Mineralogic and textural analyses of drill cuttings from the San Andreas Fault Observatory at Depth (SAFOD) boreholes: Initial interpretations of fault zone composition and constraints on geologic models

The San Andreas Fault Observatory at Depth (SAFOD) is part of the Earthscope initiative and tests fundamental questions regarding earthquake and fault mechanics (Hickman et al., 2004; http://www.icdp-online.de/sites/sanandreas/index/). In addition to numerous geophysical applications, the project also provides an opportunity to directly sample rocks related to active faults at depth. One of the primary objectives of the SAFOD project is to determine the structure and composition of the San Andreas Fault zone at depths where earthquakes nucleate. A compilation of data collected from 2002 to 2005 in two closely spaced boreholes (∼15-m map distance separation) at the SAFOD drill site, near Parkfield, California (Fig. 1),¹⁰² provides insight into fault zone properties. These data, along with measurements of in situ stress, permeability and pore pressure conditions, analyses of frictional behavior of fault zone materials, and the determination of physical properties and chemical processes in the fault zone, will help constrain the behavior of seismogenic and creeping faults.

Much of our understanding of fault zone properties in the upper 2–15 km of the Earth's crust is largely derived from studies of exhumed faults. The composition and structure of fault zones, along with the deformation mechanisms, the fluid-rock interactions, and evolution of fault zones in the context of the seismic cycle, have been examined in a variety of settings by numerous workers (Anderson et al., 1980, 1983; Caine and Forster, 1999; Chester and Logan, 1986; Chester et al., 1993; Cooper and Norris, 1994; Holdsworth et al., 2001; Schulz and Evans, 2000; Stewart et al., 2000; Wibberley and Shimamoto, 2003; Faulkner et al., 2003; Kondo et al., 2005). These studies, among many others, have revealed a great deal about the across-strike structure of fault zones, with the recognition that faults often include a damage zone, where rocks exhibit a higher than background intensity of fractures, small faults, veins, and evidence for fluid-rock interactions, but in which little slip has occurred. At the boundaries of, or within damage zones, which range in width from meters to tens of meters up to 1 km for large-displacement faults, one or more fault cores (centimeters up to 1 m thick) consisting of ultracataclasite, foliated cataclasite, clay gouge, or breccia are typically present. Within these core zones, extremely narrow slip surfaces (Chester and Chester, 1998; Wibberley and Shimamoto, 2003) may record much of the slip along the fault.

There are at least two caveats in using exhumed fault zones as a proxy for the analysis of in situ processes in seismically active faults: (1) With the exception of pseudotachylytes, no fault-related rock retrieved from exhumed faults can definitively be shown to be the product of seismic slip, and (2) post-slip alteration, during uplift or while faults are inactive at depth, may alter the textures of fault-related rocks. Observations via drilling into active faults can help to overcome these issues and can tell us much about the deformation processes within faults (Moore et al., 1995; Ohtani et al., 2000; Ikeda, 2001; Matsuda et al., 2001; Hung et al., 2005). Drill hole-based studies provide opportunities to clarify the nature of fault slip at depth and reduce the impact of overprinting, associated with surficial processes, which may obscure the primary textures and geochemical signatures in faults.

The San Andreas Fault Observatory at Depth (SAFOD) project is aimed at examining the processes of fault slip. Unlike previous fault-zone drilling projects, SAFOD provides two unique opportunities: (1) SAFOD targets a section of fault in which earthquakes currently nucleate, rather than in the upper portion of a fault, above the seismogenic region, and (2) SAFOD integrates geological studies and subsurface sampling with geophysical data to help define in detail the processes associated with a slipping fault.

In this contribution, we present the results of quantitative analyses of cuttings obtained during drilling of SAFOD as a significant method for characterizing fault-zone deformation within a seismically active fault zone. Cuttings recovered from the Pilot Hole (PH) drilled in 2002, Phase One (MH1) Main Hole drilled in 2004, and Phase Two (MH2) Main Hole (MH) drilled in 2005 were systematically examined to determine lithology and to document the distribution and style of deformation and alteration within each borehole. Detailed optical microscopic analyses of cuttings samples obtained from the SAFOD boreholes provide valuable information about the sedimentary and igneous mineral assemblages, textures, alteration products, and deformational features present within individual grains. Thin-section analyses of samples from fault zones allow for quantitative measures of mineral abundance, degree of deformation, and alteration products associated with faulted sequences throughout the boreholes. Identification of the abundance of cataclasite in the cuttings allows for the determination of the relative locations (at the meter-scale range) of damage zone and fault core, and may correlate with previously identified shear zone locations inferred from geophysical logs (Boness and Zoback, 2004, 2006).

We also use X-ray diffraction (XRD) analyses to determine the primary mineral assemblages in several samples. Our work, in combination with other whole-rock geochemical and XRD studies (Kirschner et al., 2005; Solum et al., 2006) and borehole geophysical studies (Boness and Zoback, 2006), provides constraints on the design and coring of the active San Andreas Fault zone (SAF) or Phase Three of SAFOD, planned for the summer of 2007. This work also develops a conceptual model for the geologic setting in which the target earthquakes occur and, in general, may offer insight into broader questions associated with the deformation and structure of fault zones.

The objectives of this paper are to: (1) present the results of analysis of thin-section grain mounts produced from the drill cuttings as a primary method to identify mineral assemblages and differentiate lithologic sequences throughout the PH and MH; (2) observe and describe microstructural deformation features within individual grains; (3) categorize and correlate fault zones by the presence of cataclasite and microfractures between the PH and MH; (4) discuss the implications of these results for the coring plan in 2007; and (5) evaluate the implications of our work for the study of fault-zone composition, structure, and processes. This work complements other studies, including (1) the XRD analyses of Solum et al. (2006), which describe detailed information regarding mineralogy; (2) the borehole geophysical interpretations illustrating rock properties with depth (Boness and Zoback, 2006); and (3) the geological investigations of rocks present in the region (Draper, 2007).

