Introduction

Concentration and isotopic composition of trace elements are widely used to characterize the physico-chemical conditions during fluid-rock interaction and to determine solute sources to natural waters as diverse as groundwater, river water, seawater, or hydrothermal fluids. Over the past 25 years, the rare earths and yttrium (REY) have become a powerful geochemical tool to gather information on the genesis of hydrothermal deposits, and hydrothermal fluorite deposits have been the focus of particular attention (e.g., Schneider et al. 1975; Möller and Morteani 1983; Dill et al. 1986; Eppinger and Closs 1990; Möller 1991; Lüders et al. 1993; Bau and Dulski 1995; Möller et al. 1998; Williams-Jones et al. 2000; and references therein). Moreover, Nd isotopes have been used to date fluorite mineralization events (Chesley et al. 1991, 1994; Galindo et al. 1994). Although it has been criticized (Nägler et al. 1995), this approach may gain momentum after experimental evidence recently highlighted the slow diffusion rates of REY in fluorite (Cherniak et al. 2001). In addition to radiometric dating, the study of isotopes (Sr, Pb, Nd) is often preferred over the study of solute concentrations, because concentrations in the solution are affected by element uptake and loss during fluid-rock interaction, whereas Sr, Nd, and Pb isotope ratios are altered solely by element uptake. This basic difference favors isotope studies even more when the fluid itself is not accessible and any conclusion regarding solute sources and physico-chemical conditions during fluid-rock interaction can only be derived from precipitates such as hydrothermal minerals or chemical sediments. Hydrothermal ore deposits are a prime example of a research area that relies heavily on such indirect evidence, since fluid inclusion studies are the only way to obtain direct evidence from these fossil hydrothermal systems. Hence, combining evidence from radiogenic isotopes with that from REY distribution should provide valuable insight into fossil hydrothermal systems.

Carbonate-hosted Mississippi Valley Type (MVT) deposits are important sources of base metals such as lead and zinc and locally fluorite. Studies of fluid inclusions hosted in minerals from MVT deposits (e.g., Sawkins 1966; Roedder 1967; Rogers 1977; Crocetti and Holland 1989; Masheder and Rankin 1988; Kesler et al. 1995; Viets et al. 1996) have revealed compelling evidence for the deposition of ore and gangue minerals from mildly hot (<200 °C), high-salinity NaCl-CaCl2 brines (>15 wt% NaCl equivalent) that are similar in fluid chemistry to modern oil field brines (Hanor 1979, 1995; Fritz and Frape 1987). The widely favored genetic model for MVT deposits assumes large-scale migration of the ore-forming hydrothermal fluids out of deep sedimentary basins in response to compressional tectonic events (e.g., Anderson and Macqueen 1982; Garven and Freeze 1984; Sverjensky 1986; Leach et al. 2001). Although this genetic concept appears to be generally applicable, the chemical diversity between individual MVT deposits is large and requires each deposit or ore-district to be addressed separately.

Here, we report the results of an integrated study of REY distribution and Sr-Nd-Pb isotope systematics in fluorite and its limestone host rocks from the MVT deposits in the South Pennine Orefield, England. The study aims at the characterization of potential source rocks of solutes transported by the fluorite-forming hydrothermal solution(s) and at the attempt to radiometrically date the famous Blue John fluorite mineralization. We also extend existing work on fluorite from the North Pennine Orefield (Shepherd et al. 1982; Halliday et al. 1990; Jones et al. 1991), and compare the trace element and isotope characteristics of fluorite in the southern ore district to those of its northern counterpart.

The South Pennine Orefield

The South Pennine Orefield is situated within the Derbyshire Dome in central England some 20 km southwest of Sheffield (Fig. 1). It consists of Lower Carboniferous (Dinantian) platform carbonates with interbedded basic lavas and tuffs, chert, and dolomite (Walters and Ineson 1980; Mostaghel and Ford 1986). The Carboniferous carbonate rocks can be as thick as 1,800 m and overlie Lower Paleozoic or Precambrian basement. Ordovician to Devonian strata had been eroded before Lower Carboniferous times.

Fig. 1.
figure 1

Map showing structural features, mineral veins, and sample sites in the South Pennine Orefield (Derbyshire Dome). Key to locations: 1 Treak Cliff Caverns / Castelton, 2 Mitchell Bank, 3 Dirtlow open-pit, 4 Smalldale, 5 Lady Wash mine; in white: Carboniferous strata; inset relates positions of the South and North Pennine Orefields (SPO and NPO, respectively.) to the rest of the UK

Full size image

Mineralization is particularly abundant in the central and eastern part of the Derbyshire Dome (Fig. 1). The mineralization consists to more than 90% of fluorite (Dunham 1983). Among the sulfides, galena is the most common, whereas sphalerite, chalcopyrite, and pyrite are less abundant. The mineralization occurs in four different forms, which traditionally have been classified as "rakes", "scrins", "flats", and "pipes" according to their relation to tectonic and sedimentary settings (Ford 1967). In the eastern part of the Derbyshire Dome, fluorite is the most common gangue mineral, whereas calcite is more abundant in the western part of the vein structures. Sphalerite is concentrated in the east, but galena is abundant throughout both the eastern and western parts. A detailed description of the mineralogy and paragenesis of the South Pennine Orefield is given by Ixer and Townley (1979) and Dunham (1983).

Formation of MVT mineralization in the South Pennine Orefield is thought to have occurred in multiple episodes. The mineralization in the northernmost part of the Derbyshire Dome (Treak Cliff Caverns and Windy Knoll) is assumed to be of Late Carboniferous age (ca. 290 Ma; Ewbank et al. 1995), whereas in other parts of the South Pennine Orefield, the mineralization is thought to have formed during several distinct periods starting in the Early Permian (Ineson and Mitchel 1972; Dunham 1983). For the North Pennine Orefield, Rb-Sr isochron data from fluid inclusions suggest a Triassic age of quartz that occurs within the MVT fluorite mineralization (Shepherd et al. 1982). This result is compatible with evidence from Sr and Nd isotopes in fluorite (Halliday et al. 1990). However, Brannon et al. (1991) demonstrated for hydrothermal minerals from the Viburnum Trend, SE Missouri, that there can be significant differences in Sr-isotopic compositions between host minerals and primary fluid inclusions therein, and hence, the true mineralization age(s) in the study area may still be unclear.

