Earth and Planetary Science Letters 301 (2011) 179–189

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

j ourna l homepage: www.e lsev ie r.com/ locate /eps l

Nanoscale porosity in SAFOD core samples (San Andreas Fault)

Christoph Janssen a,⁎, Richard Wirth a, Andreas Reinicke a, Erik Rybacki a, Rudolf Naumann b,Hans-Rudolf Wenk b, Georg Dresen a

a GFZ German Research Centre for Geosciences, Telegrafenberg, 14473 Potsdam, Germanyb Department of Earth and Planetary Science, University of California, Berkeley, CA, United States

⁎ Corresponding author. Tel.: +49 331 2881323; fax:E-mail address: jans@gfz-potsdam.de (C. Janssen).

0012-821X/$ – see front matter © 2010 Elsevier B.V. Adoi:10.1016/j.epsl.2010.10.040

a b s t r a c t

a r t i c l e i n f o

Article history:Received 30 April 2010Received in revised form 13 October 2010Accepted 31 October 2010

Editor: L. Stixrude

Keywords:San Andreas FaultSAFODporositypermeabilityfault rock compositionTEM

With transmission electron microscopy (TEM) we observed nanometer-sized pores in four ultracataclasticand fractured core samples recovered from different depths of the main bore hole of the San Andreas FaultObservatory at Depth (SAFOD). Cutting of foils with a focused ion beam technique (FIB) allowed identifyingporosity down to the nm scale. Between 40 and 50% of all pores could be identified as in-situ pores withoutany damage related to sample preparation. The total porosity estimated from TEM micrographs (1–5%) iscomparable to the connected fault rock porosity (2.8–6.7%) estimated by pressure-induced injection ofmercury. Permeability estimates for cataclastic fault rocks are 10−21–10−19 m2 and 10−17 m2 for thefractured fault rock. Porosity and permeability are independent of sample depth. TEM images reveal that theporosity is intimately linked to fault rock composition and associated with deformation. The TEM-estimatedporosity of the samples increases with increasing clay content. The highest porosity was estimated in thevicinity of an active fault trace. The largest pores with an equivalent radiusN200 nm occur around large quartzand feldspar grains or grain-fragments while the smallest pores (equivalent radiusb50 nm) are typicallyobserved in the extremely fine-grained matrix (grain sizeb1 μm). Based on pore morphology we distinguishdifferent pore types varying with fault rock fabric and alteration. The pores were probably filled withformation water and/or hydrothermal fluids at elevated pore fluid pressure, preventing pore collapse. Thepore geometry derived from TEM observations and BET (Brunauer, Emmett and Teller) gas adsorption/desorption hysteresis curves indicates pore blocking effects in the fine-grained matrix. Observations ofisolated pores in TEM micrographs and high pore body to pore throat ratios inferred from mercury injectionsuggest elevated pore fluid pressure in the low permeability cataclasites, reducing shear strength of the fault.

+49 331 2881328.

ll rights reserved.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction

The mechanical behavior of faults depends strongly on theinterplay of fluids and damaged fault rocks (Hickman, 1991; Hubbertand Rubey, 1959). Local variation of porosity and fault zonepermeability may influence fluid flow and effective pressure, affectingfaultmechanics (e.g. Blanpied et al., 1992; Byerlee, 1993; Janssen et al.,2004; Rice, 1992; Schulz and Evans, 1998; Sibson et al., 1975).Laboratory studies and observations of exhumed fault zone rocksindicate that porosity and permeability reduction by compaction orfracture healing may induce high pore fluid pressure, influencingfaulting and fault stability (e. g. Faulkner and Rutter, 2001; Hickmanet al., 2007; Rice, 1992). Although studies of exposed fault rockscontinue to provide important results about the interaction betweenporosity, fluid flow and fluid pressure, the available information islimited because exhumed fault rockswere altered during exhumation,

obscuring fault-related mineral assemblages and textures (Solum andvan der Pluijm, 2004).

Core samples from the San Andreas Fault Observatory at Depth(SAFOD) borehole provide a unique possibility to study the micro-structures of fresh fault rocks of an active plate-bounding fault fromseismogenic depth. A first microstructural study of SAFOD core samplesyielded porosity values of 0–18%, with an average porosity of 3% for lessdeformed shale (Blackburn et al., 2009). Unfortunately, the interpreta-tion of pore origin remains difficult because the applied methods (SEMcombined with image-processing, using thresholding techniques) didnot allow to distinguishbetween porosity formed in-situ and pore spaceformed during core recovery and sample preparation (see also Desboiset al., 2009). Toourknowledge permeability data of SAFODcore samplesis not yet available.

Here, we present an analysis of submicron pores. Since poreswith diameters b1 μm are not visible in optical thin sections we usedtransmission electronmicroscopy (TEM) imaging. In addition, commontechniques of porosity determination, such as mercury porosimetry orthe BET gas adsorption methods, were used to measure the connectedrock porosity, pore volume and pore surface areas of our samples.Porosity data were used to estimate permeability. Different pore types

180 C. Janssen et al. / Earth and Planetary Science Letters 301 (2011) 179–189

are related to sample mineralogy and fabric. Porosity, permeability andpore structure data (i.e. surface area, pore size distribution and porevolume) are used to characterize pore spaces. We discuss the results interms of fault evolution and compare our observations with those oncore material from the Chelungpu Fault drilling Project (TCDP) inTaiwan (e.g. Song et al., 2007) and the Nojima Fault drilling program inJapan (e.g. Shimamoto et al., 2001).

2. Geological setting

The San Andreas Fault (SAF) is a 1.300 km-long transform faultforming the boundary between the northwestward moving westernPacific plate and the eastern North American Plate (Fig. 1). The SAFODdrill site is located in central California at the transition between thecreeping segment of the SAF to the North and the Parkfield segment(Fig. 1a). The geology of the SAFOD drill site (Fig. 1b) is characterizedby the presence of arkosic sedimentary rocks on the southwesternside of the fault and the presence of Great Valley sedimentary rocksnortheast of the fault (Springer et al., 2009). Based on modal content,Bradbury et al. (2007) identified at least four major geological units inthe SAFOD drill holes (Fig. 1c). After passing through near-surfaceQuaternary and Tertiary sediments and the subjacent Salinian granite,arkosic sediments were encountered beneath the Buzzard Canyonfault (Fig. 1b–c). Further east (approximately 1200 m NE of the drillsite), the lithology changes abruptly from arkosic sediments of theSalinian terrane (Pacific plate) to claystones and siltstones of theGreat Valley/Franciscan terrane (North American plate). The litho-logical boundary possibly marks an ancestral trace of the SAF (Zobacket al., 2007a). Multiple faults were crossed during drilling, includingtwo actively creeping strands of the SAF, revealed by casingdeformation at measured depths of 3192 m and 3302 m (Bradburyet al., 2007). These active fault traces are referred to as the southwestdeforming zone (SDZ) and central deforming zone (CDZ), respective-ly. They were the principal targets for coring during Phase 3 in 2007(Zoback et al., 2010).

