The MDC phlogopite was cut from a 10×7×1.5 cm crystal, sourced from the Bekily region, Madagascar (Redaa et al., 2021). It is taken to be the equivalent in geological source and age as the Mica-Mg powdered reference material distributed by the Centre de Recherches Petrographiques et Geochimiques (CRPG; Govindaraju, 1979, 1995). The expected age of 519.4±6.5 Ma used in this study, is the average reported Rb–Sr, K–Ar and U–Pb age of the region (Hogmalm et al., 2017; and references therein). The initial 87Sr/86Sr ratio of 0.72607±0.00070 was measured from two diopside analyses from the Mica-Mg sample (Morteani et al., 2013). More recent ID-MC-ICP-MS results from the Mica-Mg powdered samples give a relatively imprecise date of 521±24 from 12 aliquots (Jegal et al., 2022) and 524.9±3.6 from 1 aliquot (Redaa et al., 2022), indicating some uncertainty in the batch-to-batch reproducibility of the Rb–Sr age from the Mica-Mg powdered sample.

The Kola phlogopite was sourced from the Kovdor Massif, located on the Kola Penisnsula, Kovdorsky District, Russia. The cm-scale phlogopite book was purchased from a mineral dealer and the exact location is unknown. Similar sized phlogopites characteristic of the deposit are formed from high temperature ultrabasic pegmatites within a carbonatite-bearing alkaline-ultrabasic complex (Amelin and Zaitsev, 2002; Krasnova, 2001). A narrow range of TIMS U-Th-Pb dates have been reported from the Kovdor Massif by Amelin and Zaitsev (2002): ca. 378 Ma (baddeleyite, 238U/206Pb lower intercept), 377.52±0.94 Ma (zircon 208Pb/232Th), 377.5±3.5 Ma (apatite, 238U/206Pb lower intercept) and 380.6±2.6 Ma (apatite and calcite, 238U/206Pb lower intercept). Baddeleyite dates of 377±4 and 383±3 have also been reported (Bayanova, 2006). Phlogopites have been dated by conventional Rb–Sr, giving a date of 378.3±1.5 Ma (Amelin and Zaitsev, 2002), recalculated using the decay constant of Villa et al. (2015) and initial 87Sr/86Sr of 0.70368±0.00055. A weighted mean date based on U–Pb and Rb–Sr analyses of 378.1±1.2 was used as the reference age in this study (Table 1).

The Robins Folly muscovite was sourced form undeformed pegmatites in the Hartz Range metaigneous province, Northern Territory, Australia, emplaced during the late stages of the Alice Springs Orogeny (450–300 Ma). The sample was collected near the Ruby Mines from the same location as described in Mortimer et al. (1987). This study reported a whole rock Rb–Sr date of 320±22 Ma (recalculated using the decay constant of Villa et al., 2015) with an initial 87Sr/86Sr ratio of 0.7445±0.0028. Additional model muscovite and biotite dates from the same area are reported as 327–337 Ma and 329±2 Ma respectively (Mortimer et al., 1987), suggesting up to 5 % age variability between samples.

The Högsbo muscovite was sampled from the Högsbo pegmatite quarry in Gothenburg, Sweden. The crystallisation age of the pegmatite, constrained by a columbite 207Pb/206Pb TIMS date of 1029.7±2.0 Ma (Romer and Smeds, 1996), was used as the reference in this study, along with an initial 87Sr/86Sr ratio of 0.72000±0.00010 (Hogmalm et al., 2017). In-situ Rb–Sr dates have been reported as 1037±11 Ma from the Högsbo nano-powder pellet (Hogmalm et al., 2017), and 1035±7 and 1039±7 Ma from muscovite crystals (Rösel and Zack, 2022). More recently, an ⁴⁰Ar–³⁹Ar muscovite date of 1025.2±2.2 Ma, and in-situ Rb–Sr dates between 1023.0±2.4 and 1024±7 Ma, have been reported in Olierook et al. (2026). Significantly younger in situ Rb–Sr dates have also been reported within the same muscovite crystals: 1003±6 (Rösel and Zack, 2022) and 981±8 Ma (Olierook et al., 2026), suggesting some alteration of the pegmatites is recorded in the muscovite.

