Introduction

A key event early in Solar System history was the accretion of protoplanets and their subsequent differentiation. This led to the formation of layered bodies composed of a metallic core, overlain by a silicate mantle and capped by a basaltic crust. Understanding the timing of protoplanet accretion and differentiation are fundamental ‘known unknowns’ in our Solar System’s history1. Additional accurate data on the early Solar System chronology from a wider array of objects, sampling different reservoirs, is crucial for refining models and determining the processes (i.e. oligarchic growth or pebble accretion2,3) of rocky planet formation and how the larger planets, including Earth, were formed.

The measurement of nucleosynthetic stable isotope ratios from terrestrial rocks, meteorites and sample return missions reveals a clear divide between the carbonaceous and non-carbonaceous groups. This dichotomy has been interpreted as representing the outer and inner Solar System reservoirs, respectively4,5. Calcium-aluminium-rich inclusions (CAIs) are refractory components found in undifferentiated meteorites formed by direct condensation from the solar nebula6, resulting in their radiometric dates acting as the time zero age for our Solar System. High-precision Pb-Pb dating of CAIs gives a canonical Solar System ‘zero age’ of 4567.3 ± 0.16 Ma7, while recent statistical estimations have revised this age to 4568.36 ± 0.20 Ma8.

182Hf–182W chronology of iron meteorites indicates that protoplanets in the non-carbonaceous (NC) reservoir, interpreted as originating in the inner Solar System, differentiated ∼1–2 Ma after CAIs, while those in the outer Solar System, the carbonaceous chondrite (CC) region, experienced core formation ∼1 Ma later9. Evidence suggests this variance is due to a higher water fraction in the CC iron meteorite parent bodies. This led to a delayed melting and metal segregation onset due to the dilution of 26Al, combined with increases in thermal inertia and effective heat dissipation by porous water convection9. Therefore, the production of fully differentiated bodies in the outer Solar System, with a thin pyroxene–olivine–rich basaltic crust, a thick olivine–rich mantle and an iron–rich core, is expected to have occurred later than in the inner Solar System10. However, evidence of these ancient crusts and mantle materials is scarce, likely due to the higher rate of planetary collisions of the early Solar System and the subsequent incorporation of these ancient protoplanets into the terrestrial planets and the challenges in identifying them as meteorites on Earth.

Evidence of a primordial crust from one of these ancient inner Solar System protoplanets has been recently discovered11,12,13,14. Based on thermal models, Erg Chech 002 (EC 002) formed as a result of partial melting and rapid crystallisation on the surface of a small protoplanet (20–30 km in diameter15), while 26Al-26Mg chronometry indicates the sample formed ∼2 Ma after the formation of the CAIs7,8,11,16. In contrast, to date, the oldest basalts from the CC region, NWA 6704 and NWA 2976, show younger 26Al–26Mg ages of 4563.07 ± 0.24 and 4563.30 ± 0.21 Ma, respectively, ~5 Ma after the formation of the CAIs17,18. This chronological evidence, albeit limited, among the ungrouped basaltic achondrites, supports the idea of earlier protoplanet formation in the NC region.

Here, we report new age data from a recently recovered brecciated olivine–rich ungrouped achondrite (dunite). This sample, Northwest Africa (NWA) 12264, formed on a first–generation differentiated protoplanet in the outer Solar System19. We demonstrate that it is the oldest magmatic rock from the outer Solar System analyzed thus far and provides crucial empirical constraints on the timing of differentiation in the most ancient protoplanets that formed beyond the snowline. It demonstrates simultaneous accretion and differentiation processes operating in the inner and outer Solar System, challenging the long-held paradigm of delayed planet formation beyond Jupiter.

