Trace element release from Bowland Shale into spring water in Lancashire Open Access

Supplementary material: Raw LECO (TOC % and S %) data for the Crimpton Brook Bowland Shale sample analyses are available at https://doi.org/10.6084/m9.figshare.c.6379094

The mid-Carboniferous (late Visean–early Namurian) Bowland Shale Formation, Craven Group, consists of hemipelagic–pelagic shale that outcrops across central Britain and Ireland (Earp et al. 1961; Fewtrell and Smith 1980; Waters et al. 2009; Andrews 2013; Newport et al. 2018). The shale passes northwards into successions dominated by shallow-water limestone and coal. The shale is dark grey to black, and it contains organic carbon concentrations up to 9.29 wt% (average 2–3 wt%) (Andrews 2013; Parnell et al. 2016; Fauchille et al. 2017; Hennissen et al. 2017; Clarke et al. 2018). The Bowland Shale Formation contains anomalously high selenium (Se) contents throughout the outcrop (Parnell et al. 2016). These concentrations (max Se = 42.0 ppm; mean Se = 21.1 ppm) are more than ten times the global mean shale concentration of 1.3 ppm Se (Stüeken et al. 2015). Even most metalliferous black shales have Se concentrations an order of magnitude lower than the Bowland Shale Formation values (Mitchell et al. 2012). It is likely that the shale may have acted as a source of metals during the formation of proximal sulfide mineralization in the Forest of Bowland (Wadge et al. 1983; Hughes 1986; Jones et al. 1994).

Weathering of the Bowland Shale and equivalents liberates Se and other trace elements into the environment. This follows the incorporation of anomalous levels of trace elements into pyrite during early diagenesis, and the subsequent mobilization upon oxidative alteration of the pyrite at the surface. In particular, Se, arsenic (As) and molybdenum (Mo) released from the Bowland Shale become enriched within ochreous precipitates (Parnell et al. 2018; Armstrong et al. 2019). These elements are potentially toxic to livestock in regions of Bowland shale outcrop (Fleming and Walsh 1957; Webb et al. 1966; Brogan et al. 1973; Rogers et al. 1990). The liberation and concentration of trace elements from the Bowland Shale has been described from weathered sea-cliffs (Armstrong et al. 2019) and landslides (Parnell et al. 2018). Surface oxidation and liberation of elements has also been suggested where groundwater aquifers interact with the Bowland Shale and emerge as springs at the surface (Parnell et al. 2016). This account reports data from a spring at Crimpton Brook, Lancashire, described as a ‘sulfur spa’ in OS maps dating back to the 1850s (Ordnance Survey 1850).

Sulfurous springs occur across the Carboniferous Craven Basin (Bowland Sub-basin), Lancashire and Yorkshire (Oliver 1882; Murphy et al. 2014) (Fig. 1). The springs include historically important ‘spas’ (sulfurous springs) at Skipton, Broughton, Clitheroe, Fooden and Wigglesworth (Short 1734; Elliot 1781; Granville 1841). It has been suggested that the spring waters are related to Carboniferous hydrocarbons (Murphy et al. 2014), and possibly Carboniferous deep evaporites by analogy with springs to the south in Derbyshire (Gunn et al. 2006). Deposits from Crimpton (Fig. 2) were studied to assess:

1. Whether the Bowland Shale Formation at the locality was a source of anomalous Se and As.

2. Whether sulfurous deposits were derived from the Bowland Shale Formation, rather than a deep source.

3. Whether the sulfurous deposits conferred trace elements from the Bowland Shale Formation into the environment.

Crimpton Brook is located in the Ribble Valley District, within the Forest of Bowland Area of Outstanding Natural Beauty, Lancashire, approximately 4 km SE of Dunsop Bridge. The headwaters of Crimpton Brook originate as a series of three to four small springs and gulleys draining from Marl Hill Moor to the south and east, which converge over a c. 250 m transect. Outcrops of dark grey-black, pyritic Bowland Shale are exposed in stream-cut sections, where the shale is commonly weathered and oxidized (grid reference SD 6829 4703; 260–240 m a.s.l.). The Crimpton Brook section of Bowland Shales is a thin sliver preserved above a discontinuity surface on the Bollandoceras hodderense beds (Wadge et al. 1983). Stratigraphically, the outcrop is Pendleian stage (E₁b) Upper Bowland Shale Formation, with the nearest marine band that of T. pseudobilinguis (E. pseudobilingue) (Aitkenhead 1992).

