up to 1 ms1. Microstructures were analyzed from short displacement (

26 S.A.F. Smith et al. / Earth and Planetary Science Letters 413 (2015) 2536

mst

faslismencudasuwG(Gin

nabyexKiSmenanmdefaKoathethceseThnaanto

thdyinth(eMexgoexstth

mitciseteofe.rohicoinandy

2.

2.

VRohopabadudianroplboto

stDmrithisusat

inwthX-scofasetsuFic.

2.

huofgucota35plremmDstso3thditosh

immitexdiic strains must occur at fault irregularities such as bends and epovers (Sibson, 1986; Pavlis et al., 1993).Laboratory experiments have demonstrated that bare rock sur-

ces and gouge layers experience dynamic weakening when the ip velocity and sliding displacement approach values character-tic of earthquakes (Di Toro et al., 2011). A variety of physical echanisms have been proposed to explain the dynamic weak-ing behavior observed in the laboratory and postulated to oc-r in natural faults. In particular, mechanical and microstructural ta collected from experiments performed on solid rocks (bare rfaces) are consistent with the activity of ash heating and eakening at asperity contacts (Rice, 2006; Beeler et al., 2008;oldsby and Tullis, 2011; Kohli et al., 2011), silica gel lubrication oldsby and Tullis, 2002; Di Toro et al., 2004) and frictional melt-g (Di Toro et al., 2006; Nielsen et al., 2008).Gouge layers deformed at high velocities typically show a rrow (99 wt% calcite, with

S.A.F. Smith et al. / Earth and Planetary Science Letters 413 (2015) 2536 27

Fipaanbeinthampogrre

nogrmfu7wIncotounthjethonan

icrcacag. 1. Experimental set-up and sample assembly for gouge experiments. (a) SHIVA apparatus with main components labeled. (b) Scale diagram of gouge holder with main rts labeled. Details of calibration tests can be found in the Supplementary Information of Smith et al. (2013). The gouge layer (yellow) is contained between the outer d inner rings that slide over a base disc located in the stationary side (sliding contacts in red). Normal stress (n) is applied to the gouge layer by the loading assembly hind the axial column (Di Toro et al., 2010). Normal stress on the sliding rings is modulated by inner and outer springs. (c) Photograph of calcite gouge layer (35/55 mm t./ext. diameters) prior to an experiment. A thin layer of high-temperature grease is applied to the sliding surfaces of the rings to reduce friction. (d) Scale diagram showing e geometry of the annular gouge layer. Where in contact with the gouge layer, the rotary and stationary pieces have surface roughness with wavelength of 400 m and plitude of 200 m. The dashed line indicates the region where localization typically occurs in the gouge layers (see text for details). (e) Optical photomicrograph in plane larized light (main image) and backscattered scanning electron microscope image (inset) of calcite gouge compacted to 15 MPa without shear. The starting material has a ain size of

28 S.A.F. Smith et al. / Earth and Planetary Science Letters 413 (2015) 2536

FiSlithgoofanin0.0

foCath(s20

3.

3.

itatin(ipoeagoneaxg. 2. Evolution of shear stress, slip velocity and axial shortening during slideholdslide experiments on calcite gouge (s753) and solid cylinders of calcite marble (s758). p is reported on a log scale to highlight the mechanical behavior in the early stages of slip, but similar plots in Figs. 47 show slip values on a linear scale. (a) Slide 1: e onset of weakening in solid marble (grey arrow) occurred after c. 0.002 m of slip at a slip velocity of c. 0.1 ms1 (grey star on slip velocity curve). Instead, the calcite uges showed a prolonged phase of strengthening prior to peak shear stress and dynamically weakened (red arrow) after 0.2 m of slip at a much higher slip velocity 1 ms1 (red star on slip velocity curve). Slide 2: the onset of weakening in both solid marble and calcite gouge occurred after 0.002 m of slip at a slip velocity of 0.1 ms1. (b) During slide 1, compaction was negligible in the solid marble, but a phase of dilation occurred in the calcite gouge layer prior to peak stress (between 0.1d 0.2 m slip). Dilation ended in the gouge layer at peak stress and was followed by compaction of 150 m. During slide 2, 50 m of further compaction took place the calcite gouge layer after 0.05 m of slip. In the solid marble, a short phase of dilation lasting 1 mm was recorded just prior to peak stress (between 0.001 and 02 m). (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.)

re each experiment. The experimental procedures for the solid rrara cylinders were the same as those detailed above, except at a different sample holder was used to grip the solid cylinders ample procedures for solid cylinders described in Di Toro et al., 10; Niemeijer et al., 2011).

Results

1. Dynamic weakening in calcite gouges and solid cylinders

Fig. 2a shows the evolution of shear stress and slip veloc-y for two representative slideholdslide experiments performed 8.5 MPa normal stress and 2.25 ms1 max. slip velocity us-g (i) a 2 mm-thick layer of calcite gouge (red data, s753) and i) solid cylinders of calcite marble (grey data, s758). Slip is re-rted on a log scale to highlight the mechanical behavior in the rly stages of slip. For clarity, only the slip velocity data for the uge experiment are shown, but the slip velocity evolution was arly identical in both experiments. Fig. 2b shows variations in ial displacement during the two experiments, where positive

changes indicate compaction and negative changes indicate dila-tion.