3. Samples

We analyzedmicrostructures of four samples from SAFOD phase IIIcores (S1, S2, S3 and S4; see also Photographic Atlas of the SAFOD

Fig. 1. Location map of the study area. (a) The San Andreas Fault with the SAFOD drill sitehistorical earthquakes are indicated. The dotted line characterizes the creeping segment. (b)1999). (c) Simplified depth profile of the SAFODMH (Main Hole) with different rock lithologBuzzard Canyon fault, SAF-San Andreas Fault.

Phase 3 Cores 2010, for detailed descriptions of cores). The sampleswere recovered from different core sections located close to or atsome distance to zones of active deformation (Fig. 1c). Themineralogical composition of all samples is documented in Table 1.All depth reported for our samples are measured depth (MD) and besynchronized to the Phase 2 Baker-Atlas Open-hole logs (Zoback et al.,2010; electronic supplement).

Sample S1 (3141 m MD) was taken from a fractured, grayish-redto brownish sandstone (Hole E, Run 1, Section 6), which belongs to asequence of arkosic sedimentary rocks with interbedded shales andsiltstones. The matrix is composed of coarse- to very coarsesubrounded grains with visible feldspar and quartz particles (severalmm in diameter, Fig. 2a). This section of arkosic rocks is crosscut byseveral mesoscale faults (Photographic Atlas of the SAFOD Phase 3cores, 2010; Springer et al., 2009). The sample position is close to afault-contact between silt- and sandstone but at a distance from theactive fault trace (SDZ) of about 50 m.

Samples S2, S3 and S4 belong to the Great Valley sequence (seeBradbury et al., 2007). Sample S2 was collected at 3189 mMD (Hole G,Run 2, Section 4). This position of the core is at 3 m distance to theactive SDZ. The strongly foliated shale cataclasite is composed of abrown, fine-grained calcite-bearing clay matrix (grain sizeb1 μm)containing quartz and feldspar clasts (Fig. 2b; Photographic Atlas ofthe SAFOD Phase 3 cores, 2010). Pervasive shearing is defined by darkseams in the matrix and preferred orientation of grains. Abundantpressure solutions seams and authigenic clay minerals indicateextensive fluid–rock interaction and dissolution–precipitation pro-cesses (Fig. 2b, see also Gratier et al., 2009; Hickman et al., 2008;Schleicher et al., 2009). The sample contains several calcite veingenerations, with the latest one overprinting the fault-related fabric.

Sample S3 is from 3300 m MD (Hole G, Run 4, Section 2), i.e. itroughly coincideswith the CDZ and thus represents an actively creepingportion of the SAF. The sample shows a polished slip surface withslickensides. Sample S3 consists of a dark-brown fractured and fine-grained scaly matrix with a higher percentage of illite-smectite (I-S)compared to sample S2 (Fig. 2c). In addition, chlorite is a majormineralconstituent of the sample (Table 1). Similar to sample S2, seams ofinsoluble material (pressure solution relicts) and authigenic clayminerals suggest considerable activity of dissolution–precipitationprocesses.

. The arrow shows the sense of plate movement (Hickmann et al., 2004). Some majorGeological map of the drilling site (modified after Bradbury et al., 2007 and Dibblee et al.,ies and sample positions (synchronized to the Phase 2 Baker-Atlas Open-hole logs). BCF-

Table 1Mineralogical composition of samples S1–S4, determined by X-ray diffraction.

Minerals Sample 1(wt.%)

Sample 2(wt.%)

Sample 3(wt.%)

Sample 4(wt.%)

Quartz 36 50 30 26Plagioclase 22 13 10 22Illite/Smectite 0 22 38 47Orthoclase 4 0 0 0Microcline 17 0 0 0Calcite 0 5 3 0Analcime 0 1 0 5Chlorite 0 0 12 0Haematite 0 0 3 0Laumontite 21 0 0 0Amorphous material 0 9 4 0

181C. Janssen et al. / Earth and Planetary Science Letters 301 (2011) 179–189

The fourth sample (S4) was taken from fractured massivesiltstones at 3307 m (Hole G, Run 5, Section 4). The distance to theCDZ is about 5 m. This unit contains amixture of rubblymicrobreccias,some with scaly fabric, and numerous sub-angular matrix blocks(Fig. 2d).

4. Methods

4.1. TEM

TEM was performed using a FEI Tecnai G2 F20 X-Twin transmis-sion electron microscope (TEM/AEM) equipped with a Gatan Tridiemenergy filter, a Fishione high-angle annular dark field detector (HAADF)and an energy dispersive X-ray analyzer (EDX). In general, contrast in

Fig. 2.Microstructures of samples S1–S4. (a) Sample S1: Angular quartz and feldspar grains (matrix of quartz and clay (grain sizeb1 μm). The matrix is fractured with calcite filling. (c) Ssolution seams. (d) Sample S4: Quartz and feldspar grains lie in a matrix of highly commin

HAADF images depends on chemical composition (Z-contrast imaging)and sample thickness. Porosity is always imaged as dark contrast. InTEM bright field images porosity is imaged as bright contrast becauseof absent diffraction contrast or reduced mass absorption contrast.

Porosity studies in fault rocks of active fault zones are commonlyaffected by many uncertainties due to the fine pore size and difficultsample preparation (Desbois and Urai, 2009). To exclude preparationinduced damage, the samples for TEM studies were prepared with afocused ion beam (FIB) device (FEI FIB200TEM) at GFZ. 26 TEM foilswith the dimensions of 15×10×0.150 μm were sputtered from thinsections using accelerated Ga-ions (Wirth, 2004, 2009). It must beemphasized, that FIB sample preparation of TEM foils does not create“porosity”, and it does not widen preexisting pores or holes during thesputtering process, which certainly occurs during conventional Argonion milling. The original pore surface is protected from sputtering andthus preserved by redeposition of sputtered material together withgallium. Using this technique, we identified porosity without damagedown to≥2 nm. Smaller pores were not observed. In TEM images, theidentified pores were manually outlined and porosity was estimatedusing image-processing techniques (ImageJ 1.38, 2010). The equiv-alent pore radius rP was calculated from the pore area Ap:

rp =

ffiffiffiffiffiffiAp

π

sð1Þ

It is assumed for this calculation that the 2D porosity equals the 3Dporosity as shown by Blöcher and Zimmermann (2008) for isometricmicrostructures.

fragments) in a sandy matrix. (b) Sample S2: Cataclasite composed of very fine-grainedample S3: Dark-brown fractured and fine-grained scaly clay matrix with dark pressureuted clay. Notice the large (white) calcite patches in the right half.

