Groundwater silcrete linked to brine migration in a continental rift: An alternative to the near-surface model of silcrete

Silcretes are indurated layers of silica-cemented soil, sediment, or rock formed by near-surface silica precipitation (<100 m depth) related to weathering processes, groundwater flow, and diagenesis (Summerfield, 1983; Milnes and Thiry, 1992; Nash and Ullyott, 2007). Layers are well-indurated and consist of ~85–95 wt% SiO₂ in the form of overgrowths on detrital quartz, pore-filling cement, and/or replacement of clays and carbonate (Summerfield, 1983; Shaw and Nash, 1998; Nash et al., 2004). Two types of silcrete are recognized: pedogenic silcretes, which formed at and directly below subaerially exposed surfaces, and groundwater silcretes, which formed at or just below the top of the water table (Thiry et al., 1988; Milnes and Thiry, 1992; Nash and Ullyott, 2007). Pedogenic and groundwater silcrete commonly co-occur in sedimentary successions with similar microfabrics and mineralogy. However, pedogenic silcrete is differentiated from groundwater silcrete by the disruption of primary sedimentary fabrics and development of vertical soil profile organization, jointing, and peds (Thiry and Milnes, 1991; Milnes and Thiry, 1992; Nash and Ullyott, 2007).

The origin of pedogenic silcrete is relatively well understood from numerous documented examples in both modern and ancient environments. Under acidic, near-surface groundwater conditions, silica is liberated by the dissolution of clays, feldspars, and other silicates and then transported in pore fluids aided by eluviation-illuviation in the vadose zone. Silica precipitation is then driven by supersaturation promoted by evaporation. Silica precipitates in intergranular space or replaces detrital carbonate, clay, or evaporites (Summerfield, 1983; Khalaf, 1988; Thiry, 1999; Webb and Golding, 1998; Kendrick and Graham, 2004; Nash et al., 2004; Nash and Ullyott, 2007). In contrast, the origin of groundwater silcrete is much less well understood. Like pedogenic silcrete, silica dissolution is thought to be facilitated by near-surface weathering of silicates, followed by the substantial downward and lateral translocation of silica through the vadose zone to the water table. The exact chemical and physical processes leading to silica precipitation are poorly understood, but evaporation, water table fluctuations, and outflow along springs are considered key controls (Thiry et al., 1988; Shaw and Nash, 1998; Nash and Ullyott, 2007). Mass balance calculations by Thiry et al. (1988) suggest that silcrete precipitation could occur in meteoritic water with relatively low silica concentrations (~6 ppm; Morey et al., 1962) at realistic rates of groundwater flux over time periods of thousands of years. Nevertheless, this requires a remarkably stable near-surface groundwater transport system wherein sustained and balanced silicate weathering and groundwater recharge rates coincide with sustained local conditions at the water table that act to saturate and precipitate dissolved silica.

Notably, most investigations do not consider exogenous crustal or deep basin silica sources like high temperature, high pH, silica-bearing fluids with potential silica concentrations of ~100–600 ppm (Morey et al., 1962; Dove and Rimstidt, 1994), which could dramatically increase silica flux leading to supersaturation at the water table. Similarly, direct proxies for fluid composition or temperature, like fluid inclusion thermometry, have not been used to constrain the origins of silicifying fluids in groundwater silcretes. Also, no model has yet demonstrated evidence for the role of faults in the transport of silica, despite the fact that brine and hydrothermal fluid migration along faults is well documented in modern basins (e.g., Warner et al., 2012; Hollinsworth et al., 2019) and ancient rifts (Lampe et al., 2002; Hollis et al., 2017).

Here, we test near-surface groundwater silcrete models with field- and laboratory-based observations of an ~8–142-cm-thick, silicified horizon in Cambrian strata of the Potsdam Group, deposited over the Neoproterozoic St. Lawrence Rift system near and within the Ottawa Graben. This silicified horizon, interpreted by Selleck (1978) as a pedogenic silcrete, shows a progressive thickening, silicification, and brecciation toward numerous intrabasinal faults underlain by Mesoproterozoic shear zones. These field relationships, described in detail below, suggest a role of faults as conduits for silicifying fluids. Notably also, fluid migration along faults leading to the diagenetic precipitation of apatite, calcite, sulphides, hematite, and monazite in Potsdam strata have been documented in nearby parts of the St. Lawrence Rift along the northern margin of the Adirondacks (Selleck, 2005; Allaz et al., 2013; Gall et al., 2017). To test this hypothesis, the paragenesis of silica cements and thermometry and petrography of their fluid inclusions are documented to constrain the potential sources and migration pathways of the silica-bearing fluids for this silcrete.

