1. Introduction
The continental lithospheric mantle (CLM) is defined in part by its mainly conductive heat flow and represents an important petrological and geochemical terrestrial reservoir (McKenzie and Bickle, Reference McKenzie and Bickle1988; Pearson and Nowell, Reference Pearson and Nowell2002). Many studies of the CLM utilize seismological and laboratory-based experimental approaches, aimed at characterizing lithological properties and processes, as well as the nature of the underlying lithosphere-asthenosphere boundary (LAB). However, mantle xenoliths, carried by magmas that traverse the CLM to upper crustal or surface levels, offer opportunities for focussed petrological snapshots of the lithospheric mantle at the time of transport (e.g. Canil, Reference Canil2025). Cargoes of mantle xenoliths can inform on long-term mantle heterogeneity as represented by the CLM, and, given the host lavas or magmas can be asthenospheric melts, insights can also be gained into the mantle source controls on intraplate magma composition (Nixon, Reference Nixon1987; Pearson and Wittig, Reference Pearson, Wittig, Holland and Turekian2014).
The passive margin of NW Ireland and Scotland is underlain by continental lithospheric mantle associated with the eastern portion of the North Atlantic Craton (NAC; Menzies and Halliday, Reference Menzies and Halliday1988; Upton et al. Reference Upton, Stephenson, Smedley, Wallis, Fitton, Wilson, Neumann, Davies, Timmerman, Heeremans and Larsen2004, Reference Upton, Downes, Kirstein, Bonadiman, Hill and Ntaklos2011; Kirstein et al. Reference Kirstein, Dunai, Davies, Upton, Nikogosian, Wilson, Neumann, Davies, Timmerman, Heeremans and Larsen2004; Hughes et al. Reference Hughes, McDonald, Faithfull, Upton and Downes2015). The NAC comprises Archaean domains stitched together by zones of Palaeoproterozoic reworking, including present-day exposures in eastern and western Greenland (Bridgwater et al. Reference Bridgwater, Watson and Windley1973). In NW Scotland, where the NAC is represented by the Lewisian Complex, insights into the nature of the CLM have come from a rich diversity of mantle xenoliths from dykes associated with regional Permo-Carboniferous extension (Upton et al. Reference Upton, Stephenson, Smedley, Wallis, Fitton, Wilson, Neumann, Davies, Timmerman, Heeremans and Larsen2004, Reference Upton, Downes, Kirstein, Bonadiman, Hill and Ntaklos2011; Hughes et al. Reference Hughes, McDonald, Faithfull, Upton and Downes2015). Much rarer mantle xenolith-bearing dykes of Palaeogene age, such as the ∼45 Ma Loch Roag monchquite occurrence (Menzies et al. Reference Menzies, Halliday, Palacz, Hunter, Upton, Aspen and Hawkesworth1987; Faithfull et al. Reference Faithfull, Timmerman, Upton and Rumsey2012), are important for the same reasons but also because they offer insights into an early stage of the interaction of the CLM with the proto-Iceland mantle plume. The onset of melting associated with the mantle plume is generally accepted to be related to the early stages of rifting of the NE Atlantic Ocean and establishment of the ∼60 Ma British and Irish Palaeogene Igneous Province (BIPIP; Saunders et al. Reference Saunders, Fitton, Kerr, Norry, Kent, Mahoney and Coffin1997). Numerous studies of the petrology and geochemistry of the flood basalts and igneous centres of the BIPIP have discussed the nature of the mantle source of the magmatism, and the recognition that the wider NE Atlantic Igneous Province, including the BIPIP, hosts important base and precious metal magmatic mineralisation, has magnified the importance of understanding the composition of the underlying mantle (see Bell and Williamson, Reference Bell, Williamson, Smith and Strachan2024, and references therein).
In this study, samples from a mantle xenolith-bearing dyke of Palaeogene (Eocene) age in County Donegal, NW Ireland, are examined. The goal of the study is to characterize the composition and nature of the CLM as represented by these xenoliths and evaluate whether the petrological signature(s) of mantle plume interaction can be identified. Previously reported He isotope evidence from xenoliths in this dyke is consistent with mantle plume involvement (Kirstein and Timmerman, Reference Kirstein and Timmerman2000), but the extent to which major and trace element compositions of the rocks are overprinted is unknown. It has been suggested that the deep (garnet-bearing) keel of the CLM under the NAC was lost during one or more of the multiple tectonic events experienced by the rocks of NW Scotland and Ireland (Menzies and Halliday, Reference Menzies and Halliday1988), so another goal of the present study is to place some constraints on the thickness of the CLM and depth to the LAB in NW Ireland.