There also has been some debate of the origin of the mineralizing fluids. Ford (1976) postulated a possible origin of the ore-forming fluids in the North Sea Basin. In contrast, Robinson and Ineson (1979) favored less distant sources, namely the adjacent Edale and Widmerpool Gulfs, and the Gainsborough Trough (Fig. 2), whereas Atkinson (1983) suggested the source of the fluids were Permo-Triassic basins. Ewbank et al. (1995) studied the relationship between bitumens and mineralization. They suggested that the mineralization in the area of Castleton (Fig. 1) is related to the dewatering of the Edale Gulf during the Late Carboniferous, whereas in other parts of the South Pennine Orefield, basinal brines from the east and from the west started to penetrate the Derbyshire Dome in Early Permian times. Combined studies of noble gases and halogens in fluid inclusions hosted in calcite and fluorite from the Dirtlow and Hucklow Edge Rakes in the NE part of the Derbyshire Dome (Kendrick et al. 2002) support an origin of the mineral-forming fluids from small local basins such as the Widmerpool and Edale Gulfs. However, the concentration of 40Ar is too high to result from fluid interaction with a K-poor limestone alone, suggesting significant water-rock interaction with Namurian shales. 4He data for these fluid inclusions suggest that the fluids had been stored at depth for 40 to 50 Ma before they have been trapped in fluorite. This is compatible with a Late Carboniferous to Early Permian age for the mineralization if the fluids were derived from Namurian shales (Kendrick et al. 2002).

Fig. 2
figure 2

. Sketch map showing Dinantian to early Namurian paleogeography superimposed on the present-day geology (after Ewbank et al. 1993)

Full size image

Primary fluid inclusions hosted in fluorite from the South Pennine Orefield show homogenization temperatures in the range between 70 and 155 °C (Rogers 1977). These temperatures must be pressure-corrected by about 10 °C to obtain the true trapping temperatures, assuming a sedimentary cover of about 1 km at the time of mineral formation (Rogers 1977). The salinity of the inclusions varies between 10 and 30 wt% NaCl equivalent with a predominance of values around 25 wt% NaCl equivalent (Rogers 1977). Pseudosecondary fluid inclusions in fluorite from MVT deposits locally contain oil (Roedder 1967). However, Moser et al. (1992) have demonstrated that fluorite from Windy Knoll and Treak Cliff Caverns (Fig. 1) contains petroleum-bearing fluid inclusions and that these inclusions differ in salinity as well as by their homogenization temperatures from those trapped in typical MVT fluorite from the South Pennine Orefield. These authors assume that the hydrocarbons in the inclusions may have been derived from local host rocks and migrated with the fluorite-forming fluids from nearby sedimentary basins. Low-salinity fluid inclusions in fluorite from the Treak Cliff Caverns are also reported by Ewbank et al. (1995).

The δ34S values of galena from the South Pennine Orefield range from −23 to +6.5‰; younger galena tends to have heavier δ34S values than older precipitates (Robinson and Ineson 1979). The δ34S values of barite vary between +4 and +22.5‰. Robinson and Ineson (1979) explain the variation of the δ34S values of galena and barite by different H2S sources and mixing of fresh water sulfate with connate seawater sulfate. In the same study, carbon and oxygen isotopic compositions of vein calcites and carbonate host-rocks were investigated. The results reveal that the δ13C values of vein calcites are very similar to those of the host carbonates and suggest carbon supply from the Carboniferous limestones into the veins.

Lead isotope studies on bulk samples and leaching solutions from potential source rocks, i.e., Namurian shales and sandstones, and Old Red sandstones suggest Viséan-Namurian shales as a probable source of Pb in galena from the South Pennine Orefield (Jones et al. 1991; Jones and Swainbank 1993). However, since the ore-forming fluids may have mineralized both the carbonate rocks and the sandstones, this conclusion is not well constrained (Jones and Swainbank 1993).

The source of fluorine in MVT deposits is still unclear. For the Illinois-Kentucky fluorite district Plumlee et al. (1995) suggested that magmatic fluorine was admixed to basinal brines. For the North Pennine Orefield it has been suggested that during a post-Permian period of tensional stress related to pre-Atlantic rifting, fluids penetrated alkaline rocks at deeper crustal levels where magmatic fluorine was added (Sawkins 1966; Halliday et al. 1990). Such a model is very unlikely for the South Pennine Orefield, because noble gas and halogen data of fluid inclusions show no evidence of magmatic volatiles in the mineral-forming fluids but rather support the conventional basinal brine model (Kendrick et al. 2002). Hence, the most probable sources for fluorine are evaporated seawater and/or Namurian shales (Kendrick et al. 2002).

Sample origin

The samples studied here originate from various localities in the northern and north-eastern part of the South Pennine Orefield (Nos. 1 to 5 in Fig. 1). The Treak Cliff Caverns (No. 1 in Fig. 1) are the type locality of the so called "Blue John" fluorite, a variety of light blue, purple, and white striated fluorite (Fig. 3) that occurs in veins in about 340 Ma old (Asbian), crinoide-bearing reef limestones some 800 m west of the village of Castleton (Fig. 1). Blue John fluorite has been mined since at least the early 19th century and is famous for its use as an ornamental stone (Ford 1967). Rather similar fluorite (that can come in a yellow color) has also been sampled from the Mitchell Bank (No. 2 in Fig. 1). Fluorite, massive limestone, and recrystallized sparry calcite were collected in the Dirtlow open-pit (No. 3 in Fig. 1). The Dirtlow open-pit is located to the north wall of the Dirtlow rake (Fig. 1), which is an approximately 6-km-long, NE-striking vein structure with a width of 3 to 10 m. A detailed description of the ore body is given by Butcher and Hedges (1987). Fluorite samples from this locality are mostly blue or purple, but some clear samples have also been found. White fluorite samples were collected from a vein at Smalldale (No. 4 in Fig. 1). For comparison, fluorite was also sampled from the dump of the Lady Wash Mine (No. 5 in Fig. 1) about 5 km south of the Dirtlow open-pit, where fluorite is associated with abundant sphalerite and galena.