The Cambrian to Lower Ordovician Potsdam Group is a composite siliciclastic unit unconformably overlying ca. 1 Ga Grenville Province rocks across the Ottawa Embayment, a fault-bounded basin of flat-lying Paleozoic strata that fills the southern branches of the St. Lawrence Rift system, including the Ottawa Graben (Kumarapeli and Saull, 1966; Fig. 1). Ediacaran (590–577 Ma) dyke swarms and related alkaline intrusions constrain the timing of initial rifting of the St. Lawrence Rift system (Kamo et al., 1995; Cawood et al., 2001; Allen et al., 2010; McCausland et al., 2011) that culminated with seafloor spreading of the proto-Iapetus by 554 Ma (Kumarapeli et al., 1989). Faults in the southern St. Lawrence Rift and Ottawa Graben were subsequently reactivated repeatedly from the Paleozoic to the Cenozoic (Crough, 1981; McHone and Butler, 1984; Hogarth et al., 1988; Rimando and Benn, 2005; Salad Hersi and Dix, 2006), including during Potsdam sedimentation (Sanford and Arnott, 2010; Lowe et al., 2018).

Recent stratigraphic and provenance investigations of the Potsdam Group across the Ottawa Graben reveal a complex internal stratigraphy consisting of three unconformity-bound allounits (Allounits 1–3; Fig. 2). These record successive episodes of terrestrial and shallow marine sedimentation punctuated by erosion and unconformity development. Notably, provenance and stratigraphic data suggest these breaks in sedimentation coincided with local fault reactivation that modified basin morphology and sediment routing throughout the Ottawa Graben (Sanford and Arnott, 2010; Lowe et al., 2018). The silcrete horizon documented here caps strata of the middle allounit, termed Allounit 2, and consists of Upper Cambrian braided and ephemeral fluvial quartz arenite and quartzose conglomerate (Figs. 2–4). Based on its stratigraphic position, formation of this silcrete coincides with terminal Cambrian basin inversion and unconformity development in the Ottawa Graben (Sanford and Arnott, 2010; Lowe and Arnott, 2016; Lowe et al., 2017).

The silcrete is confined to an area south of the St. Lawrence River in northwestern New York State, USA, where it crops out along the margins of the Frontenac Arch, a linear topographical high of Grenville basement striking northwest–southeast, and separates the Ottawa Embayment from the Appalachian Basin (Figs. 1 and 3). Here, paleogeographic reconstructions of Allounit 2 strata (Lewis, 1963; Otvos, 1966; Wolf and Dalrymple, 1984; Lowe and Arnott, 2016) and detrital zircon assemblages (Lowe et al., 2018) reveal sourcing from the east and sedimentation by northwest- to southwest-flowing braided and ephemeral streams confined between northeast-trending, fault-bounded ridges (Fig. 3). The influence of these fault-bounded ridges on Late Cambrian fluvial sedimentation is indicated by local paleoflow oriented sub-parallel to faults, boulder talus and debris flow deposits adjacent to faults (Figs. 3–4), soft sediment deformation, and thick dune and bar cross-sets that record deep flow depths in laterally confined floodplains (Lowe and Arnott, 2016).

The silcrete horizon that caps quartz arenite strata of Allounit 2 crops out over an area of ~0.3 km² in northwestern New York State along the Frontenac Arch (Figs. 1 and 4–5). Here, it cross-cuts a conformable facies boundary between ephemeral and braided fluvial strata interpreted to record a change in climate from arid to humid conditions (Lowe and Arnott, 2016; Fig. 4). The horizon is 8–142 cm thick, with the thickest parts (> 130 cm) located north of two prominent northeast-trending faults: Black Lake and Wellesley Island faults (Fig. 5A). Moreover, most of the exposed silcrete occurs in the fault block between the Black Lake and Chippewa Bay faults (Fig. 5A) and overlies a major shear zone in the Grenville Province termed the Black Lake Shear Zone (Figs. 3 and 5).

Silcrete texture varies across this area and includes nodular, massive, locally brecciated, and pervasively brecciated forms (Fig. 5B). Nodular forms are generally limited to the northwestern edge of the silcrete horizon north of the Wellesley Island Fault in a series of north–south-oriented road cuts over an east–west strike length of ~1900 m (Fig. 5B). Here, quartzose nodules form a 40–120-cm-thick horizon at the top of Allounit 2 strata with abrupt lateral variations in thickness. Nodules are 4–14 cm in diameter and rounded with either a spherical or bedding-parallel elliptical shape (Fig. 6). Primary stratification including planar and cross-laminae are preserved in nodules. Nodule abundance increases upwards from ~10% near the base to 80% at the top of the horizon, where nodules coalesce to form agglomerated nodule clusters (Figs. 6A–6B). The outer surfaces of nodules are commonly highlighted by a 0.1–10 mm rind of poorly indurated iron oxide and clay-cemented sand (Figs. 6A–6B). The thickness of these rinds increases upward in the horizon as the concentration of nodules increases, resulting in a patchwork of well-cemented nodules surrounded by a moderately cemented clay- and iron oxide–rich sand in the upper ~20–30 cm of nodular horizons (Fig. 6A).

Massive silcrete is the most widespread form, cropping out extensively between the Chippewa Bay and Black Lake faults (Fig. 5B). It is characterized by an 8–105-cm-thick, silicified horizon with a sharp, undulating base with up to 15 cm vertical relief (Fig. 7). Primary depositional structures like cross-stratification and planar lamination are preserved within the silcrete. Locally, the silcrete horizon consists of two parts; a basal ~50–90 cm layer of massive silcrete is overlain sharply by a 6–13-cm-thick, more indurated layer, which is termed here the capping layer (Fig. 7).

