Vol 448 | 16 August 2007 | doi:10.1038/nature06064 LETTERS Talc-bearing serpentinite and the creeping section of the San Andreas fault Diane E. Moore1 & Michael J. Rymer1 The section of the San Andreas fault located between Cholame Andreas fault (SAF). In 2005, drilling successfully crossed the active Valley and San Juan Bautista in central California creeps at a rate trace of the SAF at ,3 km vertical depth12, where the measured as high as 28 mm yr21 (ref. 1), and it is also the segment that yields temperature is ,112 uC (ref. 13). The drillhole terminated in sedi- the best evidence for being a weak fault embedded in a strong mentary rocks of the Great Valley Sequence (K. McDougall, personal crust2–5. Serpentinized ultramafic rocks have been associated with communication) east of the fault. Since then, a portion of the well creeping faults in central and northern California6–8, and serpen- casing has been actively deforming in response to creep on a fault tinite is commonly invoked as the cause of the creep and the low strand12. Serpentine was identified in X-ray diffraction patterns of strength of this section of the San Andreas fault. However, the cuttings14 collected at the eastern margin of the zone of active frictional strengths of serpentine minerals are too high to satisfy deformation (Supplementary Fig. 1). Aeromagnetic surveys15 indi- the limitations on fault strength, and these minerals also have the cate the presence of a flat-lying slab of serpentinite at .3 km depth on potential for unstable slip under some conditions9,10. Here we the northeast side of the fault (Fig. 1). This body may be $2 km thick, report the discovery of talc in cuttings of serpentinite collected and it abuts the fault for 50–60 km (ref. 15). The serpentinite slab is from the probable active trace of the San Andreas fault that was probably part of the Coast Range ophiolite, the oceanic basement on intersected during drilling of the San Andreas Fault Observatory which the sediments of the Great Valley Sequence were deposited. at Depth (SAFOD) main hole in 2005. We infer that the talc is Serpentinized ultramafic rock has a relatively low density compared forming as a result of the reaction of serpentine minerals with to the overlying rock column, and a fault intersecting such a rock unit silica-saturated hydrothermal fluids that migrate up the fault provides the pathway for the migration of serpentinite to shallower zone, and the talc commonly occurs in sheared serpentinite. This depths16. The Table Mountain serpentinite17 east of Parkfield (Fig. 1) discovery is significant, as the frictional strength of talc at elevated is an extrusive body that formed as a result of the diapiric rise of temperatures is sufficiently low to meet the constraints on the shear low-density serpentinite from the deeply buried slab along faults that strength of the fault, and its inherently stable sliding behaviour is served as ‘fissure feeders’17. The serpentinite associated with the consistent with fault creep11. Talc may therefore provide the con- active trace in the SAFOD drillhole14 and outcrops of serpentinite18,19 nection between serpentinite and creep in the San Andreas fault, if fault gouge (Fig. 1) suggest that the same process is occurring along shear at depth can become localized along a talc-rich principal-slip the SAF. surface within serpentinite entrained in the fault zone. Serpentinite has been suggested as a possible cause of creep, The SAFOD drillsite is located 14 km northwest of Parkfield in because of its close association with creeping faults in central and central California (Fig. 1), along the creeping section of the San northern California6–8. The SAF creeping section coincides with the a Figure 1 | Distribution of serpentinite along the SAF CALIFORNIA creeping section. a, The creeping section lies between areas of the fault that ruptured during great Serpentinite body Table Mountain at >3 km depth serpentinite earthquakes in 1857 and 1906. N on east face Serpentinite occurs in rare surface Calaveras fau exposures of the fault18,19 and in the lt Hollister Paicines fault probable active trace of the fault 1857 surface encountered at ,3 km vertical ace 1906 surf Parkfield break depth in the SAFOD drillhole12,14. break San Juan San Andreas fault SAFOD Cholame The extrusive serpentinite at Table Bautista Serpentinite Mountain17 is derived from the outcrops same serpentinite body that abuts b the fault on the northeast side at ) .3 km depth15. b, Recently updated –1 30 creep rates measured at distances of NW SE 10 m to 1 km from the fault1. Total 20 offset rates along the San Andreas 10 Intermediate system in the creeping section are scale (±1 km) considered to be between 28 and 0 34 mm yr21 (ref. 1). Creep rate (mm yr Creep 0 50 100 150 Distance along the SAF from San Juan Bautista (km) 1US Geological Survey, 345 Middlefield Road, Mail Stop 977, Menlo Park, California 94025, USA. 795 © 2007 Nature Publishing