182 C. Janssen et al. / Earth and Planetary Science Letters 301 (2011) 179–189

4.2. Mercury intrusion porosimetry

The connected porosity was measured using mercury intrusionporosimetry (MIP). The samples were dried in a vacuum oven andthen placed in a container (dilatometer), mounted in the apparatusand evacuated for 30 min prior to themeasurement. Gas pressure wasapplied stepwise up to 400 MPa to inject mercury into the specimenpore space. This allows detection of a minimum pore size with 3.7 nmin diameter. The volume of mercury that intrudes into the pore spacedue to gas pressure increase from Pi to Pi+1 equals the volume of poreswith size range ri to ri+1. Pore radii as function of applied pressure aredetermined by Washburn's equation (Washburn, 1921), assuming atube model for the pore space:

rp HG = −2⋅γHG⋅ cosϑHG

pð2Þ

where γHG is the surface tension of mercury and ϑHG is the wettingangle of mercury. We used values for ϑHG (140 °) and γHG (480 mN/m). For the given pressure range, we calculate a tube diameter rangingfrom 360 μm to ≈4 nm.

The evaluation of pore-body-to-throat ratio (BTR) is based on thehypothesis that a hysteresis in mercury porosimetry curves isattributable to the pore shape. Intrusion of mercury into a cavity iscontrolled by the size of the pore throat radius while the radius of thecavity and its connectivity controls extrusion of mercury from thecavity (Fig. 9a). Determination of pore radii from intrusion andextrusion curves for the same mercury volume allows estimating arelation between pore bodies and pore throats (Webb, 2001).

Permeabilities are estimated from mercury injection curves (Katzand Thompson, 1987). The Katz and Thompson model introduces afractal percolation model to predict the permeability of disorderedporousmedia. They recognized thatmercury injection defines the firstconnected path and, hence, the diameter of the smallest pores on thatpath. This allows estimations of the permeability of rocks saturatedwith a single liquid phase (Cerepi, 2004).

4.3. BET methods

The BET theory explains the physical adsorption of gas moleculeson a solid surface and serves as the basis for analyzing the specificsurface area (Brunauer et al., 1938). After drying samples at 40 °C–100 °C (depending on clay content of the samples), adsorption anddesorption gas measurements were performed on all samples usingnitrogen (N2) with an ASAP 2010 Micromeritics gas adsorptionmachine. Surface areas and volumes of pores were calculated usingthe BET equation and the Kelvin equation, respectively (Sing et al.,1985). The amount of gas adsorbed at a given pressure allows todetermine surface areas. Accuracy and precision of this method areinfluenced by sample pretreatment and the relative pressure range.

4.4. X-ray diffraction

Powder X-ray patterns were collected using a Siemens D5000powder diffractometer with Cu Kα radiation, automatic divergent and

Table 2Statistical data of the porosity/permeability and pore structures of SAFOD core samples. AllBaker-Atlas open-hole logs (Zoback et al., 2010; electronic supplement).

SampleNr.

Sample depth(m)

Porosity Hg(%)

Porosity TEM(%)

Specific surface area BET(m2/g)

Pore vo(cm3/g)

S1 3141 4.7 1.0 0.62 0.005S2 3189 2.9 2.5 10.78 0.011S3 3300 6.8 5.3 8.47 0.016S4 3307 4.5 3.5 1.96 0.009

antiscatter slits and a secondary graphite mono-chromator withscintillation counter. The diffraction datawere recorded from 4° to 75°2Θ with a step width of 0.02 and a counting time of 4 s per step. Thegenerator settings were 40 kV and 30 mA. An internal standard (ZnO)was used to quantify the content of amorphous Material in thesamples. The Rietveld algorithm BGMN was used for quantitativeanalysis (Bergmann et al., 1998). The software BGMN is able to handlethe model for the turbostratic disorder of smectites (Ufer et al., 2004,2008).

5. Results

5.1. TEM observations of pores

Pore space is commonly subdivided into primary and secondaryporosity (Choquette and Pray, 1970). Primary porosity results fromdepositional voids between grains and particles and secondaryporosity forms during burial and diagenesis due to dissolution and/or fracturing. Here, we distinguish (1) four in-situ pore types (I–IV)describing pore spaces likely formed during deformation of thesamples but prior to coring and (2) two pore types (V–VI) withunclear origin. Apparently, one part of latter was formed during corerecovery or sample preparation. In-situ pores were found in all foursamples (see later discussion). The in-situ origin of pores wasestablished using TEM observations and is based on following criteria:(1) open pores are lined with newly formed sheet silicates (type I andfew of type V; see also Schleicher et al. 2010), (2) authigenic clayminerals form a flocculated particle association (card house fabric)with open space between clay particles (type II), (3) high-aspect ratiopores due to leaching (type III), (4) nanoscale pores are filled withnon-structured amorphous (gel) material (type IV).

The total porosity estimated from TEM microphotographs rangesbetween about 1% and 5.3% (Table 2). However, only about half of thepores imaged with the TEM are clearly identified as in-situ pores. TheTEM porosity in samples S2–S4 increases with increasing clay content(Fig. 3). Micrographs shown in Figures 4 and 5 are selections of typicalregions investigated in this study. The distribution of pore typesestimated from TEM images is presented in Figure 6.

5.1.1. Intergranular porosity (Type I)Type I pores represent voids between grains and fragments and

were found in all four samples (Fig. 4a). They typically have anequivalent pore radius N100 nm, and represent the most commontype of in-situ pores estimated from TEM micrographs (17% of allpore types; Fig. 6). The pores are commonly sub-angular and do notdisplay a preferred alignment. They are mainly located around largerquartz and feldspar grains or grain-fragments (Fig. 4b). The pores arelined with newly formed sheet silicates or partly filled withamorphous material (Janssen et al., 2010) indicating porosityformation during or after deformation but clearly prior to coring(Fig. 4b–c). In TEM images, the pores appear to be unconnectedthough this is only 2-dimensional view. In some places, illite plateletsformed in the open pore space indicating that the pores formed inplace (Fig. 5f).

depth reported for our samples are measured depth (MD), synchronized to the phase 2

lume N2 adsorption Permeability(Hg porosimetry) [10−21(m²)]

Mean equivalent pore radius(nm)

55,000 1665 128

30 206280 169

Fig. 3. Relationship between porosity estimated in TEM microphotographs and claycontent (illite, smectite and chlorite) measured by XRD illustrates that porosityincreases with increasing clay content.