Locally the massive silcrete horizon is brecciated, especially within ~1 km of the Black Lake and Chippewa Bay faults and exposures in the southwest near the Indian River fault (Fig. 5B). Here, vertical fractures extend upward into the base of the massive silcrete, forming a complex conical network of fractures and breccia that abruptly splays out to a maximum width of ~110 cm, here termed conical breccia (Fig. 8). These features become more abundant toward the northwest in the fault block between the Chippewa Bay and Black Lake faults (Figs. 4 and 5B). At all locations, breccia cones exhibit a crude concentric zoning (Fig. 8), which from the margin inward consists of three distinct zones. Along the margins is a 3–10-cm-thick, matrix-rich breccia (mBx) consisting of 0.3–6 cm, angular to subangular, massive silcrete clasts in a 10–65% matrix of massive, distinctively dark gray weathered sandstone (Fig. 8). This zone transitions sharply to a 3–12-cm-thick layer of jigsaw-breccia (jBx) consisting of 0.3–10 cm angular, interlocking, in situ silcrete clasts with ≤10% dark-weathered sandstone matrix filling a dense network of fractures around the clasts (Fig. 8). The silcrete clasts lack an identifiable preferred orientation, and many appear to have collapsed downward into the matrix (Fig. 8). This breccia is succeeded inward and upward by a 5–80-cm-wide core of massive but locally fractured silcrete (mS; Fig. 8). Most breccia cones do not penetrate the overlying massive silcrete but instead are deflected along its base to form a 2–6-cm-thick, matrix-rich breccia horizon that extends horizontally for 15–75 cm before either terminating gradationally into jigsaw breccia and massive silcrete or connecting to a 5–20-cm-diameter vertical breccia pipe truncated along the Allounit 2–3 contact.

More pervasive brecciation of the silcrete is limited to the northwesternmost sections between the Chippewa Bay and Wellesley Island faults (Fig. 5B). Here, the silcrete horizon is comparatively thin—only 15–23 cm thick—with pervasive brecciation that is laterally continuous over at least 1.2 km (Figs. 3 and 5B). The pervasively brecciated horizon consists internally of alternating sheets of chaotic, matrix-rich breccia (mBx, 0.2–0.7 m wide); interlocking-jigsaw, matrix-poor breccia (jBx, 0.3–1.1 m wide); and rare 0.3–1-m-wide, massive silcrete blocks (Fig. 9). The jigsaw and matrix-rich breccia are like those in breccia cones of thicker silcrete horizons to the south, except here they consist of angular clasts that range from 0.5 cm to 16 cm wide and subangular clasts that range from 0.5 cm to 6 cm wide. These alternating breccia sheets are locally inclined ~15° to the northwest (Fig. 9).

Standard optical petrography was undertaken using a polarizing microscope to determine silcrete compositions and textures. Optical photomicrographs of silcrete samples were used to determine the apparent intergranular volume and %-cement using ImageJ, an open-source image editing software. Framework grains were traced and filled in, black and color thresholds were adjusted to eliminate grains using the Global Threshold function, and residual intergranular volume was calculated using the Analyze Particles function. Percentages of other framework and matrix components were estimated visually.

Backscattered electron (BSE) imaging of thin sections was undertaken using a JEOL JSM-7100F scanning electron microscope (SEM) at Memorial University of Newfoundland, Canada, equipped with a Deben Centaurus CL detector for CL imaging and a ThermoFisher energy dispersive spectrometer (EDS) for elemental analysis. Fine-grained textures and mineralogical composition of detrital and authigenic accessory phases were determined using BSE imaging and EDS based on their electron densities and characteristic EDS spectra.

Panchromatic cathodoluminescence (CL) imaging of polished thin sections was undertaken using the JEOL JSM-7100F SEM at Memorial University of Newfoundland equipped with a Deben Centaurus CL detector, under an accelerating voltage of 15kV and a working distance of 13 mm. Based on the detection of visible and ultraviolet radiation (~300–700 nm), CL was used to reveal growth patterns in syntaxial quartz overgrowths not visible using optical or BSE imaging. In quartz, crystal lattice defects and/or impurities (e.g., substitution of Ti⁴⁺ or Ga⁴⁺ for Si⁴⁺) can result in CL as excited electrons in the crystal lattice return to their ground states (Frelinger et al., 2015).

Following CL imaging, five representative samples were selected to identify fluid inclusions and constrain the salinity and homogenization temperatures of fluids trapped in C1. Fluid inclusions were not found in C1.5 or C2. Double polished fluid inclusion wafers (~100 μm thick) were prepared from these samples, and petrographic description of fluid inclusions was carried out using the fluid inclusion assemblage (FIA) method and the criteria outlined in Goldstein and Reynolds (1994) and Goldstein (2003). Care was taken to focus on primary FIA trapped during cement precipitation. Numerous FIAs were also observed in detrital quartz grains but were not included in this study since they represent inherited fluid signatures.