2. Inver dyke geological setting
The mantle xenoliths studied here come from an alkali basalt dyke at Inver, southern Donegal (Figure 1). The dyke intrudes lower Carboniferous rocks of the Bundoran Shale and Ballyshannon Limestone Formations and has previously been reported to be exposed over 30 m of width and 400 m of length (Gallagher and Elsdon, Reference Gallagher and Elsdon1990). The main mantle lithologies present are spinel lherzolites and spinel harzburgites (Gallagher and Elsdon, Reference Gallagher and Elsdon1990; Shaw and Edgar, Reference Shaw and Edgar1997), as well as a variety of other igneous (cumulate) and upper crustal (sedimentary) lithologies. Gallagher and Eldson (Reference Gallagher and Elsdon1990) originally suggested that the age of the Inver dyke was Permo-Carboniferous, based on its similarities with xenolith-bearing dykes of central and northern Scotland of that age (e.g. Upton et al. Reference Upton, Aspen and Hunter1984). However, Kirstein and Timmerman (Reference Kirstein and Timmerman2000) combined K-Ar and Ar-Ar geochronology to date the Inver dyke to 43 ± 0.4 Ma. This age is significant because it means that the mantle melting associated with production of the Inver dyke is linked to the opening of the North Atlantic caused by the mantle plume presently underneath Iceland. The samples studied here come from the same disused quarry described by previous workers (Gallagher and Elsdon, Reference Gallagher and Elsdon1990; Shaw and Edgar, Reference Shaw and Edgar1997; Kirstein and Timmerman, Reference Kirstein and Timmerman2000), approximately 3 km north of Inver (Figure 1). Xenoliths range in size from <1 cm to >30 cm diameter, with a tendency for smaller xenoliths such as the ones studied here to be angular and larger ones to be rounded. Away from the locality, the dyke is very poorly exposed, and no sense of its geometry or extent could be determined during fieldwork associated with the present study.
(a) Island of Ireland with the locality of the geological sketch map in (b) highlighted. The approximate area of outcrop of the Palaeogene Antrim basalts is also shown. (b) Geology of Inver, southwest Donegal, with the position of the xenolithic dyke shown. Also shown are other Palaeogene minor sheet and plug intrusions. This figure with panels (a), (b) and (c) is adapted from Gallagher and Elsdon (Reference Gallagher and Elsdon1990). The grey grid lines represent the Irish National Grid, with most of the map area falling within letter code G, and the easternmost part falling within letter code H.
Figure 1. Long description
Panel A: A map of the island of Ireland with a highlighted region indicating the location of the detailed geological sketch map in Panel B. The approximate area of the Palaeogene Antrim basalts is also marked. Panel B: A detailed geological map of Inver, southwest Donegal, showing the position of the xenolithic dyke and other Palaeogene minor sheet and plug intrusions. The map includes various geological features such as Carboniferous limestone, pre-Carboniferous rocks, and Paleogene minor intrusions. Key locations such as Inver, Donegal, Lough Eske, Killybegs, and Ballyshannon are labeled. The map also shows roads, faults, and lithological boundaries. The grey grid lines represent the Irish National Grid, with most of the map area falling within letter code G, and the easternmost part falling within letter code H.
3. Analytical techniques
3.a. Electron microprobe analysis
Mineral chemical analysis was carried out using a JEOL JXA-8230 electron microprobe at the University of Ottawa, and the complete data tables can be accessed in the electronic supplement. Backscattered electron micrographs together with compositions of silicate minerals (olivine, orthopyroxene and clinopyroxene) and spinel were determined with a beam energy of 20 kV, a beam current of 40 nA and a beam diameter of 5 µm. Elemental values were acquired using crystals LIF for Fe, Mn and Zn, PETJ for Ca and K, PETL for Ti, TAP for Mg, Al, Na and Si and LIFL for Ni, Cr and V. The electron microprobe was calibrated prior to analysing the samples by using the following standards: albite for Na, chromite for Cr, Al and Mg, diopside for Ca, sanidine for Si and K, haematite for Fe, rutile for Ti, tephroite for Mn, vanadinite for V, pentlandite for Ni and gahnite for Zn. Peak and off-peak count times were 20 seconds each for all elements except Ca and Al, for which 40-second peak and off-peak count times were used. Relative analytical uncertainties for the major and minor elements of interest in this study generally range from 0.1–5% (1σ), respectively. Ferric iron contents in spinel were calculated following Droop (Reference Droop1987).
3.b. Laser ablation-inductively coupled mass spectrometry
Inver olivine, orthopyroxene and clinopyroxene were analysed for trace element abundances using laser ablation-inductively coupled plasma mass spectrometry (LA-ICPMS). The full table of data is provided in the electronic supplement. The in situ LA-ICPMS analyses were carried out at LabMaTer (UQAC) using a 193 nm RESOlution laser ablation system (Applied Spectra, California) equipped with a S-155 double volume ablation cell (Laurin Technic), attached to an Agilent 7900 ICPMS. A laser beam size of 55 µm in static spot mode was used, with a pulsing rate of 15 Hz and a fluence of 5 J/cm2. The He-Ar carrier gas was measured for 30 s and 10 s and used for background correction, followed by 4 s to 16 s of static spot data acquisition. The data were processed using the 3D-TE-DRS module (Paul et al. Reference Paul, Petrus, Savard, Woddhead, Hergt, Greig, Paton and Rayner2023) in IOLITE v4 software (Paton et al. Reference Paton, Hellstrom, Paul, Woodhead and Hergt2011). The calibration was achieved by combining 6 reference materials: G_NIST610, G_NIST612, G_NIST616, G-Probe6, G-GSD1G and G_GSE-1g. No internal standard was used, and fully quantitative results were achieved by an appropriate normalization as demonstrated by Savard et al. (Reference Savard, Dare, Bédard and Barnes2023). Where electron microprobe data exist for the same points as the LA-ICPMS spots, there is good agreement between both techniques (see example comparisons for Ca and Ni in olivine, provided in the electronic supplement). Most trace elements had better than 5% relative standard deviation (RSD; 2σ), and, in particular, Ca and Al in olivine had RSD values at or close to 5%. Average Longerich detection limits for Ca and Al in olivine over the analytical session were ∼30 ppb and ∼6 ppm, respectively.
