Concretionary cementation of a Scottish Middle Jurassic sandstone by hot, Paleocene fluids: a clumped isotope study

Carbonate concretions are common features in sedimentary rocks (McBride 1988; Wilkinson 1991, 1992, 1993; Mozley and Burns 1993; Selles-Martinez 1996; Raiswell and Fisher 2000; McBride et al. 2003; and others). Whereas mudstone-hosted concretions have typically initiated during microbially mediated early diagenesis (e.g. Raiswell and Fisher 2000, and references therein), the formation mechanism and timing of sandstone-hosted concretions are much less well understood; this is despite their importance in reservoir rocks, where they can reduce porosity and impede fluid flow, lowering reservoir quality (e.g. Morad et al. 2010; Nyman et al. 2014). Although reasons for nucleation, distribution (Bjørkum and Walderhaug 1990) and growth rates (Wilkinson and Dampier 1990) of sandstone-hosted concretions have been proposed, traditional geochemical approaches to understanding pore fluid evolution (e.g. Wilkinson 1993; Noh and Lee 1999; McBride et al. 2003) have been limited because pore fluid compositions and the timing of cementation are not usually known. These limitations have thus required reasonable but usually unconstrained assumptions to allow interpretations.

With the advent of clumped isotope palaeothermometry, the requirement for assumptions about fluid $δ$¹⁸O values has been removed (Ghosh et al. 2006). Furthermore, the abiotic sparry calcites that cement sandstones are excellent material for clumped isotope analysis, crystallizing slowly under shallow to deeper burial conditions where temperatures >20°C (Defliese and Lohmann 2016; Purvis et al. 2020) ensure that isotope exchange reaction rates are fast, and equilibrium between dissolved inorganic carbon (DIC) and calcite is attained (see Daëron et al. 2019).

Clumped isotopes have been used to investigate a number of sedimentary diagenetic scenarios, including investigating pore fluid sources and temperatures for cements in fault systems (e.g. Swanson et al. 2012; Bergman et al. 2013), identifying reactivation of syndepositional fracture networks in carbonate platforms (Budd et al. 2013) and to help better understand carbonate recrystallization and alteration (e.g. Huntington et al. 2011; Suarez and Passey 2014; Sample et al. 2017; Stolper et al. 2018; Guo et al. 2021). Investigations of cementation specifically have included the utility of early meteoric diagenetic cements in palaeoclimate studies (Defliese and Lohmann 2016) and cementation histories in mudrock-hosted concretions (Loyd et al. 2012, 2014; Dale et al. 2014; Heimhofer et al. 2017; Paxton et al. 2021).

The application of clumped isotopes to carbonate-cemented sandstones should significantly improve our understanding of the temperatures at which cementation proceeded (e.g. Methner et al. 2016; Purvis et al. 2020; Cao et al. 2023), provide insights into porewater evolution during concretion growth (e.g. Fan et al. 2014) and help better constrain the burial depth of cementation through the link between temperature and geothermal gradients (e.g. Cui et al. 2019; Cui and Liu 2021; Purvis et al. 2020; Jimenez-Rodriguez et al. 2022).

In this paper we present clumped isotope data from calcite cements in concretions from the Middle Jurassic Valtos Sandstone Formation of the Great Estuarine Group of the Inner Hebrides (Figs 1 and 2). These concretions are well suited to such an approach, their age and geological context making them excellent analogues for diagenetic processes operating in subsurface Jurassic North Sea reservoirs. The sedimentology and diagenesis of the sandstone have been studied extensively (Hudson 1964; Tan and Hudson 1974; Harris and Hudson 1980; Hudson and Andrews 1987; Harris 1992; Wilkinson 1992, 1993) and models for concretion growth have been offered (Wilkinson and Dampier 1990; Wilkinson 1993). However, until now the diagenetic interpretations have relied on a number of assumptions regarding the burial temperature (and depth) of concretion growth and $δ$¹⁸O values of pore fluids from which the calcite cements formed (Hudson and Andrews 1987; Wilkinson 1993). With the clumped isotope temperatures of concretionary calcite cement determined in this study, we are now in a position to calculate pore fluid $δ$¹⁸O values from which the cements precipitated using an appropriate palaeotemperature equation (Epstein et al. 1953; Craig 1965; Anderson and Arthur 1983; Kim and O'Neil 1997), thus allowing more rigorous testing of diagenetic models (combined fluid, burial and temperature histories) of concretion growth in the Valtos sandstone. The approach and results should prove instructive for wider studies on sandstone cementation.

The Bathonian Valtos Sandstone Formation (hereafter Valtos Formation; Fig. 1) is a thick sandstone unit of fluvio-deltaic origin (Harris 1992), intercalated with the mainly muddy and sandy sediments of the Great Estuarine Group (GEG). Fossils in these sediments mostly indicate a low-salinity, brackish and occasionally marginal-marine lagoonal setting (Hudson 1963, 1980). Deposition was focused in two half-graben, the Sea of the Hebrides and the Inner Hebrides basins, best defined by the deltaic sandstone units, the older Elgol Sandstone Formation (Harris and Hudson 1980; Harris 1989) and the younger Valtos Formation (Fig. 1).

The Valtos Formation crops out on the isles of Skye, Raasay, Eigg and Muck (Fig. 2) with some lateral variation between localities (Harris and Hudson 1980). In this study we concentrated on metre-scale spherical concretions from the type section (Harris and Hudson 1980) at Valtos, in Trotternish (northern Skye) and from Laig Bay on Eigg (Harris 1992). The Valtos Formation lithologies are broadly medium- or coarse-grained concretionary sandstones often containing a small, brackish-tolerant, burrowing bivalve (Neomiodon sp.; Hudson 1963, 1980). Sandstone units are typically capped by Neomiodon biosparite limestones, with further Neomiodon limestones and silty shales interbedded with the sandstones (Harris and Hudson 1980). Neomiodon shells were aragonite, and some original aragonite is preserved in shells deposited in mudstones (Hudson and Andrews 1987), but in the sandstones they are found only as neomorphic calcite within calcite-cemented concretions or limestone beds. The surrounding uncemented or lightly cemented sandstones are unfossiliferous (Hudson and Andrews 1987), suggesting that here total bioclast dissolution occurred.

Near the base of the Valtos type-section (bed 3 of Harris and Hudson 1980), just above a lower tabular concretionary unit, spherical concretions >1 m in diameter occur (Fig. 3 and Supplementary material Fig. S1), which are the primary focus at this locality; however, we also sampled a concretion higher in the sequence. At Laig Bay on Eigg we again sampled metre-scale concretions in Division E (Hudson and Harris 1979; Harris 1992) toward the top of the formation (Fig. S2). At both Valtos and on Eigg we also sampled veins of calcite that cross-cut (post-date) the concretions, Some calcite veins and decimetre-wide zones of nodular cements also post-date Paleocene igneous dykes, being common subparallel to dyke margins on Eigg (Fig. 4 and Supplementary material Fig. S3) and filling fractures within a dyke at Valtos (Fig. S4). Details of all sample locations are given in the Supplementary material.

