Earth and Planetary Science Letters 333–334 (2012) 200–210

Contents lists available at SciVerse ScienceDirect

Earth and Planetary Science Letters

0012-82

http://d

n Corr

3450 U

E-m

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

Signature of coseismic decarbonation in dolomitic fault rocks ofthe Naukluft Thrust, Namibia

Christie D. Rowe a,b,n, Ake Fagereng b, Jodie A. Miller c, Ben Mapani d

a Earth and Planetary Sciences, University of California Santa Cruz, 1156 High St, Santa Cruz, CA 95060, USAb Department of Geological Sciences, University of Cape Town, Private Bag X3, Rondebosch 7701, South Africac Department of Geological, Geographical and Environmental Sciences, Stellenbosch University, Private Bag X1, Matieland 7602, South Africad Geology Department, University of Namibia, Private Bag 13301, Windhoek, Namibia

a r t i c l e i n f o

Article history:

Received 17 December 2011

Received in revised form

17 April 2012

Accepted 19 April 2012

Editor: T.M. Harrisonunlikely to form in carbonate-hosted faults, and other evidence for seismic slip must be identified.

Keywords:

dolomite

Naukluft Nappe complex

fault weakening

pressurization

fault rocks

1X/$ - see front matter & 2012 Elsevier B.V.

x.doi.org/10.1016/j.epsl.2012.04.030

esponding author at: Earth and Planetary

niversity St., Montreal, QC H3A 0E8, Canada.

ail address: christie.rowe@mcgill.ca (C.D. Row

a b s t r a c t

Unequivocal geological signatures of seismic slip are rare, exceptionally so in carbonate-hosted faults

where carbonate minerals dissociate at temperatures lower than those required for producing a friction

melt. This thermal dissociation leads to significant fault weakening by increased fluid pressure and/or

nanoparticle lubrication, preventing further heating of the fault surface. Pseudotachylyte is therefore

We studied the lower Cambrian Naukluft Thrust which crops out in central Namibia. It contains a

cataclastic dolomite fault rock, referred to as ‘‘gritty dolomite’’, which we interpret as a signature of

coseismic carbonate dissociation and subsequent fluid–rock interactions. The fault was active at

ambient temperatures below 2001C.

‘‘Gritty dolomite’’ contains: rounded, low aspect ratio dolomite clasts with a uniform Fe-rich

dolomite coating, euhedral to subhedral magnetite, quartz, and K-feldspar in a fine-grained, massive to

laminated carbonate matrix of particulate dolomite and crystalline calcite cement. The fault rock

texture, combined with evidence of injectites of gritty dolomite into the wallrock, indicates the

cataclasite deformed as a fluidized granular flow. At seismic slip velocities, frictional heating caused

dissociation of dolomite to CO2 and Ca-, Fe- and Mg-oxides. This release of CO2 decreased the pH of the

pore fluid in the fault, causing dissolution and rounding of dolomite clasts within an inertial grain flow,

and precipitation of carbonate coatings and euhedral silicates and oxides during subsequent cooling

and CO2 escape.

Examples of similar rocks having some, if not all of these characteristics have been described from

other carbonate-hosted faults. The geological setting of the Naukluft Thrust is unique in spatial extent

and quality of exposure, allowing us to eliminate alternative hypotheses for sources of CO2 to drive

fluidization.

& 2012 Elsevier B.V. All rights reserved.

1. Introduction

Fault rock assemblages reflect the mechanical processes bywhich faults accommodate deformation. Particular fault rocks aretherefore associated with particular fault slip styles. For example,earthquake slip can cause significant heating of fault rock (Rice,2006), which in silicate rocks leads to the formation of pseudo-tachylyte (Sibson, 1975), a lithified friction melt recognized as theonly unequivocal evidence for seismic slip preserved in the rockrecord (Cowan, 1999). However, pseudotachylyte is extremelyscarce in carbonate fault rocks (only reported by Vigan �o et al.,2011, at depth � 10 km), because fast slip in carbonate rocks at

All rights reserved.

Sciences, McGill University,

Tel.: þ1 514 398 2769.

e).

low pressure causes thermo-mechanical dissociation before melt-ing temperatures are reached (e.g. Han et al., 2007a,b). Therefore,a geological record of mineral dissociation, rather than melting, isa more useful indicator for paleoseismic slip in carbonate rocks.

Identification of carbonate fault rocks that have experiencedmineral dissociation related to seismic slip will not only allow foridentification of past earthquakes, but can also be used to inferco-seismic properties of carbonate faults. Carbonate dissociation is asignificant weakening mechanism along carbonate-hosted faults.Breakdown of carbonate minerals creates sudden localised spikes inCO2 pressure, driving local overpressure and reducing fault-planeeffective stress (e.g. Billi and Di Toro, 2008; Sulem and Famin, 2009).In addition, experiments show that carbonate dissociation producesnanoparticles of oxides, facilitating nanoparticle lubrication of theexperimental faults (De Paola et al., 2011; Han et al., 2007a,b; Hanet al., 2010, 2011).

C.D. Rowe et al. / Earth and Planetary Science Letters 333–334 (2012) 200–210 201

Textural and geochemical products of carbonate dissociationhave been directly observed (Hewitt, 1988) and inferred in longrunout landslides (Anders et al., 2010; Beutner and Gerbi, 2005),and in carbonate-hosted faults (Smith et al., 2008, 2011) based ontextural evidence for fluidization of granular fault rock. However,these examples may be explained by influx of volcanic CO2 (Smithet al., 2008) or groundwater (Beutner and Gerbi, 2005) and do notconstitute direct evidence for in situ fluidization driven directly byearthquake slip. Here we describe carbonate fault rocks from theNaukluft Thrust, the basal decollement surface of the NaukluftNappe Complex, Namibia (Korn and Martin, 1959; Viola et al.,2006). In an effort to understand carbonate fault processes, weconsider the mineralogy and the macro- and microstructure of thefault rocks, discuss possible signatures of seismic slip, and addressthe geological record of dynamic fault weakening mechanisms.

2. Geological context

The Naukluft Thrust is the basal decollement of an exhumedforeland nappe complex, exposed in a klippe of the southern flank

Windhoek

0

0 10 km

20°S

20°E30°S

Namibia

Naukluft 9 National Park

Office

0 2 km

Type LocalityFigure 2A

Footwall RocksNama Grp. LimestoneNama Grp. ShaleOlder metamorphic and igneous rocks

Mapped fault trace(this study)no grittydolomite

Previously mapped trace of the Naukluft Thrust

Gritty d

olomite

on N

auklu

ft Thru

st

Recent sediments and evaporite pans Neoproterozoic-Cambrian Nama Grp.Naukluft Nappe ComplexDamara BeltKaoko BeltOlder metamorphic and igneous rocks

Fig. 1. (A) Location of Naukluft Nappe Complex in Namibia, southwestern Africa (redr

Complex (simplified from Korn and Martin, 1959). Main trace of Naukluft Thrust in w

dolomite occurs is expanded in C. (C) Occurrence of ‘‘gritty dolomite’’ in Naukluft 9 are

geology from Sylvester (2009). Outcrop locations indicated with white boxes correlate

of the Damara Belt, central Namibia (Fig. 1A). The hanging wall tothe thrust is the Naukluft Nappe Complex, a structural stack ofvariably folded carbonate and siliciclastic metasedimentary rocksof Meso- to Neoproterozoic age. This nappe complex wasemplaced southward over Mesoproterozoic metamorphic andigneous rocks of the Kalahari Craton and onto the flat-lying� 560 to�520 Ma shallow marine sediments of the Nama Groupduring the late Neoproterozoic to early Cambrian Damara Oro-geny. The cumulative displacement accommodated on the Nauk-luft Thrust has been estimated by previous workers at� 502100 km (Hartnady, 1978; Martin et al., 1983). The max-imum burial temperature of the hanging wall is estimated as� 350 1C from illite crystallinity and metamorphic assemblages(Ahrendt et al., 1978). The metamorphic grade and likely deposi-tional age of the nappes increase to the north, in structurallyhigher rocks. The footwall Nama sediments are essentially unme-tamorphosed where cut by the thrust, and the maximum burialtemperature estimated from illite crystallinity is o200 1C(Ahrendt et al., 1978).