The G17560 polylithionite was sourced from the South Australian Museum collection from a non-specific location near Itinga, Minas Gerais, Brazil. This would place it within the Araçuaí District in the Eastern Brazilian Pegmatite Province. Similar gem-quality polylithonite specimens have been found in an artisanal mine within the Taquaral pegmatite field. Two different pegmatite suites have been identified in this region with zircon U–Pb dates of nearby granite bodies ranging from 520–480 and 530–500 Ma (G5 and G4 supersuites; Pedrosa-Soares et al., 2011). There are no published Rb–Sr dates for these pegmatites, nor measured 87Sr/86Sr initial ratios. For this study, the initial was assumed to be 0.710±0.015 (assumed for enriched magmatic rocks; Rösel and Zack, 2022). The polylithionite contains very high Rb (∼1.4 wt %) and negligible common Sr, making the Sr component highly radiogenic. Hence, the choice of initial Sr ratio has no significant effect on the calculated Rb–Sr date.

All micas were sub-sampled from large mica crystals and mounted in epoxy blocks perpendicular to cleavage. Mounting in this orientation produces the most consistent spot-to-spot ablation compared to mounting parallel to cleavage (Larson et al., 2025).

The FK-N K-feldspar is used as an international trace element reference material, distributed by CRPG (Govindaraju, 1984), sourced from Tamil Nadu Province, India. Feldspar chips from the original source were obtained and mounted for analysis. The FK-N powdered sample has been dated by Jegal et al. (2022), giving an Rb–Sr date of 512±30 Ma. Co-genetic pegmatites and leucogranites from the region have also been dated giving a Rb–Sr date of 521±9 Ma and initial 87Sr/86Sr ratio of 0.7093 (Pandey et al., 1993).

The Bohus K-feldspars were separated from the Bohus granite, located in south-west Sweden. The granite massif compromises of several granitic intrusions, and has been commercially quarried since the 1840's. The granite intrudes into the Dal Group (1100 to 1050 Ma) at the end of the Sveconorwegian orogeny. It has a Rb–Sr date of 910±35 Ma and an initial Sr ratio of 0.711, and K–Ar model-ages range from 861 to 680 Ma (Skiöld, 2010). A higher precision zircon U–Pb date of 920±5 Ma is regarded as the crystallisation age of the Bohus granite (Eliasson and Schöberg, 1991).

Analyses were conducted at Adelaide Microscopy, Adelaide University, using a RESOlution-LR 193 nm excimer laser system coupled to an Agilent 8900x ICP-MS/MS. The interface tubing between the laser sample chamber and the ICP torch was cleaned prior to each analytical session and was not used for the ablation of high Ca materials (e.g. NIST glasses, 8.15 wt % Ca) to minimise the potential for Ca-contamination. A pure SiO₂ glass was also ablated at the start of each session to mobilise any residual, previously ablated aerosol particles in the tubing (Gilbert et al., 2014). A Teflon bulb (∼20 mL) was used as the signal smoothing device rather than a squid, given the latter is difficult to clean and remain Ca free. Carrier gas flow rates were 350 mL min⁻¹ He, mixed with 890 mL min⁻¹ Ar, and the addition of 6 mL min⁻¹ N₂ to enhance sensitivity. The laser parameters for all sessions were: 3.5 J cm⁻² fluence, 5 Hz repetition rate, and 100 µm spot size. All instrument parameters, including ICP-MS/MS lens tuning, are provided in Table S1.