The burial history of the GEG has been discussed by Hudson and Andrews (1987) and Morton (1987) and is updated and summarized in Figure 5. Briefly, the GEG was buried <500 m prior to uplift before the Late Cretaceous (Hudson 1983; Morton 1987), probably between 140 and 130 Ma (Fig. 5) based on apatite fission-track analysis (AFTA; Holford et al. 2010). Evidence for significant sedimentation in the Late Cretaceous is lacking and preserved outcrops do not exceed c. 25 m thickness (Hudson 1983; Harker 2002). A phase of early Cenozoic uplift and erosion between 65 and 60 Ma (Holford et al. 2010), combined with continuing eustatic sea-level fall in the Maastrichtian–Danian (Haq 2014), removed most of the thin Upper Cretaceous sediment and in places exposed underlying Jurassic rocks as a regolith (Fig. 5).

In the Paleocene, regional igneous activity associated with the British Tertiary magmatic province buried the Mesozoic sediments rapidly (within 1.6 myr; Hamilton et al. 1998) under hundreds of metres of lavas; the Trotternish lava pile is estimated to have c. 1200 m thickness (England 1994; Emeleus and Bell 2005). Intrusive sills and dykes were also emplaced during this igneous activity, including the Little Minch Sill complex of Trotternish, emplaced sometime after lava deposition (Gibson 1990; Emeleus and Bell 2005), whereas dyke emplacement probably occurred throughout the igneous event (England 1994; Emeleus and Bell 2005).

The Skye Central Complex post-dates emplacement of the Skye Lava Group (Emeleus and Bell 2005) and its thermal effects are spatially limited (Fig. 2) to a hydrothermal alteration zone about 6.5–10 km out from the centre of the complex (Taylor and Forester 1971; Forester and Taylor 1977; Lewis et al. 1992). Outside this zone, the thermal effects of localized contact alteration on Jurassic sediments are surprisingly limited and easy to recognize in localities where they have occurred. Organic matter in unaltered Jurassic rocks is extremely immature (Thrasher 1992; Lefort et al. 2012), with some vitrinite reflectivity (R₀) values well below 0.5% (Hudson and Andrews 1987; Fyfe et al. 2021); in mudrocks, molluscan aragonite is still preserved (Hudson and Andrews 1987; Holmden and Hudson 2003). In such immature shales, the thermal influence of smaller igneous intrusions on organic maturity is typically proportional to intrusion width (Bishop and Abbott 1995) and is marked in the field by darker coloration and decreased shale fissility. However, in parts of east Trotternish, particularly around Lealt and south of Staffin, intrusions of the Little Minch Sill Complex are thick and numerous enough to have caused substantial local thermal alteration.

R₀ values <0.5% in some samples of the Lealt, Duntulm and Kilmaluag formations of the GEG in Trotternish suggest burial temperatures around 50°C (see also Hudson and Andrews 1987) whereas analcime zone zeolites from the lowest part of the Trotternish lava pile (resting on GEG sediments; King 1976) suggest local post-lava hydrothermal temperatures between 75 and 90°C. There is no evidence that this hot water penetrated the clay-rich formations of the GEG. AFTA maximum temperatures from the GEG of northern Skye are between 85 ± 10 and ≥110 ± 10°C (Holford et al. 2010), but have been measured only in the sandstones of the Valtos and Duntulm formations. The likelihood of our samples having undergone post-depositional solid-state bond reordering (Passey and Henkes 2012; Henkes et al. 2014; Shenton et al. 2015; Stolper and Eiler 2015; Hemingway and Henkes 2021) is low; such resetting in calcite typically requires subsurface temperatures in excess of 100°C for more than 100 myr, which for the GEG is not possible, except where local igneous thermal alteration occurred.

The large spherical concretions formed from a single generation of ferroan calcite, with no inter-crystal zoning (Wilkinson 1993), growing concentrically outward from an (assumed) initial nucleus. At the centre of the concretion, crystals are equant, c. 0.25 mm wide, and they increase in both size and elongation radially to >50 mm (Wilkinson 1993), with a poikilotopic texture that is clearly visible in hand specimen. Neomiodon shells are preserved as neomorphic calcite and the amount of shell material preserved decreases from centres to the edges of concretions (Wilkinson 1993). Pressure dissolution of shells is common, some being reduced to ‘films’ between quartz grains (Hudson and Andrews 1987). As shells are preserved only within concretions, pressure solution must have occurred before concretionary cementation (Hudson and Andrews 1987). Point counting (Wilkinson 1989) shows that the host sandstone retains c. 30% porosity (minus cement porosity), reduced from an estimated original porosity of 45–50%. At Valtos, some concretions show compaction of thinly bedded host sediments around their margins (Fig. 3b). This indicates that some burial compaction continued after concretion growth (Hudson and Andrews 1987; Wilkinson 1993), the host sandstone framework being c. 20% more compacted than the concretions (Wilkinson 1993). Curiously, other concretions show no clear evidence of post-concretion compaction.

Samples were taken where the sandstones showed no visible or petrographic–geochemical evidence of thermal alteration. Sandstone calcite cements from the concretions, vein-filling sparry calcite and calcite amygdales in lavas were drilled from fresh surfaces, taking care to avoid heating and pressure. Core plugs of concretion VR1 [NG 516 642] (Supplementary material Fig. S1) were re-sampled along the vertical transect of Wilkinson (1993). For each VR1 measurement between 9.45 and 10.19 mg of sample was reacted, to achieve a gas volume proportionate to the yield of a pure carbonate sample of 4 mg. All VR1 samples were measured 3–4 times and sample M8605 was analysed once each day (10 measurements) to establish a representative distribution of error. Most non-VR1 samples were also analysed as 4 mg or equivalent gas yield although some were measured as 6 mg or equivalent gas yield; the larger resulting gas volumes from 6 mg did not affect the resulting $Δ$₄₇ values.

Samples were digested in a common acid bath containing c. 105% phosphoric acid heated to 87°C. The released CO₂ was dried by sublimating at about −115°C, then passing through two spiral traps held at −115°C, freezing with liquid nitrogen into a manometer. Potential hydrocarbon and chlorocarbon contaminants were removed by cryodistilling through a 12 cm × 4 mm i.d. glass tube packed with Porapak Q ion exchange resin and held at −20°C. The resultant gas was transferred directly to the mass spectrometer. Full details of the gas preparation have been given by Paxton (2022).