The SAFOD drill holes lie 1.8 km southwest of the surface trace of the active strand of the San Andreas Fault, at the northeastern end of the seismogenic Parkfield segment, and adjacent to the creeping segment to the northwest (Fig. 1A). At this point along the SAF, the fault experiences 1–2 cm/yr of creep over a zone ∼10 m wide at the surface (Zoback et al., 2005; Murray and Langbein, 2006). The SAFOD site lies north of the 1966 M_w 6.0 southeast-propagating Parkfield rupture segment, and is also at the northern edge of the fault segment of the 2004 northwest-directed M_w 5.9 rupture Park-field earthquake. To the northwest, the SAF has a creep rate of 2.5 (Titus et al., 2005) to 3.9 cm/year (Argus and Gordon, 2001). Numerous small earthquakes (M_w 0–M_w 2.0) are located in this region at depths as shallow as 2–3 km (Chavarria et al., 2004; Nadeau et al., 2004; Thurber et al., 2005).

Geologically, the site lies in a complex zone of contractional and strike-slip deformation (Fig. 1A). The southern Coast Ranges here are composed of a granitic block west of the SAF, interpreted to be Salinian granitic rock, and the Franciscan block to the northeast (Dickinson, 1966; Page et al., 1998). The Franciscan block is comprised of southeast-plunging anticlinoria cored by serpentine bodies and metasedimentary rocks of the Franciscan Formation and unconformably overlain by unmetamorphosed sedimentary rocks of the Great Valley sequence (Ross, 1978). Geologic mapping of the area indicates that folded and faulted Tertiary through Jurassic rocks are present in surface exposures east of the SAF (Dibblee 1971; Sims, 1990; Rymer et al., 2003). To the southeast of the fault, seismic reflection and refraction studies reveal a step-like feature in the P-wave velocities across the site, with a shallow, high-velocity region likely underlain by Salinian rocks to the southwest, and a low-velocity region to the east of the SAFOD boreholes (Fig. 1B; Thurber et al., 2004; Zhang and Thurber, 2005; Hole et al., 2006). Hole et al. (2001, 2006) use seismic data to show a moderately northeast-dipping transition from high- to low-velocity rocks at the SAFOD site (Fig. 1B).

The vertical PH was drilled to a depth of 2.2 km in 2002, and the MH was drilled in two phases in 2004 and 2005. The MH is vertical to a depth of 1500 m, where the hole begins its deviation in a N 35° E bearing, with an ultimate angle of inclination of ∼55° at 2070 m measured depth (MD) or 1970 m true vertical depth (TVD) (Fig. 1B). This angle was maintained to the bottom of the hole measured at 3067 m TVD or 3987 m MD (Fig. 1B). The PH is 22.25 cm in diameter, and the MH is 31.15 cm diameter to a depth of 3050 m and 20.95 cm from 3050 m to the total depth (TD) of 3987 m. Phase One and Phase Two drilling included direct sampling of 24 m of short cores acquired at 1476, 3056, and at 4028 m MD, and 52 percussive sidewall cores 2.4 cm in diameter by 1–3 cm in length. Thus, the available cuttings represent a continuous and complete sampling over the entire interval of rocks encountered in the borehole.

The boreholes were drilled with a mud-based rotary drilling system using carbide and diamond tipped, tricone drill bits. Drill cuttings are the coarse to fine, sand-sized rock particles created from the cutting action of the drill bit pads and are circulated to the surface via the drilling mud system. Cuttings mixed with bentonite-based drilling mud continuously stream across the shaker table of the drilling rig. Approximately 0.5 kg of cuttings were collected for every 3 m MD, 1 kg collected every 30.48 m MD, and ∼3 kg collected every 91.4 m MD along the SAFOD drill holes. More closely spaced cuttings were collected continuously in areas where real-time drilling information, such as changes in drilling rate or the presence of gas, indicated the presence of zones of geological interest. Near real-time analyses of the on-site washed cuttings performed by commercial mud loggers give a basic lithologic description of the rocks encountered in the boreholes. These analyses focus on macroscopic surface features of the grains, such as color and estimated grain size, and do not include information regarding the intensity or cause of deformation. Also, the mud logging does not allow for quantitative estimates of composition or degree and types of alteration. Other data of interest that give an added context to our work include rate of penetration (ROP) data, which measure the rate of the drill bit advance and provide insight into rock strength at the bit, and near real-time gas analyses.

The drill cuttings examined in this study were washed in the laboratory in a 140-mesh sieve (∼0.1 mm diameter) to remove drilling mud, followed by a magnetic separation of the cuttings on a magnetic plate to remove drill bit fragments from the cuttings. Samples were decanted with distilled water to separate mud additives, which primarily consist of crushed walnut shells. A mechanical riffle-style sample splitter was used to obtain representative samples of the washed cuttings from each depth interval. Each of these samples was then sieved to the 2-mm fraction for further analyses. Thin-section grain mounts were made at ∼30.48-m (100-ft) intervals within the PH to a depth of 2164 m, and in the MH from 670 to 3048 m MD. From 3048 m to 3985 m MD, sample spacing ranges from 0.3 to 33 m to capture the variations in composition and texture associated with drilling breaks or lithologic changes interpreted from the mud logs or the wireline logs. Closely spaced samples (<3 m) were collected concurrently with drilling on-site by our research group and are not part of the archived SAFOD collection. Nomenclature for drill-hole measurements typically expresses the location along the borehole in MD along the wellbore path. Drilling coordinates in the United States are registered in feet, and we convert all data to the metric system. True vertical depths (TVD) correct the MD values using the borehole deviation survey of the hole.

