Age and Geochemistry of High Arctic Large Igneous Province Tholeiitic Magmatism in NW Axel Heiberg Island, Canada

The Cretaceous High Arctic Large Igneous Province (HALIP) in Canada involved extrusion of continental flood basalts (CFBs) at 130–120 Ma and 100‐95 Ma and emplacement of an extensive sill and dike network that intersected the Carboniferous to Paleogene Sverdrup Basin. In this paper, we present new 40Ar/39Ar ages, major and trace elements, and Sr‐Nd‐Pb isotope ratios for HALIP lava, dikes, and sills from Bukken Fiord, NW Axel Heiberg Island, Canadian Arctic Islands. Our best constrained 40Ar/39Ar ages yield a weighted average of 124.1 ± 1 (2σ) Ma, coincident with the first pulse of tholeiitic CFB magmatism in the Arctic‐wide HALIP as exemplified by Isachsen Formation flood basalts on Axel Heiberg Island. The Bukken Fiord samples are plagioclase and clinopyroxene‐phyric tholeiitic basalts, are relatively evolved (3.2–6.5 wt% MgO), and share similar major and trace element compositions to typical HALIP tholeiites. Initial 143Nd/144Nd ranges from 0.51260 to 0.51291 and initial 87Sr/86Sr ranges from 0.70362 to 0.70776, while measured 206Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb range from 18.614 to 19.199, 15.534 to 15.630, and 38.404 to 39.054, respectively. The most primitive sample in this study has Sr‐Nd‐Pb isotope signatures that suggest an enriched plume‐derived mantle source for HALIP tholeiites. Most samples, however, possess relatively radiogenic isotope signatures that can be explained by moderate degrees of assimilation of Sverdrup Basin sedimentary rocks. Magma‐crust interaction in the HALIP plumbing system was likely widespread and may have increased the environmental impact of the HALIP, particularly if crustal carbon was volatilized.


Introduction
Large Igneous Provinces (LIPs) are the result of prodigious planetary melting events and are characterized by voluminous outpourings of flood basalt (e.g., Black et al., 2021;Ernst et al., 2019;Mittal et al., 2021).The subterranean portions of LIPs are comprised of extensive sill and dike swarms representing former plumbing systems that supplied melts to the surface.These plumbing systems are envisaged as transcrustal networks of interconnected ramifying magma bodies that form a link between the mantle (where magma is generated by partial melting) and the crust, where magma crystallizes and differentiates (Black et al., 2021;Cashman et al., 2017).Some of the hallmarks of LIPs include their association with global climate perturbations and biotic crises (e.g., Bond & Grasby, 2017;Svensen et al., 2019) and their association with natural resources such as economic-grade mineral deposits and hydrocarbons (e.g., Ernst & Jowitt, 2013).
The extrusive components of continental LIPs are known as Continental Flood Basalts (CFBs).CFBs are typically sub-alkaline (tholeiitic) and commonly display geochemical signatures that indicate the involvement of Abstract The Cretaceous High Arctic Large Igneous Province (HALIP) in Canada involved extrusion of continental flood basalts (CFBs) at 130-120 Ma and 100-95 Ma and emplacement of an extensive sill and dike network that intersected the Carboniferous to Paleogene Sverdrup Basin.In this paper, we present new 40 Ar/ 39 Ar ages, major and trace elements, and Sr-Nd-Pb isotope ratios for HALIP lava, dikes, and sills from Bukken Fiord, NW Axel Heiberg Island, Canadian Arctic Islands.Our best constrained 40 Ar/ 39 Ar ages yield a weighted average of 124.1 ± 1 (2σ) Ma, coincident with the first pulse of tholeiitic CFB magmatism in the Arctic-wide HALIP as exemplified by Isachsen Formation flood basalts on Axel Heiberg Island.The Bukken Fiord samples are plagioclase and clinopyroxene-phyric tholeiitic basalts, are relatively evolved (3.2-6.5 wt% MgO), and share similar major and trace element compositions to typical HALIP tholeiites.Initial 143 Nd/ 144 Nd ranges from 0.51260 to 0.51291 and initial 87 Sr/ 86 Sr ranges from 0.70362 to 0.70776, while measured 206 Pb/ 204 Pb, 207 Pb/ 204 Pb, and 208 Pb/ 204 Pb range from 18.614 to 19.199, 15.534 to 15.630, and 38.404 to 39.054, respectively.The most primitive sample in this study has Sr-Nd-Pb isotope signatures that suggest an enriched plume-derived mantle source for HALIP tholeiites.Most samples, however, possess relatively radiogenic isotope signatures that can be explained by moderate degrees of assimilation of Sverdrup Basin sedimentary rocks.Magma-crust interaction in the HALIP plumbing system was likely widespread and may have increased the environmental impact of the HALIP, particularly if crustal carbon was volatilized.
Plain Language Summary Throughout the Earth's history, there were episodes when extremely large amounts of magma were generated deep in the Earth.As this magma worked its way to the surface, much of it stalled and solidified in the Earth's crust and some erupted.These vast magmatic regions are known as Large Igneous Provinces (LIPs) and one of the most remote of these is the High Arctic LIP (HALIP).We dated a suite of magmatic rocks from a locality at Bukken Fiord in the Canadian Arctic Islands and analyzed them for their elemental make-up to unravel their history.We found that Bukken Fiord magmas formed around 124 million years ago from an upwelling thermal anomaly deep in the Earth and that they later interacted with crustal rocks that existed before the magma was formed.These crustal rocks contained volatile elements which could have been liberated as greenhouse gases when they were heated by the invading HALIP magma, potentially impacting the environment.DEEGAN ET AL.
Although the HALIP has a large areal extent, particularly voluminous onshore exposures occur in the Canadian Arctic Islands, where intrusive and extrusive HALIP igneous rocks form part of the stratigraphy of the Sverdrup Basin (Figure 1; e.g., Evenchick et al., 2015;Saumur et al., 2016Saumur et al., , 2022;;Williamson, 1988;Williamson et al., 2016).The Sverdrup Basin contains up to 13 km of Carboniferous to Paleogene sequences of marine and non-marine sedimentary rocks, including carbonates, mudstones, and sandstones (Embry & Beauchamp, 2019).It can therefore be thought of as a "volcanic basin," a concept that has been applied to similar volcano-tectonic environments where LIP plumbing systems intersected sedimentary basins such as the Vøring and Møre basins in the Norwegian Sea (Svensen et al., 2004) or the Karoo Basin in South Africa (Svensen et al., 2015).Erupted materials in the Sverdrup Basin have been estimated at ca. 20,000 km 3 (Williamson, 1988), but the volume of dikes and sills is thought to be 3 to 5 times greater such that the total magma volume is likely greater than 100,000 km 3 (Saumur et al., 2016).Although some of the extrusive components may have been lost to erosion, the presence of extensive sill and dike swarms has led some authors to suggest that the Canadian HALIP was a predominantly intrusive event (cf.Saumur et al., 2016).
For this study, 11 samples of broadly mafic (basic) igneous rocks were collected from near Bukken Fiord on NW Axel Heiberg Island in 2007 (Figure 1).We present whole-rock major and trace element geochemistry and Sr-Nd-Pb isotope data for these samples, which comprise (a) three dikes (AHI-1, 3, and 4) that appear to have fed into a prominent sill system, (b) five sills (AHI-5 to 9) associated with dikes AHI 1-4, and (c) two lava flows, one of which was sampled at its base and its core (AHI-10 and 11, respectively) and another that was sampled at its core (AHI-12).Based on field relationships, it is likely that the lava flows were fed by magma supplied from the dike and sill system observed in the area.We also obtained 40 Ar/ 39 Ar ages from plagioclase for five of the samples where pristine mineral separations could be obtained (dikes and lava flows).Important to note is that the dikes and sills were emplaced into Triassic sedimentary rocks (mainly shale and sandstone) belonging to the Sverdrup Basin, while the lava flows are interleaved with thin bands of Triassic sandstone and coal.