Results

Petrography and microstructures

NWA 12264 is a cataclastic breccia composed of approximately ∼95 vol% olivine (up to 2 cm in diameter) within a sparse olivine–rich matrix, with highly variable chemistry (Fo45.1–89.3)20. While the sample exhibits highly variable Fo contents, the dominant olivine (~85 wt%) has a Fo content of 77.7 ± 0.3, which is similar to other meteoritic dunites from the Moon, 4 Vesta and the angrite parent body21,22,23. In terrestrial terms (based on the olivine content >90%), it would be classified as a dunite—a rock generally found in the Earth’s mantle. While some olivine grains display extensive crystal plastic deformation (>15°), others show evidence of fresh growth and recrystallisation (Fig. 1). The sample also contains minor chromite–pyroxene symplectites, pigeonite (Wo5±2.5) with augite exsolution lamellae (Wo41.2±2.8) and diopside (Wo43.7±0.5). Plagioclase (An86.4±7.8) is crystalline, with no maskelynite present, indicating shock pressures <15 GPa24. Silica polymorphs diffract as tridymite, indicating high temperatures of >870 °C25 (Fig. S1). Phosphates occur as euhedral grains (~50 µm), often associated with other interstitial minerals (pyroxene and chromite). The phosphates show little to no crystal plastic deformation (<5°) and no evidence of recrystallisation (Fig. 2).

Fig. 1: Backscatter electron (BSE), inverse pole figure (IPF) and grain reference orientation deviation (GROD) maps of olivine within NWA 12264.
figure 1

The BSE (A) map depicts the petrological context. The IPF map (B) displays various sub-grains with differing orientations in the olivine, indicative of recrystallisation. The GROD map (C) displays little to no deformation in sub-grains, indicative of fresh growth in olivine.

Full size image
Fig. 2: Backscatter electron (BSE), inverse pole figure (IPF), and grain reference orientation deviation (GROD) maps of phosphates within NWA 12264.
figure 2

The BSE (A, D) map depicts the petrological context. The IPF maps (B, E) display a single orientation and no evidence of recrystallisation. The GROD maps (C, F) display little to no deformation, indicating a lack of shock-induced deformation.

Full size image

Pb-Pb systematics

A total of 10 in-situ Pb–Pb isotopic analyses were carried out on phosphates within NWA 12264 (Fig. 3). Based on regression through Canyon Diablo Troilite (CDT) Pb isotope composition (which assumes that phosphates incorporated primordial Pb akin to that in the protosolar nebula26), NWA 12264 phosphates yield a Pb–Pb isochron date of 4569.8  ±  4.6 Ma, with a mean squared weighted deviation (MSWD) of 0.62. Regression through Stacey & Kramers (S & K) modern terrestrial Pb isotope composition27 (which assumes that common Pb in the analyzed phosphates is terrestrial contamination) yields an identical date of 4569.8  ±  4.6 Ma, with an MSWD of 0.61. Modelled 207Pb/206Pb ages produce a weighted mean of 4571 ± 4.6 Ma. All errors are 2σ. All data are available in Supplementary Data 1.

Fig. 3: Pb/Pb isochron diagram for NWA 12264.
figure 3

CDT and MT correspond to the Pb isotope compositions of Canyon Diablo Troilite (CDT26) and Modern Terrestrial (MT27), respectively.

Full size image

Al–Mg systematics

A total of 38 analyses were carried out, with 24 on feldspar, 10 on pyroxene and 4 on olivine grains. All three minerals define a single coherent 26Al isochron with an initial 26Al/27Al ratio of (1.18 ± 0.25) × 10−6, MSWD = 1.52 (Fig. 4). This ratio translates into a closure age of the 26Al–26Mg system of 4564.44 ± 0.30 Ma (i.e. 3.86 ± 0.22 Ma after CAIs assuming 4568.36 ± 0.2 Ma as the formation age of CAIs8). The calculated initial δ26Mg* (δ26Mg*0 = + 0.18 ± 1) is consistent with the average composition of the pyroxenes and olivines. All errors are 2σ. All data are available in Supplementary Data 1.

Fig. 4: Al–Mg internal isochron defined by feldspars, pyroxenes and olivines of NWA 12264.
figure 4

The error bars correspond to 2σ on the ratios.