A total of two hundred and sixty-six thin-section grain mounts from samples in both boreholes were analyzed for mineral assemblages and fault-related textural analysis using a modified Gazzi-Dickinson method (Dickinson, 1970) with individual counts taken incrementally every 0.5 mm on an equally spaced, 300-point grid pattern (Dickinson, 1970). At each individual point, the composition and textural feature were recorded, with the primary subcategories designated for minerals, alteration products, and cataclasite. The individual point counts were recorded as modal percents of the total and were cataloged by measured depth. The recorded abundance of individual minerals, cataclasite, and altered fabrics was plotted as a function of depth, correlating lithology and shear zones to the observed petrology (Fig. 2),²⁰². The categories used to examine the samples were: cataclasite, altered cataclasite, sedimentary lithic fragments, volcanic lithic fragments, mica, calcite, plastically deformed quartz, monocrystalline quartz, plagioclase, sanidine, microcline, altered feldspar, deformed feldspar, opaque grains (interpreted to be oxide minerals), symplectite, chlorite, olivine, and amphiboles. Data files containing the raw data and photos of the thin sections are available in the electronic appendix.

X-ray diffraction analysis was performed on thirty samples with an X Pert Pro Diffractometer system running at 45KV/40 Ma with copper tubing. X pert Data Collector and X Pert High Score software were used for data analyses to determine mineral compositions present in the deeper portions of the MH section from 3078 to 3864 m MD.

To simplify the display of results, we summarize and divide the cuttings data into seven main categories: quartz, feldspar, lithic fragments, oxide minerals, cataclasite fragments, and total percentages of altered and/or deformed crystal fragments. It is important to note that due to the nature of drilling and mud circulation processes, there are inherent limitations to the geologic interpretation of drill cuttings (Winter et al., 2002). The potential limitations include: (1) mixing of cutting sample may occur as the drilling fluid is circulated along the side of the drill string; (2) samples taken at spaced intervals may not effectively represent sharp transitions observed in some of geophysical borehole or image-log analyses (Boness and Zoback, 2004; Draper et al., 2005) because any sample represents cuttings over some finite interval of rock at the drill-bit tip region; (3) thin-section mounts of drill cuttings represent an extremely small portion of the total sample collected at the SAFOD site; and (4) analysis of grains with optical microscopy does not allow for the characterization of fine-grained rocks in both the protolith and the fault zones (see Schleicher et al., 2006).

Valuable geologic information relating to the subsurface lithology, and distinct mineral assemblages through identified sequences, are obtained from the analysis of thin-sectioned drill cuttings. When merged with other data sets, including detailed X-ray diffraction analysis (Solum et al., 2006), the results from these data enable us to determine the lithology and structure encountered in the borehole (Pechnig et al., 1997; Winter et al., 2002), and to constrain the location of potential areas of interest during the continuous coring program proposed for 2007.

We summarize the lithologies of the rocks cut by the SAFOD Pilot Hole and Main Hole (Figs. 2,²⁰² and 3) as seen in the cuttings samples, followed by a discussion of the alteration and deformation features as a function of depth throughout each hole. We also present microscopic observations of the major rock types, fault and fracture characteristics, and distinct textures, with a brief review of X-ray diffraction data on several samples.

On the basis of the modal content of the cuttings examined, we identify four major lithologies in the SAFOD drill holes: (1) Quaternary and undifferentiated Tertiary sediments; (2) granitic rocks; (3) arkosic sedimentary rocks; and (4) fine- to very fine-grained sedimentary rocks (Figs. 2,²⁰² and 3). The Quaternary and Tertiary sediments occur over the interval of 0–760 m. The granitic rocks are subdivided into a granite with a quartz content of 35% to 55% from 760 to ∼1450 m MD and a granodiorite (quartz ∼20% of total, feldspar, both altered and unaltered, 30%–50% of the total) from ∼1450–1920 m MD. We distinguished the granitic rocks above 1450 m from the granodiorite below on the basis of the percentage of quartz grains and the abundance of ferromagnesian minerals, specifically biotite and hornblende. The arkosic rocks include an upper sequence from 1920 to 2550 m MD, separated by a clay-rich zone from 2530 to 2680 m MD, and a finer grained lower sequence from 2680 to ∼3150 m MD. The deeper section of the borehole is characterized by a fine-grained, quartz-feldspar-rich siltstone from 3150 to 3550 m MD, and a very fine-grained siltstone to shale from 3550 to 3987 m MD.

The uppermost sedimentary sequence (0–760 m MD) is likely the Pliocene Paso Robles and late Miocene Santa Margarita formations, exposed at the surface and encountered in the subsurface northwest, west, and southwest, of the SAFOD drill site (Dibblee, 1973; Durham, 1974; Graham et al., 1989; Dibblee et al., 1999; Thayer and Arrowsmith, 2005). The sedimentary fragments from these Quaternary/Tertiary deposits are characterized by fine-grained angular to subangular grains composed mainly of quartz and plagioclase in a very fine-grained matrix. Volcanic lithic fragments are abundant throughout this sequence and commonly have a highly altered, fine-grained to glassy ground-mass (Figs. 2A, 2B). Calcite-rich cements are observed throughout both the PH and MH over this interval.