Microthermometric fluid-inclusion analyses at Memorial University of Newfoundland used a Linkam THMSG600 heating–freezing stage mounted on an Olympus BX51 microscope. The heating–freezing stage was calibrated using synthetic H₂O and CO₂-H₂O fluid inclusion standards, with calibration measurements at the melting point of CO₂ (−56.6 °C ± 0.1 °C), melting point of H₂O (0.0 °C ± 0.1 °C), and critical point of H₂O (374.1 °C ± 0.5 °C). Following procedures outlined by Shepherd et al. (1985), the temperature of homogenization (T_h), eutectic melting temperature (T_e), hydrohalite melting temperature (T_m (hh)), and final ice-melting temperature (T_m (ice)) were measured in two-phase (liquid + vapor) inclusions. For inclusions where T_m (hh) and T_m (ice) were observed, fluid salinities (eq. wt% NaCl + CaCl₂) and molar Na/(Ca + Na) ratios were calculated using the Excel spreadsheet of Steele-MacInnis et al. (2011) for the H₂O-NaCl-CaCl₂ system and reported in eq. wt% NaCl+CaCl₂. Salinities were also calculated using T_m (ice) and the equation of Bodnar (1993), which is used to calculate salinities as equivalent wt% NaCl (eq. wt% NaCl). Homogenization temperatures were recorded from four to six fluid inclusions in each FIA, but care was taken to ensure that all inclusions in FIA homogenized over the same temperature range to record any evidence of post-entrapment reequilibrium or modification (e.g., necking and stretching; Fall and Bodnar, 2018).

Silcrete nodules consist of well-sorted, medium- to coarse-grained, subrounded quartz grains cemented by 10–50-μm-thick syntaxial quartz overgrowths (Fig. 10A). These overgrowths fill nearly all of the intergranular volume, which ranges from 28% to 35% of the apparent sample volume (31.5% average) with only ~2–6% porosity (Fig. 11). Point contacts dominate, and broad or sutured grain contacts are absent (Fig. 10A). Framework mineralogy is dominated by quartz that constitutes 97% to >99% of the detrital grains. Rare detrital potassium feldspar, partly altered to kaolinite, makes up to 2% percent of the framework in some samples. Accessory detrital phases include fine- to medium-grained tourmaline and fine- to very fine-grained zircon and rutile. Accessory authigenic phases associated with quartz overgrowths rimming grain boundaries include rare pyrite, illite, and ilmenite grains.

The rind that envelops silcrete nodules (Figs. 6 and 10) varies in thickness from 0.2 mm to 3 mm near the top of the nodules to 10 µm to 200 µm near and at the base of the nodules. The thicker rind near the top of the nodules consists of sand grains dispersed in a pore-filling matrix of kaolinite, iron oxide, and quartz silt and exhibits high intergranular volume of ~25–30% (Figs. 10B and 11). In contrast, coatings near the base are 10–20-µm-thick stylolites with evidence of significant pressure dissolution of adjacent quartz grains and consist of very fine-grained TiO₂ and silt-sized apatite and zircon (Fig. 10C). The inter-nodular sandstone surrounding the nodules has the same detrital composition as the nodules, but it has lower intergranular volume and contains ~2–10% interstitial kaolinite and iron oxide, and ~5% quartz overgrowths, some of which fill large voids (Fig. 10D).

Samples from the massive horizon are compositionally similar to the interior of nodules and consist of rounded detrital quartz grains cemented by syntaxial quartz overgrowths. Apparent intergranular volume is the same as that in nodules, ranging between 28% and 35%, (Fig. 12A). Sutured contacts are absent, but point contacts are common and some broad grain contacts are present. The framework and accessory detrital and authigenic phases in the massive horizon are the same as in the nodules described above except that feldspar is absent.

Several samples were selected from underlying and adjacent exposures of Allounit 2 sandstone to establish a baseline for host rock composition and texture compared with the silcrete. These samples are quartz arenite with quartz grains making up ~97% to >99% of the framework mineralogy (Fig. 12B). Feldspar is rare, and where present it makes up only ~2–3% of the framework mineralogy and is partly altered to clay minerals. These samples have an intergranular volume of 14–25%, which is filled with a mixture of ~5–20 μm quartz overgrowths, ~2–10% interstitial kaolinite and clay-sized iron oxide, rare carbonate cement, and ~1–3% porosity. Point and sutured grain contacts are common. Notably, quartz grains are also commonly fractured, unlike in the silcrete (Fig. 12B).

Several samples were taken from conical breccia at the boundary between the jigsaw and matrix-rich breccia zones. These samples reveal clasts that are mineralogically and texturally identical to samples of the massive silcrete, which confirms their origin as intraformational clasts. The outer boundaries of the clasts tend to be broken along grain boundaries, but some sharply truncate grains and C1 overgrowths (Fig. 12C). The matrix consists of medium-grained sandstone with a moderately well-sorted framework of quartz grains surrounded by ~10–30% clay-sized iron oxide and kaolinite and admixed quartz and heavy mineral silt that is similar to the top part of the rinds that coat the nodules described above (Fig. 12D). Point and sutured contacts are common, and quartz overgrowths are rare. Compared to the grains in the adjacent massive silcrete clasts, grains in the breccia matrix are subangular, and some have jagged edges that indicate breakage and/or dissolution. Internal fractures in grains are common.