Full size image

O-isotope compositions

A total of 19 analyses were performed on olivine grains with differing Fo contents (Fo45-8920). All measured olivine grains demonstrated consistent values with an average oxygen isotopic composition of δ17O = –3.42 ± 0.55 ‰, δ18O = 0.65 ± 0.49 and Δ17O = –3.76 ± 0.40‰ (Fig. 5). All errors are 2σ. All data are available in Supplementary Data 1.

Fig. 5: Oxygen isotope systematics of NWA 12264.
figure 5

Data for South Byron, Milton, NWA 7822 and Eagle Station Pallasite44,45,59. CCAM line, CK, CO and CV data60. PCM line61. Source Data are provided as a Source Data File.

Full size image

Discussion

NWA 12264 exhibits chromium-oxygen (Cr/O) isotopic signatures that unequivocally place it within the carbonaceous reservoir, strongly implying its formation occurred in the outer Solar System19. This means that dates from NWA 12,264 provide essential new information on the timing of planetary evolution beyond the orbit of Jupiter.

The dates obtained in this study are ancient, with the Pb–Pb age derived from phosphates overlapping with CAI formation, while the 26Al–26Mg age is ~4 Ma post-CAI formation. The discrepancy between our two independent age estimates is not entirely unexpected. Past work dating individual chondrules within chondritic meteorites or different minerals within volcanic achondrites produced similar discordances with older Pb-Pb ages28,29. It has been previously argued that a heterogeneous distribution of 26Al in the early Solar System could result in discrepancies in previously measured achondrites30, which may be the reason for the discrepancy in NWA 12264. Alternatively, studies have demonstrated that complex geological histories of extraterrestrial samples, including overprinting of thermal and shock metamorphism, are likely to have affected the time of closure of the two chronometers, undermining their synchronicity31. This is particularly important when evaluating the 26Al–26Mg systematics of plagioclase, which can be easily reset due to the fast self-diffusion of Mg in its crystal lattice32. In the case of NWA 12264, the sample has experienced shock metamorphism and intense brecciation, as evidenced by the reasonably high degrees of misorientation and recrystallisation of olivine. This could imply that our derived 26Al–26Mg age for NWA 12264 is disturbed, implying shock resetting and, potentially, the date for the breakup of the parent protoplanet. Conversely, the lower boundary of the uncertainties of the Pb–Pb age may represent the initial magmatic crystallisation of the sample, thus dating the timing of differentiation. This idea is supported by the lack of recrystallisation and very low degrees of deformation in the phosphates themselves (Fig. 2). However, the closure temperature of Pb in phosphate33 is lower than that of Mg in plagioclase32 and thus, if the shock has reset the Mg in plagioclase, one would assume it would also reset the Pb-Pb age of the phosphate. Although we do note that both Pb and Mg may have behaved differently in the early Solar System compared to idealised experimental lab setups. It is therefore difficult to confidently determine the cause of the age discrepancy between the Pb-Pb age and 26Al–26Mg age for NWA 12,264, however, in either scenario, both the Pb–Pb and the 26Al–26Mg are older than (and outside the uncertainty of) all previously measured outer Solar System basalts, i.e. NWA 6704 and NWA 2976 (4563.07 ± 0.24 and 4563.30 ± 0.21 Ma, respectively17,18]), indicating that the parent body of NWA 12264 accreted, differentiated and broke apart within a rapid timeframe (Fig. 6).

Fig. 6: Absolute chronology of the early Solar System.
figure 6

Chronological events (CAIs8; SBT differentiation9) are defined against achondrites. U-corrected Pb-Pb ages are provided for angrites, due to the potential 26Al heterogeneity in the early Solar System. The error bars correspond to 2σ. Source Data are provided as a Source Data File.