Fig. 3
figure 3

. Sample of "Blue John" banded fluorite from the Treak Cliff Caverns (1 in Fig. 1) referred to as "Blue John reference sample" throughout this paper. Numbers indicate analyzed sections of the sample (see text for details)

Full size image

To improve the data base for comparison of fluorite from the South and North Pennine Orefield (Fig. 1), we also studied fluorite from the Frazer's Hush Mine in the northern district, where mostly Pb and Zn have been mined from the so-called Groverake.

The Blue John fluorite "reference sample"

After reviewing the samples of fluorite from various locations in the South Pennine Orefield, we decided to focus our isotope study on a set of nine samples that originate from nine adjacent bands in a single specimen from the Treak Cliff Caverns (Fig. 3). These fluorite bands were deposited in sequence and therefore provide some insight into the chemical evolution of the fluorite-forming solution with time. We chose this specific specimen because it appears to be representative of Blue John fluorite with respect to texture, color, and trace element composition. The banding is primarily due to color that changes between white and shades of purple-blue.

At the base of this sample is a light-blue fluorite band (No. 9 in Fig. 3) that after a sharp contact is followed by a band that is deep purple-blue (No. 8). Band No. 7 shows a very light-blue fluorite and band No. 6 again a deep purple-blue fluorite. The contacts between these color bands are well defined. Along the contact between band No. 6 (the uppermost band of the blue subsection Nos. 9 to 6) and No. 5 (the lowermost band of the white subsection Nos. 5 to 1), well-developed euhedral crystals of deep purple-blue fluorite (No. 6) are overgrown by white fluorite (No. 5). This suggests that band No. 6 precipitated in an open fracture and might indicate a hiatus between two episodes of mineralization. The relatively thick white fluorite band is arbitrarily subdivided into four samples (Nos. 5 to 2). The white fluorite bands 1 and 2 are separated from each other by a thin layer of light-purple fluorite. Band No. 1 contains minor amounts of calcite and is overgrown by grey sparry calcite (Fig. 3).

Analytical methods

Subsamples of fluorite, calcite and host rock were cut using a diamond saw and finely ground in an agate mortar. The analytical procedure is described in detail elsewhere (Dulski 2001). About 0.1 g of sample powder was dissolved using mixed acid digestion (HF/HClO4 for host rock samples, HF/HNO3 for fluorite) under pressure and finally filled up to 50 ml volume with 0.5 mol l−1 HCl (dilution factor 500). Calcite was dissolved in HNO3. Prior to analysis, Ru and Re were added to aliquots of the solutions as internal standards for drift correction, and the mixtures were diluted by a factor of ten or two, depending on the REE concentration of the specific sample. Solutions prepared following the above procedure show a dilution factor of 5,000 or 1,000, and are about 0.5 mol l−1 in HCl. ICP-MS measurements have been performed using an ELAN 5000A quadrupole ICP mass spectrometer (Perkin-Elmer/SCIEX, Canada). Samples were measured in batches of five. Each batch was preceded by a 10 ng ml−1 calibration solution, two acid blanks and a procedure blank. Quantitative determination of element concentrations was performed applying external calibration using two calibration solutions that bracketed each batch of samples. Calibration solutions are matrix-matched using a decomposition solution of supra pure grade CaF2 (Merck, Germany). To minimize the influence of drift effects on analytical precision relatively small batches of samples were analyzed within one analytical run. Interference corrections are routinely applied to correct analyte isotopes for molecular and isobaric interferences as described by Dulski (1994). The precision of the method has been tested by ten replicate analyses of our fluorite in-house standard C5-FL. For Y and the REE, the precision is better than 3% with the exception of Eu (5.3%). Since fluorite reference samples with recommended REY concentrations are not available, accuracy was checked by comparison of data for in-house standard C5-FL obtained (1) after decomposition with AlCl3 followed by ICP-MS analysis, (2) after decomposition with HF + HClO4 followed by ICP-MS and ICP-AES analysis, respectively, and (3) by instrumental neutron activation analysis (no sample decomposition). The results are in very close agreement with a maximum deviation from the mean value of less than 10%, except for Lu for the ICP-MS measurement after AlCl3 decomposition.

Fluorite samples for Pb analysis were dissolved with HF in Teflon autoclaves. Pb was separated using anion exchange resin Bio Rad AG1-X8 (100–200 mesh) in 0.5-ml Teflon columns by an HCl-HBr ion-exchange chemistry adapted from Tilton (1973) and Manhès et al. (1978). Pb was purified by a second pass over the column. Pb was loaded together with H3PO4 and silica gel on single Re-filaments (Gerstenberger and Haase 1997). The isotopic composition of Pb was determined at 1,200–1250 °C on a Finnigan MAT262 multicollector mass spectrometer using static multicollection. Instrumental fractionation was corrected with 0.1%/a.m.u. as determined from repeated measurement of lead reference material NBS 981. Accuracy and precision of reported Pb ratios is better than 0.1% at the 2-sigma level. Total procedural blanks are better than 30 pg Pb. Fluorite samples for Sr and Nd isotope analysis were dissolved with a HF-HClO4 mixture in Teflon beakers on the hot plate. The samples were split by weight and a mixed 149Sm-150Nd tracer was added to the smaller fraction. Sr was separated using standard cation exchange techniques. Sr was loaded on single Ta-filaments and its isotopic composition was determined on a VG 54-30 Sector multi-collector mass-spectrometer using dynamic multicollection. 87Sr/86Sr data are normalized with 86Sr/88Sr=0.1194. Repeated measurement of Sr standard NBS 987 during the measurement period gave 0.710249±0.000004 (2σ, for 12 measurements). Total procedural blanks are better than 50 pg Sr. Nd and Sm were separated from the other REE using standard cation exchange techniques (HDEHP-coated Teflon). Nd was loaded on double Re-filaments and its isotopic composition was measured on a Finnigan MAT262 multicollector mass-spectrometer using dynamic multicollection. 143Nd/144Nd data are normalized with 146Nd/144Nd=0.7219. Repeated measurement of the La Jolla Nd standard during the measurement period gave 143Nd/144Nd=0.511850±0.000004 (2σ, for 12 measurements). Total procedural blanks are <50 pg Nd. Analytical uncertainties for Sr and Nd are reported as 2σm. Recalculation of the initial isotopic composition used concentration data determined by ICP-MS (Table 1).