CL imaging of samples from silcrete nodules and the massive silcrete revealed up to three generations of optically continuous quartz overgrowths termed C1, C1.5, and C2. These generations were identified based on luminosity domains that outlined early zoned luminous grain-rimming overgrowths (C1), an intermediate luminous void-filling cement (C1.5), and a late non-luminous, void-filling cement (C2). Both C1 and C1.5 cements luminesce but with a clear distinction in their relative timing, morphology, thickness, and zoning. C1 is a grain-coating cement consisting of ~10–40-μm-thick, euhedral or subhedral quartz overgrowths and occurs in all samples of nodules and massive silcrete. It exhibits concentric zoning consisting of ~2–10-μm-thick alternating zones of high and low luminosity (Figs. 13A–13C). Though similarly luminous and optically continuous, C1.5 discordantly succeeds C1 in CL, filling most of the remaining interstices and occluding spaces up to 120 μm in diameter (Fig. 13C). Where present, it is significant, making up 15–25% of the sample volume. Unlike C1, C1.5 is entirely euhedral and exhibits sector zoning consisting of 10–40-μm-thick trapezoidal zones of varying luminosity (Fig. 13C). Also, C1.5 is not ubiquitous like C1; it is only recognized in one nodule sample and one sample from the massive horizon, and in the less silicified sandstone surrounding nodules in the nodular silcrete horizon. Here, C1.5 fills larger voids not already occluded by clay and iron oxide (Fig. 10D). C2 is a non-luminescing euhedral quartz overgrowth cement that postdates optically continuous C1 and C1.5, filling any remaining void space (Figs. 13A–13C). Where C1.5 is absent, C2 is a major constituent of the silcrete, filling almost all of the intergranular volume and comprising up to 25% of the sample volume.

Several samples of matrix breccia were also imaged using CL. This revealed that many of the subangular quartz grains were rimmed or partly rimmed by C1 cement. The C1 cement appears to be locally truncated by corroded margins, which suggests that a phase of silica disaggregation and dissolution occurred during or after brecciation (Fig. 13D).

In all samples, FIA are common in C1 quartz overgrowths within 10–45 µm of the margins of detrital quartz grains, which indicates that fluids were trapped during the precipitation of C1 cement. Away from the grain boundaries, C1.5 and C2 are typically clear with no observable FIA. Fluid inclusions in C1 cement have remarkably similar petrographic and microthermometric characteristics in all samples. At room temperature, fluid inclusions in C1 cements are two-phase (liquid + vapor) and rich in liquid (~5–15% vapor). They are generally small (< 5 μm), but rare inclusions of up to 20 μm wide are also observed (Fig. 14).

Microthermometric analyses were obtained from 58 primary fluid inclusions from 15 FIAs (Figs. 15–16, Table A2). Upon cooling to −100 °C, inclusions freeze at ~−70 °C to −80 °C, forming a brown granular ice, which is consistent with high-salinity fluids (> 20 eq. wt% NaCl; Wilkinson, 2017). Some low-temperature metastable phase transition was observed, but first melting (T_e) was observed close to the eutectic temperature of the H₂O-NaCl-CaCl₂ system (−52 °C; Fig. 16). T_m (hh) occurred prior to ice melting between −40.1 °C and −36.8 °C and final ice melting (T_m (ice)) between −31.2 °C and −23.2 °C. Where T_m (hh) and T_m (ice) were observed in the same inclusions, salinities of 22.7–25.8 eq. wt% NaCl+CaCl₂ were calculated with the Na/(Na + Ca) ratio and ranged from 0.11 to 0.15. For inclusions where only T_m (ice) was observed, calculated salinities range from 24.5 eq. wt% to 29.4 eq. wt% NaCl. Note, however, that salinity estimates derived using T_m (ice) and the equations of Bodnar (1993) likely overestimate true salinities by ~2–3 wt%. Upon further heating, total homogenization to the liquid phase occurred between 120.2 °C and 151.6 °C. Homogenization temperatures of all inclusions in individual FIAs was generally < 5 °C (Fig. 15), which is consistent with little or no post-entrapment reequilibrium of the inclusions (Fall and Bodnar, 2018). However, some FIAs with higher T_h values show a wider range of temperatures (up to 10 °C), which indicates that necking or stretching of inclusions may have occurred in these FIAs.