183C. Janssen et al. / Earth and Planetary Science Letters 301 (2011) 179–189

5.1.2. Card house porosity (Type II)In samples with a high content of authigenic clayminerals (sample

S2, S3 and S4) clay flakes display edge-to-face and face-to-facecontacts forming card house structures with pore spaces in betweenthe flakes as illustrated for sample 3 in Figure 4d. These features aresimilar to those found in fault gouge samples of the Nojima Fault(Fig. 7 in Kobayashi et al., 2001). The equivalent pore radius variesbetween b50 nm and 100 nm. The pore type represents about 4% ofthe total porosity estimated from TEM micrographs (Fig. 6). Poresbetween clay flakes connected with an edge-to-face contact are largerthan pores between flakes in face-to-face contact (Kawamura, 2001).

5.1.3. Mouldic porosity (Type III)Mouldic pores typically result from fresh water leaching of the

rock matrix (mainly quartz and feldspar grains). Type III pores werefound in samples S2 and S3. Mouldic pores are generally ellipsoidaland elongated in shape (Fig. 4e), with high-aspect ratios and smoothinner pore surfaces. In some places, they are partly filled withamorphous material (Janssen et al., 2010). Similar to type I and IIpores, mouldic pores are not visibly connected. These pores contributeabout 10% to the total porosity (Fig. 6).

5.1.4. Spherical porosity (Type IV)Type IV pores are frequently observed in the fine-grained matrix

(largely comminuted quartz grains) of sample S2. The equivalentradius of spherical pores is typically b50 nm. Single pores arepreferentially aligned forming bands parallel to the foliation of sheetsilicates or they form agglomerates (Fig. 4f). The pores are filled withamorphous material as verified by electron diffraction patterns. Thispore type represents ca. 7% of total pore volume (Fig. 6). Kobayashi etal. (2001) described a similar pore type in a fault gouge from drill coresamples from the Nojima Fault.

5.1.5. Inter-clay layer porosity (Type V)Inter-clay layer porosity is themost common pore type comprising

about 32% of the total porosity (Fig. 6). Type V pores were observed insamples S2, S3 and S4. Inter-clay layer pores occur between flakes ofauthigenic illite-smectite (I-S) minerals (Fig. 5a–b). These voids areextremely thin (b30 nm) and oriented parallel to the I-S layers with alength between 100 and 500 nm. Some of these pores are assumed toresult from dehydration and/ormechanical damage of samples duringcore recovery and sample preparation. However, in a few cases newlyformed sheet silicates grow into the pore space indicating that at leastsome pores may have formed in-situ (Fig. 5a).

5.1.6. Fracture porosity (Type VI)Fracture porosity is the second most frequent pore type (30%;

Fig. 6). Pores associated with fractures were found in all samples, buttheir formation remains in part unclear. Fractures occur along thevein-host rock boundary (Fig. 5c), as grain boundary cracks (Fig. 5d)and as en-echelon tension gashes (Fig. 5e), crossingmultiple grain andfragment boundaries. They often have complex shapes, with jaggededges and branches of crack-like voids extending into the surroundingmatrix. Healed fractures are partly interconnected and were probablyformed after vein formation but prior to coring (Fig. 4c). Tensiongashes are thin (b50 nm) and elongated in shape (Fig. 5e).

5.1.7. Pore fluid chemistryInformation on the composition of pore fluids filling the pores of

the fault rocks is limited. TEM-EDX analyses of amorphous porefillings indicate a high silica content (Janssen et al., 2010). Chemicalanalyses of fluids from the SAFOD wells show a Na–Ca–Cl type waterwith a salinity of 20,000 mg/l representing the typical composition offormation water (Thordsen et al., 2005). This preliminary informationimplies that the original formation water is partly preserved ormodified by chemical exchange with fault rock minerals.

5.2. Porosity and permeability measurements by MIP

Theporosity estimated frommercuryporosimetrydatavarybetween2.9 and 6.8% (Table 2) and are thus within the same range as thoseestimated from TEM images (1%–5.3%). The sample with the highestporosity (S3) was collected in the vicinity of the CDZ. High-porositysamples also show the largest pores (Table 2). Sample permeabilitiesestimated from mercury injection curves for cataclastic samples (S2, S3and S4) are low, ranging from 5×10−21 to 2×10−19 m2 (Table 2).The fractured fault rock (sample S1) displays a higher permeability(5.5.×10−17 m2). This result suggests that the clay-rich fault coremay act as a barrier for fluid migration. Fluid transport may occurpreferentially along to the fault rather than across it (see discussion).

5.3. Pore structure inferred from BET, MIP and TEM investigations

Specific surface area (SSA), pore volume, pore volume distributionand pore sizes, estimated with the BET method, are listed in Table 2and plotted in Fig. 7a. The SSA values range from 0.62 m2/g in sampleS1 to 10.78 m2/g in sample S2, showing that the sample with thelowest average (equivalent) pore radius (S2) displays the highestspecific surface area. The low SSA values in sample S1 and S4 (Table 2)are probably related to lithology and to a larger grain size compared tosamples S2 and S3 (Fig. 2, see also samples description). The porevolume is inhomogeneously distributed among pores with differentsize (Fig. 7a). In sample S2, the largest pore volume is in the smallestpores with a diameter b10 nm. In samples S1 and S4, the largest porevolume is in pores with diameters between 40 and 50 nm. The poresize distribution curve of sample 3 is bimodal with pore diametermaxima at ~2.5 and 30 nm.

Pore sizes estimated from TEM images reveal a log-normal poresize distribution (Fig. 7b). The pore sizes, expressed in equivalentradius, vary between b50 nm and 650 nm. The pore size distributionestimated optically (TEM) reveals a similar trend as the pore sizedistribution estimated from the BET method. The smallest pore sizeshave been measured in the extremely fine-grained matrix of sampleS2, while larger pores occur predominantly in samples S3 and S4.