Samples were collected from in situ outcrop adjacent to Crimpton Brook, 100–200 m upstream (SE) of the marked ‘sulfur spa’ (Fig. 2). The available outcrop is extensively weathered and oxidized, with evidence of sulfate and iron oxide alteration products on bedding surfaces and fractures. Samples collected for analysis were chosen for their limited state of weathering and oxidation, although minor oxidative weathering cannot be discounted.

Measurements were made using elemental analysis of Bowland Shale samples, laser ablation inductively coupled plasma-mass spectrometry (LA-ICP-MS) analysis of pyrite in the shale, isotopic analysis of pyrite and alteration products, and whole-rock analysis of shale and alteration products.

Trace element distributions across sample surfaces were analysed using spatially resolved LA-ICP-MS at the University of Aberdeen Trace Element Speciation Laboratory (TESLA). Sample surfaces were cut using a diamond-edged rock saw perpendicular to bedding and subsequently prepared by flattening and polishing using carborundum and alumina polishing compounds. Analysis was performed using a New Wave laser ablation system UP213 nm using an output of 1 J cm⁻² and 10 Hz repetition rate and linked with an Agilent 7900 mass spectrometer. Ablation was performed in a grid pattern across the sample surface with a round ablation spot size of 100 µm, a line spacing of 100 µm and a 50 µm s⁻¹ ablation speed. A 15 s warm-up was applied before ablation start and a 15 s delay was applied between each ablation line. The following isotopes were monitored (dwell time): ⁵⁷Fe (0.001 s), ⁶⁵Cu (0.001 s), ⁷⁵As (0.05 s), ⁷⁸Se (0.1 s), ⁸²Se (0.1 s), ¹²⁵Te (0.1 s), ¹²⁶Te (0.1 s). Parameters were optimized using NIST Glass 612 to obtain maximum sensitivity and to ensure low oxide formation. The ratio ²³²Th¹⁶O⁺/²³²Th⁺ (as 248/232) was used to assess oxide formation and maintained under 0.3%. A reaction cell was utilized to remove Se analysis interferences using hydrogen gas. The international sulfide standard MASS-1 (USGS, Reston, VA) was used for quantitative sample calibration prior to analysis. Element concentration maps were produced using SigmaPlot in pyritic black shale.

S-isotope analysis was conducted at the stable isotopes facility at SUERC, East Kilbride, using standard extraction and analysis methods (Robinson and Kusakabe 1975). Bulk powders of crushed precipitate and pyrite extracted from black shale samples were analysed for their S-isotopic composition. For each sample, 5–10 mg was heated to 1100°C in a vacuum line with an excess of copper(I) oxide (Cu₂O) to convert all available S in the sample to SO₂. Using condensation and pentane traps, the SO₂ gas fraction was isolated from all other extraneous gas phases (N₂ and CO₂). The SO₂ fraction was subsequently analysed for ³⁴S and ³²S concentrations using a VG Isotech SIRA II mass spectrometer. The mass spectrometer was calibrated using four reference standards composing sulfides and sulfates (SUERC internal standards CP-1 and BIS and international reference standards NBS-123 and IAEA-S-3) with a reproducibility of <0.2‰ for each standard. The δ³⁴S for each sample was calculated relative to the Vienna-Canyon Diablo Troilite reference using standard corrections.

Elemental analysis was performed on samples of secondary mineralization and black shale from Crimpton Brook at ALS Laboratories, Loughrea, Republic of Ireland, using analytical methods ME-MS41 and ME-MS41L. 0.5 g of pulverized sample was digested using a 75% aqua regia solution (3:1 HCl:HNO₃) in a graphite heating block before cooling and neutralized using deionized water. The digested solution was then analysed by inductively coupled plasma-atomic emission spectrometry (ICP-AES) and inductively coupled plasma-mass spectrometry (ICP-MS) with results corrected for spectral interelement interferences. Whole-rock data are calculated based on original sample weight before digestion. This method represents a partial digest, including full digestion of sulfides, organic matter, clays and oxides, while some silicates remain undissolved in the residue. ME-MS41 and ME-MS41L follow the same analytical method and are directly comparable, with improved detection limits using ME-MS41L.