Sample CO₂ was analysed for $δ$⁴⁵–$δ$⁴⁹ on the MIRA dual-inlet isotope ratio mass spectrometer (IRMS) at the University of East Anglia (UEA), using the operating and data handling protocol described by Dennis et al. (2019). Carbonate standards ETH1, ETH3 and UEACMST were also prepared and measured every day to provide an estimate of long-term measurement uncertainty, and to determine the transfer function between the local reference frame and the carbon dioxide equilibrium scale of Bernasconi et al. (2018), ‘$Δ$₄₇ (CDES-25)’. $Δ$₄₇ values used for ETH1 and ETH3 at 87°C acid digestion were empirically derived at UEA; ETH1 = 0.218‰ and ETH3 = 0.613‰. If projected to represent acid reaction at 90°C, transfer functions created using the UEA-derived ETH values versus those from Bernasconi et al. (2021) produced sample $Δ$₄₇ (CDES-90) values that were within measurement error of one another. To convert $Δ$₄₇ values derived by reaction at 87°C to ‘$Δ$₄₇ (CDES-25)’, an acid fractionation factor of 0.062‰, empirically determined at UEA, was added.

For Sr isotopes, carbonate powders were dissolved in twice the stoichiometric volume of 1.0N HCl required to completely dissolve the sample. After the effervescence ceased, the solution was centrifuged, and the supernatant was separated from any residues and dried down on a hot plate. The sample was purified from Ca, Rb and other trace elements using conventional cation exchange chromatography. The purified Sr aliquot was then loaded onto single Ta filaments with phosphoric acid and Ta-gel to increase the ionization efficiency of Sr in the mass spectrometer. The ⁸⁷Sr/⁸⁶Sr measurements were performed using a thermal ionization mass spectrometer (Thermo-Elemental Triton instrument) operating in static multi-collection mode. An exponential law was used to correct the measured ⁸⁷Sr/⁸⁶Sr ratios for instrumental mass fractionation against measurements of ⁸⁸Sr/⁸⁶Sr with a true ratio of 8.375209. Isobaric interference of ⁸⁷Rb on ⁸⁷Sr was monitored using ⁸⁵Rb, and corrections were made if necessary. The external reproducibility of ⁸⁷Sr/⁸⁶Sr is better than ±15 ppm (2$σ$⁠), based on multiple runs of SRM 987 performed throughout the course of this work, which yielded a mean ⁸⁷Sr/⁸⁶Sr ratio of 0.710259.

The calcite–water isotope fractionation factor calibration of Kim and O'Neil (1997) was used for all samples, being appropriate for abiotic calcite cements. The measurement uncertainty for $δ$¹⁸O_CARBONATE, $δ$¹³C and $Δ$₄₇ is reported as ±1 standard error of the pooled standard deviation unless stated otherwise. Pooled standard deviation was used wherever sample compositions were similar.

Temperature and $δ$¹⁸O_FLUID values were calculated using the mean $Δ$₄₇ and $δ$¹⁸O_CARBONATE values with full propagation of the measurement and calibration uncertainties. The $δ$¹³C_CARBONATE and $δ$¹⁸O_CARBONATE values are reported on the Vienna Pee Dee Belemnite (VPDB) scale and $δ$¹⁸O_FLUID values on the Vienna Standard Mean Ocean Water (VSMOW) scale. All data, calculations, sample replicates and pooled standard deviation-derived standard errors are given in the Supplementary material.

Stable, clumped and Sr isotope results for all samples are given in Tables 1 and 2 and plotted in Figures 6–8. Here we describe the clumped isotope data for the large concretion VR1 at Valtos in detail, as this (Table 1 and Fig. 6) helps to frame the discussion of other samples. From the centre to the edge of VR1 (Fig. 6), covarying changes in measured $Δ$₄₇-derived temperatures and calculated porewater $δ$¹⁸O values follow the zonal pattern of changing $δ$¹⁸O_CARBONATE values identified by Wilkinson (1993). In Zone 1 (0–35 cm), $Δ$₄₇-derived temperatures are 45 ± 6°C at the centre of the concretion (0–24 cm), increasing to c. 50°C between 24 and 31.5 cm. The corresponding $δ$¹⁸O_FLUID values are between −11.1 ± 1.1 and −12.6 ± 1.1‰ VSMOW, increasing to −7.7 ± 0.9‰ VSMOW near the edge of Zone 1. In Zone 2 (40–55 cm) $Δ$₄₇-derived temperatures increase to 63 ± 8°C, with a peak value of 73 ± 8°C. The corresponding $δ$¹⁸O_FLUID values range between 0.0 ± 1.2 and +2.1 ± 1.1‰ VSMOW. The temperature slightly decreases to 57 ± 7°C in Zone 3 (c. 58–60.5 cm), the outer part of the concretion, with $δ$¹⁸O_WATER values decreasing as well, to −4.4 ± 1.1‰ VSMOW (Fig. 6).

For comparison, a concretion from bed 38 (Dun Dearg, Valtos cliffs) gave a $Δ$₄₇-derived temperature of 64 ± 9°C and a $δ$¹⁸O_FLUID value of −1.8 ± 1.4‰ (Table 1 and Fig. 7), which is like Zone 2 in VR1. In another example, the centre of a large concretion on Eigg gave a $Δ$₄₇-derived temperature of 79 ± 8°C and a $δ$¹⁸O_FLUID value of +4.9 ± 1.1‰ VSMOW, which is also reminiscent of Zone 2 in VR1. However, 63 cm further from the centre of this concretion, the temperature increased to 98 ± 9°C and $δ$¹⁸O_FLUID value increased as well, to +7.2 ± 1.2‰ VSMOW, which is more extreme than observed in Zone 2 in VR1, whereas the outermost sample from this concretion gave a temperature of 68°C and $δ$¹⁸O_FLUID value of +2.7 ± 1.1‰ VSMOW (Table 1 and Fig. 7).

Other examples of calcite sparry cement data from the Valtos Formation on Skye and Eigg are also shown in Figure 7. Most calcite cements in veins and vugs that post-dated concretions at Valtos gave temperatures between 40 ± 10 and 52 ± 10°C and $δ$¹⁸O_FLUID values between −12.7 ± 1.7 and −15.4 ± 2.5‰ VSMOW. Exceptions include a vein from the base of a large concretion (189-1; Table 1), which gave a temperature of 67 ± 10°C and a $δ$¹⁸O_FLUID value of +4.1 ± 1.5‰ VSMOW, and a vein from the margin of a Tertiary dyke (189-6; 24 ± 8°C; −16.5 ± 1.6‰ VSMOW). On Eigg, a nodular concretion from a cemented sandstone horizon adjacent to a Paleocene dyke margin (169-1; Table 1) gave a temperature of 42 ± 9°C and fluid value of −14.6 ± 1.7‰ VSMOW, comparable with veins from the Valtos Formation that post-date dyke intrusion.

$δ$¹⁸O_FLUID values for Paleocene Hebridean meteoric waters that post-dated lava emplacement and cooling were calculated from calcite amygdales in the Eigg Lava Formation (SA-1 and SA-2). The $Δ$₄₇ temperatures were within error at 28 ± 6 and 27 ± 6°C, with $δ$¹⁸O fluid values (−13.7 ± 1.2 and −15.1 ± 1.1‰ VSMOW) comparable with the more negative values in the concretions and vein calcites (Table 1 and Fig. 7). These samples gave the lowest $δ$¹³C values measured in this study (−6.1 and −5.8‰ VPDB) whereas $δ$¹⁸O_CARBONATE values of −16.6 ± 0.1 and −17.8 ± 0.1‰ VPDB are comparable with those of many of the calcite cements measured here.

