Introduction

Hydrogen isotope studies of altered rocks and minerals have enabled identification of cold climate and glacial conditions1,2,3. These studies demonstrate that ancient glacial conditions can be recorded and preserved in δD values within volcanic glass and secondary minerals, and encourage exploration for previously undocumented glaciation in the geologic record.600-m-high.

Antarctica has been in a polar position, near the South Pole, since the late Paleozoic4 and provides an ideal location to explore the potential for hydration of volcanic glass by isotopically depleted polar waters. In particular, the Butcher Ridge Igneous Complex (BRIC) of the Early Jurassic Ferrar Large Igneous Province (LIP) in Antarctica (Figs. 1, 2) contains the southernmost occurrence of abundant volcanic glass that may have been exposed to polar glacial melt water since emplacement c. 183 Ma5.

Fig. 1: Location of Butcher Ridge Igneous Complex.
figure 1

Bedrock map of Northern and Southern Victoria Land, Antarctica, showing the distribution of outcrop and location of the Butcher Ridge Igneous Complex (BRIC) and other localities referred to in the text with demonstrated mid to late-Cretaceous alteration ages. Basemap constructed using Quantarctica v3.2 from the Norwegian Polar Institute (Matsuoka, K. et al. Quantarctica, an integrated mapping environment for Antarctica, the Southern Ocean, and sub-Antarctic islands. Environmental Modeling and Software 140, 105015 (2021)) with Rock outcrop data from SCAR Antarctic Digital Database (ADD) Version 7.0.

Full size image
Fig. 2: Geological features of the Butcher Ridge Igneous Complex.
figure 2

a Simplified geologic map of the Butcher Ridge Igneous Complex (BRIC) with locations of images (b) and (c) arrowed (overlain on WORLDVIEW-2 satellite map) Modified from Nelson et al. 5. b Vitrophyre pod unit of the BRIC, person for scale lower right circled (c) well-layered rhyolite unit, and (d) close-up showing the distinct vitrophyre (VL), marginal (ML), and crystalline (CL) layers. Images modified from: Demian A. Nelson, John M. Cottle, Blair Schoene; Butcher Ridge igneous complex: A glassy layered silicic magma distribution center in the Ferrar large igneous province, Antarctica. GSA Bulletin 132(5–6), 1201–1216 (2020). https://doi.org/10.1130/B35340.1.

Full size image

The BRIC is best exposed along a 10-km-long, escarpment at the head of the Darwin Glacier, ~280 km southwest of McMurdo Station, Antarctica5 (Figs. 1, 2). The BRIC preserves abundant felsic volcanic glass within a 1-km-long dacitic vitrophyre pod unit and a well-layered rhyolite unit, both of which contain distinct meter-scale layering (Fig. 2). These unique layering features are interpreted to result from syn-emplacement alteration, hydration, and devitrification processes5. Detailed petrographic observations are presented in ref. 5, and briefly summarized here. The vitrophyre pod unit contains plagioclase, pyroxene, and oxides set in a glass matrix with veins of alteration. The well-layered rhyolite unit consists of vitrophyre, crystalline, and marginal layers (Fig. 2). Vitrophyre layers have similar mineralogy to the vitrophyre pod but include potassium feldspar. Crystalline layers are holocrystalline resulting from alteration and complete devitrification of glass within vitrophyre layers to secondary potassium feldspar and quartz with abundant oxidation observed in thin section5. The marginal layers represent an intermediate stage of devitrification between vitrophyre and crystalline layers. Marginal Layers are holocrystalline in hand sample but preserve a minor amount of heavily altered glass matrix when observed by scanning electron microscope and contain abundant alteration and devitrification textures5. The remaining units of the BRIC are massive or diffusely-layered and are almost entirely holocrystalline containing primarily plagioclase, pyroxenes, potassium feldspar, and oxides.