Group LETTERS NATURE | Vol 448 | 16 August 2007 mapped extent of Coast Range ophiolite and overlying Great Valley decreases with increasing velocity) and velocity-strengthening Sequence rocks on the northeast side of the fault7,8,20. Ultramafic (strength increases with increasing velocity) behaviour9,10 at different rocks of the Coast Range ophiolite are variably serpentinized. The temperature–pressure–velocity conditions; as a result, they can slip most extensive serpentinite body along this section is the one east of either unstably or stably, respectively, depending on the depth and SAFOD15 that is associated with the highest creep rate (Fig. 1). On the slip rate. basis of aeromagnetic and gravity surveys, a slab of serpentinite 1–1.5 We examined serpentinite grains from the washed SAFOD cut- km thick at $3 km depth extends northeastwards from the SAF a few tings that were collected at ,3 m intervals during drilling. Polished kilometres southeast of San Juan Bautista to the Calaveras fault grain mounts were prepared from cuttings samples for analysis with around Hollister20. Serpentinite continues at somewhat greater an optical microscope, scanning electron microscope (SEM) and depth20 east of the Calaveras fault in that area. Other, smaller masses electron microprobe. The serpentinite contents of the bulk cuttings, of serpentinite are present at $2.4 km depth between the San estimated from point counts of thin sections, exceeds 2% by volume Andreas and Paicines faults7. The Calaveras–Paicines faults creep at in the interval 3,319–3,350 m measured depth (MD), with a spike of rates of 6–10 mm yr21 south of Hollister21. ,8% in the 3,325 m MD sample (Supplementary Fig. 1). A powder The creeping section provides the best evidence for a weak SAF2–5. X-ray diffraction pattern of a separate of serpentinite grains shows Because the creeping section is characterized by aseismic slip and prominent peaks consistent with lizardite and chrysotile, the two microearthquakes, the apparent weakness of this segment cannot low-temperature serpentine minerals. No relict olivine or pyroxene be explained through some dynamic weakening process accompany- has been found. On the basis of the common occurrence of both ing a major earthquake, as it can for the locked sections. The restric- mesh texture after olivine and bastite texture after pyroxene, the tions on shear strength in the creeping SAF imposed by heat-flow22,23 original ultramafic rock was probably a harzburgite16, similar to and stress-orientation24,25 data are delimited in Fig. 2; also included the Table Mountain serpentinite17. The pseudomorphic mesh and are the frictional strengths of synthetic fault-gouge materials9–11,26 bastite textures have been extensively modified by recrystallization, prepared by grinding and sieving rock or mineral separates. The brecciation and shearing. strength experiments were conducted in a triaxial machine fitted with The serpentinite contains numerous calcite- and some quartz- an internal furnace, at various combinations of temperature, confin- filled veins, possibly resulting from focused fluid flow within the fault ing and fluid pressure, and sliding velocity. For a given mineral to zone. Talc replaces serpentine minerals along the vein walls (Fig. 3a, control the behaviour of the creeping section, it must be very weak as b), and it fills narrow cracks that extend into the serpentinite from the well as characterized by stable shear. The frictional properties of the wider veins. Talc also forms along the foliation in sheared serpenti- serpentine minerals do not satisfy the weakness criterion and under nite grains (Fig. 3c, d). The talc-forming reaction is: certain conditions do not satisfy the stability criterion. The serpent- 9 ine minerals lizardite and antigorite are not substantially weaker Mg3Si2O5(OH)4 1 2SiO2 5 Mg3Si4O10(OH)2 1 H2O than granite26 under hydrothermal conditions (Fig. 2). Chrysotile serpentine talc satisfies the heat-flow constraint to depths of ,3 km, but its strength 9,10 increases substantially at greater depths . Furthermore, all three The SiO2 comes from the dissolved silica content of heated ground serpentine minerals show both velocity-weakening (strength water (Fig. 3a, c, d) and from quartz deposited metastably in veins (Fig. 3b). Talc is stable relative to the assemblage quartz 1 serpentine 27 0 throughout the stability range of serpentine . The veins and shears with which talc is associated overprint all other textural features in Frictional strengths the serpentinite grains, suggesting that the talc is of recent origin. Talc extrapolated to SAF 2 + a b Tc Tc Pores 4 Tc Tc Sp Antigorite Granite Cc Q Tc Sp 6 + Sp Sp Sp Lizardite Tc Chrysotile Tc Depth (km) Cc Q 8 Cc Tc Pores + c d Sp 10 365 mm Tc TAL Sp C Tc Sp yr Tc 12 - 1 Tc Stress- Heat-flow orientation Tc constraint Tc constraint Cc Sp 14 Sp 0 20 40 60 80 100 Cc Shear strength (MPa) Tc Figure 2 | Shear strength versus fault depth.