4. Petrographic observations and analytical results
4.a. Petrography
Two xenoliths were selected for detailed study here, primarily on the basis of a lack of obvious post-entrainment reaction textures between the groundmass host and xenolithic minerals. These xenoliths are spinel lherzolites so are judged to be representative, as the Inver mantle peridotites have previously been shown to be dominated by this lithology (Gallagher and Elsdon Reference Gallagher and Elsdon1990). The xenoliths studied here are relatively small (∼5–10 cm2 on thin section surfaces; Figure 2(a)), and contain rounded spinel inclusions within and at the triple junctions of olivine crystals (Figure 2(b) and (c)); such spinel is not in contact or close to groundmass material. The rocks overall are equigranular, and olivine and orthopyroxene in particular are equant and granular-textured (Figure 2(d) and (e)). Olivine commonly exhibits strain lamellae and undulose extinction (e.g. Figure 2(d)). Clinopyroxene comprises ∼10 mod.% of both samples and occurs interstitially, typically with concave grain boundaries towards rounded olivine crystals (Figure 2(b) and (c)). Sulphide is sparsely present, as anhedral or rounded equant blebs on olivine grain boundaries or at the junctions between olivine and clinopyroxene crystals. The margins of the xenoliths are relatively sharp. The interiors of the xenoliths studied here do not contain the yellowish-rimmed olivine or sieve-textured spinel described by Shaw and Edgar (Reference Shaw and Edgar1997). These textures were interpreted by the latter authors as the products of post-entrainment reaction of xenoliths and groundmass. Local evidence of reaction does occur at xenolith margins, forming complex intergrowths of silicate and oxide phases, and is quite obvious under both the petrographic and electron microscopes (Figure 2(f) and (g)).
Petrography of the Inver spinel lherzolites. (a) Circular-polarized light scan of Inv-B xenolith, collected using a Zeiss Axioscan 7 at the University of Ottawa. (b) and (c) Backscattered electron micrographs showing spinel crystals in the centre of Inv-B. Note the anhedral nature of clinopyroxene occupying the interstitial spaces between olivine grains. (d) and (e) Cross-polarized light photomicrographs of granular olivine texture in Inv-A and Inv-B, respectively. Note the planar deformation feature in olivine in (d). (f) Plane-polarized light photomicrograph of olivine and clinopyroxene at the edge of xenolith Inv-A. The irregularly-shaped patch of fine-grained dark-coloured material on the right-hand side of the image is dyke groundmass. (g) Backscattered electron micrograph showing a reaction zone between the edge of xenolith Inv-A and the host dyke, between a clinopyroxene crystal and an olivine crystal. Abbreviations on all panels as follows: Ol, olivine; Opx, orthopyroxene; Cpx, clinopyroxene; Sp, spinel; Gm, groundmass (of host dyke).
4.b. Silicate and spinel major element chemistry
A summary table of the electron microprobe analyser (EPMA) data is provided in the electronic supplement. Olivine compositions in the two Inver xenoliths studied are slightly different, with averages of Fo90.1 (± 0.2; n = 49) and Fo91.1 (± 0.1; n = 26), for Inv-A and Inv-B, respectively. This difference is also reflected in NiO contents, with the more forsteritic olivine being associated with higher NiO (generally ∼3000 ppm Ni) versus lower (<3000 ppm Ni) values (Figure 3(a)). Orthopyroxene is indistinguishable between the two xenoliths, with En90.5-92.6. Clinopyroxene is diopsidic, with a ∼10 mol.% jadeite component, an average of 4.7 wt.% Al2O3 (± 0.8; n = 20), and is distinguishable between xenoliths on the basis of Inv-A having a slightly more magnesian composition (Figure 3(b)). On the basis of Mg-Fe partitioning, clinopyroxene and olivine compositions in both xenoliths are chemically equilibrated (Figure 3(c)). The silicate mineral chemical data of Shaw and Edgar (Reference Shaw and Edgar1997) are shown in Figure 3 together with the new data of this study, and the reasonably good comparison between the two datasets supports the assumption made here that Inv-A and Inv-B are representative of the mantle xenolith population more broadly.