Here we present the most depleted hydrogen isotope compositions measured in terrestrial rocks and minerals so far, down to −325 ‰ δD in BRIC glasses that contain abundant (2–6 wt.%) molecular water. We also present bulk-rock water content and in situ water distribution (H2O and OH) mapping using Fourier Transform Infrared Spectroscopy (FTIR). These data and in situ 40Ar/39Ar geochronology suggest that the extremely depleted δD values reflect incorporation of polar glacial melt waters during a Cretaceous hydration alteration event at 72 to 105 Ma and a peak at c. 91.4 Ma. The presence of polar glaciation during this time, i.e., Cenomanian through Campanian, contradicts existing paleoclimate models indicating Antarctica was ice free due to a thermal maximum and global hot greenhouse conditions6,7,8,9,10,11,12. Instead, these data support the presence of widespread polar glaciation during the Late Cretaceous13,14,15,16,17,18,19.

Results

In situ water content and speciation

The total water content of volcanic glass (H2Ot) consists of water incorporated as hydroxyls (OH) and molecular water (H2Om). A complex interplay between interface kinetics, diffusion, and re-speciation processes controls the speciation of water involved in secondary hydration of degassed silicate volcanic glass20,21,22,23. Generally, water incorporated as OH is interpreted as magmatic in origin and retained from earlier incomplete degassing, while molecular water is interpreted as the primary diffusing species and sourced from paleo-environmental meteoric waters24. However, there is evidence that magmatic water may be incorporated as H2Om and meteoric water incorporated as OH depending on the degassing and secondary hydration history of the glass25. In either case, the hydrogen isotope composition of hydrated glass represents a mixture of at least these two isotopically distinct sources that may be distinguished based on water speciation. We measured the in situ water content and speciation (i.e., molecular vs. hydroxyl) of BRIC glass using Fourier Transform Infrared (FTIR) spectroscopy on doubly polished chips (see analytical methods below) to determine if the BRIC contains molecular water from a paleo-environmental meteoric source and to assist in deconvolving the isotopic source(s).

FTIR water content maps (Fig. 3) indicate that glass within the vitrophyre pod has high molecular water contents, ranging between 3 and 8 wt.% H2Om, relatively low OH contents up to 0.8 wt.% OH and higher molecular water content correlating with lower hydroxyl content. Vitrophyre layers from the well-layered rhyolite unit also have high molecular water contents ranging from 2 to 4 wt.% H2Om and low OH contents of up to 0.8 wt.% OH. A sample of the marginal layer of the well-layered rhyolite unit has patches of high molecular water, 2 to 5 wt.% H2Om, with no detectable OH. These water contents are consistent with solubility experiments for H2O in rhyolitic glass of 3–5 wt.% at 175–375 °C26 and indicate comparable or lower temperatures were likely involved during secondary hydration and alteration of the BRIC.

Fig. 3: Fourier transform infrared spectroscopic data for Butcher Ridge Igneous Complex rocks.
figure 3

Fourier Transform Infrared Spectroscopy (FTIR) water speciation map data for vitrophyre pod, vitrophyre layer, and marginal layer samples of the Butcher Ridge Igneous Complex. a Molecular water (H2Om) content, (b) hydroxyl (OH) content, and (c) transmitted light image.

Full size image

These textural observations, water content, and speciation data demonstrate that BRIC glasses preserve high amounts of molecular water and low hydroxyl content, consistent with secondary hydration and minimal preservation of magmatic water or incorporation of secondary waters as OH. Therefore, hydrogen isotopes incorporated into BRIC glass during secondary hydration likely contain critical paleoclimate information.

Bulk water and hydrogen isotopes

We measured the bulk-rock water contents and hydrogen isotope composition (analytical methods described below, Supplementary Data 1, Fig. 4) of samples of the vitrophyre pod, vitrophyre layers, and marginal layers to constrain the δD value of the incorporated paleowater as a paleoclimate indicator (Methods as described in Martin et al. 201727). We also measured the bulk water and δD values of crystalline layers of the well-layered, diffusely-layered, and massive units to determine the magmatic δD values to deconvolve mixing between environmental and magmatic waters. Holocrystalline BRIC samples have an average δD value of −115 ‰ (n = 9) and vary from nearly primary magmatic (−88‰) to −140 ‰ δD with increasing bulk water content (average of 0.78 wt.% H2Ot), characteristic of fractionation during degassing6. Despite the mineralogy of crystalline samples being dominated by plagioclase, pyroxene, oxides, and secondary devitrification minerals potassium feldspar and quartz, water within crystalline samples is interpreted to be stored within hydrated secondary minerals or trace hydrated glass5. Vitrophyre pod samples, in contrast, have a much higher bulk water content (2.13 wt.% H2Ot average) and lower average δD values ranging from −250 to −325 ‰ with an average of −298 ‰. Vitrophyre layers from the well-layered unit have similar bulk water contents as vitrophyre pod samples (2.12 wt.% H2Ot average) but intermediate δD values ranging from −169 to −295 ‰ and an average of −240 ‰. Marginal layers have the highest bulk water contents averaging 4.8 wt.% H2Ot but high δD values ranging from −104 to −151 ‰ and an average of −130 ‰.