Sample Preparation
Prior to analysis, all weathered exteriors on the samples were removed using a diamond-edged saw blade at Stockholm University, Stockholm, Sweden.Any saw marks were then removed by polishing and the freshest portions of the sample interiors were selected for further processing, whereupon they were split into three aliquots: one for thin sectioning, one for crushing and pulverizing into rock powder for bulk analysis, and one for plagioclase separation for 40 Ar/ 39 Ar dating.Suitably fresh plagioclase crystals for dating were present in our three dike samples (AHI-1, AHI-3, AHI-4) and two of the lava flow samples (AHI-11, AHI-12).Plagioclase crystals were separated from the crushed material using heavy liquids and subsequently inspected under a binocular microscope.Great care was taken to only select the most pristine, inclusion-free material for analysis. 40Ar/ 39 Ar Age Dating 40 Ar/ 39 Ar step-heating analyses of plagioclase separates were carried out at the University of Florida, USA, following standard methods (e.g., McDougall & Harrison, 1999).Samples were packed in quartz tubes along with flux monitor GA-1550 biotite (98.8 + 0.7 Ma, Renne et al., 1998).Irradiation was performed at the Cadmium-Lined Inner-Core Irradiation Tube (CLICIT) facility of the Oregon State University TRIGA reactor.Samples were step heated in a Modifications Ltd double vacuum resistance furnace.The released gas was expanded into a stainless-steel clean-up line and purified with two 50 L/s SAES getters.Argon isotopes were measured using a MAP215-50 mass spectrometer in electron multiplier mode.Data reduction was performed using ArArCALC software (Koppers, 2002).Data were corrected for system blanks, machine background, and mass discrimination.Correction factors for interfering isotopes were determined from irradiated K-glass and optical-grade CaF 2 salts and the following average values were applied: ( 40 Ar/ 39 Ar) K = 2.66 × 10 −2 , ( 36 Ar/ 37 Ar) Ca = 2.70 × 10 −4 , and ( 39 Ar/ 37 Ar) Ca = 6.76 × 10 −4 .J-values for the unknowns were calculated by interpolating weighted mean J-values for multiple total fusion analyses of multi-grain splits of GA-1550 biotite.Uncertainties on plateau ages are quoted at 2σ and include uncertainties in the J-parameter, the age of GA-1550, and the 40 K decay constant, propagated via the method of Karner and Renne (1998).A 40 K decay constant of 5.543-10 y −1 was used in the age calculations (Steiger & Jäger, 1977).The data were subsequently plotted using the Isoplot v.4.15 add-in for Excel.Note that K/Ca variations were measured in order to check the validity of the 40 Ar/ 39 Ar dates.

Major and Trace Element Geochemistry
Major element oxides were determined using a Rigaku ZSX Primus II X-ray fluorescence (XRF) spectrometer at the PetroTectonics analytical facility, Stockholm University, Stockholm, Sweden.Rock powder was weighed into ceramic crucibles and heated at 1000°C for 6 hr to calculate loss on ignition (LOI).Sample powders were then mixed with a lithium borate flux in a proportion of 1:5 and fused in platinum crucibles for 10 min at 1100°C using a Phoenix VFD automated fusion machine.The fused glass beads were measured by comparison of X-ray intensities for each element with calibration lines generated from 24 international standards of known concentrations.Operating conditions were standard and comparable to other facilities (e.g., Johnson et al., 1999).As an internal quality monitor, two USGS standards (BCR 1 and AGV 2) were analyzed as unknowns at regular intervals.All Fe was measured as Fe 2 O 3 .The analytical precision, based on repeated analyses of USGS reference materials throughout the running sequence, is better than 1%-2%.
Whole-rock trace element abundances were measured by high-resolution inductively coupled plasma mass spectrometry (HR-ICP-MS) at the Pacific Centre for Isotopic and Geochemical Research (PCIGR), University of British Columbia, Canada, using a Thermo Finnigan Element2 instrument.Sample and reference material powders (∼100 mg) were digested using a mixture of concentrated sub-boiled HF and HNO 3 acids in screw-top Savillex® vials on a hotplate at 120°C for ∼36 hr, followed by re-digestion with 6M sub-boiled HCl for 24 hr.The sample solutions were agitated in an ultrasonic bath several times to ensure complete digestion.Each solution was subsequently dried, re-converted to nitrates by fluxing with concentrated sub-boiled HNO 3 for a few hours at 120°C, and dried again.Prior to analysis, the digested samples and reference materials were diluted ∼5,000x with a mixture of 1% HNO 3 + 0.05% HF, containing 10 ppb indium (In), which was used as an internal standard to monitor for instrumental and sensitivity drift.USGS reference material BCR-2 was employed for external calibration assuming the preferred values of Carpentier et al. (2013) and was run after every eight 10.1029/2023GC011083 5 of 24 unknowns throughout the analytical sequence, which also included several analyses of USGS reference material BHVO-2.All measurements were normalized to the internal standard and blank subtracted.For most elements, values obtained for the analyzed reference materials were within <5% RSD of the literature and GeoRem recommended values (Carpentier et al., 2013;Chauvel et al., 2011;Jochum et al., 2005;Raczek et al., 2001;Willbold & Jochum, 2005).Procedural duplicates and replicate measurements also exhibited excellent reproducibility, with relative standard deviations (RSD) generally less than 5% for most elements.Total procedural blanks (low ppb range) are negligible compared to the ppm concentration levels of the samples analyzed.