Horizontal and vertical centre to edge geochemical transects from concretion VR1 (Wilkinson 1993) indicated that the concretion grew concentrically. The cement was assumed to derive from the redistribution of calcium carbonate from dissolution of originally aragonite Neomiodon shells. This is supported by the preservation of shell debris within concretion bodies but not in the host sands (Tan and Hudson 1974; Hudson and Andrews 1987) and by the reduction in shell material from concretion centres to edges (Wilkinson 1993).

The initial cementation in Zone 1 (c. 14% of concretion volume) had calcite cement $δ$¹⁸O as low as −18.2‰ (Fig. 6a and Supplementary material Fig. S5), values that were difficult to explain in Wilkinson's favoured model of concretion growth under a stable burial temperature. In Zone 2 (c. 53% of concretion volume) the cements had increased Mg and Fe contents (Fig. S6) and $δ$¹⁸O_CARBONATE had higher values (Wilkinson 1993). In Zone 3 (c. 33% of concretion volume), the outermost part, cement $δ$¹⁸O_CARBONATE decreased again (Fig. 6a) along with both Mg and Fe content. The new $Δ$₄₇-derived temperatures and $δ$¹⁸O_FLUID compositions (Fig. 6b and c) clearly record the zonal centre to edge changes identified by Wilkinson (1993). This range in temperatures in a single concretion is incompatible with a stable burial temperature model and rules out post-depositional solid-state bond reordering (Passey and Henkes 2012; Henkes et al. 2014; Shenton et al. 2015; Stolper and Eiler 2015; Hemingway and Henkes 2021).

The initiation of cementation was thought by Wilkinson (1993) to be due to an influx of meteoric water into the sandbody that reduced (by dilution) any Mg inhibition to calcite nucleation in the connate pore fluids (Wilkinson 1993). Meteoric water flushing was assumed to have been a Mid to Late Jurassic event or a result of recharge after uplift in the Cretaceous (Wilkinson 1993). $δ$¹⁸O_FLUID values of c. −12‰ VSMOW from the centre of concretion VR1, although consistent with a meteoric water source (e.g. Dansgaard 1964), are much more negative than traditionally estimated Hebridean values of Jurassic meteoric water (between −4 and −7‰ VSMOW; Hudson and Andrews 1987; Searl 1992). Such estimates are incompatible with the Zone 1 (Z1) $Δ$₄₇-derived $δ$¹⁸O_FLUID values of c. −12‰ VSMOW.

There is no direct evidence of meteoric cementation in the sparse Inner Hebridean Cretaceous sediments, which in the study area are largely of Turonian age (Harker 2002); this precludes measurements or inferences regarding pore fluid composition. Cretaceous meteoric water recharge could have occurred between 140 and 130 Ma following uplift (Holford et al. 2010), but there is no concrete evidence for it. Elsewhere in the UK, Hendry (1993) interpreted pore-filling cement in English Midlands Middle Jurassic limestones to have formed from Early Cretaceous meteoric water with $δ$¹⁸O more negative than −8‰ VSMOW. Climatic models for Hebridean Middle Cretaceous palaeolatitudes also suggest negative $δ$¹⁸O values between −5.5 and −11‰ VSMOW (Table 3 and references therein). We therefore cannot rule out the possibility that Z1 cementation in concretion V1 was from pore fluids sourced from isotopically negative meteoric water during a period of pre-late Cretaceous exposure.

The involvement of meteoric water in hydrothermal reactions in zones around the Paleocene igneous centres allowed Taylor and Forester (1971) and Forester and Taylor (1977) to calculate the $δ$¹⁸O_FLUID composition of the Paleocene meteoric water as −12‰ VSMOW. These low values were corroborated by Fallick et al. (1985) and by the new clumped isotope data reported in this study (Table 1) from the Eigg Lava Formation calcite amygdales (⁠$δ$¹⁸O values of −13.7 ± 1.2 and −15.1 ± 1.1‰ VSMOW). There is thus direct evidence that VR1 initiation may have commenced from isotopically negative Paleocene pore fluids.

There are several widely dispersed localities where Paleocene lava was emplaced directly onto a land surface where the Valtos Formation sandstone cropped out (Fig. 2), including the east coast of Eigg, in the Upper Glen 1 borehole in Waternish [NG299506] (see Schofield et al. 2016; Fyfe et al. 2021), on Raasay [NG577430], at Craig Ulatoa near Prince Charles’ Cave in Trotternish [NG51047] and in Strathaird [NG553211] where the Valtos Sandstone is directly overlain by a meteoritic ejecta layer (Drake et al. 2018) of similar age to the overlying lavas. At Waterstein Head on Waternish the lavas are just above the Valtos Formation. These outcrops demonstrate that parts of the Valtos sandstone aquifer were completely unroofed by erosion, basinwide, exposing multiple points of entry (others may be present in sub-crop) for meteoric-derived fluids to enter the aquifer. Moreover, as fragmentary outcrops of thin, marine, Upper Cretaceous sediments are known close to a number of these localities (Hudson 1983), it is very likely that this phase of erosion post-dated Late Cretaceous deposition (Fig. 5), driven by Paleocene (65–60 Ma) uplift (Holford et al. 2010) accompanied by continuing eustatic sea-level fall in the Danian (Haq 2014).

Cementation of Z1 occurred at around 47°C and it is unlikely that much of this heat was related to burial (see Wilkinson 1993), given that parts of the sandstone were demonstrably at land surface with maximum burial probably less than the thickness of the sandstone itself (i.e. c. 100 m in Trotternish). However, if Z1 cementation was concurrent with Paleocene igneous activity, local sources of heated water would have been common, even before lava emplacement. The earliest evidence for volcanic activity in northern Skye is from the Portree Hyaloclastite Formation (Anderson and Dunham 1966; Bell and Williamson 2002), which although not dated radiometrically, is not likely to be much younger than 60.53 ± 0.08 Ma (the age of the Rum basic–ultrabasic pluton (Hamilton et al. 1998), clasts of which are found in the Skye Lava Group). Phreato-magmatic vent volcanism (Anderson and Dunham 1966) during subaqueous tuff and hyaloclastite deposition explains how heated groundwater may have locally accessed and migrated into the porous Valtos Sandstone aquifer. The Z1 $δ$¹⁸O_FLUID values begin around −12‰ in the concretion centre, changing to −11‰ VSMOW (Table 1), values that are identical to, or not much evolved from, previously estimated meteoric water compositions of −12‰ VSMOW (Taylor and Forester 1971; Forester and Taylor 1977); they imply a near open system (high water/rock ratio) for oxygen isotopes. The Krafla geothermal system on Iceland is a modern isotopic analogue, where meteoric groundwater (⁠$δ$¹⁸O_FLUIDc. −12‰ VSMOW) is heated in basaltic rocks under high water/rock ratios, resulting in only small changes to the measured hydrothermal fluid $δ$¹⁸O (Pope et al. 2009, 2010).