The granitic rocks encountered below these sedimentary rocks (760–1920 m MD) are likely part of the Mesozoic Salinian Block that lies west of the San Andreas Fault in the region (Dibblee, 1973; Ross, 1978; James and Mattison, 1988). We define an upper granite (quartz 40%–60%, feldspars 20%–40%) from 760 to 1450 m MD (Figs. 4A–4C), and a granodiorite between 1450 and 1920 m MD (quartz 20%–40%, feldspars modal values of 40%, 2%–5% Fe-Mg minerals (mostly hornblende), and 4%–6% biotite. Between 80%–95% of the quartz encountered in the cuttings consists of monocrystalline quartz, with a minor fraction of either polycrystalline or plastically deformed quartz (Fig. 4B) indicative of metamorphic rocks associated with the Salinian block (Ross, 1978). Good evidence for a fault does occur within the upper granite unit around 1050 m (Figs. 2,²⁰² and 3).

The lithologic break at ∼1450 m is not associated with a significant increase in alteration or cataclasite abundances (Figs. 2,²⁰² and 3); however, the texturally and compositionally abrupt change in lithology, mineralogy, and borehole and geophysical character encountered in the MH at 1920 m MD reflects a change from Salinian granitic rocks to a sedimentary sequence that consists of two types of arkosic to lithic arenites (Solum et al., 2005b, 2006; Draper et al., 2005; Draper, 2007). This sedimentary sequence is characterized by 20%–60% lithic fragments in the samples. The lithic fragments within this sequence are fine-grained, subangular grains composed primarily of quartz and mafic minerals in a very fine-grained matrix (Figs. 4D–4F). Individual grains of highly altered volcanic groundmass were also observed, but overall represented a small portion of the total grains point counted.

A broad, clay-rich zone from 2530 to 2680 m MD (Draper, 2007) divides this sequence into two packages. Differences in lithic abundance and composition, the nature of chloritic grains, and alteration products between the two packages were determined with microscopy of cuttings and core in conjunction with image-log analysis and integration of borehole-based geophysical logs (Draper, 2007). Based on XRD analyses, the lower package is enriched in chlorite and illite relative to the upper package. The lower package also contains laumontite, which is generally absent in the upper package (Solum et al., 2006). The depositional setting, age, and tectonic implications of this block of arenites are discussed in Draper (2007), who suggests that this package of rocks represents a proximal portion of a submarine fan or turbidite sequence, perhaps part of a Late Cretaceous to Early Tertiary Salinian cover sequence found west of the San Andreas Fault (Clarke and Nilsen, 1973; Graham et al., 1989; Seiders and Cox, 1992; Grove, 1993; Draper et al., 2005).

The deepest lithologic change occurs at ∼3150 m MD, where the arenites are abruptly replaced by fine-grained siltstone and shale fragments (Figs. 2B and 4G). Due to the fine-grained textures of these fragments and the binning required by the Gazzi-Dickinson method, many of these grains are classified as lithic clasts (Figs. 2B and 4H), but, in reality, they are fragments of quartz-rich siltstone to mudstone. At a depth of ∼3400 m MD, we lose most of the distinct quartz and feldspar grains in the thin sections, suggesting the rocks from ∼3400 to ∼3850 m MD are mudstones, some of which contain fossils (Fig. 4I). This depth also marks a pronounced change in clay mineralogy because below this depth chlorite concentration and crystallinity are fairly homogeneous (Solum et al., 2006). At the bottom of the hole, the percentage of quartz increases and lithic fragments decrease, suggesting the presence of a siltstone sequence. From 3850 m MD to the end of the drill hole, the gamma-ray log and optical microscopy suggest a mixed lithology of silt-stone and claystone.

We use the term alteration in this work to denote the presence of minerals such as sericite (fine-grained illite or muscovite), calcite, zeolite minerals, chlorite, and clay minerals that cannot be optically resolved and are often overprinted on pre-existing minerals (Fig. 5) or occur as fragments of what appear to be veins. Careful attention was paid to the identification of alteration phases because the heterogeneous quality of the samples collected throughout the borehole may create a bias in recognition. For example, with these samples, it is relatively easy to identify alteration in the granite and granodiorite sequence; however, it proved more difficult within the deeper, fine-grained sections, where the alteration phases may also be of detrital origin or not readily visible. We identify several zones in which alteration phases (determined on the basis of composition and texture) comprise >20% of the modal amount of grains (Fig. 2),²⁰²: (1) an ∼150-m wide zone (as sampled in the vertical drill hole) in the MH at ∼1050 m; (2) a broad region from 1600 to 2000 m MD; (3) a minor zone with greater variability between 2200 and 2500 m MD in the upper sedimentary section; (4) a section of increasing alteration from 2700 to 3000 m MD; (5) numerous zones between 3050 and 3350 m MD; and (6) a thin zone ∼3600 m MD.

Alteration products and textures include ser-icitization of feldspars and recrystallization of quartz (Figs. 5A and B), calcite replacement mineralization (Fig. 5B), calcite veins (Fig. 5C), zeolite overprints on host grains (Fig. 5D), and the formation of fine-grained clays and/or talc (Fig. 5E). Remnants of calcite veins within individual grains may indicate periods of fluid movement in or near fault zones since they are commonly attached to or found layered with cataclasite. Alteration of the Salinian rocks correlates well with the abundance of cataclasite fragments (Fig. 2B) indicating that alteration in that section is likely associated with deformation rather than broad alteration of the block.

