Fracture Mechanical Properties of Damaged and Hydrothermally Altered Rocks, Dixie Valley‐Stillwater Fault Zone, Nevada, USA

Damaged and hydrothermally altered rocks are ubiquitous in fault zones, with the degree of damage and type and intensity of alteration varying in space and time. The impact of damage and alteration on hydromechanical properties of fault zones is difficult to assess without characterizing the associated changes to rock and fracture mechanical parameters. To evaluate the mechanical properties of fault rocks from different alteration regimes, we conducted (1) double‐torsion load‐relaxation tests to measure mode‐I fracture toughness (KIC) and subcritical fracture growth index (SCI), (2) uniaxial testing to measure unconfined compressive strength (UCS) and static elastic parameters, and (3) mineralogic and textural characterization of rock from four sites in the footwall of the Dixie Valley‐Stillwater fault zone. Alteration at these sites includes acid sulfate alteration and silicification associated with active fumaroles, intense silicification after calcite and chlorite alteration in an epithermal setting, quartz‐kaolinite‐carbonate alteration from an intermediate‐depth system, and a calcite‐chlorite‐hematite assemblage containing abundant unhealed damage. Silicification is associated with high KIC, SCI, UCS, and increased brittleness, and in precipitation‐dominated settings produces fault cores that are as strong or stronger than adjacent damage zone material. Calcite‐chlorite‐hematite assemblages containing abundant unsealed microfractures are approximately 4–5 times weaker than the granodiorite protolith. Mechanical properties are not predicted by mineralogical composition alone; a key control is the accumulation of damage and degree of healing. Measures of strength increase when mineral precipitation reduces microfracture porosity to <10–15% of total microfracture area. These results show that fault‐proximal weakening or strengthening is influenced by hydrothermal setting.


Introduction
Fault-fracture networks contribute critically to fluid flow in low-porosity crystalline rock, impacting the distribution of heat, fluid, and minerals in the upper crust. Hydraulically conductive faults and fault segments are characterized by well-developed damage zones composed of opening-mode and sheared fractures (Anders et al., 2013;Brown & Bruhn, 1996;Caine et al., 1996;Eichhubl et al., 2009;Nelson et al., 1999;Sibson, 1996). The occurrence of opening-mode fractures in and around fault zones and the evolution of these flow systems from single fractures to complex interconnected networks by fracture reactivation, propagation, and coalescence provide a fundamental control on the permeability evolution of faults and fault-controlled hydrothermal and epithermal systems (Blenkinsop, 2008;Davatzes et al., , 2005Sornette, 1999).
Chemical controls on conduit evolution are well documented in field, experimental, and numerical studies, with dissolution, precipitation, and chemical alteration impacting the hydraulic properties of fault-facture networks (Berkowitz, 2002;Eichhubl et al., 2004Eichhubl et al., , 2009Moore et al., 1983;Summers et al., 1978). These effects are particularly pronounced in chemically reactive environments encountered in high-temperature hydrothermal systems (Lowell et al., 1993), where dissolution, advection, and precipitation of different mineral species in response to thermal and chemical disequilibrium and water-rock reactions lead to the development of distinct alteration assemblages (Facca & Tonani, 1967;Henley & Ellis, 1983;Sillitoe, 2010;Simmons et al., 2005;Tosdal et al., 2009). Large regions of alteration are commonly associated with active (a-d) Representative field photographs from each site. Active hydrothermal circulation within the Dixie Valley-Stillwater fault zone manifests as hot springs and fumaroles, and is utilized for electricity generation in the Dixie Valley geothermal production area (Berry et al., 1980;NREL, 2016). Quaternary faults after USGS and NBMG (2010). Geology modified from Crafford (2007). hydrothermal systems (Browne, 1978) and around fossil conduits associated with magmatic and hydrothermal ore deposits (Henley & Ellis, 1983). However, some degree of alteration is common in most fault zones, where fracture-enhanced fluid flow and mechanical grain size reduction promote chemical reactions (Bruhn et al., 1994;Solum et al., 2010).
Mineralogical and textural changes associated with fault zone damage, fluid flow, and hydrothermal alteration result in hydromechanical properties that differ from those of relatively pristine samples commonly used in geomechanical laboratory tests. For example, Seront et al. (1998) showed a decrease in porosity and permeability of argillically altered fault core samples collected from the Dixie Valley-Stillwater fault zone, Nevada, USA, with cemented damage zone rocks exhibiting a 7-115% increase in compressive strength. Wyering et al. (2014) and Siratovich et al. (2014) reported higher unconfined compressive strength (UCS) in calcite and quartz ± propylitic altered volcanic rocks than in the same rocks dominated by quartz and clay alteration in the Taupo Volcanic Zone, New Zealand. Allen et al. (2017) showed that damage, fluid-rock interaction, and mineralization near the principal slip zone of the Alpine Fault, New Zealand, resulted in reduced seismic velocity and permeability, and increased anisotropy of both properties, relative to intact country rock. Less work addresses the impact of damage and alteration on the fracture mechanical parameters fracture toughness (K IC ) and subcritical fracture growth index (SCI; Atkinson, 1984;Kobayashi et al., 1986;Major et al., 2018), despite (1) the importance of fractures and fracture growth in fault systems and (2) distinct differences between naturally altered and damaged fault rocks and relatively pristine geomechanical materials.
For opening-mode (mode-I) fractures, the stress intensity at the fracture tip, K I , may be written as a function of remote applied stress, σ r ; fracture length or height, a; and fracture geometry, Y (Brown & Strawley, 1966): Mechanical fracture propagation occurs when K I reaches a critical threshold, known as fracture toughness (K IC ). K IC is more influential during early fracture growth when fracture length, a, is small and K I is proportionally lower (Engelder et al., 1993). Cyclic loading and chemical corrosion at the fracture tip can lead to fracture growth below the critical stress intensity threshold, referred to as subcritical fracture growth. Subcritical fracture growth is quantified by the SCI parameter, which describes the relationship between fracture propagation velocity and stress intensity for fracture propagation below K IC in the empirically derived equation (Pletka et al., 1979): where V is the velocity, V * is a constant, and K 0 is a normalization factor. Subcritical fracture growth has been invoked to explain long-term strength of the crust (Anderson & Grew, 1977;Brantut et al., 2013;Rudnicki, 1980), and differences in SCI may influence fracture spatial arrangements (Olson, 1993(Olson, , 2004. Together, K IC and SCI characterize key fracture mechanical properties in a deforming rock volume. Here we present experimental measurements of K IC and SCI from doubletorsion load-relaxation (DT-LR) fracture mechanics tests of samples of  (Blackwell et al., 2007). Boiling point with depth for pure water from Haas (1971).
hydrothermally altered and damaged rocks collected from four sites in the Dixie Valley-Stillwater fault zone ( Figure 1 and Table 1). Sample sites represent a variety of hydrothermal settings, including (1) shallow acid sulfate alteration at Dixie Meadows fumaroles, (2) epithermal silicification after chlorite-calcite alteration at Dixie Comstock, (3) quartz-kaolinite-carbonate alteration and cementation in the Mirrors normal fault exposure, and (4) chlorite-calcite-hematite dominated assemblages at the Box Canyons. Alteration and deformation histories at these sites record exhumation from different temperatures, depths, and hydrothermal conditions (Figure 2) constrained by mineral assemblages, mineral textures, fluid inclusion homogenization temperatures, and geologic relationships, and illustrate the relative impact of damage-dominated versus precipitation-dominated settings on fracture mechanical behavior. We demonstrate that hydrothermal alteration, previously recognized as an important factor influencing rock mechanical properties, also impacts the fracture mechanical properties influencing initiation, growth, and coalescence in fault-fracture networks.