The transition of $δ$¹⁸O_CARBONATE from −18.5 to −8.6‰ is caused by a change in both temperature and $δ$¹⁸O_FLUID composition (Fig. 6) and not simply a change in pore fluid compositions as envisioned by Wilkinson (1993).

The $Δ$₄₇-derived temperature change between outermost Z1 and innermost Zone 2 (Z2) is from 48 ± 5 to 73 ± 8°C and is most simply explained by the introduction of heated water, with temperature sustained around 70–80°C for the duration of the cement precipitation in Z2, accounting for over half the volume of the concretion. The temperature change is preceded, and then accompanied, by a change in $δ$¹⁸O_FLUID composition (Fig. 6); Z1 averages −11.1‰ VSMOW, whereas Z2 averages +0.9‰ VSMOW, a 12‰ change, with highest values of +2.1 ± 1.1‰ VSMOW at the edge of Z2. There is also a marked increase in calcite Fe and Mg content (Wilkinson 1993; Fig. S6) that follows the temperature and $δ$¹⁸O_FLUID trends. The meteoric pore fluid of Z1 must have been replaced by a pore fluid that was relatively enriched in ¹⁸O. Despite the sharp geochemical change between zones there is no evidence for a pause in cementation; only a single generation of ferroan calcite is present (Wilkinson 1993) although the crystals become more elongated in Z2. The $δ$¹⁸O_FLUID composition then evolved progressively to more positive values over time.

Heated fluids in hydrothermal systems are known to undergo isotope exchange with surrounding rock units, and such water–rock interactions result in more positive $δ$¹⁸O_FLUID values after exchange with ¹⁸O-rich country rocks (Craig 1963; Sheppard 1986). In the case of the Valtos Formation concretions the underlying country rocks are mainly Middle and Lower Jurassic mudrocks and sandstones, and the overlying country rocks are Middle and Upper Jurassic mudrocks and sandstones followed by Paleocene basalts. Skye basalts have $δ$¹⁸O values above +5‰ SMOW (Taylor and Forester 1971; Forester and Taylor 1977) whereas mudrocks in the upper Great Estuarine Group have $δ$¹⁸O values between +10 and +17‰ SMOW (Hamilton et al. 1992).

If Z1 cementation was concurrent with Paleocene igneous activity, it is logical to attempt to explain the progressive positive change in pore fluid values towards the outer part of the concretion as a result of meteoric water percolating through, and reacting with, hot basaltic lavas. However, we cannot find any evidence to support the notion that such reactions could achieve the magnitude of oxygen isotopic exchange required; that is, to alter pore fluid $δ$¹⁸O from −12‰ (Paleocene meteoric water) to +2‰ (Z2 fluid) VSMOW. In the Icelandic Reykjanes geothermal system, there is oxygen isotopic exchange between fluid and host basalts that results in hydrothermal fluid $δ$¹⁸O around +2‰, but this happens largely from an initial seawater fluid source (Pope et al. 2009, 2010).

The alternative is that hot basinal fluids, enriched in ¹⁸O by reaction with host basinal sediments, particularly mudrocks, migrated upward. There is now growing clumped isotope characterization of such basinal fluids with temperatures and $δ$¹⁸O_FLUID values similar to those attained in our Z2 (e.g. Swart et al. 2016; Mangenot et al. 2018b; Dennis et al. 2019). In this scenario, basinal fluid flow was promoted by the progressive crustal loading of the thickening Paleocene lava pile. A 1 km thickness of lava emplaced within 1.6 myr (Hamilton et al. 1998) would rapidly increase lithostatic pressure (c. 25 MPa). In confined sediments this could promote failure of over-pressured mudrock units to release a significant volume of fluid (Cathles and Smith 1983; Cathles and Adams 2005) that then flowed upward along faults (see Wilkinson 1993) into the Valtos Formation. The main phase of basinal-fluid cementation in Middle Jurassic sediments of the Paris Basin shows broad similarity in $Δ$₄₇-derived temperature characteristics and $δ$¹⁸O_FLUID values, albeit the latter influenced by Triassic brine contributions (Mangenot et al. 2018b). The Valtos basinal fluids initially mixed with but then gradually displaced the Z1 meteoric fluids, explaining the progressive change to heavier $δ$¹⁸O_FLUID values through Z2 and possibly influencing (elongating) cement crystal morphology.

This scenario also explains why the Z2 cements have high Fe and Mg contents (about 1.5× higher) relative to Z1. Both elements will be relatively easily leached from mudrocks, with ferrous Fe solubility promoted by reducing and hot conditions (Garrels and Christ 1965; Seyfried et al. 1991). The Z2 Fe contents are c. 3×, and the Mg contents c. 8×, those in limestone-hosted sparry calcite burial cements from the overlying Duntulm Formation (see Andrews 1986; Hudson and Andrews 1987). Basinal fluid temperatures of 70–80°C suggest generation in mudrock units well below the GEG in the Lower Jurassic succession (see Morton 1987). For example, the Pabbay Shale Formation (Hesselbo et al. 1998) is likely to have been buried to around 2 km depth in north Skye and Waternish (Fyfe et al. 2021), and its pore fluids are likely to have been of marine or modified marine origin. There are no reliable data on the Paleocene geothermal gradient but in the Upper Glen borehole (Fig. 1) vitrinite R₀ values between 0.7 and 0.8% are recorded in the Lower Jurassic Pabbay and Broadford formations (Fyfe et al. 2021), values that are consistent with burial temperatures c. 80°C.

Taking all the evidence together it seems likely that, whereas Z1 of concretion VR1 began forming during the earliest stages of effusive igneous activity on Skye, Z2 was forming as the lava pile thickened. This must have occurred after 60.53 ± 0.08 Ma (see above) but before 58.91 ± 0.07 Ma (a date on Cuillin Gabbro; Hamilton et al. 1998), which post-dates extrusion of the Skye Lava Group. While concretion VR1 was growing the down-bearing pressure of the lava pile was initially not enough to overcome the supporting litho- and hydrostatic pressure in the sandstone. At some stage the lava pile was thick enough to initiate sandstone burial compaction, at least some of which occurred after the concretion had formed. Concretion growth was probably complete before eruption of the Skye Central Complexes, which post-date eruption of the Skye Lava Group (Emeleus and Bell 2005). This chronology rules out the possibility that the Z2 fluids are related to hydrothermal systems developed around the central igneous complex of Skye (Taylor and Forester 1971; Forester and Taylor 1977; Schofield et al. 2016). Similarly, influence of heat from intrusion of the Little Minch Sill complex into the Valtos Formation at Valtos cannot have been responsible for local fluid heating during concretion growth because the intrusion was emplaced after the Skye Lava Group (Anderson and Dunham 1966; Gibson and Jones 1991; Emeleus and Bell 2005).