Mineral major element chemistry for the Inver xenoliths. In all panels, the black filled squares represent data from the xenoliths of Shaw and Edgar (Reference Shaw and Edgar1997), designated S + E on the plots. Note that only compositions from xenoliths labelled as ‘primary’ from the latter study are plotted here. (a) Ni (ppm) versus forsterite (mol.%) content in olivine. Fields of olivine compositional data are shown for comparison, for kimberlitic olivine (green, Victor Pipe, Canada), olivine in picrite lavas (pink, Baffin Island, Canada), spinel peridotites (red dash line) and garnet peridotites (black dash line). All of the aforementioned fields of data are those presented in De Hoog et al. (Reference De Hoog, Gall and Cornell2010). (b) Al2O3 versus Mg# (Mg/(Mg+Fe2+) for Inver xenolith clinopyroxenes. The fields for abyssal and supra-subduction zone (SSZ) peridotites (green fill and red dash, respectively) are based on Choi et al. (Reference Choi, Shervais and Mukasa2008) and references therein. (c) Fields showing olivine Fo content versus average clinopyroxene Mg# variation for Inv-A and Inv-B samples. The solid red line illustrates the equilibrium partitioning of Fe2+ and Mg between olivine and clinopyroxene, assuming mineral melt Fe2+-Mg distribution coefficients of 0.30 and 0.23 for olivine and clinopyroxene, respectively (from Roeder and Emslie, Reference Roeder and Emslie1970). The dashed lines on either side of the red line represent the uncertainties of 0.02 for each distribution coefficient. (d) Spinel Mg# versus Cr# (Cr/[Cr+Al]) for Inver xenoliths. Most of the S + E data points lie under the Inv-A and Inv-B points. Fields for spinels of various compositions are taken from O’Driscoll et al. (Reference O’Driscoll, Leuthold, Lenaz, Skogby, Day and Adetunji2021) and references therein. Spinel compositional end members are abbreviated as follows: Mg, magnesiochromite; Cr, chromite; Sp, spinel; Hc, hercynite. The black arrow shows increments of melt depletion as represented by spinel composition.
Spinel compositions show compositional differences between both of the xenoliths studied. Values for Mg# (Mg/[Mg+Fe2+]) range between 0.73–0.81 for Inv-A, and 0.78–0.80 for Inv-B. Chromium number (Cr#; Cr/[Cr+Al]) and TiO2 range between 0.16–0.24 and 0.01–0.06 for Inv-A, and 0.15–0.18 and 0.08–0.37 for Inv-B (Figure 3(d)). The new spinel data are very similar to the compositions reported in Shaw and Edgar (Reference Shaw and Edgar1997; Figure 3(d)).
4.c. Silicate trace elements
We carried out our LA-ICPMS analyses on 42 spot locations across the two xenolith samples studied; the spots were all located in close proximity to the locations analysed using the electron microprobe. A summary of the data is provided in the electronic supplement. Olivine compositions in the Inver xenoliths bear similarities to spinel and garnet peridotites sampled as mantle xenoliths from other localities (e.g. Wang et al. Reference Wang, Su, Robinson, Xiao, Bai, Liu, Asamoah Sakyi, Jing, Chen, Liang and Bao2021; Figure 4(b)). Both xenoliths are, in general, distinguishable on the basis of their olivine trace element characteristics, although they are relatively similar. Calcium and Al contents range between 235–480 ppm and 34–82 ppm, respectively (Figure 4(a)). Within this range, Inv-A olivine tends to higher concentrations of both elements, but also has the lowest values. On a plot of Ni/Mn versus V/Sc, after Wang et al. (Reference Wang, Su, Robinson, Xiao, Bai, Liu, Asamoah Sakyi, Jing, Chen, Liang and Bao2021), both xenoliths plot within the field for spinel peridotite residues, slightly toward the reduced end of the compositional spectrum (Figure 4(b)).
Mineral trace element abundances in Inver xenoliths. (a) Olivine Al (ppm) versus Ca (ppm). (b) Olivine Ni/Mn versus V/Sc. The fields for garnet, spinel, ophiolite and layered intrusion peridotites are taken from Wang et al. (Reference Wang, Su, Robinson, Xiao, Bai, Liu, Asamoah Sakyi, Jing, Chen, Liang and Bao2021), and the bulk silicate Earth (BSE) value for Ni/Mn of 2 is from McDonough and Sun (Reference McDonough and Sun1995). Primitive mantle (PM)-normalized rare earth element diagrams for Inver (c) orthopyroxene and (d) clinopyroxene. The data plotted for Rinibar and Loch Roag clinopyroxene are averages of individual analyses reported by Hughes et al. (Reference Hughes, McDonald, Faithfull, Upton and Downes2015). The primitive mantle normalizing values were taken from McDonough and Sun (Reference McDonough and Sun1995). Symbols used are the same as in previous figures unless otherwise stated.
Figure 4. Long description
Panel A: A scatter plot shows the relationship between olivine Al (ppm) on the vertical axis and Ca (ppm) on the horizontal axis. The data points are represented by green circles. Panel B: Another scatter plot displays olivine Ni/Mn on the vertical axis and V/Sc on the horizontal axis. The fields for garnet, spinel, ophiolite, and layered intrusion peridotites are shown in different colors, with the bulk silicate Earth (BSE) value for Ni/Mn marked by a red line. Panel C: A line graph presents primitive mantle (PM)-normalized rare earth element diagrams for orthopyroxene, with data series labeled as Inv-A and Inv-B in different shades of green. Panel D: Another line graph shows PM-normalized rare earth element diagrams for clinopyroxene, with data series for Loch Roag (green and black lines) and Rinibar cpx (black line). The graphs illustrate variations in trace element abundances and their relationships in different mineral phases.