A range of reaction gas mixtures and flow rates were tested based on the previously published reaction products of Sr and Ca (Hogmalm et al., 2017; Redaa et al., 2021). The aim was to determine the optimal gas flow rates in terms of reaction product ion sensitivity and the signal to background ratio for ^40+XCa (where X represents the mass shift of the product ions, see reaction gas optimisation section below). The four gas combinations selected for detailed testing were: SF₆ only (0.77 mL min⁻¹), SF₆ and H₂ mix (0.57 mL min⁻¹ and 2 mL min⁻¹, respectively), and N₂O and H₂ mixed in two different proportions: 0.16 mL min⁻¹ N₂O and 7 mL min⁻¹ H₂; and 0.16 mL min⁻¹ N₂O and 10 mL min⁻¹ H₂. These setting are referred to as SF₆ only, SF₆ + H₂, N₂O + low H₂ and N₂O + high H₂, from here on. Using N₂O as the reaction gas without H₂ was not tested in detail due to the significant production of ⁴⁰Ar¹⁶O which causes an interference with the target ⁴⁰Ca¹⁶O reaction product. When H₂ is added, Ar is neutralised via a charge transfer reaction (Eiden et al., 1996). In our experiments, no Ar reaction products were detected (measured as ^36+XAr), indicating the charge transfer reaction is thermodynamically favourable in the reaction cell. Both SF₆ and N₂O gases were introduced into the reaction cell using the third mass flow controller (MFC3), N₂O as 99.999 % purity, and SF₆ as a mixture of 10 % SF₆ in He (referred to as SF₆ throughout). The 99.999 % purity H₂ was introduced via a dedicated MFC.

To determine the most abundant rection products for Ca and Sr, product ion scans (PIS) were conducted. The first quadrupole (Q1) was set to a mass-to-charge ratio (m/z) of 44 or 88, respectively, and Q2 scanned across the mass range with mass shifts of 0 to 150 amu. When using SF₆, both with and without the addition of H₂, the predominant reaction products were CaF and SrF (+19 amu mass shift; Fig. 1a). For reactions with N₂O plus H₂, hydroxide (OH) product ions were most abundant (+17 amu mass shift; Fig. 1b). Thus, based on the PIS results, the highest abundant reaction products of Ca¹⁹F, Sr¹⁹F, Ca¹⁷OH and Sr¹⁷OH were used in further analysis. For both Ca and Sr, measurable amounts of hydrides and oxides were also present. For Ca, heavier Ca(H₂O)OH and Ca(H₂O)F were also detected. These H₂O bearing reaction products may have been caused by a contaminant (e.g. air or moisture) in the H₂ gas lines, as these occurred without the deliberate addition of oxygen i.e. with SF₆. Additional gas line flushing is recommended to reduce the production of these molecules.

For SF₆, the PIS also showed the SF_x breakdown products at masses 51, 70, 89, 108 and 127 amu (Fig. 1). These reaction products occur across 3 masses in each experiment, reflecting the isotopic proportions of S. As these product ions are created within the reaction cell from the breakdown of the SF₆ reaction gas itself, their occurrence is independent of the ions that passes through Q1 into the reaction cell. For example, the SF₆ reaction products were also measurable at the same rates when Q1 was set to 3 amu, meaning that no ions were entering the reaction cell. For K-Ca and Rb-Sr isotopic analysis, these products do not affect the measurements; however, they can have implications for the range of trace elements monitored in the same analysis; e.g. ⁵¹V, ⁵³Cr, ⁸⁹Y, and ⁹⁰Zr will all have elevated backgrounds when SF₆ is used as the reaction gas.

The isotopes measured with dwell times (ms) in brackets, were: ²⁷Al (5), ^36+XAr (20), ³⁹K (10), ^39+XK (20), ^40+XCa (50), ^42+XCa (50), ^43+XCa (50),^44+XCa (100), ⁸⁵Rb (20), ^86+XSr (50), and ^87+XSr (50). For the reacted isotopes X represents a mass shift of either 19 (e.g. ⁴⁰Ca¹⁹F) when using SF₆ reaction gas or 17 (e.g. ⁴⁰Ca¹⁷OH) when using the N₂O and H₂ mixture. The presence of unreacted ⁴⁰Ca and ⁸⁷Sr prevents the direct measurement of the parent isotopes, so ³⁹K and ⁸⁵Rb were measured as proxies for ⁴⁰K and ⁸⁷Rb respectively and calculated assuming natural isotopic abundance (Hogmalm et al., 2017; Redaa et al., 2021).