Geologic Setting
The Dixie Valley-Stillwater fault zone is a north to northeast striking, east dipping basin-bounding normal fault system in northwest Nevada, USA ( Figure 1). The Stillwater Range is composed of Triassic phyllite, Jurassic mafic igneous and volcanic rocks of the Humboldt Igneous Complex and associated sedimentary facies, small Cretaceous granitic plutons, and Oligocene plutonic and volcaniclastic sequences, and is capped by mid-Miocene basalt (Dilek & Moores, 1995;John, 1995;Kistler & Speed, 2000;Page, 1965;Speed, 1976;Speed & Jones, 1969). The basin hosts a series of buried, nested grabens with >1.8 km of Quaternary lacustrine and alluvial fan deposits (Bell & Katzer, 1990;Okaya & Thompson, 1985). Post 8 Ma dip-slip displacement in northern Dixie Valley is between 2.2 and 2.9 km (Okaya & Thompson, 1985) and post-Oligocene slip in southern Dixie Valley may exceed 6 km (Parry & Bruhn, 1990;Thompson et al., 1967).
The region has attracted attention in part due to its location at the northern end of the Central Nevada Seismic Belt Wallace, 1984;Wallace & Whitney, 1984 Bell et al., 2004;Bell & Katzer, 1990;Caskey et al., 1996;Caskey et al., 2004;Slemmons, 1957;Wallace, 1984). Fault scarps from the 1954 Dixie Valley earthquake parallel the Stillwater Range through the southern portion of the field area, with older Quaternary fault scarps located throughout the Dixie Valley-Stillwater fault zone ( Figure 1).

Mineralogical and Textural Characterization
We evaluated bulk mineralogy of damaged and altered rocks with X-ray diffractometry using a Bruker D8 Advanced X-ray diffractometer with LynxEye detector (Table 2). Samples were prepared as randomly oriented spray-dried powders (Hillier, 1999) and scanned at~0.01°increments from 4 to 66°2-theta for 1.5-2 hr. Initial analysis of X-ray diffractometry spectra was conducted with EVA software, followed by Rietveld analysis using TOPAS 4.2 software. Methods and spectra are included in the archived data (Callahan, 2019).
Alteration products and reactants were further described with thin section petrography. Host rock porosity, microfracture porosity, and microfracture porosity occluded by mineral cements were measured using point counting of blue-epoxy impregnated thin sections at 120X magnification and approximately 400 points per sample (Table 3). Total microfracture area was calculated from the sum of points encountering   dmg microfracture porosity or cement. Thin sections were oriented approximately perpendicular to the dominant structural fabric.