Orthopyroxene and clinopyroxene are also clearly distinguishable in each xenolith on the basis of their trace element compositions (Figure 4(c) and (d)). On a primitive mantle-normalized rare earth element (REE) plot, Inv-A orthopyroxene exhibits relatively consistent patterns with 0.01–0.1 × primitive mantle (PM) light REE (LREE) contents and 0.1–1 × PM for the heavy REE (HREE). Sample Inv-B reveals two orthopyroxene analyses with even more depleted compositions (i.e. <0.01 × PM LREE) and a third with an anomalously flat pattern and all REE between 0.1–1 × PM. The clinopyroxene from the two samples is very distinct, with that for Inv-A characterized by relatively LREE-enriched (>10 × PM) compositions and HREE at 1–10 × PM. The clinopyroxene in Inv-B reveals LREE-depleted patterns (<10 × PM) and similar, though slightly elevated, HREE compared to Inv-A.
5. Interpretation and implications
5.a. Thermobarometry and the mantle source of the Inver xenoliths
There are a number of experimentally calibrated thermometers for mantle peridotites, many based on the exchange of different cations between the constituent minerals. For example, Köhler and Brey (Reference Köhler and Brey1990) developed a Ca-Fe-Mg thermometer for orthopyroxene-clinopyroxene pairs (TBKN), and Liang et al. (Reference Liang, Sun and Yao2013) a thermometer between orthopyroxene-clinopyroxene based on REE exchange (TREE). Other thermometers utilize Fe-Mg or Al exchange between olivine and spinel (TFeMg, Li et al. Reference Li, Kornprobst, Vielzeuf and Fabriès1995; TAl, Wan et al. Reference Wan, Coogan and Canil2008; Coogan et al. Reference Coogan, Saunders and Wilson2014). Köhler and Brey (Reference Köhler and Brey1990) also proposed a barometer for spinel peridotites, based on the exchange of Ca between olivine and clinopyroxene (PCa). The closure temperatures for cation exchange in these different systems are diffusion-dependent and may not yield the same results for a given sample or set of mineral pairs, potentially indicating discrete processes operating on the rocks (Canil, Reference Canil2025). For example, Fe-Mg exchange between olivine and spinel is relatively rapid and may be reset during xenolith transport to the surface. Similarly, Ca diffusion in olivine is susceptible to reheating so PCa estimates should be treated with care. Temperatures calculated using TBKN, TREE and, in particular, TAl are typically less affected by later reheating since the cations involved diffuse more slowly in the respective minerals. Information on the chemical state of the CLM can also be derived from estimates of oxygen fugacity ( f O2). The CLM has been shown to be relatively reduced, and to exhibit a relationship of decreasing fO2 with depth (e.g. Woodland and Koch, Reference Woodland and Koch2003). In mantle peridotites, fO2 can be calculated using oxybarometers based on Mg-Fe exchange in olivine-spinel (Ballhaus et al. Reference Ballhaus, Berry and Green1991, Reference Ballhaus, Berry and Green1994) and olivine-spinel-orthopyroxene (Sack and Ghiorso, Reference Sack and Ghiorso1991a, Reference Sack and Ghiorso1991b; Davis et al. Reference Davis, Cottrell, Birner, Warren and Lopez2017).
Some temperature estimates for the Inver xenoliths are shown in Figure 5(a). Values for TAl were calculated using individual olivine Al contents and assuming an average value for spinel composition for each xenolith; it was observed for these samples that the maximum variation in spinel Cr and Al contents leads to less than ±5°C difference in the calculated TAl. The ranges of TAl are 882–908°C and 865–974°C for Inv-A and Inv-B, respectively. Average values of TBKN and TREE are 946°C and 922°C for Inv-A, and 881°C and 1004°C for Inv-B, respectively. The TBKN values were calculated using average PCa values of 12.1 (±1.8; n = 11) and 14.7 (±2.5; n = 14) kbar for Inv-A and Inv-B, respectively. Values of TFeMg, calculated using the range of spinel compositions and a single average olivine composition for Inv-A and Inv-B are 1011 (±76; n = 62) and 1021 (±31; n = 28), respectively, so they lie above (or just overlap the upper end of) the range of TAl in each case (Figure 5(a)). Overall, it seems there has been some disturbance of TFeMg in each xenolith, probably as a result of a thermal overprint from incorporation and transport within the Inver dyke. Calcium in olivine appears to have been incompletely reequilibrated, again possibly during transfer to the upper crust; the full range of pressure estimates from within each xenolith is shown in Figure 5(b). However, even if the average, instead of the lowermost, values for PCa are taken with the TAl results as representative of the Inver CLM source, and considering these values in the context of a range of model geotherms (see Canil, Reference Canil2025), it seems that the Inver dyke sampled relatively cold and deep lithospheric mantle (Figure 5(b)). Indeed, the maximum calculated PCa yields depths of ∼65 km for Inv-B.
(a) Comparison of results from different geothermometers as applied to the Inver xenoliths, for TFeMg, TAl and TBKN. The uncertainties on each value represent the 1σ standard deviation of the individual calculations from each sample. Uncertainties on TAl values do not overlap any of those produced by the other thermometers. (b) Depth versus temperature plot for the Inver xenoliths, adapted from Canil (Reference Canil2025). The model geotherms are also from Canil (Reference Canil2025), in turn based on Greenfield et al. (Reference Greenfield, Ghent and Russell2013). The blue hexagon is the temperature-pressure estimate for the Inver dyke bulk rock, with uncertainties: see text for discussion. Symbols used are the same as in previous figures unless otherwise stated.