Three Sr isotope measurements from VR1 cements (Table 2 and Fig. 8) were used to investigate whether the calcites retained any isotopic fingerprint of fluid interaction between basinal sediments or lavas. All of the VR1 data (Fig. 8) are much less radiogenic than those for GEG clays (Hamilton et al. 1992) and more radiogenic than the lava values (Scarrow 1992). The VR1 data are closest to values for GEG shells and other carbonates (Andrews et al. 1987; Holmden and Hudson 2003). This outcome implies that both heated meteoric water and basinal fluids were undersaturated with respect to aragonite, allowing them to dissolve aragonite shells encountered in the basinal sediments and/or as they entered the sandstone (Wilkinson 1993), the shell-derived Sr overprinting any sediment or lava signature. This is supported by the $δ$¹³C and Sr contents of the cements (Table 1 and Fig. S6), which also have rock buffered values derived from shell dissolution (Wilkinson 1993).

Cements in Zone 3 (Z3) lithify 30% (by volume) of VR1 in a relatively thin outer zone where we have a single $Δ$₄₇ value (Table 1 and Fig. 6). This shows that there is a change in $Δ$₄₇-derived temperature and $δ$¹⁸O_FLUID between Z2 and Z3, in addition to decreases in $δ$¹⁸O_CARBONATE and Mg and Fe contents (Wilkinson 1993). The last temperature from Z2 is 73 ± 7°C, which decreases to 57 ± 7°C in Z3. The Sr isotope value for Z3 cement is not different from those for Z1 or Z2, which implies that dissolving Neomiodon shells were still the main carbonate source. The Z3 fluid was clearly cooler and its $δ$¹⁸O_FLUID value (–4.4 ± 1.1‰ VSMOW) indicates that it contained less rock-derived ¹⁸O, and dissolved Fe and Mg. This suggests that the basinal fluid source was decreasing and possibly mixing with meteoric fluids that were penetrating the lava pile through fractures. High levels of seasonal meteoric precipitation (1175–2267 mm) are indicated by fossil flora preserved within the Skye Lava Group (Poulter 2011). This mixed fluid had a composition that promoted the development of poikilotopic calcite crystals.

Cements from a sandstone-hosted concretion (EG191015-1; Table 1) from Dun Dearg (Valtos cliffs, bed 38 of Harris and Hudson 1980), stratigraphically about 40 m above VR1, gave a $Δ$₄₇-derived temperature of 64 ± 9°C, and a $δ$¹⁸O_FLUID value of −1.8 ± 1.4‰ VSMOW (Fig. 7). These values are similar to those for Z2 of concretion VR1 (Fig. 7), suggesting that this phase of ferroan spar cementation was essentially coeval with the main growth phase of VR1.

To better understand the significance of the VR1 data in a basin-wide context we studied a large spherical concretion at Laig Bay (Isle of Eigg), 75 km south of Valtos (Fig. 2). Although this concretion is similar to VR1 it comes from near the top of the formation Division E (Harris and Hudson 1980), rather than the base. Three measurements, roughly centre, intermediate and outer edge (Supplementary material Fig. S2), from this concretion gave values between 68 ± 7 and 98 ± 9°C, and isotopically enriched pore fluid $δ$¹⁸O between +2.7 ± 1.1 and +7.2 ± 1.2‰ VSMOW (Table 1 and Figs 6 and 7), comparable with those from Z2 in concretion VR1. They indicate that the high-temperature, isotopically enriched basinal pore fluids were widespread, both throughout the basin and throughout the thickness of the Valtos Formation. However, there is no evidence so far for an equivalent of the earlier Z1 meteoric fluids on Eigg, either in our data or inferred from data in Wilkinson (1992).

To put the VR1 and the Eigg concretion data into a wider paragenetic context, we discuss here clumped isotope data from vein-fed sparry calcite cements that post-dated concretion formation (Table 1 and Fig. 7).

The first (189-1; Valtos) cross-cuts a concretion of c. 2 m diameter from the same horizon as VR1 and gave a $Δ$₄₇ temperature of 67 ± 10°C and a derived $δ$¹⁸O_FLUID value of +4.1 ± 1.5‰ VSMOW. These results are similar to those for Z2 in concretion VR1 (Fig. 7), indicating that a hot, evolved fluid was still accessing some fractures. The $δ$¹³C is 0.1‰ VPDB, consistent with carbon sourced from dissolving Neomiodon shells.

The second vein samples come from the margins and interior of a small dolerite dyke at Valtos (Fig. S4), about 2 m from 189-1. These veins are the fills of intrusion-parallel fractures that opened sometime (unconstrained) during or after cooling and contraction of the dyke and probably postdate concretion formation by hundreds of thousands of years. Samples 189-5v and 189-7a (Table 1) come from two generations of calcite that filled a fracture 5 cm wide within the dyke, whereas sample 189-6 comes from a 2–5 mm thick vein filling a fracture between the wall of the dyke and the host sandstone. Samples 198-5v and 189-6 gave $Δ$₄₇ temperatures of 52 ± 10 and 40 ± 10°C with pore fluid values of −12.7 ± 1.7 and −13.8 ± 1.9‰ VSMOW respectively. The former (189-5v) calcite is intergrown with a brown fibrous smectite (possibly saponite) as identified by X-ray diffraction. The Sr isotope composition of the earliest part (sample 189-7a) of the vein was 0.709941 ± 0.000007, which overlaps the values in the VR1 cements, suggesting that carbonate was similarly sourced from dissolving GEG shell carbonate or limestones (Fig. 8). These temperatures indicate that both generations of vein calcite formed after the dyke had cooled substantially from its emplacement temperature; they are also cooler than the maximum cementation temperatures reached during VR1 (Z2) cementation. The pore fluid $δ$¹⁸O values are consistent with a fracture-fed source from isotopically negative Paleocene meteoric water that had undergone relatively low interaction (high water/rock ratio) with host rocks. The presence of smectite suggests that the water interacted with some silicate during transit, either rocks of basaltic composition or mudstones in the overlying GEG, and it may be associated with intrusion of the Little Minch Sill Complex, which post-dated lava emplacement (Gibson 1990; Gibson and Jones 1991; Fowler et al. 2004). Sample 189-6 (Table 1) was taken from the dyke margin in contact with the host sandstone and gave the coolest temperatures (24 ± 8°C) for this locality, and the most negative pore fluid value (–16.5 ± 1.6‰ VSMOW). This fracture was clearly cemented after cooling of the lava pile.