Fig. 4: Hydrogen isotope and bulk water data for Butcher Ridge Igneous Complex rocks.
figure 4

Hydrogen isotope and bulk water data for samples of the Butcher Ridge Igneous Complex (BRIC), Scarab Peak Chemical Type (SPCT) lavas, and Antarctic snow and ice28,29.

Full size image

To establish the source of the water, ∆′17O values were measured in the least-hydrated (∆18O = 6.78, ∆′17O = −0.065 ± 0.011 ‰) and lowest δD samples (∆18O = 9.24 ‰, ∆′17O = −0.087 ‰) following the methods of Bindeman (2021)30 (Supplementary Data 2). Samples plot on the terrestrial mass fractionation line, and the end-member δD values in BRIC glasses are interpreted to reflect the δD of incorporated environmental water at the time of secondary hydration31.

BRIC glasses have considerably more depleted δD values than any other known terrestrial rocks and reflect δD values of incorporated environmental water31. Importantly, there is no evidence that significant kinetic hydrogen isotope fractionation (identifiable to within ±20 ‰ δD) occurs when molecular water is mobilized through volcanic glass32. Fractionation does occur during uptake of environmental water, such that δD values of the secondarily hydrated volcanic glass may be offset by −33 ‰ from the environmental water24,25, however, even lower ΔDglass-water values down to −90‰ in some higher temperature hydration experiments have been observed due to secondary repartitioning of environmental water into the isotopically depleted OH site7. Consequently, the δD values recorded in the BRIC glasses record paleo-environmental waters isotopically depleted down to approximately −292 ‰ δD, or even if all water has repartitioned into OH, which is not evident in the FTIR data (Fig. 2), the original paleo-environmental water would still be very depleted down to −235‰. It is likely, however, that the bulk-rock measurements include a minor contribution of higher δD magmatic water. Therefore, the original paleo-environmental δD values are expected to have been more depleted than calculated here. Furthermore, secondary clays and micas have not been observed in BRIC vitrophyre samples but even if present, they would not significantly change the inferred δD of environmental water because mica-water fractionations are comparable to that of glass-water, i.e., ΔD ≤ ~100‰33.

Such low δD values from −235 to −292 have only been observed in polar snow and ice28,29. Recent glaciation in Antarctica began ~34 Ma and persists to the present with snow and ice ranging from −150 to −450 ‰ δD28,29,34 (Fig. 3). The most recent glaciation in Antarctica is therefore a potential source for hydration of the BRIC, but older glacial episodes since the Jurassic may also have been responsible. We addressed the timing of hydration using 40Ar/39Ar geochronology.

In situ 40Ar/39Ar geochronology

Secondary hydration of glass by ~2–5 wt.% water or greater has been shown to disturb the 40Ar/39Ar system by the substitution of H+ for alkalies, allowing potassium migration out of the glass structure, and/or alteration of the glass structure to promote loss of argon35,36. Argon isotope systematics in altered glass may be further complicated by uptake of atmospheric argon, excess non-radiogenic, and/or non-atmospheric argon as well as possible kinetic mass fractionation of argon during hydration35. We anticipate hydration occurred below the closure temperature of argon within glass and minerals, i.e., <500 °C26, and, therefore, hydration is the principal mechanism for disturbance of the 40Ar/39Ar system. If hydration of the BRIC glass favored loss of potassium then the 40Ar/39Ar age would be older than the emplacement age (~183 Ma). If hydration favored argon loss, then the 40Ar/39Ar age should be younger than the emplacement age and consequently reflect a partial or complete resetting of the 40Ar/39Ar age to the age of hydration involving polar glacial met water37,38.