Significant abundances of deformed and/or cataclasite grains are observed in several locations within the MH 1050, 1650–1750, 1900–2000, ∼2650–2700, 3050–3300, and ∼3650 m MD (Figs. 2,²⁰² and 3). Four main styles of cataclasite deformation are observed in this study: (1) individual grains or zones with intense fracturing and what appears to be the initial stages of grain comminution within intragranular fractures (Fig. 6A); (2) relatively unaltered cataclasite (Fig. 6B); (3) altered cataclasite (Fig. 6C); and (4) layered and highly deformed cataclasite (Fig. 6D). The undeformed/unaltered cataclas-ite is characterized by very fine-grained rounded to sub-rounded grains in a dark-gray or dark-brown, ultrafine groundmass. Cataclasite may also be found sutured to undeformed grains and/or may consist of various layers comprised of comminuted and rotated grains, fine-grained clays or unidentified matrix, iron-oxide/hydroxides, and quartz or calcite veins (Figs. 6E–6F). The altered cataclasite is characterized by the presence of very fine-grained rounded feldspar grains to a zeolite phase with indistinct extinction, fibrous or fuzzy habits, and gray-white pleochroism (Fig. 6C). The deformed cataclasite is typically foliated and very fine-grained and may occur as a multilayer sequence including mineralized microfracture surfaces.

Feldspars exhibit significant deformation in the regions of high cataclasite content (Figs. 6B and 6E), with abundant alteration and intra-granular, cleavage-controlled fractures. The amount of feldspar alteration is interpreted to be a possible product of increased fluid migration and/or compartmentalization adjacent to shear-one locations.

The occurrence of cataclasite in the PH suggests that it encountered a fault zone in the granitic sequence at 1500 m. We infer that this fault zone was intersected in both drill holes at roughly the same depth, which would imply that the fault dips shallowly northeast (Fig. 3). This relationship may coincide with the lithologic transition between the granite and granodiorite, the reduced V_p and V_s values, and changes in resistivity seen in the borehole geophysical logs at 1150–1400 m (Boness and Zoback, 2004). Observations of cataclasite, deformation features, and alteration show a decreasing trend near the bottom of the PH, as compared to an increasing trend in these features at similar depths (∼1900–2000 m MD) within the MH. The arkosic-rich sedimentary section that intersects the MH at 1920 m MD is also not documented near the base of the PH. These observations support the presence of a steeply southwest-dipping fault within the MH, which was not yet penetrated by the PH due to the completion depth (Figs. 2,²⁰² and 3). The broad distribution of the alteration and cataclasite intensities in the MH data suggests that this fault is a relatively large fault, and it may represent the down-dip continuation of the Buzzard Canyon fault (Rymer et al., 2003; Hole et al., 2006; Thayer and Arrowsmith, 2006). Deeper in the section, the presence of another broad region of intense deformation and cataclasite combined with alteration between 3000 and 3300 m MD may represent a major strand of the San Andreas Fault and surrounding damage zone, with the abundance of cataclasite noticeably decreasing below 3600 m MD in the MH.

The intensity of alteration correlates reasonably well with cataclasite abundances measured throughout the hole (Figs. 2,²⁰² and 7). A region of such correlation deep in the hole, and which may have a bearing on the SAF sensu stricto, is the zone of increased alteration and cataclas-ite abundance observed in the 3050- to 3350-m MD interval (Fig. 2),²⁰². Alteration phases identified through microscopic analysis of the cuttings include quartz, calcite, chlorite, sericite, iron oxides, serpentine, zeolites, and clays. X-ray diffraction analyses of Solum et al. (2006) document bulk compositions, relative mineral abundances, and details of clay mineralogy of this interval and throughout the entire MH. This region also correlates with increases in ROP observed at 3185–3215 and 3290 3353 m MD, both of which are also associated with changes in mudgas composition (gas interpreted to be exsolved from the formation; Wiersberg and Erzinger, 2005).

As a supplement to the microscopic identification of minerals, we used X-ray diffraction analysis to further examine the potential constituents of the cutting samples from the region of 3078–3864 m MD ⁰¹. Minerals denoted in this table have at least 5%–10% relative abundance in each sample. Note that sample material from 3520 m MD and deeper was extremely limited and resulted in weak signatures; therefore, only the identification of primary peak phases was possible. We identify several rock types based on mineralogy. Almost all samples have quartz and feldspar. Samples to a depth of 3325 m MD are relatively clay poor, contain muscovite, and have Mg-Fe-Al oxides and a scant amount of zeolites. From 3335 to 3356 m MD, the rocks contain illite and kaolinite, and, in several cases, halite, along with Mg and Fe oxides. The deepest section of the hole recovered olivine, zeolites, Ti-oxides, cris-tobolite or trydimite, ± minor sulfides. The presence of oxide phases and clay minerals indicates three zones of increased alteration: 3078–3290; 3330–3345, and 3520–3595 m MD.

Our analysis of drill cuttings from the SAFOD boreholes provides a preliminary view of the details of the lithology, alteration, and nature of deformation in the rocks encountered in the SAFOD boreholes. This work reveals aspects of the structure and composition of the volume of rock surrounding the SAF, and can be used as a guide for further studies of the fault zone continuing at the SAFOD site.

A detailed discussion and synthesis of the geologic and tectonic interpretation of the rocks encountered in the SAFOD boreholes are beyond the scope of this paper because such an analysis requires the synthesis of the data presented here and many other data sets, including, but not limited to: Sims, (1990); Thurber et al., (2003; 2004); Rymer et al., (2003); Unsworth and Bedrosian (2004); McPhee et al., (2004); Hole et al. (2001, 2006); Thayer and Arrowsmith, (2005); Boness and Zoback (2006); Solum et al. (2006); and Tembe, et al. (2006). We focus here on the rocks encountered by the borehole and their implications for coring in 2007.