Unconfined Compressive Strength and Static Elastic Characterization
We measured static elastic properties and unconfined compressive strength (UCS) with uniaxial compressive strength tests using a GCTS rock mechanics system (Table 4). Plug orientations were vertical (V), parallel to strike (H), downdip (D), or mutually perpendicular to strike and dip (P) of local range front faults. Plug diameter was 25.4 mm and average plug length was approximately 53.0 mm. Loading was conducted at a preprogrammed axial strain rate of 0.055%/min (~0.5 μm/s). UCS is reported from peak, area-corrected load. Young's modulus and Poisson's ratio were calculated from the middle portion of the loading curve where the relationship between stress and strain was approximately linear. Complete load curves are included in the archived data (Callahan, 2019).

Double-Torsion Load-Relaxation Fracture Mechanics Testing
We used double-torsion load-relaxation (DT-LR) tests to measure K IC and SCI using multiple specimens of altered rocks. DT-LR tests were conducted by repeatedly propagating a fracture down the axis of a specimen prepared as a thin rectangular wafer ( Figure 3a). All specimens were cut from the same blocks of material used for petrographic, mineralogic, and mechanical characterization. Specimens were cut so the propagation directions of induced fractures were typically vertical and/or parallel to the local structural grain, although this was limited by poor material quality in some samples. Orientation information for each wafer is included in the archived data (Callahan, 2019). Detailed descriptions of the testing apparatus, method, and data reduction procedure used here can be found in Chen et al. (2017), and are summarized below.
The DT testing apparatus consists of a base plate, specimen supports, a loading ram with internal force sensor, and a linear variable displacement transducer to record displacement ( Figure 3b). K I at the fracture tip was calculated using the equation (Williams & Evans, 1973): where P is the load supported by a prefractured specimen, W m is the moment arm of the DT apparatus, ν is the Poisson's ratio, W is the specimen width, t is the specimen thickness, and t n is the reduced thickness along an axial groove created by pulling the specimen across a recessed diamond saw prior to testing ( Figure 3c). If insufficient sample material existed for UCS tests, v was estimated from similarly altered and damaged samples. The geometric correction factor, ψ, is based on individual specimen geometry (Fuller, 1979): We tested specimens with thickness (t), width (W), and length (L) dimensions of approximately 1.8 mm × 30 mm × 75 mm, respectively, meeting dimensional requirements of~24 t < 2 W < L (Nara & Kaneko, 2005). Absolute specimen dimensions were similar to those used by Atkinson (1979b), Sano et al. (1992), and Chen et al. (2017) and were limited in part by sample size and load cell capacity. Prefractures were induced using the DT apparatus at low displacement rates (<1 μm/s) until a distinct load drop was observed, indicating the formation of an edge crack. Journal of Geophysical Research: Solid Earth DT tests for fracture toughness were conducted on prefractured specimens at fast displacement rates (180-220 μm/s) to total failure ( Figure 4a). DT-LR tests for SCI were conducted by loading prefractured specimens at low displacement rates (1-2 μm/s) until fracture propagation was indicated by a rapid drop in supported load. Displacement was stopped, and the load allowed to decay for 5-10 min. Ideal DT-LR test patterns included a high prefracture load, a subsequent plateau region of lower peak loads and load-decay curves that were used to calculate SCI, and a final load drop upon complete specimen failure (Figure 4b). A separate estimate of fracture toughness (K IC *) based on the stress intensity from peak loads in the plateau region during slow displacement tests was evaluated as a proxy for K IC . Load and displacement were recorded at 14 Hz for DT tests and 5 Hz for DT-LR tests. All tests were conducted under ambient conditions at 23-24°C. Relative humidity was not measured for all tests, but commonly ranged between 58 and 75% and could have contributed small variations to test results (Nara et al., 2012).
Fracture propagation velocity was calculated from the load relaxation curve (Evans, 1972): where a 0 is the initial fracture length and P i is the load at the start of each load-decay cycle. The correction factor for fracture front geometry, ϕ, was assumed to be 0.2 (Atkinson, 1979a;Chen et al., 2017;Williams & Evans, 1973). Because prefracture lengths were difficult to observe in these materials, we followed Chen et al. (2017) and use a 0 of 12.7 mm, with later cycles using a n = a 0 + n × 12.7 mm, a nominal value that accounts for the average number of fracture growth increments and specimen length. Variation of a 0 has limited impact on calculated fracture front velocity (Chen et al., 2017). SCI was calculated from K-V curves using an in-house LabView script for smoothing and fitting following derivations described in Holder et al. (2001). Individual specimen dimensions, peak loads, K IC , K IC *, derived SCI, and load-decay curves are included in the archived data (Callahan, 2019).