The Naukluft Thrust is nearly 100% exposed in the southernNaukluft Mountains (Fig. 1B). The shallow-dipping fault surface is

Olive Fault

Coarse blue-gray quartz arenite and gray dolostone

Red flaggy micaceous siltstoneand massive orange-gray dolostone

Blue-green shale with carbonate lenses, thin orange dolostones

Massive orange dolostone with red shales, white/purple marls

Purple and green shales:Swartzkalk series, Nama Group

Blue-gray limestone:Swartzkalk series, Nama Group

Naukluft Thrust

Olive Fault

Nama Group (foot wall)

Gritty dolomiteOther fault rocks

Naukluft Nappe Complex (hanging wall)

100 km

N

Olive Trail

GolfStop

Zebra Hill

awn from Miller, 1983; Viola et al., 2005). (B) Geological map of Naukluft Nappe

hite line. Studied areas along thrust trace are outlined in black, area where gritty

a with hanging wall and footwall lithologies, Naukluft-Namib National Park. Some

to samples shown in Figs. 2–5.

C.D. Rowe et al. / Earth and Planetary Science Letters 333–334 (2012) 200–210202

continuous, gently spoon-shaped (Korn and Martin, 1959) andgenerally marked by a dolomitized band visible for long distances(‘‘Sole Dolomite’’ Korn and Martin, 1959; Viola et al., 2006,Fig. 2A). Regionally, the fault rock assemblage (collectivelyreferred to as the Sole Dolomite) comprises carbonate mylonites,massive featureless dolostones, breccias, very fine-grained lami-nated carbonate horizons, and an unusual granular rock locallyreferred to as ‘‘gritty dolomite’’ (Korn and Martin, 1959; Violaet al., 2006). Each of these fault rocks requires different mechan-isms of formation and implies different fault behaviors duringdeformation. We have mapped fault rock distribution and varia-tion in hanging wall and footwall lithologies and fabrics along� 100 km trace length of the thrust fault surface to investigatethe relationships between these characteristics.

3. The gritty dolomite

Amongst the components of the Naukluft Thrust fault rockassemblage, the gritty dolomite is least well understood (e.g.Miller et al., 2008; Viola et al., 2006). Previous workers haveinterpreted the gritty dolomite as an unconformable sedimentaryhorizon which accommodated gravity sliding (Korn and Martin,1959), a ‘‘metasomatic mush’’ (Hartnady, 1978), a blasto-mylonite(Munch, 1978), an injected slurry of pressurized evaporite (Behret al., 1981), or a dolomitic cataclasite which was fluidized andlocally injected into wallrock during brittle deformation (Milleret al., 2008; Viola et al., 2006). All these authors noted flowtextures and common injections of gritty dolomite dikes intowallrock fractures. They all infer high strain and granular flowwithin the gritty dolomite, although the slip mechanism and theorigin and evolution of the rock assemblage is unclear. They alsoobserve that the gritty dolomite is found exclusively along themain surface of the major Naukluft Thrust decollement, and notalong less significant faults which sole into it. This observationsupports an interpretation for in situ formation of the grittydolomite as a product of shear strain over the interpretation offoreign injection into the decollement (allowed by Behr et al.,1981; Miller et al., 2008; Viola et al., 2006) which would alsoinject along other available fault surfaces. Viola et al. (2006)suggested that the gritty dolomite formed by comminution ofother fault rock members of the Sole Dolomite, but the specificmechanism, mineralogy and texture have not been explained, norhave any specific constraints on slip rates been determined byprevious studies.

The gritty dolomite forms a single layer of variable thickness(� 0:525 m thick; Fig. 2A), which occurs along only � 25 km of4100 km continuous fault trace exposure we have studied(Fig. 1B). This area is correlated to the limits of a limestone unitin the footwall (Figs. 1C and 2A), beyond which the discretedolomite horizon marking the fault plane does not contain ‘‘grittydolomite’’ but is massive or poorly developed, and the footwallrock is dominated by shales. In general, the footwall shales showdistributed deformation fabrics. The gritty dolomite is found onlyin this fault core or in injections filing fractures in the wallrockcontiguous with the fault core (Fig. 2B).

The gritty dolomite is massive or contains laminae (onlylocally parallel to layer boundaries) defined by clast size sorting,weak alignment of long axes of clasts, and color variation in thefine-grained matrix (Fig. 3A, B). This banding swirls around largerclasts, and shows inclination toward sites where fluidized grittydolomite flowed into wallrock fractures (a crack-filling injectionshown in Fig. 2B and incursion into hanging wall breccias shownin Fig. 2C). Flattened tight to isoclinal folds are common (Fig. 3B).The gritty dolomite layer also contains at least one to three fine-grained, clast-poor, laminated yellow–white crystalline carbonate

layers � 225 cm thick (Figs. 2D and 3C), typically with a sharpbase and diffuse upper transition to typical gritty dolomite. Thenumber and distribution of these fine-grained layers varies alongthe fault trace.

The framework of the gritty dolomite is composed of mostly� 1 mm monocrystalline rounded dolomite grains (range is 0.1–4 mm; Figs. 3A, D; 4A, B; 5A, B) and angular to rounded silicate lithicclasts (Fig. 3A). Euhedral crystals of magnetite ð � 0:0525 mmÞ arefound throughout the gritty dolomite (Figs. 3A; 4A, 5C), and euhedralalkali feldspar and quartz crystals overgrow and cement the grains(Figs. 3E, F; 5C, D). Some of these quartz crystals containfluid inclusions of CO2 and high-salinity brines (Fig. 5D) (Behr andHorn, 1982).

Overgrowth cements consisting of aggregates of � 100 mmK-feldspar and doubly terminated quartz crystals are concen-trated along laminae and in fold hinge regions (Fig. 3B) or occursporadically in the gritty dolomite fault rock (Fig. 3E), and over-grow and include clasts of gritty dolomite (Fig. 5C, D). Aggregatesof K-feldspar and quartz also occur as fine-grained pseudomorphsafter a euhedral phase, resembling a zeolite or feldspathoidmineral (Fig. 3F).

Each individual clast in the gritty dolomite bears a� 10220 mm coating of iron-rich dolomite. The coatings areoptically continuous with the dolomite grain nuclei (Fig. 4A, B).However, the strength and color of cathodoluminescence (CL) aredistinct and indicate that the coatings represent a single genera-tion of dolomite precipitation (Fig. 5A, B). Microprobe analysisreveals a consistent iron-enrichment (up to 5 wt.%) in the dolo-mite coatings, which may be responsible for the strong redluminescence color. Some grains have multiple layers of dolomitecoating.