Measurement of ⁴⁰Ca, the major Ca isotope (96.9 % abundance), is complicated by the presence of the major isotope of Ar (99.6 % abundance), as well as the parent ⁴⁰K. Since ⁴⁰Ar⁺ is the most abundant ion produced in the Ar burning ICP, the total number of ions passing through Q1 (set at 40 amu) and entering the reaction cell is very large, greater than the maximum detectable 8×1010 cps. The total ion load is thus likely to significantly exceed the maximum capacity of the reaction cell in terms of ensuring efficient and controlled reactions.

Product ion scans were recorded at m/z 40, both in the gas background (⁴⁰Ar) and while ablating NIST SRM 610 glass (⁴⁰Ar and ⁴⁰Ca). Figure 2 illustrates the additional complexity of measuring ⁴⁰Ca compared to the measurement of ⁴⁴Ca. There are four major observations of note. Firstly, the enhancement of the reaction gas breakdown product ions e.g. 106 cps for SF⁺ when Q1 = 40 (Fig. 2), compared to 2000 cps at Q1 = 44 (Fig. 1a).

Secondly, N₂O product ions, such as N₂O-N₂O⁺ and N₂OH-N₂O⁺, were detected on m/z 62 and 63 but were not detected at other Q1 masses (Fig. 1). This is likely due to the high density of Ar⁺, reacting with and breaking down the reaction gasses. Thirdly, N₂O and SF₆ can attenuate Ar⁺ ions via charge transfer reactions, although less effectively than H₂. The on-mass 40 amu measurement remains significant for reactions without H₂ (c. 106 cps with SF₆ and c. 10⁸ cps with N₂O; Fig. 2); however, this is lower compared to when no reaction gasses are used (above the maximum 8×1010 cps). As predicted, the addition of H₂ to both SF₆ and N₂O reduced the detectable 40 amu to <1000 cps. Fourthly, a high gas background was observed on all Q2 masses when Q1 was set to 40 amu, indicated with the dashed line in Fig. 2. The addition of H₂ to SF₆ had little effect on this background (red and orange lines, Fig. 2), whereas its addition to N₂O significantly reduced the background (light and dark blue lines, Fig. 2). Therefore, the high background is likely to be caused by the presence of Ar⁺ ions, as opposed to the reactive H⁺ ions generated from the Ar-H₂ charge transfer reaction. This background can be significant at the CaF and CaOH reaction product masses, 57 and 59 amu (highest values > 10⁴ cps in the example below), and subsequent testing detailed below aimed to optimise the reaction gas flow rates to minimise this background. Ultimately, sensitivity is a function of signal to background ratio, thus background signals need to be carefully monitored prior to analysis.

The reaction gas flow rates and proportions between the gases (SF₆, N₂O and H₂) were systematically varied to determine their effect on ^40+XCa reaction products. Count rates on m/z=57 (Q1 = 40, Q2 = 57) and 59 (Q1 = 40, Q2 = 59) were measured in the gas background and when ablating NIST610 to produce CaOH and CaF reaction products. The signal to background ratios were compared to determine the optimal flow rates for each reaction gas (Fig. 3).

As the SF₆ flow rate was increased from an initial rate of 20 % (0.38 mL min⁻¹), the sensitivity of CaF increased to a plateau at 30 % for SF₆ + H₂ (0.57 mL min⁻¹ SF₆, 2 mL min⁻¹ H₂) and 40 % for SF₆ only (0.77 mL min⁻¹), while the background decreased slightly or was stable (Fig. 3). These sensitivity plateaus occur at the flow rates that give the maximum reaction efficiency. Higher gas flow rates eventually attenuate the signal due to increased collisions in the cell. While the addition of SF₆ only can resolve the Ar interference on ⁴⁰Ca (no ⁴⁰ArF reaction), the SF₆ + H₂ mixture results in slightly higher sensitivity for ⁴⁰Ca¹⁹F and, more importantly, a significantly lower background (signal to background ratio increases from 3650 to 5770, Fig. 3).