We define four major lithologies in the borehole and suggest that at least five major faults were encountered (Figs. 3 and 8). Lithologic analysis of the cuttings supports the interpretation of borehole geophysical data acquired in the two holes (Boness and Zoback, 2004; Boness and Zoback, 2006), which suggested the presence of seven distinct lithologic changes in the area. Most of the faults appear to be steeply southwest-dipping faults, as suggested from seismic imaging of the region (Hole et al., 2001, 2006), and the location of microseismicity (Thurber et al., 2004). When correlated with borehole geophysical data (Zoback et al., 2005; Boness and Zoback, 2006), the active strand of the SAF, inferred from the location of wellbore casing deformation, location of small earthquakes, and the presence of a low-velocity zone (Zoback et al., 2005), appears to be associated with a region of significant alteration (Figs. 2,²⁰², 3, and 8) at 3300–3500 m MD. Our analysis of the cuttings documents the presence of the arkose and lithic arenites between the vertical portion of the drill hole and the San Andreas Fault. We also suggest that while a fine-grained lithology is encountered below the arenites, this contact does not correspond to the active strand of the SAF. The lithologies encountered deep within the borehole consist of well-indurated siltstones and mudstones that are part of the uppermost Great Valley sequence based on analysis of microfossils (K. McDougall, written commun., 2006), and the presence of volcanic quartz and olivine detrital grains as determined from the X-ray diffraction analyses. Thus, any subsurface geologic model of the site needs to incorporate the presence of a high-velocity arkosic sedimentary section southwest of the fault and the fine-grained lithologies northeast of the fault that are not consistent with the Jurassic Franciscan Formation (Hole et al., 2001; 2006). An added structural complexity introduced by the presence of the Great Valley sequence at the bottom of the SAFOD MH is that Franciscan rocks are exposed at the surface ∼3 km northeast of the SAF (Fig. 1A). Thus, the internal geometry of the SAF and its related structures appear to be more complex than originally interpreted from geophysical data and surface mapping (Dibblee, 1971; Sims, 1990; Page et al., 1998; Hole et al., 2001; McPhee et al., 2004; Unsworth and Bedrosian, 2004).

The steep change in the resistivity structure of Unsworth et al., (2003) and Unsworth and Bedrosian (2004) appears to correspond to the presence of a well-developed fault at the Salinian-arkosic sequence transition in the borehole and may correspond to the Buzzard Canyon fault zone mapped at the surface (Figs. 1A and 9; Rymer et al., 2003; Thayer and Arrowsmith, 2005). Increased alteration and deformation between 3300 and 3500 m MD may correspond to the SAF, which would indicate the fault dips 80–85° southwest and consists of several strands at depth. The steep westerly dip agrees with the fine-scale analysis of the earthquakes (Thurber et al., 2004; Zhang and Thurber, 2005; Ellsworth et al., 2005).

The subsurface sections in Figure 9 incorporate surface geologic mapping (Thayer and Arrowsmith, 2006; Sims, 1990) and sub-surface geophysical data (Hole et al., 2001, 2006; Catchings and Rymer, 2002; Chavarria et al., 2004; McPhee et al., 2004; Thurber et al., 2004). At least two possible interpretations for the subsurface structure are considered after review of the previously mentioned studies and recent work (Evans et al. 2005; Boness and Zoback, 2006; Solum et al., 2006; Tembe et al., 2006; Draper, 2007). The sections shown in this study are permissible, but they are by no means the only interpretations, especially within the deeper portions of the sections where the data are not well constrained.

Common to both interpretations are the following elements: (1) the surface trace of the Buzzard Canyon fault projects down dip and connects to the fault that juxtaposes granodiorite on the southwest side and Tertiary arkosic rocks on the northeast side; (2) the reverse faults in the Tertiary rocks (Thayer and Arrowsmith, 2006) represent small displacement faults; and (3) the main trace of the San Andreas Fault projects down dip at ∼83° and is intersected by the borehole as shown.

Interpretations of the sections differ at depth based on how we interpret the presence of the upper Great Valley sequence at the bottom of the SAFOD hole. In Figure 9A, we show the Great Valley sequence and younger rocks as a fault-bounded wedge with the northeastern-most fault being a subsidiary to the San Andreas Fault, as mapped at the surface at the latitude of the SAFOD project (Hole et al., 2001, 2006; M. Rymer, 2005, personal commun.). Adding to the complexity of this interpretation is the presence of the Gold Hill fault (Sims, 1990) directly to the south of the area. Sims (1990) interprets the Gold Hill fault as a steep, northeast-directed reverse fault that is cut off by the San Andreas Fault. In Figures 9A and 9B, we show the northeast strand to be the northern continuation of the Gold Hill fault, as interpreted by Hole et al. (2001, 2006) and the down-dip projection of the Buzzard Canyon fault, which may merge with the San Andreas Fault at depths greater than 5 km based on fault-zone trapped, wave studies (Shalev and Malin, 2005).

In an alternate interpretation (Fig. 9C), we show the Gold Hill fault as Sims (1990) had interpreted it—cut off by the San Andreas Fault. To explain the presence of the Great Valley sequence encountered in the bottom of the SAFOD borehole, a fault is required between the Great Valley and Franciscan rocks. The steep dip of the contact and the omission of much of the thickness of the Great Valley could be due to diapiric structure within the Franciscan, in which upward flow of the Jurassic rocks placed it against, and just to the northeast, of the Great Valley sequence (see Dickinson, 1966, for nearby field examples).

Small amounts of whole-rock core samples recovered from 1476 to 1484, 3056 to 3067, and 3150 to 3410 m MD provide information on depositional features, deformation history, and rock properties in the MH (Almeida et al., 2005; Schleicher et al., 2006; Tembe et al., 2006; Draper, 2007). Almeida et al. (2005) show that the upper cored interval from Phase One (1476–1484 m) consists of a medium-grained hornblende biotite granodiorite with leucocratic phenocrysts and weakly foliated lenses. Evidence for both low- and high-temperature deformation is present in the core, including a series of subvertical fractures and moderately dipping shears with secondary mineralization comprising centimeter-thick halos of low-grade alteration and staining of host. Abundant fracture sets with irregular cataclastic bands, up to 2 cm thick, are orientated at high angles to the core, recording multiple stages of deformation and fluid infiltration (Almeida et al., 2005).