A vein-fed sample of pure sparry calcite drilled from cemented shelter porosity within articulated Neomiodon shells in a limestone (189-8spar; ex situ block derived from c. 10–20 m above VR1 in an inaccessible part the Valtos cliffs; Table 1) gave a $Δ$₄₇ temperature of 42 ± 6°C and a $δ$¹⁸O_FLUID value of −14.5 ± 1.1‰ VSMOW. These results are essentially identical to those for a fourth post-cementation vein sample (18-4bulk; Table 1), a ferroan sparry calcite with $Δ$₄₇ temperature of 40 ± 9°C and a $δ$¹⁸O_FLUID value of −14.2 ± 1.6‰ VSMOW. Cathodoluminescence characteristics suggest that these cements (vug and vein fill) were the same phase (Paxton 2022), the cooler temperatures confirming their formation after the main phase of post-compactional limestone cementation (Hudson and Andrews 1987), probably later in the Paleocene after a period of cooling. The fluid compositions are in keeping with other assumed Paleocene meteoric water sourced values measured in this study (see Table 1 and Fig. 7).

On Eigg we also studied nodular cements sampled from a distinct, 60 cm wide zone of cementation adjacent to a Paleocene dyke (Fig. 4). This sample, taken from the fully cemented horizon (see description given by Hudson and Andrews 1987) within a few centimetres of the dyke margin, gave a $Δ$₄₇ temperature of 42 ± 9°C and a derived $δ$¹⁸O_FLUID value of −14.6 ± 1.7‰ VSMOW. In the past it has been assumed that this zone of cementation was related to the intense contact heating from the dyke (Hudson and Andrews 1987), an interpretation apparently supported by very negative calcite $δ$¹⁸O values (−13.8 and −17.2‰; Wilkinson 1992). However, the $Δ$₄₇ temperatures show that these cements precipitated from warm fluids with a negative, Paleocene, meteoric water composition. This later zone of cementation is thus not related to contact heating effects, but rather to the palaeohydrology of the zone adjacent to the dyke, with cementation focused along the dyke wall (a vertical aquitard). The temperature and fluid composition of this calcite are within the range for the calcites from veins associated with the dyke at Valtos.

Considered together, these vein-related samples demonstrate a range of temperatures and water–rock interaction, the latter controlled by fracture geometry and connectivity. It is likely that the hotter fluids were introduced first, possibly the final phase of basinal fluid contribution, although some heating at Valtos may have post-dated the Skye Lava Group, related to intrusion of the Little Minch Sill Complex, which is well developed in this part of Trotternish (Gibson and Jones 1991; Schofield et al. 2016). Vein calcites with temperatures below 40°C are much cooler than the lowest Trotternish AFTA maximum burial temperature of 85°C (Holford et al. 2010) and cooler than a more conservative estimate of maximum burial temperature. For example, a mean upper crustal geothermal gradient of 30°C km⁻¹ (as used by Hudson and Andrews 1987) would require maximum temperatures of at least 50°C (assuming 20°C surface temperature). We thus interpret these cooler temperatures to represent mineralization during basin inversion, a rapid erosional process that was probably completed <1 myr after lava emplacement (based on the 0.53 ± 0.23 myr erosion figure established for completely unroofing the Rum Igneous centre; Chambers et al. 2005). The strongly negative $δ$¹⁸O_FLUID compositions of the calcite veins concur with a Paleocene timing for such rapid erosion (see below).

A summary of the main data trends and our interpretation is given in Figure 9. It is not clear what initiated concretion growth. Zone 1 cementation in VR1 is, so far, unique; cements with similar geochemistry are not known outside Trotternish and may be rare even at the type section (VR2 of Wilkinson (1993) lacks Z1). Despite this, Z1 in VR1 appears to record the first ingress of Paleocene meteoric water into the Valtos sandstone, crucial in constraining the timing of cement growth initiation. Moreover, the Z1 fluid seems to be the first record of substantial porewater renewal since deposition, facilitated by early Paleocene surface exposure of the upper part of the sandbody. Substantial Paleocene recharge is consistent with palaeofloral data from Mull and Skye (Poulter 2011) that indicate wet climatic conditions. In all other studied concretions, fluids with geochemical compositions similar to Z2 in VR1 precipitated the first concretionary cements (e.g. VR2 of Wilkinson 1993), interpreted here as precipitates from cross-formational basinal fluids. In all cases the pore-fluid ingress caused dissolution of host sandstone shell carbonate, promoting supersaturation of porewater dissolved carbonate. It may be that concretion nucleation was aided by attainment of a critical (high) temperature in the aquifer pore system; no large concretion records a $Δ$₄₇-derived temperature below 44 ± 6°C.

As compaction of the Valtos Sandstone by a significant thickness of lava (up to 1200 m; England 1994; Emeleus and Bell 2005) was completed after the concretions formed, the rate of concretion growth can be re-evaluated. Modelling by Wilkinson and Dampier (1990) concluded that a Valtos Sandstone concretion of radius 0.5 m would have taken >6 myr to form in stationary pore fluids, and c. 4 myr with pore fluids flowing at c. 5 m a⁻¹. Numerous necessary assumptions in the model parameterization were required, including fluid temperature, fluid composition and pore fluid flow rate. Our new interpretation requires reconsideration of these parameters, notably the change in temperature from an assumed 25°C (Wilkinson and Dampier 1990) to the increasing temperatures, from c. 47 to 70–80°C, recorded here. The temporal constraints from our new data are that (1) initiation of VR1 was probably contemporaneous with the earliest (pre-lava) igneous activity on Skye (the Portree Hyaloclastite) and (2) termination of growth occurred around (or soon after) the time the Skye Lava Group was fully emplaced. Emplacement of the Skye Lava Group is thought to have occurred within 1.6 myr (Hamilton et al. 1998), followed, within <1 myr, by rapid erosion (based on Chambers et al. 2005), which removed much of the lava pile and terminated sandstone compaction. Thus concretion growth and final host sandstone compaction must have been completed in no more than c. 2.6 myr. This implies that concretion growth rate was approximately double the previously modelled fastest value. This does not invalidate the modelling approach but it does demand reconsideration of the parameterization.

The revised (Paleocene) timing of concretion growth onset explains how negative $δ$¹⁸O_CARBONATE compositions are preserved in the middle of the large concretions during their initial stages (Z1) of growth. These compositions are partly due to the calcite precipitation temperatures but strongly linked to the negative $δ$¹⁸O composition of Hebridean Paleocene meteoric water around −12‰ VSMOW. If our proposed timing of concretion initiation is correct, it also informs on the mechanism for these strongly negative meteoric water compositions. It has been known since the 1970s that meteoric water with a $δ$¹⁸O_FLUID composition around −12‰ VSMOW was involved in hydrothermal reactions around the Paleocene igneous centres (Taylor and Forester 1971; Forester and Taylor 1977). It has been inferred that these negative compositions could have resulted from orographic effects (Taylor and Forester 1971; MacDonald et al. 2019) caused by the developing high-relief volcanic landscape. However, our timing for concretion initiation, around 60.5 Ma (Hamilton et al. 1998), predates the development of volcanic topography. This implies that the negative compositions mainly derive from the latitude of the Hebridean region at that time, around 45° N according to Smith et al. (1994) and Ganerød et al. (2010). The modern temperature (and concomitant less-negative precipitation $δ$¹⁸O) latitudinal anomaly (Rozanski et al. 1993) over coastal Western Europe is caused by heat from the Gulf Stream–North Atlantic Current. This North Atlantic circulation was established after the Paleocene, possibly in the Eocene, there being clear evidence for onset or strengthening of the Atlantic meridional overturning circulation (AMOC) c. 34 Ma, at the Eocene–Oligocene transition (Hutchinson et al. 2019).