Optical and CL petrography reveal a clear sequence of overgrowth genesis, with early grain-coating C1 overgrowths succeeded locally by C1.5 and everywhere by late-stage, pore-filling C2 (Figs. 13 and 17). Notably, the higher intergranular volume of C1- and C1.5-cemented silcrete (29–35%) compared to that of the underlying and regional baseline Allounit 2 strata (~22%) suggests that C1 and C1.5 pre-date more widespread porosity loss in the Allounit 2 strata. Moreover, silcrete lacks the abundant fractured grains and sutured grain contacts present in Allounit 2 samples that indicate widespread mechanical compaction and pressure solution. C1 and C1.5 overgrowths, therefore, represent a relatively early near-surface cement that precipitated after only minor mechanical compaction coinciding with a loss of ≤ 6% porosity from a probable initial porosity of ca. 35% (e.g., Pryor, 1973; Selley, 1978). Early C1 overgrowths likely provided local structural rigidity to the sandstone framework that counteracted the effects of compaction during burial, including pressure solution. This early timing is also demonstrated by the basal contacts of C1-cemented nodules, which are marked by localized stylolites that are lined with residual insoluble TiO₂, suggesting that nodules formed prior to compaction and pressure solution (Figs. 10 and 17).

Non-luminescing C2 overgrowths are present in all samples. C2 overgrowths fill much of the remaining intergranular volume in nodules and the massive horizon and are present as a grain-rimming overgrowth cement throughout baseline Allounit 2 samples. C2 was likely coeval with or post-dated pressure solution since it is the only overgrowth cement in compacted baseline Allounit 2 sandstone, where evidence of pressure solution is widespread. Moreover, its non-luminescing character suggests that it is either purer or better structured than C1 and C1.5 (Frelinger et al., 2015), which suggests that it precipitated slowly from the direct dissolution of detrital quartz in a relatively stable subsurface diagenetic environment.

Minor (2–10%) interstitial clay- to silt-sized fines, mainly kaolinite with iron oxide and fine quartz silt, occur in the intergranular volume of baseline Allounit 2 strata and the C1.5-cemented, inter-nodular silcrete in the nodular horizon but are absent from C1-cemented silcrete. Locally, these fines form geopetal accumulations draping the tops of nodules, where they preserve an open pre-compaction, intergranular volume locally between grains (Fig. 10). This, and their absence in C1-cemented silcrete, suggests that their emplacement post-dates C1 but pre-dates compaction. Based on their geopetal form, absence in early C1-cemented silcrete, and lack of association with weathered detrital phases, these fines are interpreted to have infiltrated unconsolidated Allounit 2 strata by eluviation through the vadose zone. These fines were most likely derived from weathering and the erosion of granite in adjacent exposed Grenville basement, given the paucity of feldspar or other silicates in Allounit 2 that could have sourced these fines as residuum of detrital mineral weathering.

Field and petrographic relationships demonstrate that fracturing and associated brecciation post-dates initial silcrete formation and predates Allounit 3 sedimentation. It also truncates, and thus post-dates, C1 cementation (Fig. 12C) but has an unknown timing relative to C1.5. Later, interstitial clay and iron oxide in fractures and breccia matrix most likely accumulated during the same period of eluviation–illuviation that affected baseline Allounit 2 strata.

Selleck (1978) interpreted this silcrete horizon as pedogenic, based on its position beneath an unconformity, evidence of subaerial exposure, and localized brecciation. However, Thiry et al. (1988) later demonstrated a second type of silcrete, termed groundwater silcrete, that forms at the water table and is characterized by the preservation of primary stratification and an absence of vertical soil structures. Like groundwater silcrete, the Potsdam silcrete lacks the vertical profile organization, vertical jointing, peds, or glaebules typical of pedogenic silcrete (Nash and Ullyott, 2007) and preserves primary stratification, with brecciation and illuviation postdating silcrete formation. Accordingly, the Potsdam silcrete is interpreted to be a groundwater silcrete that formed along the water table during or after Allounit 2 sedimentation. Following its formation, the silcrete persisted and shielded Allounit 2 succession strata from erosion associated with the Allounit 2–3 unconformity. This is demonstrated in Figure 5, which shows that the areal distribution of the silcrete coincides with the distribution of Allounit 2 strata across the Frontenac Arch.

In existing groundwater silcrete models from the Paris Basin and elsewhere, silica is thought to be sourced from the weathering of detrital feldspar and/or other detrital silicates in the sediment pile or adjacent basement highs (Nash and Ullyott, 2007; see Equation [1] below). However, fluid inclusion results from C1 overgrowths that constrain silica precipitation show that these fluids have minimum trapping temperatures of ~120–150 °C and are therefore well in excess of near-surface meteoric water temperatures. Moreover, these fluids were saline, containing ~25–30% NaCl + CaCl₂, and were enriched with Ca relative to Na (Na/(Na + Ca) of 0.11–0.15). The occurrence of minor pyrite, illite, and K-feldspar associated with C1 overgrowths further suggests that these fluids contained some HS^– and K⁺. Fluid inclusions in C1 cement have remarkably similar petrographic and microthermometric characteristics in all samples and likely represent a single fluid event. Of the potential crustal fluid reservoirs, the homogenization temperatures and Na/(Na/Ca) ratios are uncharacteristic of higher-temperature metamorphic or magmatic-geothermal fluid sources (Yardley and Bodnar, 2014). However, both fluid temperature and chemistry are similar to those of documented crustal brines formed by the equilibration of externally sourced fluids with crystalline basement adjacent to or underlying continental basin sediments (Carpenter et al., 1974; Bodnar et al., 2014; Yardley and Bodnar, 2014). In this case, the most likely origin of crustal brines is meteoric water that infiltrated faults and fractures following Neoproterozoic and/or Early Cambrian rifting. These brittle faults were continuous with deeper structural fluid conduits in the pre-existing Mesoproterozoic Black Lake and Indian River Shear Zones (Figs. 4–5; Wong et al., 2011) and facilitated meteoric water infiltration to deeper levels, where it underwent extensive wall rock equilibration and heating. Considering geothermal gradients of 20 °C/km to 40 °C/km for rifted continental crust (Bertotti et al., 1999; Njinju et al., 2019), the T_h of C1 fluid inclusions suggests that fluids could have infiltrated to a depth of ~3 km, similar to fluids documented in the modern Basin and Range Province of the southwestern U.S. (Haines et al., 2016).