Isotope Geochemistry
Sr-Nd-Pb isotopic ratios were determined at the Laboratory for Isotope Geology (LIG), Swedish Museum of Natural History, Stockholm, Sweden.All sample preparation and analyses were carried out in clean laboratory conditions (class 1,000 laboratory, class 10 laminar hoods).To ensure that the isotopic compositions acquired were representative of original magmatic ratios, sample powders were acid-leached prior to digestion to remove the effects of post-magmatic alteration and surficial contamination following the method of Nobre Silva et al. (2009Silva et al. ( , 2010)).The Sr-Nd-Pb isotopic compositions were determined on the same sample digest solutions using sequential chromatography.All sample solutions were passed twice through standard HBr-HCl anion-exchange columns for Pb purification and extraction (e.g., Weis et al., 2006).The fractions containing all other sample matrix elements that were washed out during Pb chromatography were then dried down and re-dissolved in 6M HCl and further processed using AG1-X8 anion exchange resin to remove Fe.The Fe-free eluted sample was then loaded onto tandem columns of Eichrom® TRU-Spec resin followed by Sr-Spec or LN-Spec resins for the separation of Sr and Nd, respectively, following the methods of Pin et al. (1994) and Pin & Santos Zalduegui (1997).Total procedural blanks were ∼80, 160, and 800 pg for Pb, Nd, and Sr, respectively, which are negligible compared to the elemental concentrations in the samples.
Sr and Nd isotopic compositions were measured by TIMS (Thermo Finnigan TRITON) in static mode with rotating gain compensation.Sr and Nd isotopic ratios were corrected for mass fractionation assuming the natural values of 88 Sr/ 86 Sr = 0.1194, 146 Nd/ 144 Nd = 0.7219, and an exponential fractionation law.All data were normalized using the average of the corresponding standard (NIST SRM 987 for Sr and La Jolla for Nd) relative to the accepted values of 87 Sr/ 86 Sr = 0.710248 (Thirlwall, 1991) and 143 Nd/ 144 Nd = 0.511858 (Lugmair & Carlson, 1978).During the course of these analyses, the average value of the SRM-987 Sr standard was 0.710217 ± 0.000010 (n = 5) and the La Jolla Nd standard was 0.5118632 ± 0.000003 (n = 12).Pb isotopic compositions were determined by MC-ICP-MS (Micromass IsoProbe) under dry plasma conditions using a desolvating nebulizer (Cetac AridusII) for sample introduction.Analyses were conducted in static multi-collection mode, with interference corrections for 204 Hg on 204 Pb assuming natural abundances ( 202 Hg/ 204 Hg = 4.35) adjusted for instrumental mass fractionation.Instrumental mass fractionation was monitored and corrected online using a Specpure© Tl standard solution with a 205 Tl/ 203 Tl = 0.41892.For each analytical session, all sample and standard solutions were prepared with similar [Pb]/[Tl] (∼10) to ensure matrix matching with respect to Tl, and diluted with 0.5 M HNO 3 to obtain an optimal 208 Pb ion beam of ∼6 V.The standard solution NIST SRM 981 was run after every two samples to monitor the in-run drift.The fractionation-corrected Pb isotopic ratios were then normalized off-line relative to the SRM 981 triple-spike values ( 206 Pb/ 204 Pb = 16.9405, 207Pb/ 204 Pb = 15.4963, and 208 Pb/ 204 Pb = 36.7219;Galer & Abouchami, 1998), using the ln-ln method as described by Albarède et al. (2004).During the period of analysis, the SRM-981 Pb standard yielded mean values ±2SD of 206 Pb/ 204 Pb = 16.9142 ± 0.0022, 207 Pb/ 204 Pb = 15.467 5 ± 0.0018, and 208 Pb/ 204 Pb = 36.6301± 0.0029 (n = 10).

Petrography
All samples possess similar mineralogy but differ in texture, lava samples being more fine-grained than dike and sill samples.Representative thin section photomicrographs (Figure 2) include all dated samples and one altered portion of a basaltic lava flow.Ubiquitous plagioclase feldspar is the major mineral phase.Plagioclase can be fresh but is frequently altered and is present as either sub-mm sized laths in the groundmass, mm-sized phenocrysts (≤3 mm, Figure 2a), or crystal agglomerates (several mm-sized, intergrown laths, Figure 2e).The larger phenocrysts often show complex zoning patterns as well as sieve textures, as exemplified in samples AHI-1 (dike sample, Figure 2a) and AHI-11 (flow sample, Figure 2c).Clinopyroxene phenocrysts are common, relatively small in size (<200 μm), and often altered (compare Figure 2a and Figures 2b and 2d).Oxide minerals (<100 μm) and glass occupy the groundmass along with finely crystalline plagioclase and clinopyroxene.Amphibole is a notable mineral phase in one sill sample (AHI-5).Groundmass alteration is pervasive and brown-colored altered glass is common.Sample AHI-10 is a particularly altered base of a lava flow, with vesicles filled by secondary palagonite (Figure 2f).

40 Ar/ 39 Ar Dates
The results of the 40 Ar/ 39 Ar plagioclase step-heating analyses are shown in Figure 3 and the full data set is provided in Supporting Information S2.We report dates for five samples including three dikes (AHI-1, AHI-3, AHI-4) and two lava flows (AHI-11, AHI-12), all of which yielded optically fresh plagioclase suitable for analysis.Sample AHI-1 produced a date of 114 ± 2 Ma, AHI-3 a date of 125 ± 2 Ma, AHI-4 a date of 123 ± 2 Ma, AHI-11 a date of 129 ± 2 Ma, and AHI-12 a date of 122 ± 1 Ma.The date obtained for sample AHI-1 is not considered meaningful because this sample does not demonstrate compatible K/Ca variation nor does it define a robust plateau age (Figure 3a).Discarding the results for AHI-1, the remaining 40 Ar/ 39 Ar ages (Figures 3b-3d) combine to yield a weighted mean of 124.1 ± 1 (2σ) Ma.Our Bukken Fiord samples are therefore of Early Cretaceous age (Aptian), coincident with the first pulse of tholeiitic CFB magmatism in the HALIP as represented by Isachsen Fm. flood basalts on Axel Heiberg Island (Figure 3f).In the geochemical plots and discussion that follows, we compare the compositions of our samples with literature data for Isachsen Fm. basalts.

Major and Trace Element Geochemistry
The major and trace element compositions of the Bukken Fiord samples are reported in Table 1; Supporting Information S2.We describe and plot all major element oxides normalized to 100% volatile-free compositions.The Bukken Fiord samples have low loss on ignition (LOI) ranging from 0.13 to 0.17 wt.%, consistent with limited alteration.Their SiO 2 contents range from 48.6 to 53.8 wt.%; and MgO from 3.2 to 6.5 wt.%.Total alkali (Na 2 O + K 2 O) contents vary from 2.5 to 5.3 wt.% and all except two sills (AHI-6 and AHI-7) plot as sub-alkaline on a TAS diagram (Figure 4a).On an immobile element classification plot (Zr/Ti vs. Nb/Y), however, all samples are sub-alkaline basalts (Figure 4b) and show excellent overlap with literature data for Isachsen Fm. and Strand Fiord Fm. flood basalts.We attribute the data spread in the TAS diagram to alkali mobility due to minor alteration.
Selected major element oxides and trace elements are plotted against MgO wt.% in Figure 5.Although there is scatter in the data, Al 2 O 3 only shows minor variation with changing MgO, whereas FeO* and CaO tend to decrease with a decreasing MgO (Figures 5a-5c).Likewise, the compatible elements Ni (1-62 ppm, R 2 = 0.54) and Sc (24-40 ppm, R 2 = 0.57) decrease with decreasing MgO, whereas incompatible elements like Ce (33-77 ppm, R 2 = 0.42) increase with a decrease in MgO (Figures 5d-5f).Multielement primitive mantle normalized and rare earth element (REE) chondrite-normalized plots are shown in Figure 6, where it can be seen that the Bukken Fiord samples follow similar patterns to Isachsen Fm. basalts.All samples show negative Sr anomalies, with one sill (AHI-7) showing a particularly marked negative anomaly (Figure 6a).Most samples also show negative Nb anomalies.The samples are comparable to OIB with respect to the most incompatible elements (left side of the multielement diagram) but are more enriched than OIB and display flatter patterns for less incompatible elements (right side of the multielement diagram).In terms of REE (Figure 6b), all samples display sub-parallel patterns that are similar to OIB for the light REE but more enriched in the mid-to heavy REE.
In Sm/Yb(N) versus Ce space (where Sm/Yb is normalized to N-MORB using values in Sun and McDonough (1989), the samples classify as medium-to high-Sm/Yb types (Sm/Yb(N) = 2.1-2.8) with variable Ce contents (33-77 ppm) (Figure 7).Notably, two of our samples (AHI-10 and AHI-11) plot within the field of least fractionated and/or contaminated HALIP tholeiites, suggesting that they are relatively close to inferred parental melt compositions, as discussed below.The samples possess Nb/Yb ratios ranging from 2.9 to 7.0 and Th/Yb from 0.4 to 1.2.In Th/Yb versus Nb/Yb space, all samples plot above the mantle array except for AHI-11, which plots close to Enriched Mid-Ocean Ridge Basalt (EMORB, Figure 8a).In Nb/La versus Th/ Nb space, the samples overlap with enriched mantle (EM) but several possess somewhat elevated Th/Nb ratios (Nb/La = 0.7-1.1 and Th/Nb = 0.1-0.2),similar to previously reported data for Isachsen Fm. basalts (Figure 8b), and trend toward values for Sverdrup Basin sedimentary rocks.The Bukken Fiord samples have moderate Th/ La ratios of 0.1-0.2 and Ba/Th ratios of 26-93 (Figure 8c).Note that some of the literature data for Isachsen Fm. basalts extend to high Ba/Th ratios >200, but such extreme values are not present in our data set.