SEM was conducted on samples of black shale and secondary mineralization at the Aberdeen Centre for Electron Microscopy, Analysis and Characterization (ACEMAC) facility at the University of Aberdeen. Carbon-coated samples were analysed at a 15 kV accelerating voltage, 10.5 m working distance and 60 µm aperture size using a Zeiss Gemini SEM 300 Field Emission Gun Scanning Electron Microscope (FEG-SEM). A Zeiss solid state BSE detector (4 quadrant) was utilized for backscatter detection and imaging and an Oxford Instruments X-Max 80 detector combined with Aztec spectra interpretation software was used to conduct electron dispersive x-ray (EDX) analysis to determine elemental compositions. Elemental concentrations higher than c. 0.5% were detectable for elements with an atomic number of 6 or higher. Interelemental spectral interferences were accounted for during EDX analysis.

The TOC and TS contents of the powdered samples were determined using a LECO CS744 at the University of Aberdeen. Total carbon (TC) and TS concentrations were identified by combustion analysis of 0.1 mg powdered samples, using an excess of tungsten and iron chip combustion accelerators. To determine TOC, inorganic carbon was removed by dissolving 0.2 mg of a sample in an excess of 20% hydrochloric acid (HCl), producing a decarbonated sample fraction. Decarbonated samples were subsequently analysed for TOC using LECO combustion analysis.

Certified LECO standards were used to produce a multi-point calibration daily before sample analysis. Background C and S contents were accounted for using blanks and subtracted from final values. Each sample was analysed in duplicate and an average calculated to account for any analytical variability. Real standard deviation (RSD) values were calculated for each sample to ensure analytical precision. Where RSD values exceed 5%, samples were reanalysed in duplicate.

Whole-rock analyses (Table 1) of two Bowland Shale samples from Crimpton Brook were compared with the encrustations at the spring, and with coeval (basal Namurian) coal and shaly limestone facies within 50 km to the north in the Yorkshire Pennines. All reported element concentrations are an order of magnitude greater than in the coeval coal and shaly limestone, and several elements (As, Sb, Mo) are an order of magnitude greater than the global mean shale and upper crustal compositions (Turekian and Wedepohl 1961; Rudnick and Gao 2003; Hu and Gao 2008; Stüeken et al. 2015).

Orange iron oxide (‘ochre’) staining and accumulations are present on outcrops and within the stream waters (Fig. 3a), with yellow and white jarosite mineralization present on some shale surfaces. The yellow jarosite is intermixed with iron oxide. The encrustations consist of aggregates of sub-micrometre crystals, with incorporated diatoms (Fig. 4). Whole rock analyses (Table 1) show that these samples of encrustation are richer in As than the Bowland Shale, and less rich in cadmium (Cd), lead (Pb), antimony (Sb) and uranium (U) than the shale but still well above global mean values. Proximal to these outcrops is an area labelled as ‘sulfur spa’ on Ordnance Survey maps (Fig. 2), where a few unmapped seasonal springs issue from the hillside in a boggy area of moorland. These springs are often stained orange or white with oxide mineralization and the area has a sulfurous smell (Fig. 3b). Spring-waters entering Wigglesworth Beck (SD 80162, 56683), to the NE of Crimpton Brook (Fig. 1), show comparable (but more substantial) white encrustation (Fig. 3c) after draining from bedrock of the Bowland Shale.

Elemental TOC and S analysis of the pyritic Bowland shale gives a mean TOC of 3.87 ± 0.2% (n = 8) and a mean S of 6.21 ± 0.3% (Fig. 5; Supplementary Information 1). S contents in these shales are elevated in comparison to normal marine line TOC v. S ratios (Berner and Raiswell 1984), indicative of the high pyrite contents of the shale.