A latitudinal explanation for strongly negative Paleocene meteoric water compositions is also consistent with some of the Cretaceous modelling (Poulsen et al. 2007; Suarez et al. 2011) that allows for pre-Paleocene Hebridean meteoric water $δ$¹⁸O compositions around −10 to −11‰ VSMOW. Furthermore, $δ$¹⁸O_FLUID values from the majority of veins and vugs that post-date concretion growth are even more negative (minimum of −16.5 ± 1.6‰ VSMOW); it is probably these compositions that record an orographic component, driven by the high-relief volcanic landscape that had formed by around 58.9 Ma (Hamilton et al. 1998; Bell and Williamson 2002). For comparison, modern altitude effects on $δ$¹⁸O in precipitation range from −1.7 to −5.0‰ km⁻¹ (global mean −2.8‰ km⁻¹; Poage and Chamberlain 2001). Assuming Hebridean Paleocene volcanic relief around 1200 m, our observed difference of c. 4‰ between highest and lowest inferred Paleocene meteoric water $δ$¹⁸O compositions is comfortably within the modern range for altitude effects.

Our results highlight the growing realization that secure chronologies on basinal burial diagenetic cementation are crucial for the correct interpretation of paragenetic evolution based on geochemical data. It has recently proven feasible to radiometrically date suitable calcite cements using U/Pb techniques (Jahn and Cuvellier 1994; Rasbury and Cole 2009; Mangenot et al. 2018a; Roberts et al. 2020). However, such dating is difficult in many calcite-cemented sandstones owing to relatively high common Pb levels or low U/Pb ratios (Su et al. 2022); some Paleogene Hebridean calcites are known to have too little radiogenic Pb to yield an age (MacDonald et al. 2019).

The case of the Valtos Sandstone Formation concretions described here is unusual, probably unique; clumped isotope data, combined with traditional stable isotope and trace element data, have allowed an unusually precise understanding of cementation chronology because the cementation timing is genetically related to a Paleocene volcanic history that has been radiometrically dated. Ironically, although the chronology of the Paleocene Igneous Event may still be considered quite poor volcanologically, it provides a relative chronology for cementation history of comparable quality to those based on U/Pb dates (e.g. Hill et al. 2016; Beaudoin et al. 2018; Li et al. 2022).

Our new clumped isotope data show that a number of aspects of previous interpretations about cementation history in the Valtos Sandstone Formation were wrong. The studied concretions were not cemented in the Jurassic, or the Cretaceous (Hudson and Andrews 1987; Wilkinson 1993), but in the Paleocene. The sandstone aquifer was apparently not significantly renewed by inflowing porewater for almost 100 myr, a finding that is highly relevant to other sandstone hydrocarbon reservoirs or aquifers. Meteoric water that did access the aquifer was at very shallow burial depth, no more than 100 m. This groundwater was locally recharged, then heated by Paleocene phreato-magmatic (vent) activity. Subsequently, this fluid was mixed in the sandstone pore-system with upward flowing basinal fluids. The rapidly growing Paleocene lava pile caused compactional basinal fluid flow and sandstone compaction. The latter may have been in part concurrent with later growing concretions (those with Z2 geochemical characteristics) and may explain why compactional deformation is not obvious around all concretions.

The radiometric dates from the Skye Lava Group show that the concretionary cementation and final compaction was completed in no more than 2.6 myr, far more rapidly than modelled previously (Wilkinson and Dampier 1990). There is no evidence that the sandstone was a conduit for convection-driven meteoric water emanating from the Skye igneous centre hydrothermal system (Schofield et al. 2016). Such flow would necessarily have post-dated concretion formation but there is no record of it in the Valtos Formation or in the older Bearreraig Sandstone (Lewis et al. 1992; Thrasher 1992).

The clumped isotope data also demonstrate that paragenetic histories based on geochemical data where assumptions have been made about burial temperatures and/or pore fluid $δ$¹⁸O compositions should be regarded as insecure. Our new discoveries were unexpected and have forced us to rethink the burial diagenetic and geological history of rocks we thought we understood. Until now we have missed the significance of Paleocene volcanic rocks resting on an erosive contact with underlying GEG rocks. It had been concluded (Hudson and Andrews 1987) that the GEG rocks had largely completed their sedimentary diagenesis long before the arrival of Paleocene lavas. The possibility that the Valtos Formation sandstones were a high-permeability conduit that briefly transmitted hot, meteoric-sourced, shallow groundwater in the Paleocene was not considered. We now know that cement temperatures >50°C in the GEG mainly come from the large concretions of the Valtos Sandstone Formation: cement temperatures >50°C are uncommon in mudstone-dominated units (Lealt Shale and Duntulm formations; Fig. 1) above and below the sandstone (Paxton 2022); these acted largely as aquitards, with fluid flow restricted to faults, fractures and thin permeable sandy beds. The mudstone-dominated formations thus preserve the most reliable record of the overall GEG burial conditions, albeit with temperatures estimated at only c. 50–60°C. The Valtos sandstone records a brief period of hydrothermal or basinal fluid temperatures >50°C and <100°C during the Paleocene.

Core plugs from VR1 were collected by M. Wilkinson and made available from the Leicester University archive collection by V. Ward, who also provided thin sections of the plugs. E. Gallagher and M. Wakefield provided additional samples; in particular, basalt from the Eigg Lava Formation, collected from the SW side of Laig Bay. We thank J. Hudson, B. Bell, T. Anderson and J. Faithful for helpful suggestions and discussion; M. Wilkinson and an anonymous reviewer made helpful comments on the paper.

RBP: conceptualization (supporting), formal analysis (lead), investigation (lead), methodology (equal), project administration (supporting), resources (supporting), validation (equal), writing – original draft (lead), writing – review & editing (equal); JEA: conceptualization (lead), funding acquisition (lead), investigation (supporting), methodology (lead), project administration (lead), supervision (lead), writing – original draft (supporting), writing – review & editing (equal); PFD: formal analysis (supporting), funding acquisition (supporting), investigation (supporting), methodology (supporting), project administration (supporting), resources (lead), supervision (equal), validation (supporting), writing – review & editing (supporting); ADM: data curation (equal), formal analysis (supporting), investigation (supporting), supervision (supporting), validation (supporting), writing – review & editing (supporting); CH: formal analysis (supporting), writing – review & editing (supporting)

This work was funded by a Natural Environment Research Council (NE/L002582/1) studentship to R.P. through the ENVEAST Doctoral Training Partnership at UEA.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

All data generated or analysed during this study are included in this published article and its supplementary information files.