Isotope Geochemistry
The Sr, Nd, and Pb isotope data for the Bukken Fiord samples are reported in Table 2; Supporting Information S2 and displayed in Figures 9 and 10.Initial 143 Nd/ 144 Nd ratios range from 0.51260 to 0.51291 and initial 87 Sr/ 86 Sr ratios range from 0.70362 to 0.70776 (2σ uncertainties are typically in the sixth decimal place and smaller than the symbols in the figures).Most of the new analyses overlap with the datafield for 120 Ma tholeiitic basalts from Franz Josef Land and with various other samples from the Canadian HALIP (Figure 9).A single sample (AHI-11, lava flow) overlaps with Arctic MORB and Icelandic OIB.Our new data are comparable to previously published data for Isachsen Fm. and Strand Fiord Fm. flood basalts (Figure 9b), but sample AHI-11 stands out as the most isotopically primitive HALIP sample recorded to date with initial 143 Nd/ 144 Nd = 0.51291 (ƐNd = +8.5)and initial 87 Sr/ 86 Sr = 0.70362.
Since our samples display a wide range of uranium contents (from 1.0 to 21.8 ppm; see Table 1), some of which are elevated compared to mantle-derived rocks (Sun & McDonough, 1989), attempts at age-correction of Pb isotope ratios yielded spurious data whose error is difficult to quantify (cf.Rosholt et al., 1973).In the following we cite measured 206      2).In Pb-space, the Bukken Fiord samples overlap with Arctic MORB, Icelandic OIB, and 120 Ma Chukchi basalts, but some display relatively high 207 Pb/ 204 Pb ratios that scatter away from the Northern Hemisphere Reference Line (NHRL) (Figure 10a).Our samples plot roughly parallel to the NHRL on a 206 Pb/ 204 Pb versus 208 Pb/ 204 Pb isotope diagram, similar to the data fields for Arctic MORB, Icelandic OIB, and Chukchi basalts (Figure 10b), although some scatter to slightly higher 208 Pb/ 204 Pb.Our samples plot at relatively high Δ 207 Pb/ 204 Pb values, ranging from Δ 207 Pb/ 204 Pb = 3 for one lava sample (AHI-11) to Δ 207 Pb/ 204 Pb = 9 for two sill samples (AHI-6 and AHI-7), and trend toward upper continental crust (Figure 10c).

Age and Magmatic Affinity of Bukken Fiord Magmatism
The 40 Ar/ 39 Ar ages obtained here combine to yield a weighted average of 124.1 ± 1 (2σ) Ma, which places the Bukken Fiord magmas in the Aptian Stage of the Early Cretaceous (Figure 3).Early Cretaceous ages have been reported for magmatic rocks from several of the Canadian Arctic Islands, including Axel Heiberg, Ellef Ringnes, Ellesmere, and Melville (e.g., Balkwill & Haimila, 1978;Dockman et al., 2018;Estrada & Henjes-Kunst, 2013;Evenchick et al., 2015;Villeneuve & Williamson, 2006;Williamson, 1988).Many of these previous age dates were obtained using the Ar-Ar or K-Ar methods, which are known to be prone to disturbance by secondary events such as alteration (e.g., Corfu et al., 2013;Evenchick et al., 2015;Polteau et al., 2016).There are relatively few high-precision U-Pb ages available for Early Cretaceous HALIP rocks emplaced in the Sverdrup Basin.Examples include mafic intrusions on Ellef Ringnes Island that yielded U-Pb ages of 120.8 ± 0.8 and 126.6 ± 1.2 Ma (Evenchick et al., 2015) and mafic intrusions on Ellesmere and Axel Heiberg Island with U-Pb ages of 120.9 ± 0.9 and 120.3 ± 0.8 Ma, respectively (Dockman et al., 2018).Examples from further afield include U-Pb ages of 124.5 ± 0.2 and 124.7 ± 0.3 Ma for mafic intrusions on Svalbard and a U-Pb age of 122.7 ± 0.3 Ma for a mafic sill on Franz Josef Land (Corfu et al., 2013).The combined data are consistent with a period of Early Cretaceous HALIP magmatic activity extending across the Canadian Arctic, Svalbard, and Franz Josef Land that correlates with stratigraphically constrained Isachsen Fm. flood basalts on Axel Heiberg Island.Given that the 40 Ar/ 39 Ar ages of the Bukken Fiord samples are comparable to U-Pb ages from similar rocks in the region and that their K/Ca ratios show normal and systematic variation (except for sample AHI-1, see Figure 3), we suggest that they are representative of the crystallization age of plagioclase.Based on their age, composition, and location, we conclude that the Bukken Fiord samples correlate with Isachsen Fm. flood basalts and were formed by the same event.
Since LOI is low for all of the Bukken Fiord samples (<0.2 wt.%) and since they plot consistently on an immobile element classification plot (Figure 4b), we argue that the whole-rock data overall reflect primary igneous compositions.The Bukken Fiord samples moreover show coherent trends for major element oxides (e.g., Al 2 O 3 , FeO*, CaO) and compatible elements (e.g., Ni, Sc) (Figure 5) all of which are similar to those exhibited by the bulk of HALIP basalts in the literature (e.g., Bédard et al., 2021a;Naber et al., 2021).In addition, trace and rare element patterns and trace element ratios exhibited by the Bukken Fiord samples are comparable to those previously reported for Isachsen Fm. flood basalts (Figures 6 and 7).These observations lead us to suggest that Bukken Fiord magmas follow compositional trends that typify many HALIP tholeiitic basalts.

Fractional Crystallization
Within the Bukken Fiord sample suite, FeO* and CaO decrease as MgO decreases from 6.5 to 3.2 wt.%, while Al 2 O 3 shows only minor variation (Figure 5).Scatter in the data likely reflects accumulation of plagioclase in some samples, consistent with thin section observations.Overall, the major element oxides appear to be controlled by fractional crystallization of dominantly plagioclase and clinopyroxene (±olivine, ±FeTi oxides), similar to other HALIP tholeiites (e.g., Bédard et al., 2021a;Naber et al., 2021).We explored possible liquid lines of descent by performing fractional crystallization Table 1 Continued models with the Magma Chamber Simulator (Bohrson et al., 2014(Bohrson et al., , 2020) ) utilizing a rhyolite-MELTS v.1.2.0 engine (Ghiorso & Gualda, 2015;Gualda et al., 2012).The models are displayed in Figures 5a-5c and commence with the most magnesian sample in our data set (sample AHI-11 with MgO = 6.5 wt.%), assuming fractionation at 1 kbar of pressure along the fayalite-magnetite-quartz oxygen buffer.Three scenarios are shown with variable H 2 O contents of 0.1, 0.5, and 1 wt.%, of which the model with H 2 O = 0.5 wt.% is the best fit to the data.According to our modeling, the samples reflect crystallization over a temperature range of 70°C.Additional qualitative evidence for fractional crystallization includes trends of decreasing Ni and Sc and increasing Ce with decreasing MgO (Figures 5d-5f), negative Nb and Sr anomalies in the primitive mantle normalized multielement plot (Figure 6a), and relatively high Ce for a given Sm/Yb(N) in many samples (Figure 7).