Acid Sulfate Alteration and Silicification at Dixie Meadows Fumaroles
The Dixie Meadows fumarole site is located near the northern terminus of the 1954 Dixie Valley fault rupture and west of Dixie Meadows Hot Springs (Figure 1). At this site, fumarole-related alteration in Oligocene tuff is exposed in the footwall of the Dixie Valley-Stillwater fault zone. Alteration products here include native sulfur, sulfate minerals, kaolin group minerals, montmorillonite, calcite, and quartz (Kennedy-Bowdoin et al., 2004;Lamb et al., 2011;Schwering, 2013), a suite of minerals reflecting shallow acid sulfate alteration related to ongoing fumarole activity and boiling or near-boiling conditions in the shallow subsurface ( Figure 2).
We tested multiple plugs and DT specimens from two samples from Dixie Meadows fumaroles: a moderately altered tuff (sample 1) and a portion of exhumed fault core composed of weakly silicified bladed calcite and microquartz (sample 2; Figure 5 and Table 2). Sample 1 is a pale, nonwelded, devitrified tuff, with partially dissolved feldspar phenocrysts. The dominant mineral species are quartz and feldspar, with lesser kaolinite and undifferentiated clay. Vugs in pumice are commonly filled with kaolinite, whereas vugs in partially dissolved feldspar grains contain small euhedral quartz crystals. Hematite occurs as disseminated grains and as fracture fill. Pores are typically smaller than 0.5 mm, with some secondary pores in dissolved grains exceeding 2 mm. Total porosity is~7.1%, and total microfracture area is low (~0.7%; Table 3). Sample 2 is a coarsegrained fault breccia, mineralogically and texturally dominated by multiple generations of quartz and calcite, with minor hematite and sericite. Calcite occurs as bladed laths and as disseminated, fine-grained, intergrowths with microquartz. Microquartz occurs as clasts and as cement between clasts. Clast sizes range from 1 cm to <1 mm. Total porosity and total microfracture area are low (0.3% and 2.7%, respectively).

Na-Ca Alteration and Silicification at Dixie Comstock Epithermal Gold Deposit
The Dixie Comstock epithermal gold deposit is located along a north striking portion of the Dixie Valley-Stillwater fault zone (Figure 1). Several temporally distinct episodes of alteration are preserved in the Dixie Comstock area. Early and widespread sodic and calcic alteration of the Jurassic Humboldt Igneous Complex (Dilek & Moores, 1995;Johnson & Barton, 2000) is overprinted by aureoles of quartz, albite, sericite, kaolinite, and iron oxide around apophyses of Cretaceous granite (Vikre, 1993). Alteration in the mine area is dominated by silicification of the range front fault, with portions of the silicified fault core exceeding 2 m in thickness and approaching 100% quartz. Silicification overprints and entrains earlier assemblages. Intense silicification extends~300 m north and south of the mine workings, with quartz veins in the footwall and minor silicification of fault breccia extending 1.5 km along strike and at least several hundred meters downdip. Bladed calcite (this study) and liquid-and vapor-rich inclusions (Vikre, 1993) indicate that boiling conditions existed in shallow parts of the system, and fluid inclusion microthermometry indicates temperatures between 160 and 180°C (Vikre, 1993). Boiling near 170°C suggests exhumation from as shallow as 76 m (Haas, 1971; Figure 2).
Dixie Comstock samples were obtained from a distal portion of the footwall (sample 3), from the footwall behind the main mineralized deposit (samples 4 and 5), and from the silicified fault core (samples 6 and 7; Figure 6). Alteration reactions are dominated by selective replacement of feldspars and mafic minerals with chlorite, calcite, and sericite and intense silicification (Table 2). Sample 3 is a medium-grained gabbro, dominated by plagioclase and altered mafic minerals. Alteration minerals include albite, chlorite, and minor quartz, epidote, sericite, sulfides, oxides, and trace calcite. Plagioclase is partially albitized, and , whereas sample 5 has higher total microfracture area (3.6%) and lower porosity (0.7%; Table 3). Samples 6 and 7 were collected from different locations within the silicified fault core and reflect textural, mineralogical, and mechanical variations observed in this material. These samples are dominated by fine-grained, intergrown, microquartz with minor chlorite, feldspar, and plagioclase, and trace sericite, calcite, and sulfides and oxides. Both samples contain evidence of multiple generations of brecciation and quartz cementation in the form of broken and rounded clasts of earlier microquartz breccia. Total microfracture area in sample 6 is intermediate (6.9%) and porosity is low (0.5%). Total microfracture area in sample 7 is high (12.6%), but intense silicification has reduced porosity to 1.6%. In both samples, porosity is restricted to quartz lined and nearly occluded vugs and partially quartz-or calcite-cemented fractures.