Figure 5. Long description
Panel A: A scatter plot compares results from different geothermometers applied to the Inver xenoliths. The y-axis represents temperature in degrees Celsius, ranging from 850 to 1050 degrees Celsius. The x-axis categorizes the data into two groups: Inv-A and Inv-B. The plot includes three types of data points: circles for TAl, squares for TBKN, and diamonds for TFeMg. Each data point has associated error bars representing the 1σ standard deviation of individual calculations from each sample. Panel B: A depth versus temperature plot for the Inver xenoliths shows depth in kilometers on the y-axis, ranging from 0 to 120 kilometers, and temperature in degrees Celsius on the x-axis, ranging from 400 to 1600 degrees Celsius. The plot includes model geotherms based on Canil (2025) and Greenfield et al. (2013), with lines labeled for different heat flow values (60 mW, 70 mW, 80 mW, and 90 mW). The blue hexagon represents the temperature-pressure estimate for the Inver dyke bulk rock, with uncertainties. The symbols used are consistent with previous figures unless otherwise stated.
Estimates of oxygen fugacity for the Inver xenoliths suggest relatively reducing conditions. The oxybarometer of Ballhaus et al. (Reference Ballhaus, Berry and Green1991) returned Δlog f O2(QFM) values of −5.70 and −5.36 for Inv-A and Inv-B, respectively. These values seem too low to be credible. In addition to olivine and orthopyroxene compositions, the Davis et al. (Reference Davis, Cottrell, Birner, Warren and Lopez2017) formulation for fO2 in mantle peridotites requires input of the magnetite activity in spinel (spinel a Mt), calculated here using MELTS in ThermoEngine (Thermoengine, Reference Thermoengine2022). The resulting Δlog f O2(QFM) values for Inv-A and Inv-B, calculated at 897°C, 12.1 kbar and at 941°C, 14.7 kbar, respectively, are −3.2 and −2.3. These estimates are tentative: Davis et al. (Reference Davis, Cottrell, Birner, Warren and Lopez2017) highlight that the uncertainty on spinel a Mt, as a result of uncertainty on the spinel Fe3+ values used, can be the dominant contributor to overall uncertainties on fO2. Estimates of Fe3+/∑Fe for the Inver spinels are based here on stoichiometry and are on the order of 0.05. Following Davis et al. (Reference Davis, Cottrell, Birner, Warren and Lopez2017), the uncertainty on the fO2 values presented above is on the order of ±1 log unit (1σ). More accurate determination of spinel Fe3+/∑Fe, i.e. by Mössbauer spectroscopy, is needed to constrain these estimates further, but the implication that the Inver peridotites are reduced is in agreement with the olivine trace element contents (Figure 4(b)), which also hint at formation or residence in a reducing environment. The combined temperature, pressure and oxygen fugacity estimates are therefore consistent with a CLM origin for the Inver xenoliths (see Goncharov et al. Reference Goncharov, Ionov, Doucet and Pokhilenko2012 for comparison).
The extent of melt depletion experienced by the xenoliths can be calculated using olivine Fo content (Mou et al. Reference Mou, Lee and Borchardt2025) and spinel Cr# (Hellebrand et al. Reference Hellebrand, Snow, Dick and Hofmann2001). The results do not show particularly good agreement, with Inv-A yielding ∼14% and ∼8% melt depletion from olivine Fo content and Cr#, respectively, and Inv-B revealing 22% and 6% depletion from the same parameters. Given the observation made above that TFeMg values may reflect resetting by later heating, and also the microstructural position of much of the spinel analysed here as inclusions within olivine, it is suggested here that the estimates of melt depletion from spinel Cr# might be more accurate, meaning that the Inver peridotites only saw relatively limited melt extraction before incorporation into the CLM under NW Ireland. Indeed, spinel compositions are consistent with abyssal peridotite-like compositions for the Inver peridotites.
Although a low modal abundance mineral in the Inver xenoliths studied, the trace element chemistry of clinopyroxenes also provides useful insights into the nature of the CLM beneath the NAC. The clinopyroxene in both samples is unaltered and fresh in appearance, with clean grain boundaries against olivine and grain shapes suggesting pseudomorphing of an interstitial melt. Figure 4(c) indicates that the clinopyroxene is in chemical equilibrium, at least with respect to Mg-Fe2+ exchange, with olivine. One xenolith (Inv-A) exhibits pronounced LREE enrichment, whereas in the other, such a signature is not (or only cryptically) present. The LREE-enriched clinopyroxene pattern for Inv-A is quite similar in shape to ‘metasomatized’ green clinopyroxenites from the Loch Roag dyke (Hughes et al. Reference Hughes, McDonald, Faithfull, Upton and Downes2015), though REE abundances overall are slightly lower (Figure 4(d)). Together with the Zr and Hf concentrations in the Inver clinopyroxenes, which are similar for Inv-A and Inv-B at ∼1.5 × PM and ∼2 × PM, respectively (see Downes et al. Reference Downes, de Vries and Wittig2015, for a discussion of Zr-Hf systematics in clinopyroxene), it is therefore suggested that this localized (grain scale) metasomatism is a primary feature of the CLM under NW Ireland and western Scotland, rather than being associated with modification by the Palaeogene mantle plume. It is worth noting that the lack of depletion of Hf in these clinopyroxenes also points to only very low degrees of melt depletion being recorded by these rocks (Pearce Reference Pearce and Wyman1996), supporting the estimates from spinel Cr# discussed above.