The Nature of Parental Melts
The sample with the most primitive composition in our data set is AHI-11, which has an MgO content of 6.5 wt.%.AHI-11 also has a low Ce content of 33 ppm, placing it among the least fractionated and/or contaminated HALIP tholeiites in Sm/Yb(N) versus Ce space ("type p melt" using the nomenclature in Bédard et al., 2021a, Figure 7).Sample AHI-10 is also weakly fractionated and/or contaminated ("type k melt") and, together with AHI-11, forms the basis of one or more magma differentiation series with medium to high Sm/Yb ratios (Figure 7).Since shallow magmatic processes cannot significantly alter the Sm/Yb ratio of a melt, the variation in Sm/Yb among HALIP tholeiites is likely due to partial melting at various mantle depths (see Bédard et al., 2021a;Dockman et al., 2018 for discussion).Melts with medium to high Sm/Yb ratios, like those at Bukken Fiord, are thought to derive largely from extensive spinel-field mantle melting (Figure 7).As well as meeting the criteria laid out above for being among the least fractionated and/or contaminated HALIP tholeiites, sample AHI-11 is the only sample in this study that plots in the mantle array in Nb/Yb versus Th/Yb space (close to EMORB in Figure 8a).This observation is significant because the Nb/Yb versus Th/Yb diagram was designed to distinguish between basaltic melts that are relatively uncontaminated (and which form the mantle array on the plot) versus those whose compositions were modified by input of continental crust at some point in their genesis (Pearce, 2008).We therefore suggest that AHI-11 reflects the composition of parental Bukken Fiord melts and that these melts were extracted from an enriched mantle source.
The idea that AHI-11 represents a relatively unfractionated and uncontaminated parental melt composition is consistent with its initial 87 Sr/ 86 Sr, 143 Nd/ 144 Nd, and ƐNd values, which rank among the most primitive so far reported for the Canadian HALIP at 0.70362, 0.51291, and +8.5, respectively (cf.Dockman et al., 2018;Estrada, 2015;Estrada & Henjes-Kunst, 2004;Estrada et al., 2016;Kingsbury et al., 2016; see also Figure 9).In Sr-Nd space, AHI-11 overlaps with Arctic MORB and Icelandic OIB, again pointing to an enriched mantle source similar to the enriched variety of  Literature data (filtered at ≤54 wt.% SiO 2 ; gray symbols) are from the legacy data set for the Canadian HALIP (Bédard et al., 2020(Bédard et al., , 2021a)).Curves in panels (a)-(c) are Magma Chamber Simulator fractional crystallization models employing the most magnesian Bukken Fiord sample as parent (AHI-11 with MgO = 6.5 wt.%).The curves show the liquid line of descent (LLD) at QFM, 1 kbar of pressure, and with varying water contents.Trends of decreasing Ni and Sc and increasing Ce with decreasing MgO are qualitatively consistent with fractional crystallization.Oxides were normalized to 100% before plotting.
with the Arctic MORB and Icelandic OIB reference fields, while all other samples trend toward upper continental crust.The Pb-isotope evidence therefore suggests that HALIP tholeiites were sourced from enriched, plume-derived mantle melts or perhaps from enriched ambient asthenospheric mantle (Figure 11).Our findings are thus in agreement with previous workers who argued for a mantle plume origin for HALIP tholeiites (e.g., Buchan & Ernst, 2018;Drachev et al., 2018;Embry & Osadetz, 1988;Estrada & Henjes-Kunst, 2013;Tegner et al., 2011).We note that, in contrast to HALIP alkaline rocks where a HIMU component has been inferred (see Bédard et al., 2021b), there is currently no Pb-isotope evidence for a HIMU component in HALIP tholeiites.