The evolution from mildly acidic meteoric water to high-salinity, silica-bearing brines can be explained by the chemical evolution of meteoric water that infiltrated the local Grenville upper crust, which in this area consists mainly of marble, metapelite, quartzite, granite, and rare anhydrite meta-evaporites (Whelan et al., 1990; Wong et al., 2011). For example, albite and calcite dissolution by infiltrated meteoric fluids is envisaged to have occurred during interaction with granite and marble, respectively, resulting in increased salinity (addition of Na⁺ and Ca²⁺) and pH (addition of bicarbonate).

Following upward migration, a heated, silica-bearing crustal brine entered the water table from adjacent faults, similar to examples in the modern Suez and East African Rifts (Hollis et al., 2017; Hollinsworth et al., 2019; Fig. 18). This likely coincided with a decrease in silica solubility due to a decrease in temperature and pH during re-equilibration with near-surface temperature and atmospheric CO₂ (e.g., Selleck, 1978). Luminescence zoning at the micron-scale occurs in C1 overgrowths in nodules and massive silcrete and suggests temporal variations in compositional and/or kinetic conditions that controlled silica precipitation (Frelinger et al., 2015), which were possibly related to brine pulses and/or cycles of groundwater recharge and evaporation. Moreover, the presence of both nodular and massive silcrete forms suggests local variations in silica precipitation kinematics. The nodules lack evidence of concentric zonation or nucleation and therefore formed by pervasive rather than incremental concentric growth (Raiswell and Fisher, 2000), which suggests that nucleation site availability was not a factor in the formation of nodules. Instead, the formation of nodules is likely related to the immiscibility of the silicifying brine in groundwater. This brine-groundwater immiscibility resulted in localized sub-spherical pockets of interstitial, silica-bearing brine that precipitated quartz overgrowths as discrete nodules. In contrast, the massive horizon records either a degree of homogenization/miscibility with groundwater or the coalescence of an immiscible brine layer capping the groundwater aquifer. Either way, the key to the difference is in the flux of brine; it was higher in areas that formed massive horizons relative to areas with nodular forms. This is consistent with the observed areal distribution of the massive and nodular horizons relative to faults; massive horizons are widespread and overlie the Black Lake Shear Zone and associated shallow faults that acted as fluid conduits, whereas nodular horizons formed in a relatively small area at the northern edge of the underlying Black Lake Shear Zone and therefore are distal to the proposed fluid conduits (Figs. 4, 5, and 17).

Silcrete breccia is in situ, and optical and CL petrography suggest that both fracture fill and matrix are composed of disaggregated silcrete with evidence of C1 dissolution (Fig. 13D). The fractures and breccia form a continuum of genetically related deformation textures that formed through varying degrees of fracturing, fluid migration, and silcrete dissolution. The three textural zones of the conical and sheet-like fracture-breccia networks (mBx, jBx, and mS; Figs. 8–9) are similar to documented textures of shear zones with high fluid flux (e.g., Hollinsworth et al., 2019), including cataclastite (= massive breccia, mBx), heavily fractured protocataclastite (= jigsaw breccia, jBx), and protocataclastite (= fractured silcrete, mS). Moreover, the morphology of the breccia cones is similar to the “blowout faults” and associated “explosion breccias” documented by Montenat et al. (2007), which are interpreted to have formed by hydrofracturing of an indurated layer underlain by sediment with overpressurized pore fluids. As the upward-directed pore-fluid pressure overcame the confining lithostatic pressure, upward fracture and fluid propagation was facilitated by local pore-fluid stress exceeding the compressive rock strength (e.g., Engelder and Lacazette, 1990). Most fluid flux apparently passed through the massive breccia (mBx = cataclastite), where a combination of C1 dissolution, disaggregation, and shear stress from pressurized fluid resulted in nearly complete disaggregation except for remnant disorganized “survivor clasts” that appear to have mostly collapsed into the massive breccia matrix (Fig. 8). The differences noted in overall breccia form from conical to sheet-like were probably controlled by differences in stress regime, with a likely component of horizontal shear in the thinner, sheet-like breccias isolated between the Chippewa Bay and Wellesley Island faults. The dissolution of C1 quartz overgrowths coinciding with fracturing and brecciation suggests that fluids were undersaturated with respect to silica and could have been hot, high pH crustal brines like those associated with C1 and C1.5.