The matrix of the gritty dolomite is composed of tiny roundedparticles of dolomite and patchy interstitial calcite cement(Fig. 4C, D). The dolomite particles are o5 mm, usually about1 mm and are rounded or ellipsoidal. The calcite cement partiallyinfills the pores between particulate dolomite matrix and includesparticles. The discrete laminated surfaces (e.g. Fig. 3C) are mostlymade up of calcite cement and fine dolomite particles with fewlarger grains.

No evidence of crystal plastic deformation was observed in anymineral grains of the gritty dolomite. The gritty dolomite alwayscrosscuts all mylonitic shear fabrics. Locally, fractures in an oldergritty dolomite layer are infilled by injections of younger grittydolomite, indicating that the granular layer was lithified betweenslip events.

4. Insights from experiments

To interpret the processes responsible for formation of thegritty dolomite, we refer to two types of laboratory experiments:high-speed friction experiments on carbonate rocks and gouge(Han et al., 2007a,b; Han et al., 2010, 2011; Smith et al., 2010),and experiments on CO2-brine-rock reactions at P/T conditionsconsistent with a few km depth (Kaszuba et al., 2003, 2005, 2006).

The high-speed friction experiments show that at room tem-perature, low to moderate normal stress and slip velocities ofabout 1 m/s, carbonate rocks dissociate during sliding to produceCO2 gas and very fine particles (10 nm–1 mm; Han et al., 2007a,2010). These particles are oxides of the cations from the consti-tuent carbonate minerals: CaO, MgO and FeO (Han et al., 2007a).This dissociation is caused by frictional heating of the slippingsurface above the carbonate dissociation temperatures(c.500750 1C for iron carbonates, 700–900 1C for calcite anddolomite; Han et al., 2007b, and references therein). During theseexperiments, the shear stress/normal stress ratio ðmÞ of the

Fig. 2. (A) Outcrop of the Naukluft Thrust ‘‘Type Locality’’ in the Naukluft River (location shown in Fig. 1C). (B) Example of gritty dolomite injection into fracture cutting

footwall partially dolomitized mylonite (photo from � 500 m east of outcrop shown in A). (C) Gritty dolomite is � 1:522:5 m thick at this site. Granular flow has

penetrated fractures in hanging wall dolomicrite. (D) A thin sharp discrete slip surface at the base of gritty dolomite is marked by 1–3 cm of white–yellow fine grained

laminated dolomite. The uppermost footwall limestone is gradationally dolomitized below the slip surface to a depth of 0.5–1 m. (For interpretation of the references to

color in this figure caption, the reader is referred to the web version of this article.)

C.D. Rowe et al. / Earth and Planetary Science Letters 333–334 (2012) 200–210 203

slipping surface drops from m� 0:6 at initiation of sliding down tomr0:1 at seismic slip speeds ðZ1 m=sÞ. They interpreted thecause of the weakening to be either pressurization of the slipsurface by CO2 production, or nanoparticle lubrication. Han et al.(2011) tested the frictional properties of periclase gouge (MgO) toassess whether nanoparticles have sufficiently low friction toexplain the observed weakening. In the unconfined experimentsat low stress, the dissociated mineral material is very porous andthe CO2 escaped easily, but similar weakening was observed.Therefore, Han et al. (2011) determined that the lubrication wascaused by rolling of nanoparticles to make coarser compositeparticles (up to 1 mm) and the development of grooves on theslipping surface. Smith et al. (2010) ran similar high velocityexperiments on wet calcite fault gouge, and observed the forma-tion of another type of particle: accretionary grains. The grainshave a core mineral clast surrounded by a concentric mantle offine accreted particles of carbonate. These accretionary grainswere not reported in the dry experiments (Han et al., 2007a,2010). However, Han et al. (2007b) reported in experiments on

solid calcite marble, that the particles produced included not onlyCaO, but also Ca(OH)2, showing that the reactive particles of CaObond with even the small amount of water available in thelaboratory atmosphere.

Kaszuba et al. (2005)’s experiments examined the effect of CO2

introduction, and then venting, on a shale–arkose–brine mixtureat temperatures similar to the likely temperature of the NaukluftThrust during fault activity ð � 200 1C.) The system was held atconstant temperature until the rock and brine were in equili-brium. In one experiment, CO2 was injected, the system was helduntil the brine composition became stable, and then the experi-ment was stopped. In a second experiment, the rock–brine systemwas held at the same conditions without injection of CO2. Tinycrystals of analcime, a zeolite mineral, were formed in bothexperiments. In the CO2 injection experiment, carbonic acidformed, lowering the pH of the brine, which facilitated anincrease in dissolved silica. The pH of the brine then increasedsubstantially when the experiment was terminated and the CO2

was allowed to vent from the system. When examining the rock

Fig. 3. Field photos of gritty dolomite fault rock. (A) Weathered outcrop surface of ‘‘gritty dolomite’’. Dark grains are subhedral magnetite crystals (labeled M). Black

arrows point to angular to rounded lithic fragments. White arrows point to single-crystal carbonate grains. Note equidimensional, smoothly rounded rhombic to spherical

shape. (B) Fine-grained K-feldpar and quartz occurring as cement along flow banding and concentrated in a fold hinge. (C) Very fine-grained laminated crystalline

carbonate horizon within gritty dolomite layer. Note diffuse edges with gradation in concentration of dolomite grains to gritty dolomite. (D) Angular clasts of dolomicrite

wallrock (light gray; Dm) in gritty dolomite with coarse crystals of dolomite recrystallized out of the dolomicrite (dn). At the center of photo, disaggregation of

recrystallized dolomicrite releases coarse mono-crystalline clasts into gritty dolomite. Within a few millimeters of their source area, they have acquired sub-spherical

shapes (d1). (E) Doubly terminated crystals of quartz (Q) and K-feldspar (K) overgrowing fine-grained gritty dolomite. (F) Fine-grained K-feldspar and quartz pseudomorph

after a euhedral zeolite or feldspathoid mineral.

C.D. Rowe et al. / Earth and Planetary Science Letters 333–334 (2012) 200–210204

material after the experiments, Kaszuba et al. (2005) observedseveral minerals which only formed in the CO2 experiments: asecond population of larger, skeletal zeolite crystals, magnesite,siderite, and clays. It is not known whether the growth of theseminerals was caused by the brine acidity, fluid immiscibilityeffects, or the conditions of termination of the experimentincluding CO2 venting (Kaszuba et al., 2003). Precipitation ofsilica species (quartz, chalcedony or amorphous silica) is alsolikely when the fluid pH increases (Kaszuba et al., 2005, 2006).

5. Origin of clasts in the gritty dolomite

The fault core layer we have described is on average 0.5–1.5 mthick along the continuous exposure shown in Fig. 1C. As well asthe gritty dolomite, the fault core contains 2–5 cm thick very fine-grained laminated yellow–white carbonate layers with a verysharp base. These resemble the slip surfaces produced in high-velocity slip experiments on marble by Han et al. (2010, see theirFig. 11), and we interpret these discrete laminated layers to be theprinciple slip surfaces of the Naukluft Thrust. Their gradationalboundaries to the gritty dolomite indicate that the two materialsare co-deformed, or were mobile at the same time, prior to theircurrent cohesive states. The abundant injectites of gritty dolomite

into fractures in the wall rock also indicate high mobility of thismaterial during flow.