Signal and background data was collected for a range of N₂O flow rates, from 0.16 to 0.32 mL min⁻¹ (15 %–30 %) using 7 mL min⁻¹ H₂ (Fig. 4a). The sensitivity dropped with increasing N₂O flow rate, resulting in a lower signal to background ratio. The same relationship was observed when using higher H₂ flow rates (e.g. 10 mL min⁻¹). Figure 4b presents cps data using 0.16 mL min⁻¹ N₂O (15 %) and varying the H₂ flow from 5–10 mL min⁻¹ (with 10 mL min⁻¹ being the highest flow rate achievable on the mass-flow controller). Signal sensitivity and background both decreased with increasing H₂ gas flow, however, the highest signal to background ratio was achieved using a maximum of 10 mL min⁻¹ H₂. The higher H₂ gas flow rates not only effect ⁴⁰Ca sensitivity, but they also effect the sensitivity of the minor ⁴⁴Ca, where higher sensitivity is beneficial to improve overall uncertainty of the measured ratios. The two conditions of 0.16 mL min⁻¹ N₂O and 7 mL min⁻¹ H₂ (N₂O + low H₂), and 0.16 mL min⁻¹ N₂O and 10 mL min⁻¹ H₂ (N₂O + high H₂), were used for further testing of mineral samples. The 7 mL min⁻¹ H₂ flow rate is a compromise in signal to background ratios, while maintaining higher signal sensitivity.

Isotopic ratios were calibrated using a glass reference material, manufactured at the Institute for Photonics and Advanced Sensing, Adelaide University. The glasses were made from SiO₂-Na₂O-(BaO)-MgO matrix doped with known quantities of K, Ca, Sr and Rb (referred to as SNBM-1 and SNM-1). The raw materials were ball-milled for 1 h to achieve thorough mixing and then loaded into a platinum crucible. The batch materials were melted at 1500 °C for 2 h and cast into a brass mould. On the following day, the sample was annealed at 520 °C for 2 h to relieve internal stress. To improve the homogeneity of the glass, the initial glass was powdered and re-melted at 1500 °C for 2 h, followed by the same casting and annealing steps as before.

The synthetic glasses were used in preference to the more commonly used high-Ca glasses: NIST SRM 610 (8.15 wt % Ca) and BCR-2G (5.05 wt % Ca), to avoid Ca contamination from remobilisation of previously ablated material in the interface tubing (i.e. false signal). The manufactured glasses also enabled customisation of the K, Rb, Sr and Ca concentrations to match the electron multiplier detector modes (pulse or analog counting) between reference materials and samples.

Initial testing used the SNBM-1 glass with a SiO₂-Na₂O-BaO-MgO matrix doped with 2000 ppm K and Rb, and 500 ppm Ca and Sr (data Tables S3, S4, S6 and S8). However, ablating this glass with 100 µm spot size and 5 Hz repetition rate resulted in ^40+XCa being measured in analog mode, while the samples were all in pulse mode. Note that ^44+XCa was always measured in pulse mode because the natural abundance ratio between ⁴⁴Ca and ⁴⁰Ca is c.47:1. Reducing the laser beam size or repetition rate to lower the ^40+XCa signal into pulse mode caused an unfavourable reduction in precision of the ratios for the mineral samples due to lower sensitivity of the minor ^44+XCa isotope. The MDC phlogopite (519±6.5 Ma, 87Sr/86Sr_i 0.72607; Hogmalm et al., 2017; Morteani et al., 2013; Redaa et al., 2021) was used as the mineral-matched reference material to correct for matrix-dependent mass bias on 87Rb/87Sr and 40K/40Ca (Glorie et al., 2024a) via Eqs. (1) and (2), respectively. 187Rb87SrCorrected=87Rb87SrS×eλTM-1eλTR-1240K40CaCorrected=40K40CaS×eλTM-1eλTR-1 Where subscript S denoted the measured isotopic ratio of the sample corrected to SNBM-1 glass, TR is the reference age of MDC and TM is the raw measured date (Ma) calculated from the inverse isochron of the ratios corrected to SNBM-1 glass. Decay constants (λ) of 1.3972×10-5 and 5.5305×10-4 were used for the Rb-Sr and K-Ca systems, respectively (Renne et al., 2011; Villa et al., 2015). The correction to the matrix matched age reference material (MDC) was calculated in Excel.