BET adsorption isotherms show that samples S2, S3 and S4 exhibita significant desorption hysteresis (Fig. 8). Such desorption hysteresisin pore networks is usually explained by pore blocking effects (Greggand Sing, 1982; Ravikovitch and Neimark, 2002). The shapes of thehysteresis loops indicate specific pore geometries (Sing et al., 1985).The shape of the hysteresis loop for sample S2 is often associated withpores that are characterized by narrow necks and wide bodies (ink-

Fig. 4. Microphotographs of in-situ pore types. (a) TEM high-angle annular dark field (HAADF) image illustrates the distribution of interparticle porosity of sample S4. (b) TEMHAADF image of open pore spaces located around quartz and feldspar grains (sample S4). (c) A single open pore lined with newly formed sheet silicates indicating porosityformation during or after deformation (sample S4). Gallium is remaining from sample preparation. (d) Clay flakes in edge-to-face contact form card house structures with small openpore spaces between the clay flakes (sample S3). (e) Large mouldic pore with elongated shape. The open pore place is partly filled with amorphous material (sample S2). (f) TEMbright field image shows round-shaped nano-pores agglomerated or lined up like pearls on a string (sample S2).

184 C. Janssen et al. / Earth and Planetary Science Letters 301 (2011) 179–189

bottle pores). Hysteresis loops for samples S3 and S4 with steeperdesorption branches may indicate that high-aspect ratio pores formeddue to parallel alignment of clay mineral particles (Aringhieri, 2004;see also description of inter-clay layer porosity).

The effect of pore geometry on fluid flow properties of the faultrock material is expressed in pore-body-to-throat ratios (BTR),

determined for samples S1 to S4 (Fig. 9b). A significant differencebetween the BTR distribution of sample S1 and the samples S2 to S4 isobserved. The latter ones show an increase in BTR for a larger porethroat diameter from 5 nm to 140 nm, whereas sample S1 has amaximum at ~5 nm and decrease continuously to 200 nm. We expecta pore blocking effect (Ravikovitch and Neimark, 2002) due to the

Fig. 5. Microphotographs of pores with unclear origin. (a) TEM HADF image of inter-clay layer pores that occur between flakes of authigenic I-S minerals (sample S3). (b) Quartzgrain surrounded by sheet silicates with open pore spaces between flakes (sample S2). (c) Open jagged pore fractures along the boundary vein-host rock (sample S2). (d) Grainboundary fractures surrounding quartz and feldspar grains (sample S4). Open pore spaces partly filled with new formed sheet silicates. (e) Thin tension gashes crossing multiplegrain and fragment boundaries (sample S1). (f) Open pore space divided by illite (?) platelets (sample S3).

185C. Janssen et al. / Earth and Planetary Science Letters 301 (2011) 179–189

narrow pore throats in samples S2 to S4 suggesting that fluidtransport in these samples is most likely controlled by the throatdiameter of the pores. Sample S1 from the fractured rock outside thedeforming zone has larger pore (throat) radii and a significantlysmaller BRT. Hence, a higher permeability and a smaller pore blockingeffect are expected in this zone, in agreement with permeabilityestimates.

6. Discussions and conclusions

In spite of significant differences in the measured mass of TEM(ng) and MIP samples (1.5 g), the porosity estimates from TEMimages and MIP are in close agreement. This suggests that TEMmicrographs yield a representative image of microstructures andporosity. For sample S1, TEM based porosity estimates are likely too

Fig. 6. Distribution of pore types estimated in TEM images reveals a dominance offracture and inter-clay layer porosity.

186 C. Janssen et al. / Earth and Planetary Science Letters 301 (2011) 179–189

small due to the presence of larger pores not adequately representedin the TEM micrographs.

The significant adsorption–desorption hysteresis loops in BETisotherms for samples S2–S4 indicate a considerable amount of ink-bottle-type pores, i.e. relatively wide pore bodies and narrows porethroats, difficult to identify in two-dimensional TEM images. Bottle-shaped pores are common in mudstones (Hildenbrand and Urai,2003). The authors assume that large pore body to pore throat ratios,

Fig. 7. Pore size distribution. (a) BET N2 intrusion into porous sample material is used to estcurves show uni- (S1 and S2) and bimodal (S3 and S4) maxima. (b) Number of pores (N) asvary between b50 nm and 650 nm.

which were also observed in this study, may lead to an overestimateof small pore radii.

Pore morphology and variation of porosity between the foursamples are independent of sample depth but are closely related tomineralogical composition of the cataclasites and the distance to theactive fault traces. For example, the porosity increases with increasingclay mineral content (Fig. 3). Inter-clay layer pores and card housepores are closely related to authigenic clay minerals. In addition, thesample with the highest porosity (S3) is located at the margin of theCDZ, whereas the sample with greatest distance to the active fault(S1) is characterized by the lowest porosity. Further, considering therelation between porosity and the mechanisms of deformation it isobvious that porosity was enhanced by dissolution and precipitationreactions and the formation of alteration products (clay minerals).

Large pores (rpN200 nm) are typically located around quartz andfeldspar grains or fragments while smaller pores (rpb100 nm) occurbetween newly formed clay flakes. The smallest pores (rpb50 nm)often are preferentially aligned with the foliation. These pores mostlyoccur in the fine-grained matrix of foliated shales in sample S2.

imate the relation between pore diameter and pore volume. The pore size distributiona function of pore size (equivalent pore radius estimated from TEM images). The radii

Fig. 8. Types of BET hysteresis loops exhibit a variety of shapes. The adsorption anddesorption branches change from vertical to horizontal. They are not parallel. Theshapes of the different hysteresis loops can be used to identify pore structures.

Fig. 9. Pore throat to pore diameter ratios. (a) The intrusion (solid lines) and extrusion(dashed lines) curves of mercury intrusion porosimetry are used to estimate the porethroat to pore body ratio (BTR). (b) The BTR distribution of sample S1 differssignificantly from samples S2 to S4. The high BTR of these samples from the cataclasitezone indicate pore blocking, which may lead to an overpressurization during faulting.

187C. Janssen et al. / Earth and Planetary Science Letters 301 (2011) 179–189

Specific surface areas differ markedly between samples (Table 2).Large specific surface areas such as in samples S2 and S3 may enablelarger moistening surfaces enhancing fluid–rock interactions.