Following silcrete formation and brecciation, much of Allounit 2 strata across the Frontenac Arch were eroded during Late Cambrian and earliest Ordovician basin inversion (Lowe et al., 2018; Fig. 18). The silcrete horizon acted as a local well-indurated cap that prevented the erosion of Allounit 2 strata south of the Wellesley Island fault (Figs. 4–5). Succeeding Early Ordovician epeiric seaway transgression over the St. Lawrence Rift system led to clastic and carbonate marine sedimentation that continued until at least the Late Ordovician and buried Potsdam strata to a depth of at least ~1 km (Sanford and Arnott, 2010; Bédard et al., 2018). The pressure solution contacts observed throughout Allounit 2 (Fig. 12B) and at the base of the silcrete nodules (Fig. 10C) most likely occurred during this time. Pressure solution in quartz arenite typically occurs at depths of >2.5 km and temperatures of 90 °C–130 °C; the dependence on temperature can cause pressure solution at shallower depths (Worden and Morad, 2000). Notably, detrital zircon (U-Th)/He thermochronological modeling by Hardie et al. (2017) demonstrates that the Potsdam strata reached temperatures of ~100 °C–130 °C by the Devonian, which suggests either burial depths of >2.5 km or an elevated geothermal gradient that facilitated pressure solution. The latter is more likely given the absence of evidence of post-Ordovician sedimentation in this part of the St. Lawrence Rift system. Dissolved silica from pressure solution of detrital quartz in Allounit 2 strata was then preferentially transported by pore fluids to the relatively porous and uncompacted silcrete horizon, where C2 quartz overgrowths precipitated over earlier C1 ± C1.5 cements and occluded much of the remaining porosity. This proposed transport mechanism is consistent with petrographic and geochemical evidence from numerous similar sedimentary basin burial environments suggesting that silica is commonly transported from one clastic unit to another during burial diagenesis (Houseknecht, 1984; Gluyas and Coleman, 1992; Trewin and Fallick, 2000).

The groundwater silcrete horizon described here locally caps one of three unconformity-bound allounits in Upper Cambrian strata of the Potsdam Group, along the southernmost exposed extension of the St. Lawrence Rift system exposed along the Frontenac Arch, which lies south of the St. Lawrence River. Here, silcrete texture and thickness strongly support a connection between faulting and silicification through the thickening of massive horizons adjacent to Paleozoic faults and underlying Mesoproterozoic shear zones, brecciation near where brittle faults intersect shear zones, and nodules near the margins of shear zones.

Petrography and CL microscopy demonstrate that silcrete consists of three generations of quartz overgrowths. The earliest two, termed C1 and C1.5, are associated with silcrete formation. Fluid inclusion microthermometry from primary fluid inclusion assemblages trapped in C1 overgrowths suggests that silica precipitated from heated saline crustal brines sourced at depth from the Grenville basement, which supports textural and thickness observations from the field suggesting that silicifying fluids migrated along reactivated shear zones. These fluids are envisaged to have then entered the shallow Potsdam aquifer and ascended to the water table, where decreasing temperature and pH promoted quartz precipitation. Differences in silcrete form from nodular to massive reflect differences in brine flux into the Potsdam aquifer, with nodules forming at the edge of the shear zones where flux was lower, and massive horizons forming directly over shear zones where flux was higher. In areas where shallow Paleozoic faults intersect Mesoproterozoic shear zones, silcrete formation was followed by localized fracturing and brecciation caused by the violent upward-migration of overpressurized pore fluids below the impermeable massive silcrete.

This silcrete horizon then acted to resist subsequent Late Cambrian erosion, and also acted to preserve pre-compaction intergranular pore space during Ordovician burial. The latest generation of overgrowth cement, C2, was sourced from silica dissolved by pressure solution of detrital quartz in underlying and adjacent Potsdam strata and precipitated preferentially in the uncompacted and porous silcrete.

This example of groundwater silcrete highlights a unique case example with crustal silica sources and fluid migration along faults. Notably, unlike the research reported here, the documented examples that form the basis of the existing near-surface groundwater silcrete model (e.g., Thiry et al., 1988) have not been tested by detailed CL microscopy or fluid inclusion thermometry to constrain silcrete paragenesis or fluid sources. The possibility that many such groundwater silcretes precipitated from crustal brines thus expands the model of groundwater silcrete to include a new type—here termed brine silcrete—that could provide a basis for identifying evidence of syn-sedimentary deformation in similar tectonically active basins.

Provided here are data tables including locations of sections and fluid inclusion microthermometry data. Table A1 provides location data, silcrete thickness, texture, and corresponding samples. Table A2 is the raw fluid inclusion data, including thermometry data, calculated salinities, and fluid inclusion assemblages.

This work was supported by a Natural Sciences and Engineering Research Council of Canada Discovery Grant to R.W.C. Arnott. We acknowledge the memory and pioneering work of Bruce Selleck and Bruce Sanford in this region, without which this work on the Potsdam and St. Lawrence Rift system would not be feasible. Thank you also to the two anonymous reviewers for their constructive comments that ultimately improved the clarity and impact of this manuscript.