Table 1. Trace element concentrations for fluorite and carbonate from the South and North Pennine orefields (all concentrations in µg/g)
Full size table

The distribution of rare earths and yttrium (REY) is illustrated by REYSN patterns (SN: normalized to post-Archean Australian shale from McLennan 1989) where Y is inserted between Dy and Ho according to its ionic radius (for details see Bau 1996).

Results

REY and REYSN patterns

Blue John fluorite is characterized by low REY concentrations. In our reference sample, there is no systematic increase or decrease from band No. 1 to band No. 9, but the subdivision based on texture and color is also seen in the REY contents (Table 1): the Nd concentration, for example, in bands 2 to 5 (0.75–1.29 ppm) is lower than in bands 6 to 9 (1.67–1.97 ppm) and in band No. 1 (1.57 ppm). This subdivision is also evident in the Y/Ho ratios (102–119, 85–89, and 92, respectively). However, band No. 1 contains some calcite in addition to fluorite, which may explain its deviating chemical composition. Purple fluorite tends to show slightly higher REY concentrations than whitish-clear fluorite. The spread of concentrations of individual REY between the nine bands systematically increases with decreasing ionic radius from 1.7 for La to 7.6 for Lu. We emphasize that regardless of their color, all nine fluorite bands show similar REYSN patterns (Fig. 4A) that are strongly depleted in light and heavy REE compared to the middle REE (LREE, HREE, and MREE, respectively). They display strong positive YSN anomalies, small positive GdSN anomalies, and negative CeSN anomalies. There are no EuSN anomalies, but some samples show small positive LaSN anomalies. Regardless of their specific origin, the other fluorite samples from the South Pennine Orefield show REY concentrations and REYSN patterns that are almost indistinguishable from those described above for our Blue John fluorite reference sample (Fig. 4), suggesting that this sample indeed may be used to exemplify the chemical composition of fluorite in the South Pennine Orefield.

Fig. 4A–F
figure 4

. Shale-normalized REY (REYSN) distribution patterns of fluorite samples from various occurrences in the South Pennine Orefield (1, 2, 3, 4, 5 in Fig. 1). Note that all fluorite samples from the South Pennine Orefield share the same characteristics, in particular the negative CeSN and positive GdSN anomalies

Full size image

In contrast to fluorite from the southern ore-district, hydrothermal fluorite from Frazer's Hush mine in the North Pennine Orefield shows significantly higher REY concentrations (e.g., 11.3 to 83 ppm of Nd) and REYSN patterns with a strong positive EuSN anomaly, but without any negative CeSN anomaly (Figs. 5 and 6). This is in agreement with published REE data (Shepherd et al. 1982; Halliday et al. 1990; Jones et al. 1991) for fluorite from other locations in the North Pennine Orefield (Fig. 5).

Fig. 5
figure 5

. Plot of Nd concentration vs. CeSN anomaly (CeSN/Ce*SN) for hydrothermal vein fluorite. Ce*SN=0.5LaSN+0.5PrSN. BJrs Blue John reference sample from South Pennine Orefield (from Fig. 4A); SPO miscellaneous fluorite from the South Pennine Orefield (from Fig. 4B–F); FHM Frazer's Hush Mine, North Pennine Orefield (see also Fig. 6), NPO miscellaneous fluorite from the North Pennine Orefield (data from Halliday et al. 1990; CeSN/Ce*SN arbitarily set at 0.95); KIF miscellaneous fluorite from the Kentucky–Illinois fluorite district (data from Chesley et al. 1994). Note the clear distinction between fluorite from the North and from the South Pennine Orefield, and the similarity between fluorite from the South Pennine Orefield and that from the Kentucky–Illinois district

Full size image
Fig. 6
figure 6

. REYSN distribution patterns of fluorite samples from Frazer's Hush Mine, North Pennine Orefield, England (data for the Blue John fluorite reference sample, South Pennine Orefield, shown for comparison). Note the presence of a positive EuSN anomaly and lack of negative CeSN and positive GdSN anomalies in samples from Frazer's Hush (see text for further discussion)

Full size image

The limestone country-rock at Dirtlow open-pit shows REYSN patterns that are depleted in LREE and flat from the MREE to the HREE (Fig. 7). There are strong positive YSN anomalies, positive LaSN and GdSN anomalies, and negative CeSN anomalies. Such a REY distribution is typical of aluminosilicate-free post-Paleoproterozoic marine sedimentary carbonates and modern seawater (Bau and Dulski 1996; Webb and Kamber 2002). Remobilized sparry calcites show patterns that are similar to those of the limestones, except for a slight decrease from Er or Tm to Lu and a small positive EuSN anomaly (Fig. 7).

Fig. 7
figure 7

. REYSN distribution patterns of Lower Carboniferous limestone and sparry re-crystallized calcite therein from Dirtlow open-pit (3 in Fig. 1). REYSN patterns of limestone are similar to those typical of Phanerozoic to modern pure marine sedimentary carbonates

Full size image

Neodymium isotopes

With two exceptions, the fluorite bands display εNd(290 Ma) values (mineralization age of 290 Ma from Ewbank et al. 1995) in a narrow range between −8.5 and −7.8 (Table 2). The two exceptions are bands 1 and 4, which show more radiogenic values of −7.0 and −7.3, respectively. There is no correlation between the concentration and the isotopic composition of Nd. The heterogeneous initial Nd isotopic composition of Blue John fluorite from the Treak Cliff Caverns precludes the use of a Sm-Nd isochron to determine its age.

Table 2. Rb-Sr. Sm-Nd. and Pb isotope data for fluorite from Treak Cliff Cavern (No. 1 in Fig. 1) and country rock carbonates from Dirtlow open-pit (No. 3 in Fig. 1)
Full size table

The limestones from the Asbian country rock show εNd(340 Ma) values (at time of deposition) between −9.0 and −8.4 and εNd(290 Ma) values (at time of fluorite mineralization) between −9.4 and −8.9.