Quartz-Kaolinite-Carbonate Alteration at the Mirrors Fault Zone Exposure
The Mirrors site is located along a northeast striking section of the Dixie Valley-Stillwater fault zone southwest of the producing geothermal field (Figure 1). The protoliths at the Mirrors are intrusive and extrusive components of the Jurassic Humboldt Igneous Complex (Dilek & Moores, 1995;Page, 1965;Speed, 1976). Alteration includes regional sodic, calcic, and chlorite alteration, with later kaolinite, carbonate, and quartz after mafic minerals and feldspars in well-developed fault damage zone and core (Caine et al., 2010;Lutz & Moore, 1996). Physical conditions during the dominate phase of alteration are constrained by post-Miocene exhumation of <2 km (Power & Tullis, 1989, 1992 and the occurrence of quartz with kaolinite indicating temperatures <270°C (Figure 2). Ferroan dolomite, chalcedony, goethite, and barite (Lutz & Moore,

10.1029/2018JB016708
Journal of Geophysical Research: Solid Earth 1996) and chalcedony with kaolinite (this study) indicate that at least some alteration occurred at temperatures <180°C.
Samples from the Mirrors include altered and moderately silicified fault damage zone (sample 8) and fault core material (sample 9; Figure 7 and Table 2). Sample 8 is an argillic-silicic altered fineto medium-grained mafic plutonic rock. Mineralogy is dominated by quartz and kaolinite, with lesser carbonate (calcite, ankerite), sericite, and trace hematite. Primary magmatic textures are cryptic, with kaolinite replacing feldspars, calcite replacing amphibole, and degraded pyroxene. Quartz is dominantly fine grained, intergrown microquartz, with rare chalcedony-lined, kaolinite-filled vugs. Carbonate occurs as small (<50 μm), disseminated grains, as fill in thin, discontinuous fractures, and replacing amphibole grains <2 mm long. Sample 9 was obtained from the cemented fault core. Sample texture is heterogenous at the thin section scale. Cement is dominantly fine grained, interlocking microquartz, with angular to subangular clasts <2 cm long of broken calcite and ankerite veins. Macroscopic textures indicate repeated brecciation, alteration, and cementation during exhumation, with little primary texture preserved. Kaolinite, sericite, and calcite occur in the matrix, with some sericite and carbonate replacing clasts. Samples 8 and 9 both contain intermediate total microfracture area (3.2 and 5.0%, respectively), and low total porosity (0.6 and 0.3%), similar to porosity between 0.3 and 0.4% previously measured in Mirrors fault core samples (Seront et al., 1998).
Box Canyon samples (10-14) record increasing alteration and damage of granite and granodiorite ( Figure 8 and Table 2). Background samples include two plutonic phases: a less altered granodiorite (sample 10) and a more altered granite (sample 11). Quartz grains in sample 10 contain minor deformation. Plagioclase laths up to 1.5 mm in length contain small, uneven fractures, but no rotation and only minor sericitization. Biotite and amphibole are relatively pristine. All crystals are intergrown, with little interstitial space, no observed porosity, and low total microfracture area (0.2%). Feldspars in sample 11 contain patches of albite, and cloudy, vacuolized cores. Damage includes transgranular fractures with minor, early hydrothermal biotite. Total microfracture area is intermediate (3.3%) with low porosity (0.5%). Sample 12 was collected near the range front fault, but away from the chlorite-calcite-hematite altered area. Hydrothermal biotite replaces mafic minerals and occurs as fracture fill in orthoclase. Feldspars and some plagioclase laths contain cloudy, vacuolized cores, large sericite grains, and increased albitization. Porosity (2%) is comparable to 1.2% previously reported for Box Canyon damage zone

10.1029/2018JB016708
Journal of Geophysical Research: Solid Earth samples (Seront et al., 1998), and occurs in microfractures and in partially dissolved feldspars. Total microfracture area is intermediate (5.5%). Samples 13 and 14 were both collected from the most altered area in the Box Canyons site, the "ultradamaged carapace" of Seront et al. (1998), and contain among the highest microstructural heterogeneity at the thin section and specimen scale. Alteration minerals include carbonate and chlorite, with lesser sericite, epidote, and sulfides and oxides. Damage includes multiple cross-cutting cataclastic bands, broken grains, and thin, partially calcite-and hematite-filled fractures (Figure 8). Total microfracture area is high (19.4-23.3%). Porosity (2.5-5.6%) occurs in late, partially calcite-filled fractures, and at the edges of cataclastic bands.

UCS and Static Elastic Properties
We conducted UCS tests on 24 plugs from 11 samples of crystalline rock with varying types and degrees of alteration and damage (Table 4 and Figure 9). The number of successful repeat tests was limited in some samples by the amount and quality of sample material. When multiple orientations were tested, strength  Journal of Geophysical Research: Solid Earth and elastic measurements from plugs with different orientations were generally within error of one another (archived data and Figure S1 in the supporting information).
Samples with the highest UCS and Young's modulus include the minimally altered granodiorite sample (10), silicified samples (6 and 7), and the calcified sample (5). Silicified samples have the highest mean compressive strength (up to 286.5 MPa) and Young's modulus (62.8 GPa), and among the lowest Poisson's ratios (0.11-0.13). The strength of the silicified epithermal material is greater than minimally altered granodiorite collected from the Box Canyons site, and 6 times higher than the weakest altered and damaged samples. The weakest materials contain disseminated chlorite and calcite alteration and a higher percentage of open and partially cemented microfractures.