Evidence for Crustal Assimilation
Having established the composition of parental melts in the region (defined above), the potential role of crustal assimilation in melt evolution at Bukken Fiord can be assessed.Recent work by Bédard et al. (2021a) synthesized the existing geochemical data for the Canadian HALIP and employed exhaustive trace element modeling to propose that HALIP tholeiites are strongly fractionated and record almost ubiquitous assimilation of continental crust.
A useful diagram to examine this concept is a Nb/Yb versus Th/Yb plot (Pearce, 2008; see Figure 8a), where it can be seen that many HALIP basaltic samples, including those from Bukken Fiord, plot outside the gray sloping bar that defines the mantle array.Since generating basaltic compositions outside the mantle array is difficult to achieve via normal mantle melting processes, samples that plot outside this zone were most likely influenced by the addition of continental crust with high Th/Yb.To underscore this point, note that sedimentary rocks from the Sverdrup Basin (Patchett et al., 2004) plot in the upper right quadrant of Figure 8a (brown circles) with high Th/ Yb for a given Nb/Yb.Previous work showed that back-modeling the trace element compositions of HALIP tholeiites to remove the effects of fractionation and assimilation produces parental compositions within the mantle array while forward assimilation and fractional crystallization modeling was shown to reproduce the observed HALIP data (see Bédard et al., 2021a;Naber et al., 2021).Given that the Bukken Fiord data overlap exactly with Isachsen Fm. tholeiites in Figure 8a, they too likely record the effects of crustal assimilation.This conclusion is consistent with the observation that the Bukken Fiord samples possess high Ce for a given Sm/Yb (Figure 7), which is another proxy for assimilation and fractional crystallization processes.
Further trace element constraints on crustal additions to the Bukken Fiord magmas can be obtained from plots of Nb/La versus Th/Nb (Figure 8b) and Th/ La versus Ba/Th (Figure 8c).Ratios of Nb/La are generally low in continental crust and sedimentary rocks, whereas Th/Nb is generally high.Therefore, mixing between an enriched mantle-derived melt (such as EMORB or our most primitive sample AHI-11) and Sverdrup Basin sedimentary rocks will result in trends toward low Nb/La and high Th/ Nb (Wang et al., 2016).It can be seen in Figure 8b that the Bukken Fiord samples scatter along such mixing paths and reflect moderate (less than 20%) crustal input.We also note that the literature data for HALIP basalts show a good fit to these curves, which is consistent with the findings of previous workers as discussed above.As an alternative to crustal assimilation, Kingsbury et al. (2016) explored the possibility of contamination of the mantle source region.These authors suggested that high Ba/Th ratios in Isachsen Fm. basalts, some of which exceed 200, reflect the presence of Ba-rich, subduction-modified sediment in the mantle source.However, such extremely high Ba/Th ratios are not observed in our Bukken Fiord data set (max Ba/Th = 93) and our samples scatter along mixing paths between EMORB (or sample AHI-11) and Sverdrup Basin sedimentary rocks (Figure 8c).We therefore think it unlikely that the source for the Bukken Fiord magmas was contaminated with subduction-modified sediment, although we cannot rule out other Ba-enrichment processes for HALIP rocks that possess extremely high Ba/Th.Sun and McDonough (1989).Isachsen Fm. data are from the legacy data set for the Canadian HALIP (Bédard et al., 2020(Bédard et al., , 2021a)).Abbreviations: EMORB, enriched mid-ocean ridge basalt; NMORB, normal mid-ocean ridge basalt; OIB, ocean island basalt.Estrada et al. (2016) suggested that two types of crustal contaminants contributed to HALIP magma genesis, one from the lower crust with relatively low Sr isotope ratios and one similar to upper continental crust with high Sr isotope ratios (e.g., Pearya orthogneisses with 87 Sr/ 86 Sr > 0.74).Dockman et al. (2018) later reported Sr-Nd isotope data for two samples of sedimentary rocks from the Sverdrup Basin and proposed these as possible end-members in HALIP crustal assimilation models.Since Bukken Fiord sills and dikes were emplaced into the Sverdrup Basin, we tested the viability of the crustal samples in Dockman et al. (2018) as contaminants.However, we found that mixing scenarios using these sedimentary rock samples were unsuccessful because the samples have relatively high initial 143 Nd/ 144 Nd (0.51192-0.51216)and low initial 87 Sr/ 86 Sr (0.70846-0.72345),whereas feasible assimilation models would require a contaminant with significantly lower initial 143 Nd/ 144 Nd and higher initial 87 Sr/ 86 Sr ratios.Patchett et al. (2004) provided trace element and Nd isotope data for Sverdrup Basin sedimentary rocks, but unfortunately there are no corresponding Sr isotope data available.The Nd isotope data in Patchett et al. (2004) display a wide range, as shown in Figure 9a.We performed two-component mixing calculations employing sample AHI-11 as a proxy for an uncontaminated parental melt and various hypothetical crustal end-members based on data for Sverdrup Basin sedimentary rocks from Ellesmere and Axel Heiberg Island in Patchett et al. (2004) with high, medium, and low initial 143 Nd/ 144 Nd (a, b, and c in Figure 9a; see Supporting Information S2 for details).In the absence of Sr isotope data, we assigned these crustal end-members an initial 87 Sr/ 86 Sr ratio >0.72 (cf.Dockman et al., 2018;Estrada et al., 2016).We also assigned these hypothetical crustal end-members the average Sr and Nd concentrations of the Sverdrup Basin samples from Ellesmere and Axel Heiberg Island for which Nd-isotope ratios exist in Patchett et al. (2004) (i.e., 72 and 34 ppm, respectively).Our mixing models reproduce the variations in the Bukken Fiord data quite well, although several of our samples (notably sills AHI-7 and AHI-8) plot at high 87 Sr/ 86 Sr for a given 143 Nd/ 144 Nd (Figure 9b) possibly due to Sr-enrichment from minor alteration (recall that some sill samples are also mildly enriched in alkalis).The outlying sill data aside, the remaining Bukken Fiord data can be explained by less than 20% mixing with a type B or C crustal end-member.We also explored mixing in Th/Ce versus 143 Nd/ 144 Nd space and, depending on the exact end-member used, the data again reflect moderate crustal addition (Figure 9c).The results of our models are roughly in-line with the thermal limits of crustal assimilation by basaltic melts (Heinonen et al., 2022).However, we emphasize that all HALIP assimilation models are necessarily tentative because there is a large degree of uncertainty surrounding the composition of the assimilated crust due to the paucity of available data.
This study is the first to report Pb isotope data for Canadian HALIP tholeiites.Measured Pb-isotope ratios (Figure 10) are relatively radiogenic and fall mostly above the NHRL, overlapping with Arctic MORB and Icelandic OIB (reference fields drawn using the data compilation by Stracke et al., 2022aStracke et al., , 2022b)).Data for 118-112 Ma basalts from the Chukchi Borderland (western Arctic Ocean; Mukasa et al., 2020) show similarities to Bukken Fiord, but some of our samples scatter to relatively high 207 Pb/ 204 Pb for a given 206 Pb/ 204 Pb, toward the composition of upper continental crust (Millot et al., 2004) (Figure 10a).A plot of Δ 208 Pb/ 204 Pb versus Δ 207 Pb/ 204 Pb is a useful tool for identifying crustal components in basalts (e.g., Hart, 1988;Jackson et al., 2007;Figure 10c).In this diagram, average upper continental crust plots in the upper right quadrant at high Δ 207 Pb/ 204 Pb values whereas various mantle components (depleted, enriched, and HIMU mantle) plot at lower Δ 207 Pb/ 204 Pb values (Jackson et al., 2007).The Bukken Fiord samples plot at relatively high Δ 207 Pb/ 204 Pb values, ranging from Δ 207 Pb/ 204 Pb = 3 for one lava sample (AHI-11) to Δ 207 Pb/ 204 Pb = 9 for two sill samples (AHI-6 and AHI-7).The data appear to form a trend extending from mantle values toward upper continental crust values.Therefore, although quantifying crustal assimilation is hindered by a lack of Pb isotope data for potential contaminants, there are strong indications of the uptake of crustal Pb by Bukken Fiord magmas.Bédard et al., 2021a;Saumur et al., 2022).In this scheme, HALIP tholeiites are divided into low Sm/Yb types (a-b-c series), medium Sm/Yb types (d-e-f-g and k-l-m series), high Sm/Yb types (p-q-r series), and very high Sm/Yb types (field labeled "u").Cerium is employed as a proxy for assimilation and fractional crystallization (AFC) while Sm/Yb is a proxy for melting depth of parental melts.Parental melts undergoing AFC will evolve from low-to high-Ce compositions along the AFC paths shown in Bédard et al. (2021a) (not shown here for clarity).Hatched fields denote the most primitive HALIP melts and include samples AHI-10 (type k) and AHI-11 (type p) from this study, both of which are from lava flows.Sm/Yb is normalized to N-MORB using values in Sun and McDonough (1989).

Figure 8.
In summary, the Sr-Nd-Pb isotope data allow us to infer that Bukken Fiord magmas assimilated continental crust with highly radiogenic isotope ratios.Sverdrup Basin sedimentary rocks likely made up a large component of the assimilated material.Magma-crust interaction in the Canadian HALIP involving Sverdrup Basin rocks has also been proposed on the basis of stable isotope ratios (Deegan et al., 2018), but more detailed modeling and robust quantification of the process requires isotopic data for local crustal rocks that are not yet available.