We utilized in situ 40Ar/39Ar geochronology (see methods section below) targeting hydrated glass within six BRIC samples to determine if samples experienced resetting of the argon system during secondary hydration by meteoric waters (Fig. 5, Supplementary Data 3, 4). Analyses were performed on four vitrophyre pod samples, one vitrophyre layer and one marginal layer from the well-layered rhyolite. Vitrophyre pod samples yield similar 40Ar/39Ar dates with an age range of 169.8 to 72.1 Ma and a peak age of c. 91.4 Ma (n = 41). The sample of the vitrophyre layer from the well-layered rhyolite unit yielded a date range of 190.0 to 137.1 Ma (n = 5). Glass from vitrophyre samples, layers and pods are spatially heterogenous and variable within each sample and, therefore, a precise single age cannot be determined. We suspect some analyses inadvertently included a variable contribution of phenocrysts, with original 40Ar/39Ar age c. 183 Ma, and hydrated glass with much younger dates and accounts for the data for vitrophyre dates that fall between ~105 and 190 Ma. Consequently, dates for vitrophyre glass from ~72 to 105 Ma with a peak at c. 91.4 Ma provide the most reliable alteration ages determined for vitrophyre pod samples. The marginal layer of the well-layered rhyolite yielded dates from 182.8 to 175.7 Ma and a peak age of c. 177.7 Ma (n = 6), close to the emplacement age and suggest minimal resetting. In summary, 40Ar/39Ar geochronology of altered and hydrated marginal layers indicate an Early Jurassic age consistent with emplacement during Ferrar LIP magmatism and syn-emplacement alteration, hydration, and layer formation5. The Late-Cretaceous, ~72–105 Ma with a peak at c. 91.4 Ma, dates for the vitrophyre pod samples indicate hydration favored argon loss and partial to complete resetting of the original Jurassic 40Ar/39Ar age and provides our best estimate for the age of secondary hydration by polar glacial melt water.

Fig. 5: Ar/Ar geochronology data for Butcher Ridge Igneous Complex rocks.
figure 5

Kernel density plot and histogram of in situ 40Ar/39Ar geochronology data for vitrophyre pod samples (dark gray), the vitrophyre layer sample (white), and the marginal layer sample (light gray), BR12-10. Fields for previous regional Cretaceous alteration ages in the Transantarctic Mountains (TAM) from 90 to 100 Ma39,40 and the Ferrar Large Igneous Province, 183 Ma, are provided for comparison.

Full size image

Discussion

Two observations are emphasized from the combined FTIR, hydrogen isotope, and in situ 40Ar/39Ar geochronology data: (1) The BRIC experienced hydration/alteration during the Cretaceous, ~72–105 Ma with a peak at c. 91.4 Ma; and (2) Cretaceous hydration of the BRIC involved polar glacial melt water. Importantly, the age of Cretaceous alteration recorded in vitrophyre pod samples is indistinguishable from regional Cretaceous alteration previously documented by 40Ar/39Ar geochronology of secondary apophyllites (Fig. 6) in other Ferrar LIP rocks at various locations along the TAM39,40,41. 40Ar/39Ar total fusion and plateau ages for low-temperature secondary apophyllites yielded ages of 76–100 Ma from the Mesa Range in Northern Victoria Land (NVL, Fig. 1) and 95–101 Ma from Carapace Nunataks in Southern Victoria Land (SVL, Fig. 1), with a peak in the combined data at c. 95.8 Ma40. Rb-Sr model ages for the same apophyllites have reported ages as young as 94 Ma40. Furthermore, an 40Ar/39Ar plateau age of 96.7 + 0.6 Ma was determined from apophyllites for the youngest alteration event in Ferrar LIP rocks in the Prince Albert Mountains approximately ~350 km from the BRIC (Fig. 1)39. Hand samples and individual crystals of apophyllite were reported to have age variations of 24 and 14 Ma, respectively. Typical temperatures estimated for this reported Cretaceous alteration event range from 150 to 350 °C17,18,19. We suggest the similarity in age between BRIC vitrophyre pod samples and secondary apophyllites from throughout the region is strong evidence for the reliability of vitrophyre pod ages and that the vitrophyre pod 40Ar/39Ar system had been completely reset during the Cretaceous alteration event.