Fracture Mechanical Properties
We calculated K IC for 13 samples (6-21 specimens per sample) using rapid displacement to total failure (Table 5), and K IC * for 13 samples from 337 plateau loads (6-73 per sample) during slow displacement DT-LR tests (Table 6). Twelve samples were tested for both K IC and K IC *. Despite experimental work indicating loading rate dependence on K IC in some materials (Atkinson & Meredith, 1987), we observed no consistent difference between fracture toughness calculated from rapid or slow displacement tests ( Figure 10). The average standard deviation for K IC is 21%. When measurements of K IC are combined with measurements of K IC *, the average standard deviation is reduced to 17%. All figures use K IC from rapid displacement tests unless otherwise noted.
We observed significant variation in K IC between background and more damaged, altered, and cemented fault core and fault proximal material (Figure 11). The highest maximum K IC (3.84 MPa√m) and the highest mean K IC (3.20 MPa√m) were measured in silicified samples from Dixie Comstock. These K IC values are greater than relative pristine granodiorite collected from the Box Canyons site (2.08 MPa√m) and as much as 6 times higher than the weakest altered and damaged samples (0.56 MPa√m). Moderate silicification (50-60 wt% quartz) in fault core material from the Mirrors is associated with intermediate K IC values similar to unaltered granodiorite. The transect from least altered granodiorite to damage zone samples at the Box Canyons shows a reduction in K IC from >2.0 to <0.7 MPa√m. Similarly altered plutonic rocks from both Dixie Comstock and the Box Canyons, characterized by minor chlorite, calcite, hematite, ±epidote alteration, and unhealed damage, had similarly low K IC (~0.7 MPa√m). A positive correlation between K IC and UCS ( Figure 12) is consistent with the underlying mechanism of failure in UCS tests, microfracture growth and coalescence, which is in turn influenced by K IC .
We calculated SCI for 13 samples, using 4 to 19 specimens per sample (Table 7). In some tests, induced fractures propagated out of the axial groove, either due to interaction with existing microstructures or improperly balanced loading. These decay curves were not included in SCI calculations. The number of selected decay curves per sample ranged from 1 to 53. The minimum and maximum standard deviations were 23-38%, with an average standard deviation for SCI~30%. The highest mean SCIs were measured in cemented fault core samples from Dixie Meadows, Dixie Comstock, and the Mirrors (92.3-144.6). The Dixie Comstock samples show a systematic increase in SCI with alteration and cementation, from low background values to higher values in the silicified fault core. The range of mean SCI in fault damage zone material is smaller (51.6-76.0). Plots of stress intensity versus velocity (K-V) show (1) fracture propagation at lower stress intensities in weaker material and (2) steeper K-V curves in more altered, fault proximal material with greater microstructural complexity (Figure 11). Calculated fracture growth velocities (equations (2) and (5)) are generally between 10 −5 and 10 −7 m/s. This range in fracture propagation velocity is similar to values calculated for rocks under ambient conditions by other researchers (Atkinson, 1979a;Chen et al., 2017;Nara & Kaneko, 2005;Swanson, 1984;Wilkins, 1980) and is in part limited by sampling rate and higher signal-tonoise ratios at slower propagation velocities. Figure 10. Relationship between K IC and K IC *. Mean K IC and K IC * from rapid and slow displacement tests, respectively, are within error of one another internally and fall along a 1:1 line as a group (Tables 5 and 6). Samples identified by numbers. Sample 13 was not tested with rapid loading, and Figures 11-13 show K IC * for this sample. Sample 4 was not tested with slow loading. Error bars are standard deviations of n samples.

Mechanical Properties in Altered Versus Pristine Rocks
We observed both increases and decreases in K IC , SCI, UCS, and elastic properties in damaged, altered, and cemented fault zone material. Mean K IC in relatively unaltered granodiorite from the Box Canyons is similar to reported values of K IC from Westerly granite (1.74 MPa√m, Atkinson, 1984;1.79 ± 0.02 MPa√m, Meredith & Atkinson, 1985;1.43 ± 0.05 MPa√m, Nasseri et al., 2009). SCI is similar to reported values for Lac du Bonnet granodiorite (55.9, 58.5, Wilkins, 1980;56, Lajtai & Bielus, 1986) and Westerly granite (35.9-39, Atkinson, 1982;69, Swanson, 1984). However, the most altered and damaged granitic and gabbroic rocks in our data set show a significant reduction in K IC (<0.7 MPa√m), and silicification resulted in significant increases in mean K IC , SCI, and UCS (up to 3.2 MPa√m, 144.7, and 286.5 MPa, respectively). Figure 11. Structural setting, alteration, K IC , SCI, and representative K-V curves for each site. Box plots for K IC and SCI show the complete range of DT test results (Tables 5 and 7). Alteration increases from left to right at each site, with increasing alteration represented by heavier K-V curves and warmer colors. Labels show sample IDs discussed in the text. K IC * is plotted for sample 13. Alteration and unhealed damage at the Box Canyons (BC) result in a decrease in toughness, and a subsequent shift left of the K-V curves. Healing by silicification (±calcite) at the other sites results in an increase in toughness and SCI values in and near the fault core.