The main framework grains of the gritty dolomite (lithic frag-ments, monocrystalline dolomite, magnetite, K-feldspar and quartz)and the granular texture are unusual for a carbonate rock and distinctfrom both footwall and hanging wall rock. Both grain-supported andmatrix-supported textures are common. The matrix is composed ofrounded mm-scale dolomite particles and patchy calcite cement. Thelithic fragments are clearly derived from the wall rock of the diversemetasedimentary Naukluft Nappe Complex, and the older meta-morphic and igneous rocks of the footwall. Here, we interpret thegenesis of all the other mineral grains we have described.

5.1. Dolomite

Monocrystalline dolomite is present as rhombic, euhedralcrystals within lithic dolomicrite clasts (Fig. 3D). The likely sourcefor these clasts is static dolomite recrystallization in the hangingwall of the fault, observed locally where the wall rock is massive,extensively fractured dolostone (Figs. 2C and 3C). The coarseovergrowth dolomite crystals are large (several mm � 1 cm),isolated, and euhedral and often show concentric CL banding,identical to those observed as clasts in the gritty dolomite (e.g.center Fig. 5B). Clasts in the gritty dolomite include rock

Fig. 4. Photomicrographs and electron backscatter images of typical gritty dolomite. Indicator arrows, Figs. 4 and 5: black ‘‘M’’: magnetite grain. Orange ‘‘O’’: dolomite

overgrowth or grain coating. Yellow ‘‘C’’: Concentric CL banding within a monocrystalline clast. Green ‘‘F’’: feldspar grain (clast or overgrowth cement). Blue ‘‘Q’’: quartz

crystal overgrowth. (A) (plane polarized light) and (B) (cross polarized light): Typical gritty dolomite texture: poor sorting, weak to no alignment of grain long axes, grain

shape ranges from subhedral with rounded-off corners to well-rounded, nearly equant, across all visible scales. Dolomite grain coatings are typically optically continuous

with the grain cores and is difficult to observe in ppl or xpl, but can be seen to have a slightly higher density of solid inclusions in xpl (e.g., ellipsoidal extinct grain at center

right of A and B). Sample from ‘‘Olive Trail’’ outcrop. (C) Electron backscatter microprobe image of gritty dolomite sample from Olive Trail, showing rounded grains

(predominantly medium gray dolomite) with interstitial particles of dolomite and whispy calcite cement (light gray). (D) Close-up on interstitial particulate dolomite

(medium gray) and calcite cement (light gray). Dolomite particles are smaller than a few microns in size, and are loosely packed in pore spaces between framework

dolomite grains. Calcite forms patchy cements that partially fills pore space, and includes tiny dolomite particles (e.g. upper left of photo). (For interpretation of the

references to color in this figure caption, the reader is referred to the web version of this article.)

C.D. Rowe et al. / Earth and Planetary Science Letters 333–334 (2012) 200–210 205

fragments of dolomicrite containing these overgrowth crystals(Fig. 3D) or the crystals may break free of the matrix and formmonocrystalline clasts (Fig. 3D, 5).

These source crystals in the wall rock dolomite are dominantlyeuhedral, but the derivative clasts in the gritty dolomite aresubrounded to rounded (Figs. 3–5). We suggest that the crystalswere broken out from the dolomicrite and incorporated into thefault rock matrix, and then changed shape due to chemical and/ormechanical wear. This transition is shown in Fig. 3D wherecrystal-rich rock fragments can be seen caught in the process ofmechanically breaking up, and the freed coarse crystals in thegritty dolomite groundmass are slightly rounded, even at closeproximity to their presumed source.

Lin (1999) relates the roundness of clasts in fault rocks to themechanism of their evolution, and shows that clasts in purelycataclastic rocks are statistically angular, whereas in pseudotachy-lytes the majority of grains are rounded with high sphericity. This is

because fracturing of grains creates new corners, and angularity,such that by comminution alone there is a limit to how rounded andspherical the grains may become. However, the survivor grains inpseudotachylyte are subject to another mechanism of grain sizereduction: thermal abrasion or dissolution in the melt. This acts tosmooth the corners of previously angular clasts, and increaseroundness. For clasts suspended in this corrosive fluid, the effect isradially symmetrical, which drives the clast shape toward spherical.

By comparison, we suggest rounded monocrystalline grains inthe gritty dolomite achieved their shape during suspension in achemically and/or thermally erosive fluid. The fact that quartziticwallrock fragments of the same size are typically angular (Fig. 3A)indicates that the mechanism of rounding preferentially affectedthe dolomite grains. This implies interaction with fluids in whichdolomite is substantially more soluble than quartz (for examplean aqueous fluid after CO2 introduction, which substantiallyreduces pH; Kaszuba et al. (2005) and references therein).

Fig. 5. Cold cathodoluminescence photomicrographs of typical gritty dolomite. Indicator arrows, all photomicrographs: black ‘‘M’’: magnetite grain. Orange ‘O’: dolomite

overgrowth or grain coating. Yellow ‘‘C’’: Concentric CL banding within a monocrystalline clast. Green ‘‘F’’: feldspar grain (clast or overgrowth cement). Blue ‘‘Q’’: quartz

crystal overgrowth. (A) Cold cathodoluminescence photomicrograph of same thin section shown in Fig. 4A and B. Individual clasts are well-rounded and ellipsoidal to

equant. Each clast is completely coated in a bright red luminescing layer of dolomite, (not seen optically) in this case about 15225 mm thick. Pores between coated grains

contain small grains of red-luminescing dolomite. (B) Laminated gritty dolomite with thinner red luminescent grain coatings and silica cement overgrowths. Lamination is

designated by bands of finer and coarser grains (thin white arrows) and weak alignment of the long axes of more elongate grains (thick white arrows). Faint greenish

patches are weakly luminescent feldspar overgrowths. Sample from Olive Trail. (C) Gritty dolomite with large magnetite grain, concentrically zoned monocrystalline clasts,

brightly luminescent grain coatings, and overgrowth cements of quartz and feldspar similar to the larger grained examples in Fig. 3. Sample from Zebra Hill. (D) Thin section taken at

the edge of a band of silicate overgrowth cement similar to the folded bands in Fig. 3B. F: One quartz lathe in gritty dolomite. Bright red- or yellow-luminescent dolomite and black

magnetite included within dark blue quartz indicate that quartz overgrew these grains. Dark greenish areas (e.g. below quartz) are anhedral feldspar overgrowths. Sample from Golf

Stop (similar to Fig. 3E). (For interpretation of the references to color in this figure caption, the reader is referred to the web version of this article.)

C.D. Rowe et al. / Earth and Planetary Science Letters 333–334 (2012) 200–210206

5.2. Magnetite

Coarse euhedral magnetite crystals (Fig. 3A) are only found inthe gritty dolomite and not present in the surrounding wall rocks.Viola et al. (2006) suggested that the magnetite precipitated froman externally derived supersaturated fluid. We consider an alter-native hypothesis where the magnetite formed from local thermaldissociation and subsequent precipitation of fault zone material.Fe-carbonates break down to CO2 and Fe-oxide at temperatures of420–560 1C (Han et al., 2007a; Isambert et al., 2003). Han et al.(2007a) observed the precipitation of magnetite when artificialgouges containing iron-bearing carbonate minerals were disso-ciated by frictional heating during high-velocity slip experiments.In the experiments, dissociation of siderite and escape of CO2 leftunbound iron oxides which precipitated as magnetite in thesample. However, these experiments yielded very fine-grainedmagnetite, whereas the magnetite grains we observe in theNaukluft Thrust reach up to about 1 cm across. We have notobserved any solid inclusions in the magnetite grains, showingthat the magnetite did not form as late overgrowth cements.