Strontium and Sr isotopes

All fluorite bands show similar Sr concentrations between 80 and 103 ppm, except for band No. 1, which shows a high concentration of 211 ppm due to its calcite content (Table 1). In contrast, 87Sr/86Sr ratios are rather variable (Table 2; Fig. 8). Bands 2, 4, and 5 and bands 6 to 9 appear to form two separate groups with 87Sr/86Sr in a narrow range between 0.7084 and 0.7085, and between 0.7080 and 0.7081, respectively. This subdivision is the same as that suggested by texture, color, and REY concentration and distribution. Band No. 1 is less radiogenic (0.7070) and band No. 3 is considerably more radiogenic (0.7123) compared to the other bands. There is no correlation between concentration and isotopic composition of Sr.

Fig. 8a–g
figure 8

. Sr-Nd-Pb isotope data of fluorite and carbonate wall rock (all data calculated for t=290 Ma) a–c 87Sr/86Sr, 206Pb/204Pb, and 208Pb/204Pb for the Blue John fluorite reference sample (Fig. 3). Note that the variability of 87Sr/86Sr and 208Pb/204Pb in fluorite far exceeds that of the carbonate host and wall rocks. d–g Diagrams show the overall poor covariation of the isotopic compositions of Sr, Nd, and Pb. Note that the Pb isotopic composition requires at least three isotopically distinct components

Full size image

The Asbian limestones show a small range of 87Sr/86Sr ratios between 0.70799 and 0.70802 (Table 2). This is in rough agreement with what has been reported for Carboniferous marine chemical sediments (Burke et al. 1982; Veizer et al. 1999), but higher than the range of 0.7076 to 0.7078 suggested for Asbian seawater by data from brachiopods (Bruckschen et al. 1995). The recrystallized sparry calcites show 87Sr/86Sr ratios of 0.70805 and 0.70799, which are very close to those of the limestones.

Lead and Pb isotopes

The fluorite bands show low Pb contents between 0.1 and 0.6 ppm (Table 1), except for bands 1 and 2 that are significantly higher (5.0 and 2.7 ppm, respectively). Lead concentrations in fluorite from other locations in the Pennine Orefield are higher and may reach values as high as 1,100 ppm (Table 1). Lead isotope ratios in the Blue John reference sample are between 18.442 and 19.393 for 206Pb/204Pb, between 38.304 and 39.541 for 208Pb/204Pb, and between 15.624 and 15.683 for 207Pb/204Pb. After recalculation to a mineralization age of 290 Ma, this isotopic heterogeneity still persists (Table 2).

In contrast, the limestones and sparry calcites from the host-rock display only a small range of 208Pb/204Pb ratios (38.39 to 38.47), because the low 232Th/204Pb ratio did not cause significant modification of the initial ratios by in situ 208Pb growth. In contrast, there is a wide range in 206Pb/204Pb ratios (18.568 to 20.686) and 207Pb/204Pb ratios (15.652–15.903). This heterogeneity does not vanish after accounting for in situ Pb growth over the last 290 Ma (18.41–19.36 for 206Pb/204Pb and 15.62–15.68 for 207Pb/204Pb). The calculated heterogeneity for the time of mineralization may be due to post-mineralization U-Pb fractionation which results in over- or undercorrection of in situ Pb growth, or may reflect heterogeneity due to in situ Pb growth from the time of carbonate deposition to the time of mineralization.

Discussion

The low REY concentrations and the negative CeSN anomalies in fluorite from the South Pennine Orefield (Fig. 4) are similar to the REY signature of fluorite from MVT deposits in the Illinois–Kentucky fluorite district, USA (Fig. 5; data from Chesley et al. 1994). However, they are in marked contrast to the significantly higher REY concentrations and the lack of any negative CeSN or positive GdSN anomaly in fluorite from the North Pennine Orefield, some 200 km further north (Figs. 5 and 6). Another significant difference is the presence of strong positive EuSN anomalies in the north (Fig. 6). Uncoupling of Eu from the other REY and the generation of such an Eu anomaly require that a significant amount of Eu occurs as Eu(II), which is typically found in hydrothermal solutions at temperatures above 200 to 250 °C (Bau and Möller 1992). The lack of such a positive EuSN anomaly in fluorite from the South Pennine Orefield not only confirms the fluid inclusion data (e.g., Ewbank et al. 1995) and indicates relatively low temperatures of formation (<250 °C), but also suggests that even prior to fluorite precipitation the hydrothermal fluid had never experienced higher temperatures (Bau and Möller 1992). In the North Pennine Orefield, however, abundant positive EuSN anomalies in fluorite indicate that this hydrothermal fluid had been in a temperature regime that exceeded 200 °C prior to fluorite deposition. In this high-temperature environment a considerable amount of Eu occurred as Eu(II) which allowed for decoupling of Eu from its trivalent REE neighbors and eventually resulted in a positive EuSN anomaly in the fluid (Bau and Möller 1992). This profound difference between the North and South Pennine Orefields suggests that the fluids in the northern ore district had been derived from deeper crustal levels than those in the south.

A very interesting feature of the REY signature of Blue John fluorite from the South Pennine Orefield is the combination of negative CeSN anomalies with positive GdSN and occasional positive LaSN anomalies. Such a REY distribution is commonly found in marine chemical sediments, such as pure limestones and seamount phosphorites, that reflect the REY distribution in seawater. It is, however, extremely rare in other lithologies (note that none of these features is present in fluorite from the North Pennine Orefield).