Journal of Geophysical Research: Solid Earth
Silicification at the Dixie Comstock site corresponds to an approximately 600% increase in strength over the weakest footwall samples.
The observed fracture mechanical values in altered and damaged samples are within ranges previous measured in rocks in general (Atkinson, 1984;Swanson, 1984), but rock type is a poor predictor of these properties. K IC in silicified material is similar to both "black gabbro" (2.88 MPa√m, Atkinson, 1984;2.9 MPa√m, Atkinson, 1982;2.71-3.03 MPa√m, Meredith & Atkinson, 1985) and quartzite (2.1-2.65 MPa√m, Atkinson, 1984). Novaculite, which is mineralogically similar to the most silicified samples, has less than half of the strength (K IC of 1.335 ± 0.075 MPa√m) and lower SCI (25.1;Atkinson, 1980). These findings suggest that experimental analogs based solely on rock type or mineralogy are likely to be inadequate.
Instead, the reduction in toughness we observed in altered and damaged fault proximal samples is comparable to strength reductions measured in other naturally and experimentally damaged material. Meredith and Atkinson (1985) and Nasseri et al. (2009) measured reduced K IC in thermally treated Westerly granite containing experimentally induced microfractures. Siratovich et al. (2014) observed a factor of 4 reduction in UCS with a tripling of connected porosity in volcanic rocks from the Taupo hydrothermal field, New Zealand. Pola et al. (2014) reported a 45-50% reduction in UCS and Young's modulus in altered and weathered samples from dissolution-dominated volcanic environments. Heap et al. (2015) reported a correlation between increasing connected porosity and reduced compressive strength for lava and tuff with different degrees of advanced argillic (acid sulfate) alteration from the White Island volcanically hosted hydrothermal system, New Zealand.
Despite the occurrence of some degree of damage and alteration in most fault systems, systematic investigations of the impact of alteration on fracture mechanical properties are limited. Atkinson (1984) included a reference to dunite and serpentinized dunite, with serpentinization resulting in a >50% reduction in K IC , from 3.74 to 1.39 MPa√m. Kobayashi et al. (1986) reported low K IC in Ogino tuff (~0.7 MPa√m); the smectite and zeolite altered tuff samples they tested had a lower density, lower UCS, and lower Young's modulus than the material that we tested, possibly reflecting a greater degree of argillic alteration. Major et al. (2018) reported bleached sandstone from the Crystal Geyser system, Utah, USA, has a lower K IC than adjacent hematite-cemented sandstone. Figure 12. Relationship between K IC and UCS. Positive correlation between mean K IC and mean UCS suggests that microfracture growth is an underlying failure mechanism in UCS tests. Labels show samples IDs discussed in the text. Errors bars show standard deviation of n samples. In the materials that we tested, filling of 85-90% of fracture porosity with mineral cement is associated with a significant increase in all measures of toughness (UCS, K IC , SCI; Figure 13). The observation that damaged rocks can regain significant strength and resistance to fracture growth is consistent with other researchers that observed strengthening resulting from mineral precipitation (quartz, Yasuhara, 2005;alunite, del Potro &Hürlimann, 2009, andHeap et al., 2015; phyllosilicate and calcite, Boulton et al., 2017;hematite and calcite, Major et al., 2018). Our sample base is too small to address the differences in strength recovery resulting from specific fracture cement composition, texture, or distribution. However, the addition of fracture-filling mineral cements in particular hydrothermal settings will clearly impact the distribution of rock strength in hydrothermal systems.