The geothermobarometry calculations presented above offer a snapshot of the subsurface geology of NW Ireland at ∼43 Ma. It has been suggested that the LAB shallows northward from central Ireland, such that thinner lithosphere in the north is the result of thermal erosion by the proto-Iceland plume (Landes et al. Reference Landes, Ritter and Readman2007). Estimates of the depth of the LAB under Ireland are somewhat variable, but a recent study proposed a range of 76–112 km, with the deepest LAB being underneath south, central and eastern Ireland (Chambers et al. Reference Chambers, Fullea, Kiyan, Lebedev, Bean, Meere, Daly, Nicola Willmot Noller, Raine, Blake and O’Reilly2025). Using seismic data, the latter study calculated a LAB depth of ∼90 km under NW Ireland, which is easily reconcilable with the maximum depth calculated in the present study of ∼65 km. Garnet-bearing peridotite would not necessarily be expected at less than 90 km depth, so the suggestion that the deeper part of the lithosphere under Ireland has been lost, either by plume-related thermal erosion or by earlier orogenic events (Menzies and Halliday, Reference Menzies and Halliday1988), seems reasonable. Using the new EPMA data to reconstruct an approximate bulk rock composition for each of the Inver xenoliths, and employing the expression of Canil et al. (Reference Canil, Hyndman and Fode2021) for fertile, off-craton mantle, the depth of the spinel-garnet transition, if it were still present under NW Ireland, should lie at 85–87 km.
Chambers et al. (Reference Chambers, Fullea, Kiyan, Lebedev, Bean, Meere, Daly, Nicola Willmot Noller, Raine, Blake and O’Reilly2025) also computed a present-day geothermal gradient under NW Ireland of ∼35°C per kilometre, equating to ∼90 mWm−2 assuming a thermal conductivity of 2.75 Wm−1K−1. This is consistent with a relatively shallow LAB and the absence of garnet peridotite. To constrain the issue further, the aSiO2 barometer of Lee et al. (Reference Lee, Luffi, Plank, Dalton and Leeman2009) and Plank and Forsyth (Reference Plank and Forsyth2016) was used to estimate the depth of equilibration of the Inver dyke parental melt. The workflow involves an inversion to estimate the composition of the melt that last equilibrated with the mantle, from which pressure and temperature estimates are then derived. The dyke bulk rock composition of Gallagher and Eldson (Reference Gallagher and Elsdon1990) was used, together with an assumed H2O range of 1–3 wt.%. A mantle source olivine forsterite content of 90 ± 1 mol.% together with a Fe3+/∑Fe range of 0.15–0.25 were also factored in (Canil et al. Reference Canil, Hyndman and Fode2021). For a mantle source composition of Fo90, Fe3+/∑Fe of 0.20 and H2O content of 2 wt.%, the thermobarometer returns a temperature and pressure of 1326°C and 18.6 kbar, respectively. Varying H2O to 1 or 3 wt.%, or Fe3+/∑Fe to 0.15 or 0.25, collectively do not lead to changes greater than 1.4 kbar in the pressure, or ∼38°C in the temperature results; in both cases, these are smaller than the inherent uncertainties in the Plank and Forsyth (Reference Plank and Forsyth2016) method. Based on these results, a suggested depth for the LAB under NW Ireland at ∼43 Ma is therefore 59 ± 8 km (Figure 5(b)). This result fits well with the lower end of the range suggested by the xenoliths studied here, particularly Inv-B, allowing for some reequilibration in transit of Inv-A. The combined observations made here are therefore consistent with the loss of deep (garnet-bearing) lithosphere in this part of Ireland, either before or during mantle plume impingement. Thermal history reconstructions of northern Ireland, such as those of Green et al. (Reference Green, Duddy, Hegarty, Bray, Sevastopulo, Clayton and Johnston2000), showing several heating-cooling episodes during the Palaeogene, are likely to have been strongly influenced by the high heat flows associated with such a shallow LAB.
5.b. Implications for a NE Atlantic mantle plume signature and metallogenesis
Xenolith-bearing dykes are well-known from NW Scotland, spanning the major terranes including, from northwest to southeast the Hebridean, Northern Highland, Grampian Highland, Midland Valley and Southern Upland terranes (Upton et al. Reference Upton, Stephenson, Smedley, Wallis, Fitton, Wilson, Neumann, Davies, Timmerman, Heeremans and Larsen2004, Reference Upton, Downes, Kirstein, Bonadiman, Hill and Ntaklos2011). Most of these dykes are Carboniferous in age, related to a period of regional extension and high heat flow, with associated xenolith occurrences known from as far south in the UK as Calton Hill (Derbyshire). Occurrences of xenolith-bearing dykes of Palaeogene age are much less common, with one well-known example located at Loch Roag on the Isle of Lewis (Scotland; Menzies et al. Reference Menzies, Halliday, Palacz, Hunter, Upton, Aspen and Hawkesworth1987; Faithfull et al. Reference Faithfull, Timmerman, Upton and Rumsey2012). Such Palaeogene-aged dykes are important because they provide a snapshot of the lithospheric mantle following the development of the BIPIP, and in particular the potential to assess interaction of the proto-Iceland mantle plume with the underlying CLM. The Inver dyke therefore offers a valuable complement to the insights gained from the Loch Roag assemblage, since it samples lithospheric mantle located ∼400 km away and under orogenic crust associated with Grampian Highland (Dalradian Supergroup) rocks, rather than the Hebridean terrane (cratonic Archaean foreland).