The state of the fluid pressure in the San Andreas Fault zone is stillcontroversial. Zoback et al. (2007b) found no evidence for signifi-cantly elevated fluid pressure in the fault zone based on in-situmeasurement in the SAFOD drill hole. However, Holdsworth et al.(2009) suggest that high pore fluid pressure existed within the SAFODcore fault rocks, based on microstructural analysis of cuttings. Gratieret al. (2009) recently proposed that pressure solution and precipita-tion promote sealing of the pore space, inducing fluid pressureincrease. Mittempergher et al. (2009) also concluded that isolatedpatches of the fault zone may have attained pore pressures higherthan hydrostatic, triggering small magnitude earthquakes. Our TEMobservations document open pores that formed in-situ during or afterdeformation. In TEM images, many pores appear to be unconnected.

The pores were possibly filled with formation water and/orhydrothermal fluids suggesting that elevated pore fluid pressureexisted, preventing pore collapse. In addition, we found several in-situformed pores with small aspect ratios (b10−2–10−1; Fig. 4e)suggesting that significant fluid pressures must have existed in thecataclasite to keep these pores open. The effective mean stress at thesample depth assuming hydrostatic pore pressure is about 80 MPa(Hickman and Zoback, 2004). Assuming a Youngs modulus of 10 GPaand a Poisson ratio of 0.2 of the samples, we expect that pore fluidpressures exceeding hydrostatic are required for pores with an aspectratio of b10−2 (Walsh, 1965). These fluid-filled pores may form anenvironment for clay deposition, as the observed lined pores indicate,which in turn would allow the grain contacts to weaken.

Also BET and mercury injection data indicate low permeability andpore blocking effects in ink-bottle type pores. Low permeability of thecataclasites is in agreement with previous helium isotope cross-sectionstudies of Wiersberg and Erzinger (2007, 2008), who suggested thatlow-permeable gouge layers block fluid flow across the fault.

Our observations are in general agreement with findings from theNojima Fault (NF) zone drilling project from the Geological Survey ofJapan (GSJ) and the Taiwan Chelungpu Fault Drilling Project (TCDP) inspite of different borehole depths and geological and tectonic settings.The estimated porosities (0.2 to 5.7%) in drill core samples of theNojima Fault (Surma et al., 2003) are in the same range as observed inour samples and pore shapes are very similar (Kobayashi et al., 2001).The permeability structure of the Nojima Fault zone consists of a lowpermeability fault gouge zone (10−20–10−19 m2) within a high-permeability damaged zone (10−18–10−14 m2) of fault breccia andfractured host rock (Lockner et al., 2000; Mizoguchi et al., 2008). Themeasured porosity values for TCDP core samples vary in a wider range(15–18% for sandstones and 3–4% for silty-shale) than those observedhere and for NF samples (Chen et al., 2009). TCDP core permeabilityvalues range from 10−19 to 10−16 m2 for shaly siltstone samples andfrom 10−18 to 10−14 m2 for sandstones (Chen et al., 2009). The lowestvalue of 10−20 m2 was observed on samples recovered close to theshear zone (Lockner et al., 2005).

Are the observed similarities in permeability, porosity and porestructures among the three faults caused by similar processes?Probably not. The cumulative displacement that has occurred alongthe SAF is much greater than along the Nojima and Chelungpu faultsand led to wider fault gouges (Zoback et al., 2010). However, commonto all three fault zones is the presence of clay-gouge layers (Kobayashiet al., 2001; Solum and van der Pluijm, 2004; Song et al., 2007). Theexistence of clayminerals in the fault core has been identified inmanystudies as an important agent possibly controlling fault mechanicsand fluid flow (e.g. Boullier et al., 2009; Schleicher et al., 2009; Warrand Cox, 2001). Our observations suggest that the occurrence ofnanoscale porosity often related to the presence of abundant clayminerals indicates high pore fluid pressures prevailing in a lowpermeability cataclasite that is actively deformed by creep and

188 C. Janssen et al. / Earth and Planetary Science Letters 301 (2011) 179–189

dynamic rupture events. Such a high pore fluid pressure existing inthe cataclastic matrix may very effectively reduce fault strength.

Acknowledgements

We thank Andreas Hendrich for helping with the drafting of figures,Stefan Gehrmann for sample preparation, Rudi Naumann for XRDanalyses andAnja Schreiber for TEM foils preparationusing FIB technique.Thisworkwas fundedbyDFGgrant JA573/4-1. Benvander Pluijmandananonymous reviewer provided very constructive comments and sugges-tions that helped improve this paper. Special thanks are addressed to theSAFOD science team for sampling and support.

References

Aringhieri, R., 2004. Nanoporosity characteristics of some natural clay minerals andsoils. Clays Clay Miner. 52, 700–7004.

Bergmann, J., Friedel, P., Kleeberg, R., 1998. BGMN- a new fundamental parametersbased Rietveld program for laboratory X-ray sources, it's use in quantitativeanalysis and structure investigations. CPD Newsl. 20, 5–8.

Byerlee, J., 1993. Model for episode flow of high-pressure water in fault zones beforeearthquake. Geology 21, 303–306.

Blackburn, E.D., Hadizadeh, J., Babaie, H.A., 2009. A microstructural study of SAFODgouge from actively creeping San Andreas Fault Zone; implications for shearlocation models. EOS Trans. AGU 90 (52) Fall Meet. Suppl., Abstract T21B-1802.

Blanpied, M.L., Lockner, D.A., Berlet, J.D., 1992. An earthquake mechanism based onrapid sealing of faults. Nature 358, 574–576.

Blöcher, G., Zimmermann, G., 2008. Settle3D — a numerical generator for artificialporous media. Comput. Geosci. 34, 1827–1842.

Boullier, A.M., Yeh, E.C., Boutareaud, S., Song, S.R., Tsai, C.H., 2009. Microscale anatomyof the 1999 Chi-chi earthquake fault zone. Geochem. Geophys. Geosyst. (G3) 10.doi:10.1029/2008GC002252.

Bradbury, K.K., Barton, D.C., Solum, J.G., Draper, S.D., Evans, J.P., 2007. Mineralogical andtextural analyses of drill cuttings from the San Andreas Fault Observatory at Depth(SAFOD) boreholes: initial interpretations of fault zone composition andconstraints on geological models. Geospehere 3, 299–318.

Brunauer, S., Emmett, P.H., Teller, E., 1938. Adsorption of gases inmultimolecular layers.J. Am. Chem. Soc. 60, 309–319.

Cerepi, A., 2004. Geological control of electrical behaviour and prediction key oftransport properties in sedimentary porous systems. Colloids Surf. A Physochem.Eng. Aspects 241, 281–298.