Full size image

Crucially, the ages recorded by NWA 12264 are older than expected, outside the uncertainty of Al-Mg derived ages of the angrites (4563.31 ± 0.21 Ma34), some of the most ancient basalts from the inner Solar System. Although, due to the potential for 26Al heterogeneity in the early Solar System, we also provide a comparison to U-corrected Pb-Pb ages of the angrites35. The ages of NWA 12264 still fall outside of this range, but not by as much as compared to the Al-Mg derived ages. These findings push back the earliest known outer Solar System differentiation age by ~2 million years. These findings support new models36 and challenge the current paradigm, which posits that protoplanets in the outer Solar System accreted more slowly and underwent differentiation later than their inner Solar System counterparts9.

Data from the Atacama Large Millimetre Array (ALMA), particularly from the Disk Substructures at High Angular Resolution Project (DSHARP), demonstrate that many protoplanetary disks have complex structures, with multiple well-defined and narrow dust rings observed at substantial distances from the host star37,38. The presence of narrow, tightly confined rings challenges the smooth disk assumption (a monotonic decrease in dust density as a function of increased heliocentric distance out from the host star) that has historically dominated models of protoplanetary disk evolution39. While some rings have been interpreted as resulting from planet-disk interactions, especially in cases where features like probable mean-motion resonances, ‘double gaps’ or anomalous thermal signatures are identified37,40,41, their origin remains uncertain, with alternative explanations possible42. Thus, pressure bumps and dust gaps likely correspond to planet-forming regions in at least some or most sites and these features are now well documented to be forming across a wide range of heliocentric distances, including beyond the water-ice snowline. These data support the view that planetesimal formation at large heliocentric distances may have been common in the early Solar System and could help explain the apparent synchronicity in the accretion and differentiation of inner and outer Solar System bodies, as suggested by comparison with data from Erg Chech 002.

The oxygen isotope systematics of NWA 12264 (δ18O = 0.65 ± 0.49, Δ17O = –3.76 ± 0.40 ‰; 2σ) overlap with other achondritic meteorites, namely: the ungrouped pallasite Milton (δ18O = –3.1 ± 1.4, Δ17O = –3.6 ± 1.1 ‰ 2σ43) and the South Byron Trio (SBT) a grouplet of iron meteorites (δ18O = –8.4 ± 0.9, Δ17O = –3.7 ± 0.1 ‰ 2σ44) (Fig. 5). Previously, a close link between Milton and the SBT was inferred based not only on their similar O-isotope compositions but also on shared metallographic features, including similar siderophile isotopic (Mo and Ru) compositions, similar depletions of volatile elements and redox-sensitive elements (W, Mo, Fe and P relative to CI chondrite values) as well as characteristic Ni-rich metal compositions and similar cooling rates43. Thus, similarities across a large diversity of independent metrics strongly imply that both meteorites either originated from the same parent body (as chemically distinct melts) or from separate bodies as cosmochemical neighbours. Based on the O-isotope composition of NWA 12264, we suggest that this sample could represent the first silicate-dominated representative of this group and, therefore, expand the range of sampled materials from the core (SBT) and potentially the core-mantle boundary (Milton pallasite) to the mantle (NWA 12264). Further support for the above single-parent body origin can be found in the Pb–Pb age of NWA 12264 (Fig. 3), which overlaps with the metal-silicate differentiation age of the SBT meteorites (2.1 ± 0.8 Ma after CAIs44).

Rapid accretion was the primary factor controlling the size required for differentiation. During efficient accretion scenarios, modelling suggests that bodies larger than 20 km could have retained enough heat from the decay of short-lived radioisotopes to exceed the chondritic solidus45. Nevertheless, the NWA 12264 parent body likely accreted low amounts of water-ice to allow for such an early differentiation time as discussed for the SBT iron meteorites9.