Wider Implications
This study provides new isotopic evidence supporting the idea that HALIP tholeiites experienced widespread magma-crust interaction in the dike-sill network emplaced within the Sverdrup Basin.Recognizing assimilation of crustal materials is important at LIPs because it can lead to the formation of ore deposits (e.g., Hayes et al., 2015;Lesher, 2017) and influence the overall volatile budget (e.g., Black et al., 2021;Deegan et al., 2022;Svensen et al., 2015).Recent work by Saumur et al. (2022) suggested that the Early Cretaceous magmatic event was the most prospective episode of the HALIP for mineralizations of Ni, Cu, and platinum group elements (PGEs).The locality at Bukken Fiord meets many of the criteria for prospectivity laid out in Saumur et al. (2022); notably it contains tholeiitic intrusions that are locally differentiated, that originated from primitive magmas, and that record evidence for crustal assimilation.It may therefore be possible that the region in and around Bukken Fiord is prospective, but this would require further testing.(Bédard et al., 2020(Bédard et al., , 2021a)).(a) Nb/Yb versus Th/Yb diagram after Pearce (2008).Data for Sverdrup Basin sedimentary rocks (SB, brown circles) are from Patchett et al. (2004).Upward trending arrows show the effects of crustal assimilation, which causes samples to scatter away from the mantle array (gray sloping bar).(b) Nb/La versus Th/Nb diagram after Wang et al. (2016).Also shown are mixing curves utilizing Sverdrup Basin shale from Patchett et al. (2004) and either AHI-11 (curve with black tick marks) or EMORB (curve with blue tick marks; values from Sun and McDonough (1989) as end-members).(c) Th/La versus Ba/Th diagram.Extremely high Ba/Th ratios in Isachsen Fm. basalts were proposed by Kingsbury et al. (2016) to reflect the presence of subduction-modified sediments in the mantle source.However, there are no extreme values in the Bukken Fiord data and they largely scatter along mixing curves between an enriched parent and SB (data sources as above).Abbreviations: CC, continental crust; EM, enriched mantle; HIMU, highμ where μ = 238 U/ 204 Pb; PM, primitive mantle; all others as in Figure 6.Note.The full data are also provided in the Supporting Information along with procedural duplicates.
a Sr and Nd isotope ratios measured by TIMS.Data are normalized to the mean value of the SRM-987 Sr standard solution ( 87 Sr/ 86 Sr = 0.710217) relative to the value of 0.710248 (Thirlwall, 1991) and of the La Jolla Nd standard solution ( 143 Nd/ 144 Nd = 0.511863) relative to the value of 0.511858 (Lugmair et al., 1983), respectively.b 2σ is the absolute error value of the individual sample analysis.c Initial Sr and Nd and isotope ratios calculated on the basis of our   (Patchett et al., 2004).There are no 87 Sr/ 86 Sr ratios available for SB sedimentary rocks (apart from the two samples mentioned above) but here we assume values >0.72 (cf.Dockman et al., 2018).Various mixing paths are shown with hypothetical SB end-members A-B-C (details are provided in Supporting Information S2).
(b) Close-up of the box in panel (a).In Sr-Nd space, the Bukken Fiord samples overlap with Isachsen and Strand Fiord Fm. lavas (Estrada & Henjes-Kunst, 2004), except for sample AHI-11 which is more primitive.Most of our samples scatter along mixing curves with SB sedimentary rocks, but some sills show signs of Sr-enrichment from alteration.(c) Th/Ce versus 143 Nd/ 144 Nd showing data for Bukken Fiord samples compared to SB shale and siltstones (Patchett et al., 2004).Dashed vertical lines indicate the Th/Ce ratios of OIB and EMORB (after Sun and McDonough, 1989) and Upper Continental Crust (UCC, after Rudnick and Gao, 2003).Abbreviations: UCC, upper continental crust; all others as in Figures 6 and 8.
Mantle end-members are after Jackson et al. (2007).Abbreviations: DM, depleted mantle; all others as in Figures 6 and 8.Note that the Bukken Fiord data trend toward UCC and that there is no evidence for a HIMU component in HALIP tholeiites.Adloff et al., 2020;Galloway et al., 2022;Polteau et al., 2016).This study supports the idea that enriched, mantle-derived parental magmas supplied a magma plumbing system emplaced within the Sverdrup Basin (Figure 11).Magma-crust interaction in the plumbing system could conceivably have led to the release of carbon volatiles and contributed to an increase in the overall volatile budget of the HALIP (cf.Bédard et al., 2023;Deegan et al., 2022).
A similar model was proposed for the Barents Sea expression of the HALIP, where an extensive sill complex emplaced into Permian and Triassic sedimentary rocks is thought to have released up to ca. 20,000 Gt of thermogenic carbon through contact metamorphic reactions (Polteau et al., 2016).Similar amounts of carbon could be generated from sills emplaced in the Sverdrup Basin in Canada, given that the volume of magma generated by the HALIP in Canada exceeds 100,000 km 3 (Saumur et al., 2016), which is a similar order of magnitude as the volume of Barents Sea intrusives (see also Bédard et al., 2023 for discussion).If thermogenic volatiles generated through magma-crust interaction in the HALIP plumbing system reached the ocean and/or atmosphere, they may have contributed to environmental changes in the Aptian (cf.Galloway et al., 2022).

Conclusions
The main conclusions that can be drawn from our new 40 Ar/ 39 Ar ages, geochemistry, and Sr-Nd-Pb isotopic data are as follows: • Bukken Fiord magmas crystallized in the Aptian Stage of the Early Cretaceous at ca. 124.1 ± 1 (2σ) Ma.This age coincides with the first pulse of tholeiitic HALIP continental flood basalt magmatism in the circum-Arctic, as exemplified by Isachsen Fm. basalts on Axel Heiberg Island (Canadian Arctic Islands).Magmas gener ated at Bukken Fiord are geochemically indistinguishable from Isachsen Fm. basalts, leading us to suggest that they are correlative and formed during the same magmatic event.
• Bukken Fiord tholeiitic magmas evolved and fractionated during transport and emplacement in the crust.The trace element and Sr-Nd-Pb isotope signatures of the least fractionated and/or contaminated magma type from Bukken Fiord suggest that parental melts were extracted from a mantle source similar to either the enriched variety of MORB or a mantle plume head.
• The magma plumbing system supplying melts at Bukken Fiord spanned the crust and intersected the Sverdrup Basin.Parental magmas assimilated continental crust similar to sedimentary rocks of the Sverdrup Basin, thus inheriting radiogenic isotope signatures.Based on Sr-Nd isotope mass balance, assimilation was probably less than 20%.High Δ 207 Pb/ 204 Pb values for a given Δ 208 Pb/ 204 Pb in Bukken Fiord magmas are a strong indication of crustal assimilation in this portion of the Canadian HALIP.• Magma-crust interaction in the sill-dike network emplaced within the Sverdrup Basin could have mobilized significant amounts of carbon volatiles from shale and carbonate.If these volatiles were discharged to the ocean and/or atmosphere, they may have contributed to environmental changes in the Aptian (e.g., OAE 1a).
Figure 11.A schematic model for tholeiitic magmatism in the Canadian HALIP (after Bédard et al., 2021a).Primary melts were extracted from an enriched mantle or plume head at various depths and amalgamated in a magma plumbing system comprising a dike-sill swarm emplaced in the crust.Parental melts feeding magmatism near Bukken Fiord at ca. 124 Ma were extracted from intermediate mantle depths relative to the HALIP as a whole.Magma emplacement into the Sverdrup Basin facilitated magma-crust interaction and assimilation of sedimentary rocks, which modified primary Sr-Nd-Pb isotope ratios.Repeated melt injection into the magma plumbing system would have caused over-pressure and eruption of flood basalts from shallow magma reservoirs.Increased crustal heat flow and magma-crust interaction may have triggered the release of thermogenic volatiles in addition to the magmatic volatile budget.OL = oceanic lithosphere.
geochemical data from a variety of sources on magmatic rocks from the Canadian portion of the HALIP (Bédard et al., 2020).This data set forms the basis of the papers by Bédard et al. (2021aBédard et al. ( , 2021b) ) and is available as an Excel spreadsheet via the Geological Survey of Canada as an Open File (open access).The Arctic MORB and Iceland OIB radiogenic isotope data sets used for comparative purposes (Figures 9 and 10) form the basis of the paper by Stracke et al. (2022a) and are available from the GEOROC Data Repository (Stracke et al., 2022b).The additional comparative isotopic data shown in Figures 9 and 10 are reported in the respective cited publications and their Supporting materials (i.e., Dockman et al., 2018;Estrada, 2015;Estrada & Henjes-Kunst, 2004;Estrada et al., 2016;Millot et al., 2004;Mukasa et al., 2020;Ntaflos & Richter, 2003;Patchett et al., 2004).Data reduction for the 40 Ar/ 39 Ar data was performed using ArARCALC software (Koppers, 2002), which is freely available from https://earthref.org/ArArCALC/.The data were subsequently plotted using the Isoplot v.4.15 add-in for Excel, which is freely available from https://www.bgc.org/isoplot.The fractionation models shown in Figure 5 were created using the Magma Chamber Simulator (MCS), which is an open-system thermodynamic modeling tool described by Bohrson et al. (2014Bohrson et al. ( , 2020) ) and freely available at https://mcs.geol.ucsb.edu.In this study, the MCS was run on rhyolite-MELTS v.1.2.0 as described by Ghiorso and Gualda (2015) and Gualda et al. (2012) and is freely available at http://melts.ofm-research.org.Shephard for editorial handling.We are grateful to the Swedish Polar Research Secretariat (SPRS) and the Cambridge Arctic Shelf Programme (CASP) consortium for logistical support during sample collection.We thank the laboratory staff at the Swedish Museum of Natural History (NRM) and the Pacific Centre for Isotopic and Geochemical Research (PCIGR) for assistance with MC-ICP-MS and HR-ICP-MS analyses, respectively.This research was supported by Swedish Research Council (Vetenskapsrådet) Grants to F.M. Deegan (2018-04933 and 2022-04569) and V. Pease (2014-4375) and from the Circum-Arctic Lithosphere Evolution (CALE) project sponsors.This is Geological Survey of Canada contribution 20230096.