These two experiments illustrate important differences in the mechanical behavior of calcite gouge layers and solid cylinders of calcite marble at seismic slip rates. During slide 1, the solid marble initially strengthened reaching peak shear stress (5.5 MPa) after c. 0.002 m of slip (approximated by grey arrow in Fig. 2a). The solid marble then dynamically weakened to a much lower shear stress of 1 MPa after c. 1 m of slip. Dynamic weakening in the solid marble initiated at a slip velocity of c. 0.1 ms1 (see grey star on slip velocity curve in Fig. 2a).

Compaction was negligible during slide 1 in the solid marble cylinders (Fig. 2b). Regularly spaced oscillations were observed in the axial displacement (and shear stress) data with wavelengths of c. 0.125 m in the solid marble experiments and c. 0.15 m in the gouge experiments, equivalent in both cases to one full rotation of the annular samples (which have slightly different internal and external diameters as described above). These regular oscillations reect small misalignments of the gouge holder or the rotary and

S.A.F. Smith et al. / Earth and Planetary Science Letters 413 (2015) 2536 29

Fislicafacwenelayis

axof

prD(ac.imw(Ftw(wOdu

thdyarstimboshth

cotephtopa

wFineacslbl

thlemnomda

stleno

gocist

3.

m0.oftomwacofma prdatotain

3.

imdithstw

toimcoanthFisltoMth(sanroth10th

3.

17mofg. 3. Distance to onset of dynamic weakening vs. normal stress for 31 single-de and slideholdslide experiments performed on 2- or 3-mm thick layers of lcite gouge, as well as 13 single-slide experiments on solid cylinders (bare sur-es) of calcite marble. In all experiments the acceleration and deceleration rates re 7 ms2. The data for slide 1 in gouges are approximated using best-t expo-ntial decay functions (R2 = 0.81 for 3 mm thick layers, R2 = 0.99 for 2 mm thick ers). (For interpretation of the references to color in this gure legend, the reader referred to the web version of this article.)

ial columns, or slightly non-parallel sliding surfaces in the case the solid marble samples.In comparison to the solid marble, the calcite gouge showed a olonged phase of strengthening at the start of slide 1 (Fig. 2a). ynamic weakening in the gouge initiated after c. 0.2 m of slip pproximated by red arrow in Fig. 2a) at a slip velocity of 1 ms1 (see red star on slip velocity curve in Fig. 2a). The min-um shear stress obtained by the gouge layer following dynamic eakening was slightly higher than in the solid marble samples ig. 2a). The gouge layer initially compacted by 50 m, then be-een 0.08 and 0.2 m a transient phase of dilation was recorded 100 m dilation; Fig. 2b). Dilation ended once peak shear stress as reached in the gouge layer, followed by renewed compaction. verall compaction of c. 150 m was recorded in the gouge layer ring slide 1 (Fig. 2b).During slide 2, the evolution of shear stress was similar in both e solid marble and calcite gouge (Fig. 2a). In both experiments, namic weakening initiated after c. 0.002 m of slip (grey and red rows in Fig. 2a) at a slip velocity of c. 0.1 ms1 (grey and red ars on slip velocity curve in Fig. 2a). The decay from peak to min-um shear stress occurred over roughly the same slip distance in th experiments, although as observed in slide 1 the minimum ear stress obtained by the gouge layer was slightly higher than e solid marble cylinders (Fig. 2a).Compaction during slide 2 in the gouges was relatively minor mpared to slide 1, although 50 m of compaction occurred af-r c. 0.03 m of slip (Fig. 2b). In the solid marble, a short-lived ase of dilation occurred between c. 0.001 and 0.002 m, just prior peak shear stress (Fig. 2b). Following peak shear stress, com-ction of 100 m was observed.Fig. 3 summarizes the slip distance required to initiate dynamic

eakening in calcite gouge layers (e.g. grey and red arrows in g. 2) and its dependence on normal stress and gouge layer thick-ss for 31 single-slide and slideholdslide experiments with an celeration rate of 7 ms2. Also shown are data from 13 single-ide experiments on solid cylinders (bare surfaces) of calcite mar-e. The main results can be summarized as follows (Fig. 3):i) During slide 1 in calcite gouges (red and green lled symbols) e initial strengthening phase lasts between c. 3 and 30 cm. The ngth of the strengthening phase decreases with increasing nor-al stress, and it also decreases in thinner gouge layers. Above a rmal stress of 1520 MPa, the length of the strengthening phase ay remain constant with increasing normal stress, although more ta are required to conrm this.ii) During slide 2 in gouges (red and green open symbols), the

rengthening phase is much shorter, lasting a few millimeters or ss. The strengthening phase in slide 2 is independent of both rmal stress and layer thickness.iii) The length of the strengthening phase during slide 2 in uges is comparable to that observed in solid cylinders of cal-te marble (grey symbols) over the range of investigated normal resses.