The core from 3056 to 3067 m MD likewise agrees with the cuttings analysis consisting of a coarse, arkosic sandstone to pebble conglomerate with lithic fragments of granitic, sandstone, silt-stone, and volcanic clasts. A clay-rich shear zone several centimeters thick was cored at the bottom of this interval, and Solum et al. (2006) suggest this shear zone could be the southwestern active strand of the SAF (Zoback et al., 2005).

The composition of the whole-rock core from the bottom of the Phase Two drilling (depth of 4028–4036 m MD) agrees with the cuttings data obtained from this depth. The core is composed of shale with several thin beds of siltstone and very fine-grained sandstone, graded bedding, fossil fragments, and bioturbated sections. Numerous small veins, scaly fabric, and polished slip surfaces are observed in the core.

The comparison of the results of our work with the borehole geophysical data (Fig. 8) can be used to define a relationship between the physical rock properties and geophysical signatures. Zoback et al. (2005) and Boness and Zoback (2006) indicate that the zone from 3150 to 3410 m MD is characterized as a low-velocity zone, with low gamma-ray and resistivity character, with a small zone at 3295–3313 m MD where the borehole casing is actively deforming due to creep on an active strand of the SAF. Our data suggest that a fault may occur at ∼3050 m MD (corresponding to the cored fault at 3066 m), followed by a zone of significant alteration to ∼3320 m MD, where we see an increase in the amount of cataclasite in the cuttings (Fig. 8) and a decrease in alteration mineral abundances. Other faults may exist between 3500 and 3660 m, where an increase in alteration and cataclasite abundance is associated with changes in standard and neutron porosity and V_p and V_s in the geophysical logs (Fig. 8).

Our interpretation of the cuttings can also be examined in light of previous work on fault-zone composition, deformation, and structure from exhumed fault zones (Chester and Logan, 1986; Chester et al., 1993; Chester and Chester, 1998; Faulkner et al., 2003; Wibberley and Shimamoto, 2003) and applied to the understanding of the SAF at depth. The data support the presence of several fault zones characterized by brittle to semi-brittle deformation textures within grains and increased amounts (relative to adjacent samples) of microfractures, fine-grained clays, alteration phases, and zeolite minerals. These zones of deformation appear to be meters to tens of meters wide, and may actually represent damage zones surrounding narrow fault zones consisting of compacted cataclasite, ultracataclasite, and/or fine-grained gouge that may not be resolved based on cuttings analyses alone. Heterogeneous damage-zone elements are observed at the micrometer scale throughout the borehole and may provide insight into the various styles of deformation and related mechanisms at the meter to tens of meters scale. The character of cataclasite varies within cutting samples as a function of lithology and structural setting. For example, fracture surfaces may vary from a single, discrete, slip surface marked by a coating of clay or iron oxides and/or hydroxides to a complex fracture array consisting of multiple anastomosing fracture surfaces, alternating with coarser cataclasite, veins of polycrystalline quartz filling, and/or calcite alteration. The degree of hydrous phase alteration (e.g., zeolites) and mineralization also suggests that significant amounts of fluid-rock interaction occurred at some point in the history of fault and damage-zone development.

Alteration phases are associated with several of the major fault zones penetrated during drilling. Solum et al. (2006) quantify the mineral assemblages of five major faults penetrated during drilling of the SAFOD MH. As with the results of this study, those faults have highly variable mineral assemblages. A fault within the granitic sequence and a fault separating that sequence from underlying sediments contain the zeolite mineral laumontite, although that phase is present in trace concentrations in other faults. Two faults (one separating the upper and lower arkosic sequence and one at the bottom of the deeper Phase One core) contain a neoformed, mixed-layer, illitesmectite phase. These smectitic clays occur as films (Schleicher et al., 2006), and may be important for determining the mechanical properties of the fault zones that contain these phases. Laumontite is often associated with temperatures of 120–180 °C (Liou, 1971; Cho et al., 1987). The heating required to produce laumontite may be associated with the burial history of the rocks southwest of the SAF (Blythe et al., 2005), or related to hydrothermal alteration associated with faulting.

The bottom hole temperature was measured after drilling at 105 °C (http://www.icdp-online.de/contenido/icdp/front_content.php?idart=1033), and Draper (2007) incorporates apatite fission-track thermochronology (Blythe et al., 2005) and new zircon fission-track analysis to constrain the maximum temperature that the arkosic rocks experienced to 240 °C. The deformed and altered rocks observed in the borehole, southwest of the interpreted active strand of the San Andreas Fault, were in the temperature range for plastic deformation of calcite and semi-plastic and brittle deformation of quartz, and in an alteration window associated with clay-zeolite-chlorite alteration. The exhumation history of the area (Blythe et al., 2005; Draper, 2007) indicates general uplift and cooling from these maximum temperatures, consistent with the nature of alteration observed in this study.

The final phase of sampling is to acquire core across the active portion of the seismogenic part of the SAF at depth (see http://www.icdp-online.de/contenido/icdp/front_content.php). This sampling will be followed by installation of borehole seismometers to observe the SAF at depth. Based on the work presented in this paper, the borehole geophysical data sets, and location of earthquakes (Ellsworth et al., 2005), the likely target for coring in 2007 is the region between 3050 and 3450 m MD where the currently active part of the San Andreas Fault is in a fine-grained sedimentary sequence. The general target areas are (1) the low-velocity zone (Zoback et al., 2005; Boness and Zoback, 2006), where the borehole appears to be actively deforming at 3295–3313 m MD (Zoback et al., 2005; Hickman et al., 2005); (2) near where one or more M_w 0 earthquake(s) have occurred (Ellsworth et al., 2005; W. Ellsworth, 2007, personal commun.); (3) where we document a broad zone of alteration with one or more regions of increased abundance of cataclasite; and (4) where microstructures from sidewall core suggest significant deformation (Evans et al., 2005).