Chen, T.-M.N., Zhu, W., Wong, T.-F., Song, S.-R., 2009. Laboratory characterization ofpermeability and its anisotropy of Chelungpu Fault rocks. Pure Appl. Geophys. 166,1011–1036.

Choquette, P.W., Pray, C.C., 1970. Geologic nomenclature and classification of porosityin sedimentary carbonates. AAPG Bull. 54, 207–244.

Desbois, G., Urai, J.L., 2009. Imaging porosity at pore scale in SAFOD samples using thecombination of Ar-beam sectioning and SEM microscopy: preliminary results.Earthscope meeting, 12–15 May, Boise, Idaho, US.

Desbois, G., Urai, J.L., Kukla, P.A., 2009. Morphology of pore space in clay stones —evidence from BIB/FIB ion beam sectioning and cryo-SEM observations, Electronic-earth.net 4, 15–22.

Dibblee, T.W., Graham, S.E., Mahony, T.M., Blissenbach, J.L., Mariant, J.J., Wentworth, C.M.,1999. Regional geologic map of San Andreas and related faults in Carrizo Plain,Temblor, Caliente and La Panza Ranges and vicinity, California: a digital database. U.S.Geological. Survey, Open File Report 99–114.

Faulkner, D.R., Rutter, E.H., 2001. Can the maintenance of overpressured fluids in largestrike-slip fault zones explain their apparent weakness? Geology 29, 503–506.

Gratier, J.L., Richard, J., Mittempergher, S., Renard, F., Doan, M., Di Toro, G., Hadizadeh, J.,Boullier, A.M., 2009. Pressure solution as amechanism of creep and sealing in activefaults: evidence from the SAFOD samples. EOS Trans. AGU 90 (52) Fall Meet. Suppl.,Abstract T21B-1795..

Gregg, S.J., Sing, K.S.W., 1982. Adsorption, Surface Area and Porosity. Academic press, NewYork.

Hickman, S., 1991. Stress in the Lithosphere and the strength of active faults. Rev.Geophys. 759–775 supplement.

Hickman, S., Zoback, M.D., 2004. Stress orientations and magnitudes in the SAFOD pilothole. Geophys. Res. Lett. 31, L15S12. doi:10.1029/2004GL020043.

Hickman, S., Zoback, M., Ellsworth, W., Bonesse, N., Malin, P., Roecker, S., Thurber, C.,2007. Structure and properties of the San Andreas Fault in Central California: recentresults from the SAFOD experiment. Sci. Drilling (Special Issue No. 1), 29–32.

Hickman, S., Zoback, M., Ellsworth, W., Chester, J., Chester, F., Evans, J., Moore, D.,Kirschner, D., Schleicher, A., van der Pluijm, B.A., Solum, J., 2008. Structure andcomposition of the San Andreas Fault in central California: recent results fromSAFOD sample analyses. EOS Trans AGU 89 (52) Fall Meet. Suppl., Abstract T53F-01.

Hickmann, S., Zoback,M., Ellsworth,W., 2004. Introduction to special section: preparing forSan Andreas Fault Observatory at Depth. Geophys. Res. Lett. 31. doi:10.1029/2004GL020688.

Hildenbrand, A., Urai, J.L., 2003. Investigation of the morphology of pore space inmudstones — first results. Mar. Petrol. Geol. 20, 1185–1200.

Holdsworth, R.E., van Diggeln, E., Spiers, C., de Bresser, H.J., Smith, S.A., 2009. Themicrostructural character and evolution of fault rocks from the SAFOD core andpotential weakening mechanisms along the San Andreas Fault. EOS Trans. AGU 90(52) Fall Meet. Suppl., Abstract T52B-01.

Hubbert, M.K., Rubey, W.W., 1959. The role of fluid pressure in the mechanics ofoverthrust faulting I: mechanics of fluid-filled porous solids and its application tooverthrust faulting. Geol. Soc. Am. Bull. 70, 115–116.

ImageJ 1.38, 2010. Image processing and analysis in Java. http//rsp.info.nih.gov/ij.Janssen, C., Romer, R.L., Hoffmann-Rothe, A., Kesten, D., Al-Zubi, H., 2004. The Dead Sea

transform: evidences for a strong fault? J. Geol. 112, 561–573.Janssen, C., Wirth, R., Rybacki, E., Naumann, R., Kemmnitz, H., Wenk, H.-R., Dresen, G.,

2010. Amorphous material in SAFOD core samples (San Andreas Fault): evidencefor crush-origin pseudotachylytes? Geophys. Res. Lett. 37. doi:10.2029/GL040993.

Katz, A.J., Thompson, A.H., 1987. Prediction of rock electrical conductivity frommercuryinjection measurements. J. Geophys. Res. 92, 599–607.

Kawamura, K., 2001. Void ration, burial depth and change ofmicrostructure duringpelagicclay compaction. EOS Trans. AGU 82 (52) Fall Meet. Suppl., Abstract T22B-0935.

Kobayashi, K., Hirano, S., Arai, T., Ikeda, R., Omura, K., Sano, H., Sawaguchi, T., Tanaka, H.,Tomita, T., Tomida, N., Matsuda, T., Yamazaki, A., 2001. Distribution of fault rocks inthe fracture zone of the Nojima Fault at a depth of 1140 m: observations from theHirabayashi NIED drill core. Island Arc 10, 411–421.

Lockner, D.A., Naka, H., Tanaka, H., Ito, H., Ikeda, R., 2000. Permeability and strength ofcore samples from the Nojima fault of the 1995 Kobe earthquake. GSJ InternalReport No. EQ/00/1, Proceedings of the Internal Workshop on Nojima Fault Coreand Borehole Date Analysis, pp. 147–157.

Lockner, D.A., Morrow, C., Song, S., Tembe, S., Wong, T., 2005. Permeability of wholecore samples of the Chelungpu fault, Taiwan TCDP Scientific Drillhole. EOS Trans.AGU 86 (52) Fall Meet. Suppl., Abstract T43D-04.

Mittempergher, S., Toro, G.D., Gratier, J., Hadizadeh, J., Smith, S.A., Desbois, G., Spiess, R.,2009. Evidence of transient increase of fluid pressure in isolated patches of the SanAndreas Fault in SAFOD phase III cores. EOS Trans. AGU 90 (52) Fall Meet. Suppl.,Abstract T52B-04.

Mizoguchi, K., Hirose, T., Shimamato, T., Fukuyama, E., 2008. Internal structure andpermeability of the Nojima fault, southwest Japan. J. Struct. Geol. 30, 513–524.