Figure 1 .
Figure 1.Study area.(a) Polar projection map of the High Arctic (after Jakobsson et al., 2020) showing the extent of HALIP magmatism (after Naber et al., 2021).Axel Heiberg Island (AHI) in the Canadian High Arctic is labeled.(b) Close-up of box in (a) using a NAD 83/UTM zone 15N Lambert conformal conic projection and showing the location of Bukken Fiord.Crosses show the occurrences of tholeiitic intrusions (after Bédard et al., 2021a) and colored squares show the occurrences of Isachsen Fm. and Strand Fiord Fm. tholeiitic lava as well as the stratotype area for Hansen Point tholeiites (after Bédard et al., 2021a; Naber et al., 2021).AR = Amund Ringnes Island, ER = Ellef Ringnes Island.(c) Panoramic photograph from Bukken Fiord, showing Sverdrup Basin sedimentary formations intruded by numerous HALIP mafic intrusions (outlined in yellow).

Figure 3 .
Figure3.Age dates and stratigraphic context for the samples in this study.(a-e)40 Ar/ 39 Ar results for Bukken Fiord samples.Unfilled boxes are rejected steps and the height of the boxes corresponds to 2σ error.The best constrained ages (rejecting sample AHI-1) combine to yield a weighted average of 124.1 ± 1 Ma (2σ).(f) Stratigraphic column afterBédard et al. (2021b),Dewing & Embry (2007),Naber et al. (2021), andPointon et al. (2019) with depositional phases afterEmbry and Beauchamp (2019).Episodes of peak HALIP magmatism are highlighted to the right of the column.
Pb/204 Pb, 207 Pb/ 204 Pb, and 208 Pb/ 204 Pb ratios for the Bukken Fiord samples, which range from Major and Trace Element Data for Mafic Rock Samples From Bukken Fiord, Axel Heiberg Island dated in this study (see Figure 3; Supporting Information S2 datafile for results).b All Fe was measured as Fe 2 O 3 .c FeO* = all iron as FeO.

Figure 5 .
Figure5.Major and trace element variations for Bukken Fiord samples.Literature data (filtered at ≤54 wt.% SiO 2 ; gray symbols) are from the legacy data set for the Canadian HALIP(Bédard et al., 2020(Bédard et al., , 2021a)).Curves in panels (a)-(c) are Magma Chamber Simulator fractional crystallization models employing the most magnesian Bukken Fiord sample as parent (AHI-11 with MgO = 6.5 wt.%).The curves show the liquid line of descent (LLD) at QFM, 1 kbar of pressure, and with varying water contents.Trends of decreasing Ni and Sc and increasing Ce with decreasing MgO are qualitatively consistent with fractional crystallization.Oxides were normalized to 100% before plotting.

Figure 6 .
Figure 6.Trace element patterns for Bukken Fiord samples.Primitive mantle normalized multielement plot (a) and chondrite normalized rare earth element plot (b) showing data from this study compared to Isachsen Fm. rocks and various mantle end-members.Normalizing values and data for EMORB, NMORB, and OIB are fromSun and McDonough (1989).Isachsen Fm. data are from the legacy data set for the Canadian HALIP(Bédard et al., 2020(Bédard et al., , 2021a)).Abbreviations: EMORB, enriched mid-ocean ridge basalt; NMORB, normal mid-ocean ridge basalt; OIB, ocean island basalt.

Figure 7 .
Figure 7. Ce versus Sm/Yb(N) classification diagram for HALIP tholeiites (afterBédard et al., 2021a;Saumur et al., 2022).In this scheme, HALIP tholeiites are divided into low Sm/Yb types (a-b-c series), medium Sm/Yb types (d-e-f-g and k-l-m series), high Sm/Yb types (p-q-r series), and very high Sm/Yb types (field labeled "u").Cerium is employed as a proxy for assimilation and fractional crystallization (AFC) while Sm/Yb is a proxy for melting depth of parental melts.Parental melts undergoing AFC will evolve from low-to high-Ce compositions along the AFC paths shown inBédard et al. (2021a) (not shown here for clarity).Hatched fields denote the most primitive HALIP melts and include samples AHI-10 (type k) and AHI-11 (type p) from this study, both of which are from lava flows.Sm/Yb is normalized to N-MORB using values inSun and McDonough (1989).

Figure 8 .
Figure8.Trace element ratio plots.Data from Bukken Fiord are shown in comparison to literature data (gray symbols) from the legacy data set for the Canadian HALIP(Bédard et al., 2020(Bédard et al., , 2021a)).(a) Nb/Yb versus Th/Yb diagram afterPearce (2008).Data for Sverdrup Basin sedimentary rocks (SB, brown circles) are fromPatchett et al. (2004).Upward trending arrows show the effects of crustal assimilation, which causes samples to scatter away from the mantle array (gray sloping bar).(b) Nb/La versus Th/Nb diagram afterWang et al. (2016).Also shown are mixing curves utilizing Sverdrup Basin shale fromPatchett et al. (2004) and either AHI-11 (curve with black tick marks) or EMORB (curve with blue tick marks; values fromSun and McDonough (1989) as end-members).(c) Th/La versus Ba/Th diagram.Extremely high Ba/Th ratios in Isachsen Fm. basalts were proposed byKingsbury et al. (2016) to reflect the presence of subduction-modified sediments in the mantle source.However, there are no extreme values in the Bukken Fiord data and they largely scatter along mixing curves between an enriched parent and SB (data sources as above).Abbreviations: CC, continental crust; EM, enriched mantle; HIMU, highμ where μ = 238 U/ 204 Pb; PM, primitive mantle; all others as in Figure6.