The prerequisite for the formation of CeSN anomalies is the decoupling of Ce from its REE neighbors following oxidation of Ce(III) and formation of Ce(IV) compounds that behave differently from those of REE(III) during REY mobilization, transport, or fixation. The simplest explanation for the negative CeSN anomalies in Blue John fluorite is that they are inherited from the hydrothermal fluid from which the fluorite precipitated. Alternatively, the CeSN anomaly may have been produced in situ, i.e., during fluorite precipitation from a solution that did not show any anomaly. An in situ CeSN anomaly indicates that Ce was oxidized at the site of fluorite formation (i.e., it may be used as qualitative redox-proxy for the depositional environment), whereas an inherited CeSN anomaly does not provide information about the redox level of the system during precipitation. The generation of an in situ negative CeSN anomaly in fluorite requires the stabilization of Ce(IV) solution-complexes that limit the availability of Ce for incorporation into the fluorite's crystal structure. So far, any impact of Ce(IV) solution-complexes on solid-liquid partitioning has only been described for oxic low-temperature systems such as alkaline lakes where high CO3 2- activities allow for stabilization of dissolved Ce(IV) polycarbonate complexes (Möller and Bau 1993). However, hydrothermal systems per se are unlikely environments for the formation of significant amounts of Ce(IV) compounds, because with increasing temperature, the Ce4+/Ce3+ redox equilibrium moves towards increasingly higher oxygen fugacities (Bilal and Müller 1992). Moreover, REY speciation in fluorite-precipitating solutions in general will be dominated by fluoride complexes because of their high stability constants. If Ce(IV) fluoride complexes in solution exerted an important control on Ce partitioning between fluorite and hydrothermal fluids, it is hard to understand why negative CeSN anomalies in hydrothermal fluorite are so rare. Hence, we conclude that the presence of a negative CeSN anomaly in all of the fluorite samples from the South Pennine Orefield and in each of the nine successive bands of the Blue John reference sample indicates that the REYSN patterns of the hydrothermal fluid from which the fluorite precipitated was characterized by a negative CeSN anomaly. The negative CeSN anomaly of Blue John fluorite was inherited from the solution and does not indicate that the environment during fluorite formation was oxic with respect to the Ce(IV)/Ce(III) redox system. The positive GdSN (and occasional positive LaSN) anomalies are also inherited from the solution. If they were produced during fluorite precipitation, their occurrence should be as typical of hydrothermal vein fluorite as is the presence of strong positive YSN anomalies (Bau and Dulski 1995). They are, however, very rare.

The distinctive differences between fluorite from the South and North Pennine Orefields raise the question of whether REYSN patterns in general and the negative CeSN anomaly in particular may be used to trace the source of the REY. The ore-forming fluid may have acquired the negative CeSN anomaly either during REY mobilization from a source-rock in which Ce was fixed as a relatively insoluble Ce(IV) compound, it may have removed REY from a source-rock that itself was already depleted in Ce, or it may have preferentially lost Ce during fluid–wall rock interaction. Cerium in igneous rocks or clastic sediments is generally Ce(III) and the amount of Ce(IV) is extremely low. Although highly weathered rocks may occasionally show elevated amounts of Ce(IV), it is very unlikely that the REY budget of the hydrothermal fluid was dominated by REY released from such rare lithologies. Preferential Ce loss during fluid migration due to oxidative scavenging is also unlikely, because elevated temperatures at given oxygen fugacity and pH favor stabilization of Ce(III) compounds and the stabilities of Ce(IV) hydroxide complexes decrease with increasing temperature (Bilal and Müller 1992), whereas the stabilities of REY(III) fluoride solution-complexes increase with increasing temperature. This divergence can be expected to reduce Ce loss by oxidative scavenging during fluid migration at higher temperature. Abundant potential source-rocks that commonly show negative CeSN anomalies are marine carbonates. These rocks may also display small positive LaSN and GdSN, and strong positive YSN anomalies that reflect the REY distribution in the seawater from which they precipitated. In contrast, common igneous and clastic rocks do not have anomalous amounts of La, Gd, and Y. Thus, we conclude that the REYSN patterns of Blue John fluorite with negative CeSN and positive GdSN and YSN (and sometimes LaSN) anomalies suggest that the REY in the hydrothermal fluid were derived from marine carbonates. This is in agreement with the low REY content of Blue John fluorite compared to hydrothermal fluorite from other hydrothermal vein deposits. The REY concentration of aluminosilicate rocks is generally several orders of magnitude higher than that of marine limestones (Parekh et al. 1977) and would probably have produced higher REY concentrations in the hydrothermal fluid. Since the local limestones show all the features typical of the REY distribution in pure marine sedimentary carbonates, it may even be argued that the REY in the hydrothermal fluid were derived from the local country-rock. Such an interpretation would be supported by the 87Sr/86Sr ratios of fluorite bands 6 to 9, as these fall close to the isotopic composition of Sr in the Asbian limestones that host the Blue John fluorite mineralization (Fig. 8). However, when Nd and Pb isotopes are added to the picture, this model becomes unsustainable. As shown by the evolution of εNd(t) of fluorite and limestones through time (Fig. 9), the Nd in the latter is consistently less radiogenic than that in the fluorite, which suggests different sources of Nd (and the other REY). Lack of significant element input from the limestone host-rock into the hydrothermal fluid is further corroborated by the preservation of carbonate fossils, such as the crinoid fragment shown in Fig. 10. Although completely embedded in fluorite, this crinoid fragment does not even show corroded margins, indicating that the fluorite-forming hydrothermal fluid did not dissolve the adjacent carbonates but rather precipitated in cavities and open fractures. We emphasize, however, that despite a difference of up to 1.4 ε units between (adjacent) fluorite bands, the REYSN patterns of these bands are very similar (Fig. 4A), suggesting that in all of these bands the major source of at least Nd and the other REY were marine carbonates rather than igneous rocks or clastic sediments.

Fig. 9
figure 9

. Plot of the evolution of εNd(t) of the Blue John fluorite reference sample and host limestone as a function of time. Note the heterogeneity within the fluorite samples and the persistent differences between fluorite and host limestone

Full size image
Fig. 10
figure 10

. Crinoid fragment in purple fluorite from a fluorite lens in marginal reef limestone near Castleton (Fig. 1). Note that the fluorite-precipitating fluid did not dissolve or corrode the crinoid fragment