Mantle melting associated with the proto-Iceland plume probably began relatively deep, at >100 km (Kent and Fitton, Reference Kent and Fitton2000), before relatively shallow melting at a later stage in its lifecycle. However, at ∼45 Ma (Kirstein and Timmerman, Reference Kirstein and Timmerman2000; Faithfull et al. Reference Faithfull, Timmerman, Upton and Rumsey2012), dykes like the one at Loch Roag and at Inver postdate the major period of mantle melting that formed most of the BIPIP by ∼15 Ma, and most likely correspond to the widely recognized Eocene rejuvenation of small-scale magmatism associated with the mantle plume (Dobson et al. Reference Dobson, Stuart and Dempster2010). Kirstein and Timmerman (Reference Kirstein and Timmerman2000) reported combined noble gas isotope (He, Ar) data that unequivocally suggest (a) the age of the Inver dyke is Palaeogene and (b) xenoliths from the Inver dyke contain a relatively high 3He component (up to 23 R a), best explained by incorporation of primitive mantle He from the parental melt (host). Otherwise, the spinel lherzolites appear to have seen negligible modification by a mantle plume component: e.g. indicators of melt depletion are moderate-to-low, as are equilibration temperatures, which might not be expected if significant interaction with a plume by thermal erosion or the passage of fluxes of asthenosphere-derived melts had occurred at depth. These combined observations therefore agree well with Kirstein and Timmerman (Reference Kirstein and Timmerman2000), such that the Inver xenoliths were likely derived from unmodified CLM and transported as largely passive cargo. Incorporation of 3He from the dyke magma was probably an inevitable consequence of the relatively high diffusivity of He in such environments.
There appears to be a preference for PGE mineralization to be associated with events wherein mantle plumes interact with CLM (Groves and Bierlein, Reference Groves and Bierlein2007; Maier and Groves, Reference Maier and Groves2011). Clearly the parental melts to PGE(and Ni-Cu)-rich ore deposits have relatively high PGE concentrations, but whether this characteristic is a function of the high degrees of melting by the associated plume events, or interaction with metasomatized CLM, has been an area of discussion for the BIPIP (e.g. O’Driscoll et al. Reference O’Driscoll, Day, Daly, Walker and McDonough2009; Hughes et al. Reference Hughes, McDonald, Faithfull, Upton and Downes2015). Rare sulphide was observed in the Inver xenoliths, typically spatially associated with spinel or clinopyroxene, but also along olivine grain boundaries or triple junctions. The occurrence of this sulphide in the interstitial spaces of the peridotite microstructure suggests it was a late-crystallizing phase, possibly related to formation of the LREE-enriched clinopyroxene in Inv-A or another metasomatic event. Hughes et al. (Reference Hughes, McDonald, Faithfull, Upton and Downes2015) and O’Driscoll et al. (Reference O’Driscoll, Day, Daly, Walker and McDonough2009) postulated that the PGE mineralization observed in magmatic rocks of the BIPIP could have been the result of incorporation of an ancient metal-enriched CLM end member by the ascending asthenospheric melts. Using combined Pt-Os and Re-Os isotope measurements of Rum layered intrusion sulphides, Day and O’Driscoll (Reference Day and O’Driscoll2019) showed, in particular, that Rum sulphides carried up to 4% of a high Pt/Os crustal contribution that could be consistent with long-term residence and isolation of sulphides in ancient lithosphere, before reprocessing by plume-derived melts during asthenosphere-lithosphere interaction. The observations presented here on the Inver xenoliths support the idea that the CLM on which the BIPIP was constructed was fertile, locally metasomatized over 100s of kilometres (i.e. similar signatures observed in Loch Roag pyroxenites; Figure 4(d)), and might therefore have been an important contributor to mineralization in the BIPIP. The Inver dyke itself testifies to LAB depths of ∼59 km, suggesting significant thermal erosion of the CLM occurred, so despite the fact that the Inver xenoliths record only limited interaction with the proto-Iceland plume, CLM was clearly available to contribute to metallogenesis in the BIPIP igneous centres.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S001675682610096X
Acknowledgements
Stephen Daly first told me of the existence of the Inver Dyke nearly 20 years ago, and Sadbh Baxter shared material which allowed me to develop my interest in these rocks. Owen Sullivan and Glenn Poirier are thanked for their assistance with data collection on the electron microprobe at uOttawa. I am also grateful to Dany Savard at LabMaTer (UQAC) for his help with the LA-ICPMS analyses. Current research support from the Natural Sciences and Engineering Research Council of Canada (NSERC Discovery Grant) and from the Newmont Chair in Economic Geology (University of Ottawa) is acknowledged. Dante Canil and an anonymous reviewer provided very constructive comments on an earlier draft of the manuscript. Sarah Sherlock is thanked for editorial handling.
Open access