Fig. 6: Regional summary of Ar/Ar alteration ages from Victoria Land, Antarctica.
figure 6

Kernel density plot and histogram of in situ 40Ar/39Ar ages for glass from vitrophyre pod samples of the Butcher Ridge Igneous Complex (BRIC) compared to total fusion and plateau 40Ar/39Ar ages for secondary apophyllite minerals within Jurassic Ferrar Large Igneous Province rocks from Northern and Southern Victoria Land that constrain the timing of regional Late-Cretaceous alteration39,40.

Full size image

These data from the BRIC are, therefore, consistent with previous data reported for the region and support a Late Cretaceous, ~72–105 Ma peak at c. 91.4 Ma, alteration event in the TAM that reset the argon system within the BRIC. Cretaceous alteration within the TAM is generally attributed to either deep burial in the Mesozoic Victoria Basin41,42, an elevated thermal gradient combined with circulating fluids related to a failed rift, a mantle plume39, or increased fluid circulation during uplift and denudation of the TAM40,43. Hydrated glass within the BRIC yields variable and spatially heterogeneous ages comparable to variations seen in apophyllites. Considering these variations, we suggest the peak age, c 91.4 Ma, is the most reliable age owing to its consistency with other Late-Cretaceous alteration ages from the region. Younger ages for vitrophyre glass and apophyllite samples, down to c. 72 and 76 Ma, may indicate that alteration was a prolonged process beginning around 100 Ma with hydration reaching near solubility limits for a majority of BRIC vitrophyre glass at c. 91.4 Ma but continuing for millions of years and ending at ~72 Ma. This hypothesis is consistent with a model for Cretaceous alteration by deep burial in the Mesozoic Victoria Basin that persisted until exhumation in the Cenozoic41,42.

The latter observation from these data, that hydrothermal fluids involved in Cretaceous alteration of the BRIC were sourced from polar glacial melt water, relies on the long-term stability of hydrogen isotopes within hydrated glass and resistance to re-equilibration with modern or post-alteration water22. Consequently, the extremely depleted δD values in the BRIC glasses were either obtained during the Cretaceous or during hydrogen isotope re-equilibration that did not reset the 40Ar/39Ar age. If BRIC glasses had δD values overprinted by modern or post-Cretaceous alteration water, then we would expect similar δD values in both vitrophyre and marginal layer samples, both of which have high molecular water contents (Fig. 3). The significant difference in δD values between vitrophyre, −168 to −324 ‰, and marginal layers, −104 to −151 ‰, combined with their difference in 40Ar/39Ar age, c. 91.4 Ma versus c. 178 Ma, strongly suggests BRIC glasses avoided post-Cretaceous alteration and re-equilibration of δD by water or proton diffusion (Figs. 3, 4). Instead, their δD values coupled with field and textural observations record the composition of water during hydration. Marginal layers resisted hydration during the Cretaceous alteration event because they were already saturated with molecular water during Jurassic syn-emplacement hydration5.

Other glassy Early Jurassic Ferrar volcanic rocks in the Mesa Range (Fig. 1), referred to as the Scarab Peak Chemical Type (SPCT), ~700 km north of the BRIC also have a Cretaceous alteration age, c. 103 Ma apophyllite Rb-Sr model age, and depleted δD values, −201 to −243 δD ‰44. Although these δD values are considerably higher than those in the BRIC, we argue that they are also likely indicative of polar glacial melt water. Extremely depleted δD values in rocks that are 700 km apart implies regionally extensive glaciation in Antarctica during the Cretaceous c. 90–100 Ma.

These ages coincide with the Cenomanian-Turonian thermal maximum with hot greenhouse conditions representing the warmest climate of the past 140 million years and poles conventionally thought to be ice free9,45. There remains considerable debate as to whether polar glaciation was possible during this time, but mounting evidence suggests polar glaciation may have occurred during the Late Cretaceous13,14,15,16,17,18,19. These data from the BRIC, therefore, provide yet another line of evidence indicating the presence of Late-Cretaceous polar glaciation in Antarctica from 72 to 105 Ma, and most likely c. 91.4 Ma during the height of hot greenhouse conditions. Furthermore, the results suggest the BRIC and associated rocks are an ideal setting in which to investigate hydration and stable isotope proxies given the significant difference between magmatic and meteoric water stable isotope compositions.