LA-ICP-MS mapping showed that Se residence is strongly concentrated in crystals of pyrite (Fig. 6). Measurements of the pyrite yielded a mean composition of >50 ppm Se. At a fine scale the pyrite consists of numerous framboids, up to 10 µm in size (Fig. 4a).

Pyrite samples from the Bowland Shale have δ³⁴S compositions of −25.9, −25.9 and −27.9‰. Four samples of jarosite encrustations in the stream have compositions of −33.7, −23.6, −19.8 and −3.2‰. For regional context, the composition of sulfate evaporites of Lower Carboniferous age in northern England is in the range +15 to +22‰ (Crowley et al. 1997).

The whole-rock contents of trace elements in the Bowland Shale are high because the shale contains large amounts of pyrite, in which most trace elements reside. The TOC v. S data (Fig. 5) demonstrate that the S content is in excess of that normally expected in modern marine carbonaceous sediments (Berner and Raiswell 1984), due to the unusually high quantity of pyrite, and therefore trace elements. Diagenesis has minimal impact on the TOC/S ratios and corresponding trace element contents of sedimentary rocks, due to the stability of sulfides formed in seafloor sediments, and therefore comparison of Carboniferous shales to modern marine sediments is valid. These TOC v. S data are indicative of marine depositional conditions for the Bowland Shale at Crimpton Brook, with an average TOC/S ratio of 0.62, which is within the range identified for Carboniferous marine shales (0.5–5), though distinctly sulfur-rich (Berner and Raiswell 1984). Based on these data, there is no evidence of non-marine depositional conditions in the Bowland Shale strata analysed at Crimpton Brook, which would be characterized by a significantly higher TOC/S ratio (>10).

The coeval coals and shaly limestones have low S contents, and accordingly have low trace element contents. Unlike other trace elements, vanadium (V) levels in the shale are lower than global averages, because V resides in silicates/oxides, rather than in sulfides. The concentrations of Se, As, Cd and thallium at Crimpton Brook are all greater than the mean values determined for the Bowland Shale Formation at Mam Tor, Derbyshire (Parnell et al. 2018). The encrustations contain more As than the parent shale, which already contained high levels of As. Other elements, including Cd, Se, Mo, Pb, Sb and U, are also anomalously high in the encrustations, although lower than in the shale.

The LA-ICP-MS mapping of the shale at Crimpton Brook shows that Se is concentrated within the pyrite, as recorded elsewhere in the Bowland Shale outcrop (Parnell et al. 2018). Residence in the pyrite indicates that the Se will be mobile when the shale becomes weathered and the pyrite is oxidized. S liberated from altered pyrite is highly mobile as sulfuric acid, while liberated Se may occur in elemental form or as selenide (Se²⁻) ions under low Eh (reducing) acidic conditions, or as water soluble selenate (⁠$SeO42−$⁠) and selenite (⁠$SeO32−$⁠) under oxidizing conditions (Belzile et al. 1997; Strawn et al. 2002). These Se species can be adsorbed onto iron oxide or incorporated into sulfates such as jarosite (Harada and Takahashi 2008).

The light S isotopic composition of the pyrite at Crimpton Brook is typical of an origin by microbial sulfate reduction (Machel 2001). The framboidal texture (Fig. 4a) is typical of the low-temperature microbial formation of iron sulfides (Wacey et al. 2015). The incorporation of As into pyrite is also mediated by microbial sulfate reduction (Fischer et al. 2021; Gao et al. 2021). This is the origin of much pyrite in black shales. The jarosite could be related to alteration of the pyrite, or from hydrogen sulfide sourced from deep evaporites. Three of the four samples of jarosite have light isotopic compositions similar to the pyrite, but much lighter than the Lower Carboniferous evaporites in the region. We regard the remaining sample as aberrant. Pyrite is a proximal source, while evaporites would be a distal source of S. Overall, the sulfurous precipitate is attributable to alteration of Bowland Shale pyrite. A similar comparison of jarosite and pyrite in the Bowland Shale equivalent at Ballybunion Bay, Co. Kerry, Republic of Ireland (Armstrong et al. 2019), also suggests that pyrite is the source of S in the jarosite (Fig. 7).