Photographic Atlas of the SAFOD Phase 3 Cores, 2010. version 4 www.earthscope.org.Ravikovitch, P.I., Neimark, A.V., 2002. Experimental confirmation of different

mechanisms of evaporation from Ink-bottle type pores: equilibrium, pore blocking,and cavitation. Langmuir 18, 9830–9837.

Rice, J.R., 1992. Fault stress states, pore pressure distributions, and the weakness of theSan Andreas Fault. In: Evans, B., Wong, T.F. (Eds.), Fault mechanics and TransportProperties in Rocks. Academic press, New York, pp. 475–503.

Schleicher, A.M., Warr, L.N., van der Pluijm, B.A., 2009. On the origin of mixed-layeredclay minerals from the San Andreas Fault at 2.5–3 km vertical depth (SAFODdrillhole at Parkfield, California). Contr. Min. Petrol. 157, 173–187.

Schleicher, A.M., van der Pluijm, B.A., Warr, L.N., 2010. Nanocoatings of clay and creep ofthe San Andreas fault at Parkfield. Geology 38, 667–670.

Schulz, S.E., Evans, J.P., 1998. Spatial variability in microscopic deformation andcomposition of the Punchbowl fault, southern California: implications formechanisms, fluid-rock interaction, and fault morphology. Tectonophysics 295,223–244.

Shimamoto, T., Takemira, K., Fujimoto, K., Tanaka, H., Wibberley, C.A.J., 2001. NojimaFault Zone probing by core analyses. Island Arc 357–359.

Sibson, R.H., Moore, J.M., Rankin, A.H., 1975. Seismic pumping — a hydrothermal fluidtransport mechanism. J. Geol. Soc. Lond. 131, 653–659.

Sing, K.S.W., Everett, D.H., Haul, R.A.W., Moscou, L., Pierotti, R.A., Rouquerol, J.,Siemieniewska, T., 1985. Reporting physisorption data for gas/solid systems withspecial reference to the determination of surface area and porosity. Pure Appl.Chem. 57, 603–619.

Solum, J.G., van der Pluijm, B.A., 2004. Phyllosilicate mineral assemblages of the SAFODPilot hole and comparisons with an exhumed segment of the San Andreas Faultsystem. Geophys. Res. Lett. 31. doi:10.1029/2004GL019909.

Song, S.R., Kuo, L.W., Yeh, E.C., Wang, C.Y., Hung, J.H., Ma, K.F., 2007. Characteristic of thelithology, fault-related rocks and fault zone structure in TCDP Hole-A. TerrestrialAtmos. Oceanic Sci. 18, 243–269.

Springer, D.S., Evans, J.P., Garver, J.L., Kirschner, D., Janecke, S., 2009. Arkosic rocks fromthe San Andreas Fault observatory at depth (SAFOD) borehole, central California:implications for structure and tectonics of the San Andreas Fault zone. Lithosphere1, 206–226.

Surma, F., Geraud, Y., Pezard, P., 2003. Porosity Network of the Nojima Fault Zone in theHirabayashi Hole (Japan). EGS – AGU – EUG Joint Assembly, Abstracts from themeeting held in Nice, France, 6–11 April 2003, abstract #12946.

Thordsen, J.J., Evans, W.C., Kharaka, Y.K., Kennedy, B.M., van Soest, M., 2005. Chemicaland isotopic composition of water and gases from the SAFOD wells: implications tothe dynamics of the San Andreas Fault at Parkfield, California. EOS Trans. AGU 86(52) Fall Meet. Suppl., Abstract T23E-08.

Ufer, K., Roth, G., Kleberg, R., Stance, H., Dorman, R., Bergmann, J., 2004. Description ofX-ray powder pattern of turbostratically disordered layer structures with a Rietveldcompatible approach. Z. Kristallogr. 219, 519–527.

Ufer, K., Stanjek, H., Roth, G., Dohrmann, R., Kleeberg, R., Kaufhold, S., 2008. Quantitativephase analysis of bentonites by the Rietveldmethod. Clays ClayMiner. 56, 272–282.

Walsh, J.B., 1965. The effect of cracks on the compressibility of rock. J. Geophys. Res. 70,381–388.

Warr, L.N., Cox, S., 2001. Clay mineral transformation and weakening mechanismsalong the Alpine Fault, New Zealand. In: Holdsworth, R.E., Strachan, R.A.,Magloughlin, J.F., Knipe, R.J. (Eds.), The nature and tectonic significance of faultweakening: Geol Soc., London, Spec. Publ., vol.186, pp. 85–1001.

189C. Janssen et al. / Earth and Planetary Science Letters 301 (2011) 179–189

Washburn, E.W., 1921. The dynamic of capillary flow. Phys. Rev. 17, 273–283.Webb, P.A., 2001. An Introduction to the Physical Characterization ofMaterials byMercury

Intrusion Porosimetrywith Emphasis on Reduction and Presentation of ExperimentalData. Micromeritics Instrument Corporation, Norcross, Georgia (USA).

Wiersberg, T., Erzinger, J., 2007. A helium cross-section study through the San AndreasFault at seismogenic depth. G-cubed 8. doi:10.1029/2006GC001388.

Wiersberg, T., Erzinger, J., 2008. On the origin and spatial distribution of gas atseimogenic depth of the San Andreas Fault from drill mud gas analysis. Appl.Geochem. 23, 1675–1690.

Wirth, R., 2004. A novel technology for advanced application of micro- andnanoanalysis in geosciences and applied mineralogy. Eur. J. Min. 16, 863–876.

Wirth, R., 2009. Focused Ion Beam (FIB) combined with SEM and TEM: advancedanalytical tools for studies of chemical composition, microstructure and crystalstructure in geomaterials on a nanometre scale. Chem. Geol. 261 (3–4), 217–229.

Zoback, M.D., Hickman, S.H., Ellsworth, W., 2007a. Proposal for the National ScienceFoundation. Earth Scope, Facility Operation and Maintenance.

Zoback, M.D., Hickman, S.H., Ellsworth, W., Kirschner, D., Penell, N.B., Chery, J., Sobolev,S., 2007b. Preliminary results from SAFOD Phase 3: implications for the state ofstress and shear localization in and near the San Andreas Fault depth in centralCalifornia. EOS Trans. AGU 88 (52) Fall Meet. Suppl., Abstract T13G-03..

Zoback, M.D., Hickman, S.H., Ellsworth, W., 2010. Scientific drilling into the San AndreasFault Zone. EOS Trans. AGU 91, 197–204.