The ^40+XCa signal was collected in pulse mode in MDC, compared to analog in the SNBM-1 glass. As the ^40+XCa in all other samples were measured in pulse mode, correcting the 40K/40Ca ratios to MDC also incorporated a pulse-analog component to the correction. The pulse and analog detector modes have been observed to drift relative to each other in some Rb-Sr sessions (Glorie et al., 2024a). To assess this, the SNBM-1 glass (analog) and MDC phlogopite (pulse) were both measured every 20–30 analyses throughout each analytical session. After correction to the SNBM-1 glass, no residual time-dependent drift was apparent in the MDC isochrons (see data tables in the Supplement).

The Kola phlogopite was used to calculate a pulse-analog correction for 44Ca/40Ca using the offset between the calculated, unanchored 44Ca/40Ca intercept and the expected ratio of 47.15±0.02 (Russell et al., 1978). Kola was used in preference to MDC for the pulse-analog correction for 44Ca/40Ca as it gave higher variability in measured 44Ca/40Ca ratios between analyses, leading to more robust isochron regression and calculated initial 44Ca/40Ca ratio. For the initial method validation, the pulse-analog correction was robust enough for comparative testing to optimise the reaction gas flows; however, it is not recommended as a routine protocol due to differences in mass-bias drift between pulse and analog modes (Glorie et al., 2024b).

A second glass, SNM-1, was made to address the pulse-analog issues identified in the initial testing. It had a SiO₂-Na₂O-MgO matrix doped with 2000 ppm K, 1000 ppm Rb, 100 ppm Ca and 500 ppm Sr. These concentrations were chosen such that ³⁹K and ⁸⁵Rb were measured in analog detector mode and ^40+XCa and ^87+XSr were measured in pulse mode at 100 µm spot size. This second glass better matched the detector mode in the samples to avoid the need for additional pulse-analog correction and is recommended for all future analyses. Analyses of the micas were repeated with the two most promising reaction gas mixtures (SF₆ + H₂ and N₂O + low H₂, data Tables S5 and S7) using the SNM-1 glass to validate the initial data sets without requiring an additional pulse-analog correction.

The synthetic glasses were used to correct for instrument drift, down-hole fractionation and matrix independent mass-bias using the LADR data processing software (Norris and Danyushevsky, 2018). The reference values used for normalisation of the measured isotopic ratios for both glasses were calculated from the nominal concentrations, assuming natural isotopic abundances and molar proportions.

All uncertainties are reported at the 95 % confidence interval. The individual measurement uncertainties include the excess scatter of the ratios measured in the glass reference material as calculated in LADR. The MDC age correction factor was applied to the measured 87Rb/87Sr and 40K/40Ca ratios, and the calculated MDC age uncertainty (typically ∼2 % for K–Ca and ∼1 % for Rb–Sr) was propagated onto the uncertainty of the sample isochron dates. The long-term reproducibility of Rb–Sr dates at the Adelaide Microscopy isotope facility is ∼1.5 %, which is similar to 1.6 % reported in Olierook et al. (2026), and 1.4 % in Rösel and Zack (2022). This additional systematic uncertainty was propagated onto the final dates: 1.5 % for Rb-Sr and a more conservative 2 % for K–Ca as a realistic estimate, in the absence of long-term data.

Either the SNBM-1 or SNB-1 glass was used as the primary refence material for 86Sr/87Sr ratios without further correction, as isotopic ratios of the same element are not subject to matrix-dependent fractionation (Redaa et al., 2021; Gorojovsky and Alard, 2020) and all Sr isotope measurements were in pulse detector mode.