In Han et al. (2007a)’s high-velocity slip experiments, the hightemperature transient is associated with a peak in CO2 produc-tion, but there is no significant increase in fluid pressure as theCO2 escapes from the gouge. In the geological example of theNaukluft Thrust, fluid pressure may have persisted much longerwithin the fault zone (seconds to minutes instead of microse-conds in the experiments), dilating the fault, and thereforemagnetite crystals were allowed space and time to grow. Thus,it may have been possible for larger, euhedral grains to develop.Occurrence of large, euhedral magnetite in association with iron

carbonates is also reported from the Sevvattur carbonatites inSouthern India (Ramasamy et al., 2001), where magnetite occur-rence has been related to pressure drops in an ascending magmaunder low f O2

conditions. It is also possible that the high CO2

concentration has an effect on the relationship between diffusionrate and nucleation rate that causes fewer, larger crystals relativeto the experimental laboratory conditions where CO2 is efficientlydrained from the sample.

The alternative model with an external supersaturated fluid isrejected because (1) there is no clear source for an Fe-saturatedfluid, (2) no obvious cause for supersaturation of such a fluid inthe inferred ambient conditions, and (3) no clear reason why apercolating, external fluid should lead to precipitation of scat-tered euhedral magnetites within the fault rock only, and not alsoin surrounding fracture systems, or localized along permeablefracture planes.

5.3. Silicates

The silicate cements and overgrowths in the gritty dolomite(Figs. 3B, E, F; 5C, D) may be explained by solubility changesobserved in experiments on rock–brine–CO2 systems at similartemperature conditions (Kaszuba et al., 2005, 2006). Introductionof CO2 lowered pH, contributing to supersaturation. When the pHrecovered to more neutral values, possibly caused by CO2 escape,amorphous silica and quartz precipitated from the remainingaqueous fluid. In the experiments, which used an Na-rich brine,experiments with CO2 resulted in the precipitation of largecrystals of analcime ðNaAlSi2O6 � H2OÞ. We suggest that thepseudomorphic quartz–alkali feldspar aggregates (Fig. 3F) result

C.D. Rowe et al. / Earth and Planetary Science Letters 333–334 (2012) 200–210 207

from precipitation of a similar feldspathoid or zeolite in responseto solubility changes as pH varied with CO2 production andescape (or alteration of such a phase to K-feldspar andquartz7water; Goodwin, 1973). These silicates have also pre-viously been suggested to derive from a supersaturated, externalfluid (Viola et al., 2006), but the supersaturation would stillrequire rapid slip or advection to drive a significant change insolubility, and the source of such a fluid is uncertain.

5.4. Matrix

The solid products of carbonate dissociation (CaO, MgO andFeO) may not be stable at shallow crustal conditions over the c.500 MY since the Naukluft Thrust was active. The fine-grainedmatrix of particulate dolomite is morphologically similar to therolled nanoparticle aggregates observed in friction experiments(Han et al., 2010), and we interpret these as the remnant oxideparticles, recrystallized to stable carbonates by bonding with CO2.This recombination may have happened immediately post-seis-mically, when the temperature of the fault dropped, or may havecontinued over an extended period of time. Similarly, the inter-stitial calcite cements are texturally the latest-growing compo-nent of the gritty dolomite. These may have formed from the CaO(or CaOH) produced by carbonate dissociation, or may have beenintroduced later by groundwater.

6. Accretionary grains and granular flow rheology

Accretionary rounded grains (Figs. 4 and 5) have been inter-preted as evidence of ‘‘fluidized’’ granular flow in multiplesettings: clay–clast aggregates found in the slip zone of the1999 Chi-Chi earthquake (Boullier et al., 2009) and replicated inexperiments (Boutaread et al., 2008; Ujiie et al., 2011); in thedolomite rocks of the Heart Mountain slide (Anders et al., 2010;Beutner and Gerbi, 2005); in gouge of the Nojima Fault (Otsukiet al., 2003) and the Alpine Fault (Warr and Cox, 2001); in naturaland experimental ‘‘armored’’ calcite grains from the slip surfacesof faults in limestone (Smith et al., 2008, 2010, 2011); and inindustrial production of cement clinkers (Yuko et al., 2000). In theexperiments performed by Boutaread et al. (2008), the coefficientof friction dropped below 0.2 coincident with the formation of therounded accretionary grains.

These accretionary grains all have radially symmetric fine-grained mineral coatings, in non-foliated gouges. Multi-layeredcoatings, indicating multiple episodes of slip and accretion, were

Fig. 6. Schematic of carbonate dissociation during an earthquake on the Naukluft Thr

partially dolomitized calc-mylonite in the footwall. Second panel shows thermal dissocia

with resulting fluid and oxide dust. Last panel shows the Naukluft Thrust fault rock ob

(Fig. 3C) preserves slip and frictional dissociation surface, and total thickness of gritty d

inertial granular flow following CO2 release. (For interpretation of the references to colo

reported by Beutner and Gerbi (2005) and Smith et al. (2011).By comparison with the studies above, we interpret thecoated dolomite grains within the gritty dolomite (Fig. 5) asaccretionary grains coated during multiple episodes of fluidizedgranular flow.

Laminations, folding, and off-fault injections of gritty dolomiteindicate that the gritty dolomite flowed as a granular material.‘‘Fluidization’’ is commonly invoked in massive or laminated faultgouges with unsorted grains which are poorly or not aligned.However, ‘‘fluidization’’ is non-specific as to whether the grainsare in contact with one another during flow (grain slidingdominated) or in the grain-inertial regime (transient grain colli-sions) (Monzawa and Otsuki, 2003). ‘‘Fluidization’’ is also non-specific with regards to slip rate; however, based on observationsof mixed pseudotachylyte and fluidized cataclasite, Brodsky et al.(2009), Rowe et al. (2005), and Ujiie et al. (2007) argued thatfluidized cataclasites are likely to record fast slip. Smith et al.(2008) have also argued for seismic slip rates in the formation offluidized carbonate fault rocks along the Zuccale Fault. In all thesestudies, deformation in the grain-inertial regime is consideredmore likely at higher slip rates.

The rounded morphology of monocrystalline dolomite grainsindicates that they were not in contact with one another duringinferred chemical wear by a suspending fluid. If they had been, wewould expect selective etching on grain boundaries meeting fluid-filled pore space, resulting in increasingly tortuous grain bound-aries (not seen: Fig. 4C, D). Further, the radially concentric,constant-thickness coatings of Fe-rich dolomite found on everyobserved grain (Fig. 5) indicate that the grains were not in contactwith one another during precipitation of the grain coatings. Ifthey had been, the cement would line pore-spaces rather thangrains. An analog may be found in foods with hard sugar coatings;these are produced by tumbling food nuclei in a drum whilespraying sugar solution into the drum (e.g. candy coated popcorn,Green and Smith, 1963). Just like candy coatings on popcorn,mineral coatings would build up asymmetrically and thicken incrannies if the grains were not rapidly tumbled with the coatingmaterial (in this case, nanoparticles of oxides as observed inlaboratory studies; De Paola et al., 2011; Han et al., 2011).