NWA 12264 stands as a single representative of a former outer Solar System planetesimal’s mantle. However, some potentially unrecognised, exciting additional samples exist. NWA 7822, for example, is reported to contain a similar proportion of olivine (>90 vol%) and interstitial pyroxene and plagioclase to that of NWA 12264. However, NWA 7822 displays major differences in chemistry, chromium isotopic compositions and distinct oxygen isotopic compositions (Δ17O = –4.066 ± 0.119‰ 2σ46). This implies that NWA 7822 formed on a distinct parent body from NWA 12264 that also experienced core–mantle differentiation, indicating that at least two distinct bodies in the outer Solar System experienced extensive differentiation, supporting the existing evidence found in iron meteorites from the CC reservoir9. This study indicates that protoplanet-scale differentiation processes in the outer carbonaceous region may be more prevalent than previous evidence suggested.

Materials and methods

Petrography and microstructure

Following established methods47, for petrographic examination, a polished resin block of NWA 12264 was coated with carbon using a Safematic CCU-010 Compact Coating Unit (<5 nm). The block was then investigated using a Zeiss Crossbeam 550 with an Oxford Instruments Symmetry 2 EBSD detector at The Open University. High-resolution energy dispersive X-Ray spectroscopy (EDS) mapping was collected using an Oxford Instruments Ultim Extreme (P, Cl) and an Oxford Instruments Ultim Max detector (all other major elements). The sample was tilted to 70° and an electron beam was used to generate EBSD ‘maps’, consisting of electron backscatter diffraction patterns (EBSPs) acquired at step sizes of 400 nm. The beam conditions used for both EDS and EBSD analyses comprised an incident beam ranging between 1 and 2 nA current and a 20 kV accelerating voltage at a working distance of 12 mm.

Secondary ion mass spectrometry

Phosphates were measured in-situ following established methods23 using a CAMECA IMS1280 large-geometry ion microprobe at the NordSIMS facility, Swedish Museum of Natural History. An Oregon Physics H201 RF Plasma source generated an O2− Gaussian beam of ~2 nA with an impact energy of 23 kV, which was rastered over 5 × 5 μm areas. Before analysis, a ~15 × 15 μm rastered area was pre-sputtered for ~80 s to remove the gold coat and to minimise surface contamination. During analysis, the field aperture limited the field of view on the sample to a 7 μm square, further discriminating against surface and/or grain boundary-hosted contamination. All four Pb isotopes (204Pb, 206Pb, 207Pb and 208Pb) were measured simultaneously using low-noise (<0.003 cps) multichannel ion counting detection at a nominal mass resolution of 4860 (M/ΔM). In the Pb–Pb routine, the magnetic field was locked using an NMR field sensor in regulation mode, and 18 cycles of 20-s integrations were performed. Inter-detector yields in both routines were determined using repeated analyses of the USGS BCR-2G reference glass48. Age calculations were performed using Isoplot-Ex v. 4.1549 and assume the decay constant and 238U/235U ratio (137.794 ± 0.225) recommendations50; in inverse Pb/Pb space, error correlations are negligible and have been ignored. The source of nonradiogenic Pb in these early Solar System samples is difficult to constrain accurately, so regression calculations are presented with two ages. The first assumes the nonradiogenic Pb is terrestrial contamination introduced during both post-fall residence on Earth as well as laboratory processing and represented by model modern terrestrial Pb27; the second assumes an initial Pb component represented by initial solar system Pb CDT26, which is unlikely to have evolved to significantly more radiogenic compositions at the ages determined herein. Arbitrary 1σ uncertainties of 2% and 1%, respectively, have been assigned to these reference values in the calculations.

Nanoscale secondary ion mass spectrometry

The Cameca NanoSIMS 50L at the Open University (Milton Keynes, UK) was used to measure the Mg isotopes of plagioclase, pyroxene and olivine and the O isotopes of olivine in the studied resin block of NWA 12264, following established methods51.

Mg isotopic analyses were carried out using an O− ion beam from the Duo Plasmatron source. Before analysis, a probe current of 16 kV and 500 pA was used to pre-sputter each area (8 × 8 μm) for 3 min to remove the carbon coat and surface contamination and to achieve sputter equilibrium. The sizes of areas were reduced to 4 × 4 μm for analyses.