The best analogs for the SAF at SAFOD are faults in fine-grained sedimentary rocks, but because of their poor preservation potential, relatively few strike-slip analogs exist. Most studies of exhumed faults in fine-grained rocks are from a variety of tectonic settings (Vrolijk and van der Pluijm, 1999; Warr and Cox, 2001; Yan et al., 2001; Faulkner et al., 2003; Heermance et al., 2003; Wibberley and Shimamoto, 2003; Kondo et al., 2005; Solum et al., 2003, 2005a; Holland et al., 2006) or from faults sampled by drilling (Moore et al., 1995; Hung et al., 2005). Faulkner et al. (2003) draw on geophysical data from the Parkfield area to suggest the overall structure and composition of the Carboneras fault is analogous to the SAF. The Carboneras fault has an estimated strike-slip offset of 40 km, and cuts a wide range of rock types, including crystalline rocks, phyllosilicate-bearing metamorphic rocks, and Tertiary sedimentary rocks. The fault zone is up to 1 km wide, with a broad damage zone interspersed with narrow, anastomosing, clay-rich, gouge zones and very localized, clay-rich, slip surfaces (Faulkner et al., 2003).

Most detailed studies of fault-zone structure and composition in phyllosilicate-bearing gouge and foliated cataclasite indicate that slip occurs along millimeter-thick surfaces, on which neomineralized clay grains grow (Schleicher et al., 2006). Deformation mechanisms consist of slip or glide along cleavage surfaces. Clay-forming reactions may suggest several processes—a significant number of fluid-rock interactions have occurred (Evans, 1988; Vrolijk and van der Pluijm, 1999); anisotropy of permeability may develop (Morrow et al., 1984; Evans et al., 1997); and significant variation in spatial and temporal pore-fluid pressures, mechanical properties, and textures should be anticipated at depth (Warr and Cox, 2001; Wibberley, 2002; Faulkner and Rutter, 2003; Faulkner et al., 2003; Chester et al., 2005; Holland et al., 2006).

The data presented in this paper, along with analyses of field analogs and experimental data on permeability, porosity, and mechanical properties of fault-related rocks in fine-grained sedimentary rocks, suggest that the coring effort for SAFOD in 2007 may encounter a range of fine-grained, fault-related material, including brecciated, fractured, and vein-bearing, damage-zone rocks, in which narrow slip surfaces and mineralization and alteration products are common. Pore-fluid pressures may be variable, and recovery of portions of the core may be difficult due to the nature of damage and high degree of fragmentation within the rock. Defining the main slip surface and differentiating between creeping and seismically slipping faults may be challenging and may require careful observations of the cored material coupled with analysis of borehole data. Additional complexity may result from the variability in V_p and V_s values for the faulted rocks, which could make locating the target earthquakes difficult when using short, source-receiver distances.

We integrate results from point counts and microstructural analyses of thin sections from cuttings samples with interdisciplinary research from the SAFOD site to delineate the lithological and structural setting in the subsurface at this location. Four major lithologic packages are identified: (1) Quaternary and Tertiary sediments (0–760 m MD); (2) granitic rocks (760–1920 m MD); (3) arkosic to lithic arenites (1920–3150 m MD) separated by a clay-rich zone (2530–2680 m MD); and (4) fine-grained to very fine-grained interbedded siltstones, mud-stones, and shales (3150–3987 m MD).

Fault zones are associated with abundances of cataclasite and various alteration products, and are located at similar depths as inferred from borehole geophysical data, including density- or porosity-based logs. The point-count percentage of cataclasite and microstructural deformation features are used to locate several fault strands and related damage zones within the MH at 1050, 1650–1750, 1900–2000, ∼2650–2700, 3050–3300, and ∼3650 m MD. The currently active portion of the San Andreas Fault, where the borehole intersects the fault at 3300 m, consists of fine-grained cataclastically deformed rocks with significant alteration and the presence of very narrow, clay-lined, slip surfaces at the micrometer scale.

Zones of alteration occur at 1050, 1600–2000, 2200–2500, 2700–3000, 3050–3350, and 3600 m MD. Overall, alteration is easier to identify in the granitic sequence and occurs within several of the fault zones. Compositional variations and increases in the amount of alteration vary between fault zones and may be indicative of fluid compartmentalization related to the sub-surface lithological and structural architecture.

This work also develops a conceptual model for the geologic setting in which the target earthquakes occur and, in general, may offer insight into broader questions associated with the deformation and structure of fault zones and may be used to provide constraints on the design and coring of the active San Andreas Fault zone or Phase Three of SAFOD, planned for the summer of 2007.

Funding for this research was provided by National Science Foundation-Earthscope grant EAR-0454527 to Evans and an REU supplement for Corey Barton's work. We thank the 2002, 2004, and 2005 SAFOD on-site teams for diligent sample collection, and especially thank Naomi Boness and Amy Day-Lewis for their tireless efforts in coordinating the on-site science teams. Many thanks to Bill Ellsworth, Steve Hickman, and Mark Zoback for their efforts to bring SAFOD to fruition and for many discussions regarding SAFOD site geology and geophysics. Discussions with Fred and Judith Chester, David Kirschner, Diane Moore, Ben van der Pluijm, and Susanne Janecke are greatly appreciated. We thank Stephanie Carney for assistance with computer graphics. Thanks to John Hole and Cliff Thurber for providing the seismic velocity model figures, and to Casey Moore and Dave Dewhurst for thorough reviews of earlier versions of this paper.