7. Fluid chemistry, fluidization and seismic slip

Fluidized granular flow in faults can be enhanced by the presenceof a viscous pore fluid which transfers momentum and helps supportgrains (e.g. Kirkpatrick and Shipton, 2009). In the Naukluft Thrust,

ust. First panel shows mature thrust before earthquake: Dolostone breccias with

tion in the wake of an earthquake rupture front, and suspension/interaction of clasts

served today. Two to three centimeter thick laminated microcrystalline dolomite

olomite ð � 0:525 mÞ preserves volume that was pressurized and mobilized as an

r in this figure caption, the reader is referred to the web version of this article.)

C.D. Rowe et al. / Earth and Planetary Science Letters 333–334 (2012) 200–210208

active at shallow crustal depths, and young marine sedimentsforming the footwall, aqueous fluids would have been available fromcompaction dewatering of underthrust sediments, as well as diage-netic reactions. The advection of water along the thrust is recordedby coeval evidence of seawater-facilitated dolomitization along thefault (Miller et al., 2008). However, equilibrium pore fluid flow wouldnot cause suspension of granular materials or chemical erosionrequired during suspension to produce the observed grain roundingand sphericity. It would also not explain the change from radialerosion to radial accretion which produced the coated grains of thegritty dolomite.

To create the observed textures, the gritty dolomite must havebeen sufficiently dilated that grains were not in consistent contactwith one another during a period of chemical erosion, continuingthrough a period of chemical/mechanical accretion. This dilationis consistent with high fluid pressure and high grain velocity(Bagnold, 1954): conditions expected during granular flow andinjection by fluidization. Dry (fluid-absent) fluidization at highshear speeds could cause dilation, but not chemical reaction alongthe surfaces of the grains. The acute chemical erosion of thedolomite grains (but not the silicate grains) indicates that thefluid was preferentially reactive with carbonate. The addition ofCO2 to pore water brine could create such a fluid, with majorimplications for mineral dissolution, precipitation, and transport(Kaszuba et al., 2006).

The background temperature along the fault (between slipevents) must have been at or below the maximum temperatureexperienced by the wall rocks (o200 1C Ahrendt et al., 1978). Wehave no direct constraints on the pressure, but the low meta-morphic grade of the footwall shales, as well as highly localizeddeformation along the fault, and structural position in the toe ofthe foreland fold belt of the orogen, are all consistent withshallow crustal conditions (we assume r5 km, or lithostaticpressure r100 MPa).

At these conditions, fast frictional heating during earthquakeslip provides all the necessary mechanisms to explain each texturaland mineralogical element of the gritty dolomite we havedescribed. A schematic model is shown in Fig. 6. Panel A showsthe Naukluft Thrust (Sole Dolomite) as seen in Fig. 2. The hangingwall dolostone is fractured and locally brecciated, and the footwalllimestone is dolomitized near the contact (yellow–orange) with anarrow gradation to unaltered, undamaged footwall (blue). InPanel B, an earthquake rupture (red line) propagates along thefault surface. The transient frictional heating on the slip surfacecauses localised thermal dissociation of dolomite, resulting in therelease of CO2 and nanoparticles of CaO, MgO, and FeO (repre-sented by white cloud emanating from slipping surface) (Han et al.,2007a). The CO2 reduces the pore water pH, resulting in increasedcarbonate and silicate solubility. The nanoparticles of oxides mayform the sticky grains which accreted to the surface of suspended,tumbling clasts (similar to the aggregation of nanoparticles observedin experiments by Han et al., 2010). Panel C follows one dolomitegrain during suspended flow, while it is dissolved, or chemicallyeroded, in this fluid, becoming more spherical. As the temperaturedrops after the passage of the slip pulse, and CO2 is able to escapefrom the system through cracks in the wall rock (or is consumed bythe retrograde reconstitution of carbonate minerals), the pH in porewater would increase toward neutral. This change in pH wouldcause the deposition of quartz and feldspars or zeolites. The increasein Fe-content of the dolomite grain coatings (Fig. 5) is consistentwith precipitation during an increase in fluid pH (Katz, 1971).

The mineralogy compares favorably to the products of high-velocity slip experiments, but in these experiments, CO2 andparticles produced by decarbonation may freely escape the slipsurface. In a seismogenic fault at depth, these products would beconfined. The production of fluid, and thermal pressurization of

pore waters and newly produced CO2, would accentuate the slipweakening effects by decreasing the effective normal stress in thefault zone (Sulem and Famin, 2009). This pressurization is thelikely driving force which caused the injections of gritty dolomiteinto both hanging wall and footwall rocks.

By comparison to other fluidized fault rock assemblages onother faults (Rowe et al., 2005; Smith et al., 2008; Ujiie et al.,2007), shear experiments on carbonate gouges (De Paola et al.,2011; Han et al., 2007a,b; Han et al., 2010; Smith et al., 2010) andwith the added evidence of chemical erosion and grain coatingwhile clasts were in suspension and experiencing flow in aninertial regime, the gritty dolomite is interpreted as a fault rockindicative of seismic slip in a carbonate-hosted fault. The mineralassemblage and fault rock texture is explained with the addedconsideration of solubility fluctuations occurring in a H2O–CO2

fluid at rapidly changing pressure/temperature conditions duringand immediately after frictional heating by seismic slip (Fig. 6).

7.1. Uniqueness of the Naukluft Thrust ‘‘gritty dolomite’’

The fault rock described in this paper is relatively unusual, butthe key textural features are common to other carbonate faultrocks. The Naukluft Thrust may be an ideal case where thejuxtaposition of multiple features enabled us to develop a con-ceptual model of the process, which may be applied to otherfaults where some subset of these features is observed.

As noted above, the gritty dolomite only occurs along thesection of the Naukluft Thrust where dolomitized calc-myloniteforms the local footwall to the fault, although hanging walllithology is quite variable (Fig. 1C). This may be attributable tothe relative strength of dolostone, which has a much higherplastic yield strength than calcitic limestone or clay-rich shaleat low temperature ðo200 1CÞ. We speculate that this increasedstrength prevented plastic creep during interseismic periods,thereby promoting frictional slip as the principal mechanism offault motion. Therefore, the assemblage we observe shows noplastic overprint and primary textures are preserved.

This dolostone also contains iron, which formed magnetiteafter reaction. In the Tre Monti Fault, Italy, which is hosted incalcitic limestone, calcite-coated grains have been observedwhich are morphologically similar to the ones reported here,however, the core grains are not noticeably rounded (Smith et al.,2011). This texture can be explained if the grains of carbonategouge did not interact with a low-pH aqueous fluid prior to beingcoated in the fine particles produced by dissociation. This idea issupported by laboratory experiments on wet calcite gouge whichproduced coated grains but no rounding (Smith et al., 2010).Similarly, angular grains with concentric carbonate coatings havebeen observed in long-runout landslides such as the HeartMountain Detachment, Wyoming (Anders et al., 2010; Beutnerand Gerbi, 2005). In both the experiments and the landslide case,water was present but likely limiting, if not confined in the slipsurface in sufficient volume to suspend and corrode the individualcarbonate grains. No magnetite is reported in the Tre Monti faultgouge or analog experiments, consistent with the lack of iron-bearing minerals in the host rock (Smith et al., 2010, 2011).