For plagioclase (27Al/24Mg > 300), a probe current of 500 pA was used for analyses with an analysis time of 20 min. Secondary ions of 24Mg+, 25Mg+, 26Mg+ and 44Ca+ were collected using electron multipliers, and 27Al+ was measured using a Faraday cup to accommodate for the high 27Al+ counts. For pyroxene and olivine (27Al/24Mg < 0.05), a 5 pA probe current was used for analyses with an analysis time of 4 min. All five secondary ion species of 24Mg+, 25Mg+, 26Mg+, 27Al+ and 40Ca+ were collected simultaneously using electron multipliers. The mass resolving power was set to over 12,000 (Cameca definition52) for all Mg isotopic analyses. 27Al/24Mg ratios were corrected using sensitivity factors defined for each measurement session using the Al/Mg measured by EPMA on the standards.

Miyake Jima anorthite (Al/Mg = 396.3) was used as the standard for plagioclase, while San Carlos olivine and Shallow Water enstatite were used as the standards for olivine and pyroxene, respectively. The Mg isotopic composition of the standards is assumed to be identical to the terrestrial value52. The typical total counts for plagioclase were 1.7 × 107 for 24Mg, 2.2 × 106 for 25Mg, 2.4 × 106 for 26Mg, 7.7 × 109 for 27Al and 1.6 × 108 for 44Ca. For olivine, the total counts were 1.7 × 107 for 24Mg, 2.2 × 106 for 25Mg, 2.4 × 106 for 26Mg, 1.3 × 104 for 27Al and 7.9 × 104 for 40Ca and for pyroxene the total counts were 1.3 × 107 for 24Mg, 1.7 × 106 for 25Mg, 1.8 × 106 for 26Mg, 3.9 × 105 for 27Al and 3.0 × 105 for 40Ca.

A Cs+ ion beam was used to measure the O isotopes of olivine. Before analysis, each area was pre-sputtered with a 16 kV and 100 pA Cs+ probe for 4 min over an area of 7 × 7 mm to remove the carbon coat and surface contamination and to achieve sputter equilibrium. Analyses were performed with a 100 pA Cs+ probe rastered over 5 × 5 mm in ‘spot’ mode, with each scan of the area lasting 0.54 s. An electron flood gun was used for charge compensation. Six different secondary ion species were collected simultaneously, with 16O− measured on a Faraday cup while 17O−, 18O−, 30Si−, 26Mg16O− and 56Fe16O− were measured on electron multipliers. A mass resolving power of ~11,000 (Cameca definition52) was used, which is sufficient to resolve the 16OH− interference from 17O−. The 16OH− signal was measured for 10 s at the start and end of each measurement. Each analysis lasted approximately 9 min, providing a total of ~1.5 × 1010 counts for 16O−. Analyses were corrected for instrumental mass fractionation against a standard sample of San Carlos olivine (Fo90), analyzed before and after each block of unknown samples. Analytical uncertainty (all 2σ), incorporating internal precision and external precision from standard replicates, is typically ± 1.2‰ for δ17O, ± 0.7‰ for δ18O, and ~±1.2‰ for Δ17O. San Carlos olivine (Fo90), olivine from the Eagle Station pallasite (Fo80), and an olivine with a composition of Fo72 were used for matrix correction, and offsets of 0.6–2.9 per-mil between the sample and San Carlos olivine were applied to account for differences in the Fe/Mg of the samples of olivine to the standard.

All positions of raster pits for both Mg and O isotopes were checked and verified using the SEM following analyses (Figs. S2–S5). Spots that sampled a mix of phases, as well as spots affected by holes/cracks, were rejected.