Full size image

Compared to the REY, Pb isotopes tell a completely different story about the fluorite-forming hydrothermal fluid and the rocks it interacted with. Thorium concentrations in the fluorite are below 0.06 ppm (the lower limit of determination by ICP-MS; Table 1) and cannot explain the variation of 208Pb/204Pb ratios between the fluorite bands (Table 2). Furthermore, the initial 208Pb/204Pb ratios of Blue John fluorite fall outside the range defined by the limestone host-rock and the sparry calcites (Fig. 8), which excludes the local host-rock as the only or dominant source of Pb. The initial 208Pb/204Pb ratio of some fluorite bands is higher than that of the carbonates whereas that of other bands is lower (Fig. 8). The variation of Pb isotopic ratios between the fluorite bands indicates that the initial isotopic composition of Pb in the hydrothermal fluid changed with time. If the hydrothermally leached Pb does not differ from the isotopic composition of the bulk rock, this variation implied the involvement of two different source rocks, such as limestone units of different age and initial Pb. However, when Pb is mobilized from clastic sediments or igneous rocks, it is mobilized by the sequential dissolution of different minerals, and hence its isotopic composition changes with time even though it is derived from a single source rock. Derivation of Pb from clastic sediments would be in agreement with earlier results that suggested Pb in fluorite from the South Pennine Orefield is derived from Visean–Namurian shales (Jones et al. 1991; Jones and Swainbank 1993), a view that is compatible with the origin of Ar and He in fluid inclusions (Kendrick et al. 2002). Strong decoupling of Pb from REY and from Sr is also suggested by plots of 208Pb/204Pb versus εNd(t) or 87Sr/86Sr (Fig. 8).

Although our results clearly show that Sr, Nd, and Pb isotopes cannot be used to determine isochron ages for Blue John fluorite, age information may be obtained indirectly by comparison of the isotopic signature of the fluorite to that of potential source-rocks. If it can be shown that the fluorite-forming hydrothermal fluid received Sr or Nd input from a specific source-rock the age of which is known, this source rock's age provides a maximum age for the fluorite mineralization. Indeed, this approach has been applied previously to MVT deposits in the North Pennine Orefield (Halliday et al. 1990).

The marine residence time of Sr is longer than the mixing time of the oceans, and the Sr isotopic composition of open-ocean seawater (and of marine carbonates), therefore, is independent of location. Thus, the evolution of the Sr isotopic composition of seawater through time follows a unique reference line and the age of a pure marine limestone may be approximated by comparison. The 87Sr/86Sr ratio of fluorite that precipitated from a hydrothermal solution which mobilized Sr from pure marine limestone reflects the Sr isotopic composition of this source-rock and must, of course, be younger than this source. In contrast to Sr, Nd isotopes cannot be used this way. Due to the short marine residence time of Nd, the Nd isotopic composition of seawater (and of marine chemical sediments) may show strong regional differences. Since the Nd isotope ratio of seawater is strongly affected by input from regional and local clastic material (atmospheric dust and river particulates), the εNd values of Phanerozoic sedimentary carbonates are often very similar to those of related clastic sediments. Thus, there exists no "global" evolution for the Nd isotopic composition of seawater.

An even more important limitation results from the fact that the isotopic composition of Sr, Nd or Pb in a hydrothermal fluid or any other natural water does not necessarily reflect the isotopic composition of the respective element in the bulk source rock from which it was mobilized. While congruent dissolution and trace element release from individual minerals may occur, the poly-mineralic nature of most aluminosilicate rocks usually precludes congruent trace element release from bulk rocks. In clastic sediments, the individual minerals may not only show a wide range of element concentrations and ratios, such as Sm/Nd, Rb/Sr, U/Pb, and Th/Pb, but may have been derived from multiple sources with different ages. Since dissolution rates (and element release rates) are mineral-specific, congruent dissolution of aluminosilicate rocks in general and of clastic sediments in particular is the very exception, and the isotopic composition of a solution that mobilized trace elements may differ substantially from the bulk composition of the source rock. However, the mono-mineralic composition of pure sedimentary carbonates can be expected to result in congruent Sr release and close similarity between the Sr isotopic composition of solution and source rock.

If the fluorite-precipitating hydrothermal fluid was in chemical equilibrium with the carbonate lithology from which most of the Sr was derived, there would have been only minor reaction with any other carbonate the fluid might have encountered along its migration path. In this case, the isotopic signature of the source would have been preserved in the hydrothermal fluid. Any variation of the isotopic composition during fluorite precipitation, therefore, would indicate a change of the dominant fluid source with time. Such a shift in fluid source may reflect the exhaustion of an aquifer or the progressive response of deeper aquifers to compressional tectonics. In contrast, if the fluid migrates from its source to the place of fluorite deposition across a wide variety of different lithologies, the fluid will be in geochemical disequilibrium with its immediate wall-rock after each lithological change. In this case, there will be a constant readjustment of the fluid's composition. This adjustment will be most pronounced in early fluids, but will become less significant in later fluids, as these re-equilibrate with rocks that have already been altered during earlier fluid-rock interaction. In such a scenario, the Sr isotopic composition of the first crystallized fluorite is dominated by Sr that originates from the host-rock, because this is the last unit the fluid equilibrated with. Later fluids preserve their source signature to a larger extent, which may be reflected in the larger variability of the Sr-isotopic composition.

The lack of major host-rock dissolution (Fig. 10) by the fluorite-precipitating fluid at Treak Cliff suggests that the hydrothermal solution was in or close to equilibrium with a carbonate lithology. This suggests that the isotopic composition of Sr in Blue John fluorite reflects the composition of Sr in a carbonate source-rock.

Assuming that the Sr in the Blue John fluorite reference sample (except bands 1 and 3) was exclusively derived from marine carbonates and considering that Ordovician to Devonian strata had been largely eroded prior to the Lower Carboniferous, Lower Carboniferous to early Permian limestones appear to be the most likely source-rocks for Sr. Comparison of the 87Sr/86Sr ratios of the Blue John fluorite reference sample with the evolution of Sr isotopes in seawater (Burke et al. 1982; Veizer et al. 1999) suggests that Sr in bands 6 to 9 (0.70803 to 0.70812) originates from Carboniferous/Early Permian, Early Triassic, or Late Triassic carbonates; the somewhat more radiogenic 87Sr/86Sr ratios of bands 2, 4, and 5 (0.70837 to 0.78850) can only be derived from Lower Carboniferous carbonates. The isotope data suggests, therefore, that the Sr in the Blue John reference sample was initially (bands 6 to 9) derived from Upper Tournaisian (Ivorian) limestones and may subsequently (bands 2, 4, 5) have changed towards a Lower Tournaisian (Hastarian) limestone source. The very unradiogenic Sr in calcite-bearing band No. 1 either suggests that this Sr originated from much younger limestones (either ca. 250 or 160 Ma old, which is incompatible with a supposed mineralization age of 290 Ma) or that it was (partly) derived from mafic sills that occur in the Lower Carboniferous country-rock (Walters and Ineson 1980; Mostaghel and Ford 1986). The 87Sr/86Sr ratio of band No. 3 (0.71225) is much too high for limestone-derived Sr, and points towards an aluminosilicate source-rock. We conclude that the isotopic data are most compatible with Lower Carboniferous limestones as the dominant source of Sr in the reference sample of Blue John fluorite.