The matrix textures we observe in the Naukluft Thrust (Fig. 4C,D) are very similar to the � 1 mm particulate carbonate matrixreported by Vigan �o et al. (2011) as carbonate mylonites. Theirinterpretation was supported by the detection of amorphousaluminosilicate material found in the matrix with the carbonateparticles. They interpret this amorphous material as glass, aproduct of silicate melt. However, we note that amorphousaluminosilicate material is produced at lower temperature infriction experiments on clay-rich gouges, without melting. Thedehydration of clay minerals is seen in experiments by Boutaread

C.D. Rowe et al. / Earth and Planetary Science Letters 333–334 (2012) 200–210 209

et al. (2008) and Ujiie et al. (2011) to facilitate the formation ofarmored grains (CCAs) which are coated in non-stoichometricvery fine grained material derived from the breakdown of clay.Therefore, frictional heating to clay dehydration and carbonatedissociation temperatures ð4800 1CÞ may explain their observedassemblage without requiring melting. These temperatures maybe more difficult to achieve in carbonate rocks, as the frictionalheat production is lower than for silicates at the same slip rate. Iftrue carbonate pseudotachylyte exists, it must be very rare, sinceas noted by Vigan �o et al. (2011), dissociation is more likely thanmelting if water is locally available. Subsequent studies on theNaukluft Thrust rocks will investigate the possibility that the fine-grained slip surfaces (Fig. 3C) are carbonate frictional melts ashypothesized by Vigan �o et al. (2011).

8. Conclusion

We have described a unique fault rock assemblage of mineralsand structures (‘‘gritty dolomite’’), which we interpret as diag-nostic of CO2–H2O–rock interaction at transiently high tempera-ture ð4800 1CÞ, found along the slip surface of the carbonate-hosted Naukluft Thrust. High-temperature (co-seismic) fluid–rockinteraction during inertial grain flow is recorded by rounded, lowaspect ratio carbonate grains. Post-seismic cooling and fluidescape from the fault is recorded by the precipitation of even-thickness carbonate grain coatings, silicate cements and euhedraloxides. This fault rock is therefore interpreted as a record offrictional carbonate dissociation during seismic slip. If this inter-pretation is correct, it records a qualitative fluid evolution andtemperature path during slip, and represents a distinct fault rockassemblage which can be used to identify coseismic slip surfaceson dolomite-hosted faults, and address the co-seismic processesoperating in carbonate-hosted faults.

Acknowledgements

Thanks to Nils R. Backeberg, Lynise Esterhuizen, Carly Faber,and Fernando Y. G. Sylvester, for assistance, helpful conversationsand braai in the field, and to Steve Smith for helpful discussions.We particularly appreciate the very thoughtful and thoroughcomments from John Kaszuba and the contributions from theeditor and an anonymous reviewer. We thank Parks Namibia forsampling permissions and hospitality. Funding from the SouthAfrican Regional Cooperation Fund for Scientific Research andTechnological Development (NRF South Africa) to J.A. Miller, NSFMARGINS Award OCE-0840977 to E.E. Brodsky and C. Rowe andUniversity of Cape Town start-up funds to A. Fagereng. PrideMangeya and Kate Naude appear in Fig. 2A along with JodieMiller.

References

Ahrendt, H., Hunziker, J.C., Weber, K., 1978. Age and degree of metamorphism andtime of nappe emplacement along the southern margin of the Damara Orogen/Namibia (SW-Africa). Geol. Rundsch. 67, 719–742.

Anders, M.H., Fouke, B.W., Zerkle, A.L., Tavarnelli, E., Alvarez, W., Harlow, G.E.,2010. The role of calcining and basal fluidization in the long runout carbonateslides: an example from the Heart Mountain Slide Block, Wyoming andMontana, USA. J. Geol. 118, 577–599, http://dx.doi.org/10.1086/656383.

Bagnold, R.A., 1954. Experiments on a gravity-free dispersion of large solid spheresin a newtonian fluid under shear. Proc. R. Soc. London Ser. A: Math. Phys. Sci.225, 49–63.

Behr, H.J., Ahrendt, H., Schmidt, A., Weber, K., 1981. Saline horizons acting asthrust planes along the southern margin of the Damara Orogen (Namibia/SW-Africa). Geol. Soc. London Spec. Publ. 9, 167–172.

Behr, H.J., Horn, E.E., 1982. Fluid inclusion systems in metaplaya deposits and theirrelationships to mineralization and tectonics. Chem. Geol. 37, 173–189.

Beutner, E.C., Gerbi, G.P., 2005. Catastrophic emplacement of the Heart Mountainblock slide, Wyoming and Montana USA. Geol. Soc. Am. Bull. 117, 724–735.

Billi, A., Di Toro, G., 2008. Fault-related carbonate rocks and earthquake indicators:recent advances and future trends. In: Landlowe, S.J., Hammler, G.M. (Eds.),Structural Geology: New Research. Nova Science Publishers, Inc., p. 24.

Boullier, A.M., Yeh, E.C., Boutareaud, S., Song, S.R., Tsai, C.H., 2009. Microscaleanatomy of the 1999 Chi-Chi earthquake fault zone. Geochem. Geophys.Geosyst. 10, 25, http://dx.doi.org/10.1029/2008GC002252.

Boutaread, S., Calugaru, D.G., Han, R., Fabbri, O., Mizoguchi, K., Tsutsumi, A.,Shimamoto, T., 2008. Clay-clast aggregates: a new textural evidence forseismic fault sliding? Geophys. Res. Lett. 35, 5.

Brodsky, E.E., Rowe, C.D., Meneghini, F., Moore, J.C., 2009. A geological fingerprintof extremely low viscosity fault fluids mobilized during an earthquake.J. Geophys. Res. 114, 14, http://dx.doi.org/10.1029/2008JB005633.

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

Goodwin, J.H., 1973. Analcime and K-feldspar in tuffs of the green river formation,Wyoming. Am. Mineral. 58, 93–105.

Green, J., Smith, P.S., 1963. Sugar-Coating Process. U.S. Patent 3,094,947. FiledFebruary 12, 1959, Ser. No. 792,697.

Han, R., Hirose, T., Shimamoto, T., 2010. Strong velocity weakening and powderlubrication of simulated carbonate faults at seismic slip rates. J. Geophys. Res.115. http://dx.doi.org/10.1029/2008JB006136.

Han, R., Hirose, T., Shimamoto, T., Lee, Y., Ando, J.i., 2011. Granular nanoparticleslubricate faults during seismic slip. Geology 39, 599–602, http://dx.doi.org/10.1130/G31842.1.

Han, R., Shimamoto, T., Ando, J.i., Ree, J.H., 2007a. Seismic slip record in carbonate-bearing fault zones: an insight from high-velocity friction experiments onsiderite gouge. Geology 35, 1131–1134, http://dx.doi.org/10.1030/G24106A.1.

Han, R., Shimamoto, T., Hirose, T., Ree, J.H., Ando, J.i., 2007b. Ultralow friction ofcarbonate faults caused bythermal decomposition. Science 316. http://dx.doi.org/10.1126/science.1139763.

Hartnady, C.J.H., 1978. The stratigraphy and structure of the Naukluft NappeComplex. Precambrian Research Unit Annual: University of Cape TownDepartment of Geology 14/15, 163–170.

Hewitt, K., 1988. Catastrophic landslide deposits in the Karakoram Himalaya.Science 242, 64–67.

Isambert, A., Valet, J.P., Gloter, A., Guyot, F., 2003. Stable Mn–magnetite derivedfrom Mn–siderite by heating in air. J. Geophys. Res. 108, 2283, http://dx.doi.org/10.1029/2002JB002099.