Implications for Strength Distribution in Fault-Hosted Hydrothermal Systems
Just as hydrothermal systems commonly contain systematic spatial variations in alteration products related to temperature, fluid chemistry, and fluid-rock ratios (Nishimoto & Yoshida, 2010;Sillitoe, 2010;Simmons et al., 2005;Tosdal et al., 2009), the spatial distribution of mechanical properties in and around fault-fracture conduits is expected to change with the dominant alteration mechanism and from competition between the accumulation of damage versus healing. Where precipitation of strong minerals outpaces deformation, fault rocks will experience interseismic strengthening. Conversely, regions with a lower rate of mineral precipitation relative to deformation will tend to undergo progressive weakening. A comparison of the Dixie Comstock and Box Canyons results suggests a depth-dependent inversion between the mechanical properties of fault rocks and host rocks in fault-hosted hydrothermal systems ( Figure 14). In the shallow, precipitation-dominated epithermal setting at Dixie Comstock, samples from the thick, silicified fault core are up to 6 times stronger than chloritized footwall samples and more resistant to fracture growth. The exhumed chlorite-calcite-hematite assemblage preserved at the Box Canyons shows a reduction in UCS and K IC near the fault. The reduction in fracture mechanical properties with depth, in particular, could be exacerbated by increasing temperature and changes in fluid chemistry (Atkinson, 1979a;Atkinson & Meredith, 1981;Balme et al., 2004;Funatsu et al., 2004;Karfakis & Akram, 1993;Meredith & Atkinson, 1985;Nara et al., 2013Nara et al., , 2014Nara et al., , 2017Rostom et al., 2012). However, chemically aided fracture growth is sensitive to rock composition, and has not been thoroughly characterized in similarly altered fault rocks.
The distribution of mechanical properties with depth in hydrothermal systems may have implications for fault zone architecture and hydraulic properties. In portions of hydrothermal systems dominated by dissolution and the accumulation of unhealed damage, we expect deformation to become increasingly localized. In contrast, rapid advection, cooling, and enhanced precipitation in shallow portions of fault-hosted hydrothermal systems, where conditions approach the boiling point with depth curve (Figures 2 and 14), may inhibit fracture growth in the primary conduits and promote fracture growth in adjacent material, ultimately contributing to large volumes of fractured rock and distributed alteration in shallow portions of these systems.  Measures of strength and resistance to fracture propagation (UCS, K IC , SCI) are generally lower in the fault core than in adjacent rock, but in shallow, precipitation dominated hydrothermal regimes cementation of the fault core by silicification (±calcification) increases rock strength relative to less cemented damage zone and protolith. Increased dissolution in the nearsurface acid sulfate environment is associated with a loss of strength.

10.1029/2018JB016708
Journal of Geophysical Research: Solid Earth 6.3. Broader Impacts of Alteration, Damage, and Healing Hydrothermal alteration and the preservation or healing of damage impacts several other geologic systems of particular interest to society, including mineral deposits, volcano-magmatic systems, and fault systems more generally. Moir et al. (2013) showed that dilatant damage around mineralized fault zones could be related to contrasting mechanical properties in different host lithologies, a correlation which was also observed around the Alpine Fault in New Zealand (Williams et al., 2016). Our results suggest that the distribution of fracture strength parameters, fracture growth, and fracture permeability may be heterogeneous in both space and time around mineralizing faults due to mechanical changes resulting from hydrothermal alteration. Rock weakening caused by acid sulfate and argillic alteration in volcanic edifices is linked to flank collapse (del Potro & Hürlimann, 2009;Lopez & Williams, 1993;Reid et al., 2001;Watters et al., 2000). However, cementation by quartz or alunite in these environments may result in local densification and strengthening, with physical and mechanical contrast between different alteration products potentially influencing where flank collapse ultimately occurs. Opening-mode fractures are fundamental parts of fault initiation in crystalline rock (Crider, 2015) and persistent elements in damage zones in mature fault systems (Davatzes et al., 2005;Wilson et al., 2003). Subcritical fracture growth in damage zones has been implemented in preseismic and postseismic behavior of fault systems (Anderson & Grew, 1977;Brantut et al., 2013;Rudnicki, 1980). However, because some degree of hydrothermal alteration is common in large seismogenic faults (e.g., Bruhn et al., 1994;Parry et al., 1991), geomechanical tests of pristine material alone may not completely characterize the fracture mechanical properties of fault zones. The existence of a healing threshold, where sealing of 85-90% fractures results in a significant increase in material strength in our samples, has been documented in other fault settings (e.g., Williams et al., 2016) and may represent a useful approximation of strength recovery in faults and hydrothermal systems.

Conclusions
We combined unconfined compressive strength tests, mineralogical and textural characterization, and double-torsion load-relaxation fracture mechanics tests to measure strength, elastic, and fracture mechanical properties in exhumed suites of rocks from different hydrothermal alteration regimes preserved in the footwall of the Dixie Valley-Stillwater fault zone. The alteration regimes include (1) a shallow acid sulfate regime dominated by quartz and calcite precipitation; (2) an epithermal regime dominated by intense silicification after earlier sodic and calcic alteration; (3) a moderate depth and temperature silicic-argillic regime dominated by quartz, kaolinite, and carbonates; and (4) a retrograde alteration regime dominated by chlorite-calcite-hematite assemblages and unhealed damage. Based on the alteration assemblages that we tested, the mechanical contrast between fault core and host rocks changes between precipitation-versus damage-dominated regimes. Compared to minimally altered granodiorite, silicified fault rocks are stronger and more resistant to fracture growth, whereas fault damage zone samples containing chlorite-calcitehematite alteration and abundant unhealed fractures are one third to one fifth as strong. Sealing of >85-90% of microfracture porosity by quartz and/or calcite in the fault zone in the epithermal environment is associated with a significant increase in UCS, K IC , and SCI above values of altered protolith in the footwall. The mechanical properties that we measured in altered, damaged, and healed fault zone samples are not readily predicted from geomechanical tests of pristine rocks.