The general conclusion of a marine limestone source for both Sr and REY is somewhat unexpected. While limestones and shales show similar Sr contents, shales are significantly higher in Nd than limestones. Hence, trace element release from clastic sediments can be expected to make a much stronger impact on the isotopic composition of dissolved Nd than on that of dissolved Sr. Moreover, the capacity of a fluoride-rich hydrothermal fluid for Nd mobilization and transport is particularly high, because the high stability of Nd fluoride complexes reduces the mineral/solution partition coefficients and allows for increased partitioning of Nd into the fluid. However, the Blue John fluorite reference sample indicates the opposite trend. Band No. 3 is anomalously rich in radiogenic 87Sr that can only be derived from Rb-bearing aluminosilicates, but its εNd(t) value is very similar to those of the other fluorite bands. While this alone could be explained by the close similarity between the isotopic composition of Nd in limestones and associated shales, the REYSN pattern of band No. 3 indicates that Nd (and the other REY) are not derived from aluminosilicates. Release of REY from aluminosilicates adds REY without Ce, La, Gd, and Y anomalies, and if this REY input is significant relative to the REY concentration in the fluid, any anomaly the fluid initially showed will be erased. Considering the relatively low REY content in Blue John fluorite, shale-derived REY in band No. 3 should have caused somewhat higher LREE concentrations and smaller (or loss of) anomalies of Ce, La, Gd, and Y in band No. 3 compared to the other bands. However, this is not observed. Instead, it appears that in all nine fluorite bands marine sedimentary carbonates were the sole source of REY, despite episodic additional input of Sr from aluminosilicates (band No. 3). This observation indicates that release of REY from the aluminosilicates that provided the Sr found in band No. 3 was negligible, and suggests that despite high fluoride concentration in the fluid, the REY were most likely fixed in secondary phases immediately upon their release from the primary aluminosilicate minerals. This decoupling of Sr and Nd observed in band No. 3 is further evidence for incongruent trace metal release from bulk aluminosilicate rocks.

The above discussion demonstrates that Sr and Nd may be uncoupled from each other during fluid-rock interaction and that the isotopic compositions of Sr and Nd in natural waters may differ from those of their respective bulk source rocks. We emphasize that, therefore, comparison of the isotopic composition of hydrothermal fluids (or of hydrothermal minerals) with that of potential bulk source-rocks in plots of εNd vs. 87Sr/86Sr (commonly used in igneous geochemistry to describe source rocks and mixing relationships) cannot generally be utilized to characterize solute sources of hydrothermal fluids or groundwaters.

Conclusions

The results of our comparative study of low-temperature hydrothermal fluorite from the MVT deposits in the South and North Pennine Orefield, England, highlight significant differences between these two districts. The REYSN patterns suggest that the hydrothermal fluids from which Blue John fluorite in the South Pennine Orefield precipitated did never experience temperatures significantly above 200 °C. This is in marked contrast to the >200 °C suggested by the REYSN patterns in the North Pennine Orefield, and may indicate a shallower depth of the hydrothermal system in the south as compared to the north. The REYSN patterns of Blue John fluorite from the South Pennine Orefield further suggest that the REY are derived from relatively pure marine limestones, whereas Pb isotopes indicate that Pb has been mobilized from aluminosilicate rocks. The Sr isotopic composition of fluorite suggests that Lower Carboniferous (Tournaisian) limestones were the most important Sr source. All isotope systems studied indicate that the host-rocks of the fluorite mineralization (Asbian limestones) did not contribute to the trace element load of the fluorite-forming hydrothermal fluid. This is in agreement with the occurrence of uncorroded fragments of carbonate host rocks, including fossils within the fluorite. With regard to the element sources, the isotopes of Nd provide only inconclusive evidence. We emphasize that neither Nd nor Pb or Sr isotopes can be used to determine isochron ages for the Blue John fluorite.

Our integrated study of the REY distribution and the Sr-Nd-Pb isotope systematics provide evidence for that the different solutes present in a hydrothermal solution may be derived from different sources. The results indicate that any interpretation that is based exclusively on either Sr or Pb or Nd isotopes or REY distribution, for example, is unlikely to provide a basis for a full understanding of the genesis or history of a mineralization or hydrothermal system. This should be typical of natural waters in general (including groundwater and river water) and not be confined to hydrothermal fluids. While migrating towards the site of ultimate mineral deposition, the hydrothermal fluid(s) that produce MVT deposits encounter a range of different lithologies that may or may not leave their mark(s) on the composition of this solution. The physical and chemical properties of both these rocks and the fluid determine the extent to which (or if at all) a specific element partitions from the rock into the fluid or vice versa. Since these parameters not only change along the fluid's pathway but may also change with time (for a specific element, for example, a system may reach the exchange (or isotopic) equilibrium between fluid and mineral surfaces), a change in chemical or isotopic composition between individual bands of banded hydrothermal deposits does not necessarily indicate a change in the fluid's pathway or the onset of admixture of another fluid. The data we presented here from Blue John fluorite also provide strong evidence against the common notion that the chemical and isotopic composition of a hydrothermal fluid is usually controlled by the "last" rock it encountered.

Although hydrothermal fluids (and those that form MVT deposits, in particular) may have had a complex history, the study of different elements and isotopic systems in the minerals that precipitate from these solutions are a way to trace the sources of specific elements and can provide insight into the chemical and physical evolution of these fluids. We are confident that this will ultimately lead to a better understanding of ore-forming processes and hence, allow for the development of increasingly successful exploration strategies.