Methods

FTIR

For each doubly polished chip, a transmission Fourier Transform Infrared Spectroscopy (FTIR) absorbance map was generated using short-duration analyses (<1 s) on a Thermo Fisher Nicolet iN10-MX spectrometer at USGS Menlo Park. Data were collected with a 35 µm square aperture that was traced across the sample in a 50 µm grid. For each analysis, peak heights at 5230, 4520, and 3570 cm−1 were measured. The thickness of each chip was measured with a micrometer and the wt. % water values were calculated following the methods outlined in Zhang et al.46.

Hydrogen isotopes

Samples of clean glass, ~1 mm in size, were separated under a microscope from any contamination products. Samples were crushed and sieved, and the 50–150 µm fraction was used for analysis as it provides the best size for water extraction27; Crushed material was washed several times and sonicated in deionized water and ethanol for 15 min to remove dust and particulates. Analyses were performed on a Thermo Fisher high-temperature convertor-element analyzer (TC/EA) with a glassy carbon reactor and crucibles interfaced with a Thermo MAT-253 isotope ratio mass spectrometer (IRMS) housed at the University of Oregon. Two to 3 mg of sample were loaded into Ag capsules depending on H2O content of the glass. In all cases, samples were dried overnight in a hot 130 °C vacuum drying oven. The TC/EA furnace was operated at 1450 °C and gas chromatographs were operated at 70 °C. After vacuum drying, samples were immediately placed into an autosampler carrousel under a Helium atmosphere and dropped one by one into a graphite crucible and placed in the 1450 °C furnace where they were melted to liberate H2O which was immediately quantitatively reduced with CO and H2 gases. These gases pass through and are separated in gas chromatographic column held at 70 °C.

Graphite crucibles were removed between analytical sessions to reduce backgrounds and memory effects. Helium carrier gas flushes extracted volatiles to the IRMS at rates of 80–120 mL/min. Reference waters sealed in Ag tubes and micas were run concurrently for H2O and δ2H calibration. Standards run concurrently with the unknown included:

University of Oregon, Butte Montana muscovite (BUD, δ2H = −151‰), as well as USGS Mica biotite (USGS57, δ2H −91‰) and muscovite (USGS58, δ2H −28‰) are used47, and SMOW (0‰) and GISP (−195‰) waters sealed in Ag capsules. Correction for instrumental (TC/EA) mass fractionation within analytical sessions were 9–11‰ and included a linear fit between SMOW (0‰) and GISP (−195‰), as well as intermediate in δ2H values solid standards. The offsets in the lower and upper ranges were similar and so extrapolation of δ2H data to the lower δ2H range measured in this study is robust.

Water concentrations were determined by knowing water concentration (3.5 wt%) in the mica standards and used integration the peak areas of masses 2 (H2) and 3 (δ2H). Errors on δ2H are estimated to be in the vicinity of 2 permil while water in the vicinity of 0.05‰27,48. Bulk water concentrations determined by TC/EA agree with bulk-rock loss on ignition (LOI) values determined by XRF.

Oxygen isotopes

For oxygen isotope analyses we used two analytical set ups following the methods of Bindeman (2021)30. For triple O analyses, we fluorinated glasses and UWG2 garnet standard with a laser using BrF5 reagent in a single analytical session that preceded a session where we run many mantle olivine standards.

Generated O2 gas was first put through a boiling Hg to remove F2 gas excess, then it was put through an 8 ft long GC column under 10 ml/s He flow and frozen on another zeolite getter. This GC purification procedure removes NFx gases that affect mass 33 (17O/16O) during analysis as is done at the University of Oregon lab30. O2 gas was then run 48 times in a dual inlet mode against a well-calibrated gas to obtain absolute D′17O and d18O values on a MAT253 mass spectrometer connected to the vacuum line. Precision on d18O is ±0.1‰ and on D′17O is ±0.01‰.