The low-Ca glasses (SNBM-1 and SNM-1) were used in preference to high-Ca NIST610, to minimise the potential for carry over of Ca between reference glasses and samples. However, contamination was still seen in some ablation signals either from the remobilisation of previously ablated material (predominantly affecting the lower sensitivity isotope ⁴⁴Ca) or surface contamination on the samples (predominantly affecting ⁴⁰Ca, the most abundant common Ca isotope). This was especially apparent in low-Ca micas and for the initial data sets. To remove surface contamination, the samples were repolished and cleaned using distilled water, and 5–10 pre-ablation shots were used to clean the surface prior to sample ablation. During data processing signals were cropped where possible to avoid Ca spikes and surface contamination. Outlier analyses with high ⁴⁴Ca giving a young age bias, were excluded from the isochrons. These contamination issues were most prevalent in the initial data sets (Tables S4 and S6) and minimised in the later after sample cleaning (Tables S5 and S7). Example ablation signals are provided in Supplement Figures.

A relatively long counting time was used for the low abundance ⁴⁴Ca, to reduce the impact of high uncertainty at low count rates (cps): 100 ms, where the minimum detectable 1 count equates to 10 cps. The signals were also cropped during data processing to avoid any zeros in the ⁴⁴Ca signals to ensure accurate uncertainty calculations. Average background subtracted signal sensitivities (cps) are provided in Supplement Data tables.

All data were plotted as inverse isochrons (plotting parent/daughter versus stable/daughter isotope ratios) to minimise uncertainty correlations and improve data visualisation (Li and Vermeesch, 2021), using isoplotR (Vermeesch, 2018). The 87Sr/86Sr initial ratios used for plotting sample isochrons were either obtained from the literature when known (Table 1) or an assumed value based on the recommendations by Rösel and Zack (2022). A 40Ca/44Ca initial value of 47.15±0.02 was used for plotting all K-Ca isochrons, as the range of naturally occurring Ca isotopic ratios can be considered invariant within the uncertainty of the LA-ICP-MS/MS measurements (Russell et al., 1978). This is a significant benefit of the K–Ca method because it allows direct calculation of dates for every analysis without a need to assume initial ratios.

Both Rb/Sr and K/Ca isotope ratio pairs have contrasting element volatilities with Rb and K being the more volatile (Jackson, 2008). Hence, down-hole fractionation (DHF) between the elements during ablation is likely and has been described elsewhere for Rb/Sr (Gorojovsky and Alard, 2020; Redaa et al., 2021; Glorie et al., 2024a). The DHF effects were minimised in this study by using a relatively large laser spot size (100 µm) and 5 Hz repetition rate to reduce the depth to diameter ratio of the crater. Nevertheless, DHF was still observed for both isotope pairs (Fig. 5).

For Rb/Sr there is very little difference in DHF between the materials after the first 10 s of ablation, with the exception of Bohus feldspar which shows a more variable average signal curve during the ablation (Fig. 5a, Table S9; Lloyd et al., 2025). The relative differences between materials are more pronounced for K/Ca ratios, especially during the first 10 s of ablation (Fig. 5b), with up to 10 % DHF, compared to 5 % for Rb/Sr. The SNM glass, MDC, G17560 and Kola micas all have relatively flat K/Ca ratios which is likely to be representative of laser induced DHF. The more variable profiles primarily reflect variations in common Ca rather than differences in ablation behaviour. This was most pronounced for feldspar, where the fewest number of signals were averaged due to naturally heterogeneous common Ca within the ablation volume of the feldspars, producing more variable average K/Ca. The main outlier is Högsbo, for which the steeper curve is attributed to minor common Ca contamination on the surface, which potentially penetrated along cleaves into the subsurface. Despite 7 pre-ablation shots, the common Ca signal was highest at the start of each ablation before decaying to a stable level. When pre-ablation shots were recorded, the Ca signal was significantly higher than during the main ablation, compared to major elements which had similar sensitivity, indicating surface contamination of Ca. For both causes of variable common Ca, the most stable part of the signal was selected to calculate dates, although the whole signals are used to generate the ratio curves in Fig. 5. Typically, the first 10 s of ablation was not included in the signal selection to avoid common Ca contamination and to minimise DHF differences between matrices (see Supplement Figures).