2. Microstructural evolution of calcite gouge layers

A series of experiments was performed at 8.5 MPa nor-al stress with increasing total displacements in the range of 010.35 m to provide insights in to the microstructural evolution the calcite gouge layers during the transition from strengthening dynamic weakening. Observations from three of these experi-ents are summarized below (Figs. 4, 6, 7). All three experiments ere performed with 3 mm-thick gouge layers, imposing a target celeration rate of 7 ms2 and a target maximum slip velocity 1.1 ms1. An additional small displacement (0.028 m) experi-ent was performed at a higher normal stress of 17.3 MPa with dolomite strain marker constructed in the calcite gouge layer ior to shearing (Fig. 5). In part a of each of Figs. 47, the grey ta curves show the shear stress evolution of experiments taken larger displacements, which serve to illustrate the typical dis-nce required for the onset of dynamic weakening (also see data Fig. 3).

2.1. 0.01 m slip (s784)Due to the small displacement in this experiment, the max-um slip velocity obtained was 0.27 ms1 (Fig. 4a). The total splacement was approximately an order of magnitude lower than e c. 0.1 m required to initiate dynamic weakening at this normal ress (grey curve in Fig. 4a; data in Fig. 3). Compaction of c. 50 mas measured after 0.01 m of slip (Fig. 4a).The calcite gouge layer contains a well-dened shear band up

20 m thick (between white arrows in Fig. 4b), dened in SEM ages by a much ner grain size and more compact appearance mpared to the surrounding gouge layer (Fig. 4b, c). The grain size d overall appearance of the bulk of the gouge layer (i.e. outside e shear band) are similar to the gouge starting material (compare gs. 1e and 4b). As observed in all gouge experiments that reached ip velocities >0.1 ms1, the shear band developed sub-parallel gouge layer boundaries (a Y-shear after e.g. Logan et al., 1979;arone and Scholz, 1989) and at a distance of c. 100 m from e surface roughness on the stationary side of the gouge holder ee position of dashed lines in Fig. 1d). The shear band consists of gular to sub-angular calcite grains

30 S.A.F. Smith et al. / Earth and Planetary Science Letters 413 (2015) 2536

Fibothgonarto

veco

ncogoDFigrsts(F

c.lacooflitrobydunototh

cog. 4. Mechanical data and microstructures of experiment s784, stopped after 0.01 m of slip. (a) Plot of shear stress, slip velocity and axial displacement vs. slip. The inset x shows a detail of the rst 0.012 m of slip. (b) SEM image of gouge layer. A narrow shear band of ne grain size (outlined by the white arrows) is developed close to e stationary side of the gouge holder (see approximate position of localization in Fig. 1d). The inset shows a schematic representation of the location of the preserved uge layer with respect to the original gouge layer boundaries. (c) Detail of shear band in part b showing angular calcite clasts

S.A.F. Smith et al. / Earth and Planetary Science Letters 413 (2015) 2536 31

Fiaxof(cawagelowgore

3.

inCodiwg. 5. Experiment s781 (17.3 MPa normal stress) performed with a dolomite strain marker and stopped after 0.028 m of slip. (a) Evolution of shear stress, slip velocity and ial displacement in experiment s781. (b) SEM and EDS images of the calcite gouge layer and dolomite strain marker. The small greyscale image shows an SEM mosaic the entire preserved gouge layer and a representation of the total experimental displacement. The colored image shows an EDS chemical map of Mg (dolomite) and Ca lcite) distribution in the gouge layer that was used to identify the distribution of dolomite and reconstruct the strain distribution after shearing. The dolomite marker s initially sub-perpendicular to gouge layer boundaries. (c) Tracing of the dolomite strain marker and interpretation of the three strain domains distinguished from the ometry of the marker; low-strain domain, intermediate-strain domain, and high-strain shear band. A series of R1-Riedel shears offset the edges of the strain marker in the -strain domain. (d) SEM image (location shown in parts b, c) of the high-strain shear band. The shear band is 50100 m wide and much ner grained than the adjacent

uge. The white arrow shows where one edge of the shear band is dened by a discrete surface. (For interpretation of the references to color in this gure legend, the ader is referred to the web version of this article.)

2.3. 0.2 m slip (s631)Experiment s631 was stopped after 0.2 m of slip (Fig. 6a). Dur-

g deceleration the shear stress recovered to nearly its peak value. mparison to other experiments performed under the same con-tions (grey curve in Fig. 6a; data in Fig. 3) indicates that s631 as stopped approximately mid-way through the dynamic weak-

ening phase. A total of c. 200 m of compaction was measured, although this included a transient phase of dilation between c. 0.05and 0.12 m (Fig. 6a). Dilation ended once peak shear stress was reached and dynamic weakening initiated.

Compared to experiments stopped before the onset of dynamic weakening (e.g. Figs. 5, 6), the bulk of the gouge layer has a much

32 S.A.F. Smith et al. / Earth and Planetary Science Letters 413 (2015) 2536

Fig. 6. Mechanical data and microstructures of experiment s631, stopped after 0.2 m of slip. (a) Plot of shear stress, slip velocity and axial displacement. (b) Optical pho-toscsliofinre

ntishtoobTh