D′17 O is defined relative to a reference slope 0.5305 as:

$${{{{{{\rm{D}}}}}}}^{{\prime} 17}{{{{{\rm{O}}}}}}\,=\,{{{{{{\rm{d}}}}}}}^{{\prime} 17}{{{{{\rm{O}}}}}}\,-\,0.5305\,\ast\, {{{{{{\rm{d}}}}}}}^{{\prime} 18}{{{{{\rm{O}}}}}}$$
(1)

Where d′17O are linearized as:

$${{{{{{\rm{d}}}}}}}^{{\prime} 18}{{{{{\rm{O}}}}}}\,=\,1000\,{{{{\mathrm{ln}}}}}({{{{{{\rm{d}}}}}}}^{18}{{{{{\rm{O}}}}}}/1000\,+\,1)$$
(2)
$${{{{{{\rm{d}}}}}}}^{{\prime} 17}{{{{{\rm{O}}}}}}\,=\,1000\,{{{{\mathrm{ln}}}}}({{{{{{\rm{d}}}}}}}^{17}{{{{{\rm{O}}}}}}/1000\,+\,1)$$
(3)

This procedure changes numerical values insignificantly (e.g., d18O is very close to d’18O) but allows for better linear manipulation of delta values over a large delta range.

For conventional d18O analyses, gas generated by laser fluorination was put through boiling mercury to remove excess of F2 gas, then the gas was converted to CO2 by a small C-Pt converter. Yields were measured by a Baratron gauge and then CO2 gas was run for one six-cycle analysis. Error on d18O is ±0.1‰.

40Ar/39Ar geochronology

All 40Ar/39Ar analytical work was performed at the University of Manitoba using a multi-collector Thermo Fisher Scientific ARGUSVI mass spectrometer, linked to a stainless-steel Thermo Fisher Scientific extraction/purification line, Photon Machines (55 W) Fusions 10.6 CO2 laser and Photon Machines (Analyte Excite) 193 nm laser. Argon isotopes were measured using the following configuration: 40Ar (H1; 1 × 1012 Ω resistor), 39Ar (AX; 1 × 1013 Ω resistor), 38Ar (L1; 1 × 1013 Ω resistor), 37Ar (L2; 1 × 1013 Ω resistor) and 36Ar (compact discrete dynode [CDD]). The sensitivity for argon measurements is ~6.3 × 1017 moles/fA as determined from measured aliquots of Fish Canyon Sanidine49,50.

Standards and unknowns were placed in 2 mm deep wells in 18 mm diameter aluminum disks, with standards placed strategically so that the lateral neutron flux gradients across the disk could be evaluated. Planar regressions were fit to the standard data, and the 40Ar/39Ar neutron fluence parameter, J, interpolated for the unknowns. All specimens were irradiated in the Cadmium-lined, in-core CLICIT facility of the Oregon State University TRIGA reactor. The duration of irradiation was 35 h and using the Fish Canyon sanidine (28.201 Ma50).

Standards: were placed in a Cu sample tray, with a KBr cover slip, in a stainless-steel chamber with a differentially pumped ZnS viewport attached to a Thermo Fisher Scientific extraction/purification line and baked with an infrared lamp for 24 h. Single crystals were fused using the CO2 laser.

Unknowns: disks, 3 mm in diameter and ~40 µm thick were cut from the same “thick” polished section investigated by optical microscopy. The disks were mounted using a ceramic adhesive (PELCO®) on a quartz slide placed in a stainless-steel chamber with a sapphire viewport attached to the same stainless-steel high vacuum extraction system as the CO2 laser, and baked with an infrared lamp for 48 h. For this study, a raster size of about 100 µm × 100 µm was used and ablation pits were excavated to an estimated depth of about 40 µm.

Reactive gases were removed for both the standard and unknown, after 3 min, by three GP-50 SAES getters (two at room temperature and one at 450 °C) prior to being admitted to an ARGUS VI mass spectrometer by expansion. Five argon isotopes were measured simultaneously over a period of 6 min. Measured isotope abundances were corrected for extraction-line blanks, which were determined before every sample analysis.

Detector intercalibration (IC) between the different faraday cups was monitored (in Qtegra) every 2 days by peak hopping 40Ar. The intercalibration factor between H1 and the CDD was measured with the unknowns by online analysis of air pipettes. A value of 295.5 was used for the atmospheric 40Ar/36Ar ratio51 for the purposes of routine measurement of mass spectrometer discrimination using air aliquots, and correction for atmospheric argon in the 40Ar/39Ar age calculation. Corrections are made for neutron-induced 40Ar from potassium, 39Ar and 36Ar from calcium, and 36Ar from chlorine52,53,54. Data collection and reduction was performed using Pychron55. The decay constants used were those recommended by ref. 56.