Kaszuba, J.P., Janecky, D.R., Snow, M.G., 2003. Carbon dioxide reaction processes ina model brine aquifer at 200 1C and 200 bars: implications for geologicsequestration of carbon. Appl. Geochem. 18, 1065–1080.

Kaszuba, J.P., Janecky, D.R., Snow, M.G., 2005. Experimental evaluation of mixedfluid reactions between supercritical carbon dioxide and NaCl brine: relevanceto the integrity of a geologic carbon repository. Chem. Geol. 217, 277–293.

Kaszuba, J.P., Williams, L.L., Janecky, D.R., Hollis, W.K., Tsimpanogiannis, I.N., 2006.Immiscible CO2-H2O fluids in the shallow crust. Geochem. Geophys. Geosyst.7, 11, http://dx.doi.org/10.1029/2005GC001107.

Katz, A., 1971. Zoned dolomite crystals. J. Geol. 79, 38–51.Kirkpatrick, J.D., Shipton, Z.K., 2009. Geologic evidence for multiple slip weakening

mechanisms during seismic slip in crystalline rock. J. Geophys. Res. 114, 14, http://dx.doi.org/10.1029/2008JB006037.

Korn, H., Martin, H., 1959. Gravity tectonics in the Naukluft Mountains of SouthWest Africa. Geol. Soc. Am. Bull. 70, 1047–1078.

Lin, A., 1999. Roundness of clasts in pseudotachylytes and cataclastic rocks as anindicator of frictional melting. J. Struct. Geol. 21, 473–478.

Martin, H., Porada, H., Wittig, R., 1983. Where lies the root zone of the NaukluftNappe complex?. In: Miller, R.M. (Ed.), Evolution of the Damara Orogen ofSouth West Africa/Namibia, Special Publications, vol. 11. Geological Society ofSouth Africa, pp. 199–207.

Miller, J.A., Viola, G., Mancktelow, N.S., 2008. Oxygen, carbon and strontiumisotope constraints on the mechanisms of nappe emplacement and fluid rockinteraction along the subhorizontal Naukluft Thrust central Namibia. J. Geol.Soc. London 165, 739–753.

Miller, R.M., 1983. The Pan-African Damara Orogen of South West Africa/Namibia.In: Miller, R.M. (Ed.), Evolution of the Damara Orogen of South West Africa/Namibia, Special Publications, vol. 11. Geological Society of South Africa,pp. 431–515.

Monzawa, N., Otsuki, K., 2003. Comminution and fluidization of granular faultmaterials: implications for fault slip behavior. Tectonophysics 367, 127–143.

Munch, H.G., 1978. Das Schmiermittel an der Basis der Naukluft-Decke,Sudwestafrika: the sliding layer at the base of the Naukluft Nappe South-westAfrica. Z. Dtsch. Geol. Ges. 129, 7–31.

Otsuki, K., Monzawa, N., Nagase, T., 2003. Fluidization and melting of fault gougeduring seismic slip: identification in the Nojima fault zone and implicationsfor focal earthquake mechanisms. J. Geophys. Res. 108, 18, http://dx.doi.org/10/1029/2001JB001711, 2003.

Ramasamy, R., Gwalani, L., Subramanian, S., 2001. A note on the occurrence andformation of magnetite in the carbonatites of Sevvattur, North Arcot district,Tamil Nadu, Southern India. J. Asian Earth Sci. 19, 297–304.

Rice, J.R., 2006. Heating and weakening of faults during earthquake slip.J. Geophys. Res. 111, B05311.

C.D. Rowe et al. / Earth and Planetary Science Letters 333–334 (2012) 200–210210

Rowe, C.D., Moore, J.C., Meneghini, F., McKiernan, A.W., 2005. Large-scale pseu-dotachylytes and fluidized cataclasites from an ancient subduction thrustfault. Geology 33, 937–940.

Sibson, R.H., 1975. Generation of pseudotachylyte by ancient seismic faulting.Geophys J. R. Astron. Soc. 43, 775–794.

Smith, S.A.F., Billi, A., Di Toro, G., Niemeijer, A., 2010. Principal slip zones inlimestone: natural and experimental examples and implications for theseismic cycle. Geol. Soc. Am. Abstr. Programs 42. Abstract number 173-4.

Smith, S.A.F., Billi, A., Toro, G.D., Spiess, R., 2011. Principal slip zones in limestone:microstructural characterization and implications for the seismic cycle (TreMonti fault, central Appenines, Italy). Pure Appl. Geophys. 168, 2365–2393, http://dx.doi.org/10.1007/s00024-011-0267-5.

Smith, S.A.F., Collettini, C., Holdsworth, R.E., 2008. Recognizing the seismic cycle inancient faults: CO2-induced fluidization of breccias in the footwall of a sealinglow-angle normal fault. J. Struct. Geol. 30, 1034–1046.

Sulem, J., Famin, V., 2009. Thermal decomposition of carbonates in fault zones:slip-weakening and temperature-limiting effects. J. Geophys. Res. 114. http://dx.doi.org/10.1029/2008JB006004.

Sylvester, F.Y.G., 2009. The Size, Shape and Style of a Continental Detachment:Fault Geometries and Brittle–Ductile Behaviour of the Naukluft Thrust,Namibia. BSc Honours, University of Cape Town.

Ujiie, K., Tsutsumi, A., Kameda, J., 2011. Reproduction of thermal pressurizationand fluidization of clay-ruch fault gouges by high-velocity friction experi-ments and implications for seismic slip in natural faults. In: Geology of the

Earthquake Source: A Volume in Honour of Rick Sibson, vol. 359. GeologicalSociety of London, Special Publications, pp. 267–285.

Ujiie, K., Yamaguchi, A., Kimura, G., Toh, S., 2007. Fluidization of granular materialin a subduction thrust at seismogenic depths. Earth Planet. Sci. Lett. 259,307–318.

Vigan �o, A., Tumiati, S., Recchia, S., Martin, S., Marelli, M., Rigon, R., 2011. Carbonatepseudotachylytes: evidence for seismic faulting along carbonate faults. TerraNova 23, 187–194, http://dx.doi.org/10.1111/j.1365-3121.2001.00997.x.

Viola, G., Andreoli, M., Ben-Avraham, Z., Stengel, I., Reshef, M., 2005. Offshore mudvolcanoes and onland faulting in southwestern Africa: neotectonic implica-tions and constraints on the regional stress field. Earth Planet. Sci. Lett. 231,147–160.

Viola, G., Mancktelow, N.S., Miller, J.A., 2006. Cyclic frictional-viscous slip oscilla-tions along the base of an advancing nappe complex: insights into brittle-ductile nappe emplacement mechanisms from the Naukluft Nappe Complexcentral Namibia. Tectonics 25. http://dx.doi.org/10.1029/2005TC001939.

Warr, L.N., Cox, S., 2001. Clay mineral transformations and weakening mechan-isms along the Alpine Fault, New Zealand. In: Holdsworth, R.E., Strachan, R.A.,Magloughlin, J.F., Knipe, R.J. (Eds.), The Nature and Tectonic Significance ofFault Zone Weakening, vol. 186. Geological Society of London, SpecialPublications, pp. 85–101.

Yuko, T., Ikabata, T., Akiyama, T., Yamamoto, T., Kurumada, N., 2000. New clinkerformation process by the fluidized bed kiln system. Cem. Concr. Res. 30,1113–1120.