The SNM-01 glass was used to correct for DHF in the other materials. The average curve for SNM-01 is slightly flatter compared to the micas (Fig. 5), which is likely to under-correct the minerals. However, no residual slope was observed on individual analysis ratios for the minerals within the noise of the signals, and any residual bias is assumed to be corrected as a component of the matrix-match mineral age correction (to MDC phlogopite). In the current absence of data processing software which can use isochronous mineral-matched reference materials to correct DHF directly, the suitability of SNM-01 as a DHF correction material is evaluated in this study. Differences in ablation characteristics between minerals (especially between the MDC mineral-matched reference material and K-feldspars) are likely to become more apparent at smaller spot sizes due to increased differences in DHF curves between matrices.

The accuracy of the current technique is limited by the reference K–Ca age of the matrix matched reference material. In this study, we assume MDC has the same K–Ca age as reported for Rb–Sr. This proves to be valid in accurately reproducing the ages of Robins Folly and Högsbo muscovites within 2 %–3 % uncertainty. However, independent validation of the K–Ca age would be required for the technique to develop further and to assess accuracy within 1 %–2 % uncertainty, comparable to the in-situ Rb–Sr method. Recent work by Olierook et al. (2026) has provided new Ar–Ar and Rb–Sr dates for a range of commonly used mica reference materials, and further characterisation of these micas for K–Ca would also be beneficial to the analytical community.

The low concentrations of common Ca, typical in micas, mean the analyses can be easily compromised by Ca contamination. This can occur via remobilisation of previously ablated particulates in the interface tubing, especially after ablation of high Ca glasses such as NIST SRM 610 and BCR-2G, or from surface contamination during sample preparation and polishing. The impact of contamination was minimised by using 5–10 pre-ablation shots prior to the main analysis, and by cleaning the samples with distilled water in an ultra-sonic bath to remove particulates inside past ablation craters and/or crystal defects. The effects of contamination can be seen as spiky signals, especially for the minor ⁴⁴Ca, which leads to a young age bias in contaminated data points (higher 44Ca/40Ca ratios). It may also, in part, explain the poor reproducibility of K–Ca dates between sessions (using the same reaction gas) for the highly radiogenic micas Kola and G17560 (<5 ppm common Ca). Hence, sample cleaning and pre-ablation are paramount for accurate K–Ca dates.

There is also an upper limit of allowable Ca concentrations to resolve the amount of radiogenic ⁴⁰Ca ingrowth from the common Ca. For FK-N feldspar, which is approximately 500 Ma and contains ∼300 ppm total K, the measured ⁴⁰Ca was only 1–5 % radiogenic (44Ca/40Ca = 0.02015–0.02099). Analysis using MC-ICP-MS/MS would improve the uncertainty on the 44Ca/40Ca ratio, enabling improved age precision and/or the ability to date younger Ca-bearing minerals such as feldspars. However, analysing both Rb–Sr and K–Ca simultaneously is currently only possible on a quadrupole ICP-MS/MS. An additional consideration is the pulse-analog transition of the electron multiplier detectors, which in this study was ∼2.7 Mcps, equivalent to ∼300 ppm total Ca.

One of the major advantages of quadrupole ICP-MS/MS is the ability to measure a wide range of isotopic masses in one analysis, enabling simultaneous Rb–Sr, K–Ca and trace element analyses. However, when using SF₆ as the reaction gas, there are some limitations as to the additional trace element isotopes than can be measured due to isobaric interferences from reaction gas breakdown products. Specifically causing high backgrounds of: SF on ⁵¹V, ⁵²Cr and ⁵³Cr, and SF₃ on ⁸⁹Y, ⁹⁰Zr and ⁹¹Zr. As these interferences are created within the reaction cell, they cannot be avoided. Thus, care is needed for isotope selection and interpretation of trace element data when using SF₆as reaction gas.