The occurrence of numerous sulfurous springs in the outcrop of the Craven Group (Fig. 1) suggests that they have a common source. S-isotope data from water in some of the other springs are quite variable (Murphy et al. 2014), but possibly reflect a range of waters from proximal to the source shale, as at Crimpton Brook, and distal to the shale where there is an opportunity for fractionation in the intervening aquifer. This variation may also be indicative of fluctuations in the S-isotope values of pyrite in the Bowland Shale across the region due to basin salinity changes during deposition (Gross et al. 2015), although this has not been observed.

Lower Carboniferous shale is also encountered beneath the surface of Marl Hill Moor, where the Marl Hill Tunnel carries water from Haweswater, Cumbria, to Greater Manchester. The shale in the tunnel releases hydrogen sulfide, which leaves S-rich deposits on the tunnel walls (Earp et al. 1961). The weathering in the tunnel repeats the weathering in the Crimpton Brook above, and suggests that the shale alteration is the origin of much of the sulfurous water flow in the region. Construction of the Marl Hill Tunnel and other tunnels along the route showed that water circulation was strongly controlled by faults and other fractures (Earp 1955).

A comparison of Bowland Shale data for Se with contents for other black shales that are implicated in environmental problems was made by Parnell et al. (2018), using samples from Derbyshire. One of the main environmental problems associated with elevated Se levels in bedrock, and the associated soils and drinking water is the potential for Se poisoning ('selenonis’) of humans and livestock, which can cause hair loss, nausea and pulmonary oedema, and can be fatal in extreme cases (Fordyce 2013; Wrobel et al. 2016). The Se concentration for shale at Crimpton Brook is an order of magnitude greater than that in black shales in the USA, France and Nigeria that are all considered to be potentially toxic due to the Se contamination of drinking waters, agricultural soils and aquatic environments linked to these shales (Tuttle et al. 2014; Bassil et al. 2016; Nganje et al. 2020). Among the other elements present at anomalous concentrations at Crimpton Brook, Mo has been found to be an environmental problem at other British and Irish sites on black shales coeval with the Bowland Shale (Webb and Atkinson 1965; Webb et al. 1966; Thomson et al. 1972; Brogan et al. 1973). Specifically, elevated Mo concentrations in soils (>3 ppm), can be toxic and cause severe copper deficiencies in livestock (primarily cattle), resulting in poor health and livestock growth (Thomson et al. 1972). However, the Crimpton Brook exposure is very small, and of no immediate concern given the limited surface area for liberation of trace elements and contamination of nearby drinking waters and soils. Similar rocks on a much larger scale could represent an environmental problem, as found in drainage systems from pyrite-rich deposits globally (Evangelou and Zhang 1995; Parviainen and Loukola-Ruskeeniemi 2019).

Elevated concentrations of Se, As and other elements recorded in the Bowland Shale Formation at Crimpton Brook are resident in pyrite, and are released into alteration products upon surface oxidation of the shale. The mean Se concentration of the shales is greater than that in rocks that have caused environmental problems elsewhere. Although the alteration products at the spring site are volumetrically very limited, and do not constitute a hazard, the data emphasize the need for caution in any commercial-scale working of the Bowland Shale Formation or other pyrite-rich shales.

We are grateful for help in field sampling from L. Bullock and D. Halbert. SEM in the ACEMAC Facility was supported by J. Still.

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.

JGTA: conceptualization (equal), data curation (lead), formal analysis (lead), investigation (lead), methodology (equal), writing – original draft (equal), writing – review & editing (equal); JP: conceptualization (equal), data curation (equal), funding acquisition (lead), investigation (equal), methodology (equal), project administration (lead), resources (lead), supervision (lead), validation (equal), visualization (equal), writing – original draft (equal), writing – review & editing (equal); AJB: conceptualization (supporting), data curation (equal), formal analysis (supporting), funding acquisition (equal), methodology (equal), resources (equal), validation (supporting); MP: formal analysis (equal), methodology (supporting).

The research was undertaken with the support of the Natural Environment Research Council (Grant Numbers NE/M010953/1 and NE/T003677/1).

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