Establishing age from Al-Mg systematics

Isotopic ratios measured by NanoSIMS are reported relative to the DSM-3 standard in delta notation as Eqs. 1 and 2, with \({\left(\frac{{25\atop}\text{Mg}}{{24\atop}\text{Mg}}\right)}_{\text{DSM}-3}=0.12663\pm 0.0013\) and \({\left(\frac{{26\atop}\text{Mg}}{{24\atop}\text{Mg}}\right)}_{\text{DSM}-3}=0.13932\pm 0.0026\)53.

$${\delta }^{25}\text{Mg}=\,\left[\frac{{\left(\frac{{25\atop}\text{Mg}}{{24\atop}\text{Mg}}\right)}_{\text{sample}}}{{\left(\frac{{25\atop}\text{Mg}}{{24\atop}\text{Mg}}\right)}_{\text{DSM}-3}}-1\right]\times 1000$$
(1)
$${\delta }^{26}\text{Mg}=\,\left[\frac{{\left(\frac{{26\atop}\text{Mg}}{{24\atop}\text{Mg}}\right)}_{\text{sample}}}{{\left(\frac{{26\atop}\text{Mg}}{{24\atop}\text{Mg}}\right)}_{\text{DSM}-3}}-1\right]\times 1000$$
(2)

The radiogenic excess of \({26\atop}\text{Mg}\) due to the decay of \({26\atop}\text{Al}\) (\({\delta }^{26}\text{M}{\text{g}}^{* }\)) is calculated using a mass fractionation factor of \(\beta =0.5128\) 54 as

$${{\delta }^{26}{\rm{Mg}}}^{*}=\left[\left(1+\frac{{\delta }^{26}{\rm{Mg}}}{1000}\right)-{\left(1+\frac{{\delta }^{25}{\rm{Mg}}}{1000}\right)}^{\frac{1}{{\rm{\beta }}}}\right]\times 1000,$$
(3)

using the mass fractionation law used in ref. 55. The choices of mass fractionation law across the literature are variable. However, we find that the choice of fractionation law and fractionation coefficient used in this work only results in minor changes to the derived parameters below the uncertainties, and the shifts are mainly in the \({\left({\delta }^{26}\text{M}{\text{g}}_{\text{corr}}^{* }\right)}_{0}\) value derived from the fit. Therefore, the relative age derived in this work for NWA 12264 is not particularly sensitive to this choice. This is calculated for both the sample and the standards measured in the same session. The radiogenic excess is then corrected for instrumental bias using the bracketing standards measured in the same session as NWA 12264 as

$${\delta }^{26}\text{M}{\text{g}}_{\text{corr}}^{* }=\left[\frac{\left(1+\frac{{\delta }^{26}{\text{Mg}}_{\text{sample}}^{* }}{1000}\right)}{\left(1+\frac{{\delta }^{26}{\text{Mg}}_{\text{standard}}^{* }}{1000}\right)}-1\right]\times 1000.$$
(4)

The \({\delta }^{26}\text{M}{\text{g}}_{\text{corr}}^{* }\) values are then plotted against instrumental bias-corrected \({27\atop}\text{Al}/{24\atop}\text{Mg}\) ratios. A linear trend is fit to the data using orthogonal distance regression implemented in the scipy Python package56 to enable fitting accounting for uncertainties in both x and y values. Using the following relationship:

$${\delta }^{26}\text{M}{\text{g}}_{\text{corr}}^{* }={\left({\delta }^{26}\text{M}{\text{g}}_{\text{corr}}^{* }\right)}_{0}+\,\frac{{\left(\frac{{26\atop}\text{Al}}{{}^{27}\text{Al}}\right)}_{0}\times 1000}{{\left(\frac{{}^{26}\text{Mg}}{{24\atop}\text{Mg}}\right)}_{\text{DSM}-3}}\times \frac{{27\atop}\text{Al}}{{24\atop}\text{Mg}}$$
(5)

(caption of Fig. 1 in ref. 57), the intercept of the line corresponds to the initial radiogenic excess \({\left({\delta }^{26}\text{M}{\text{g}}_{\text{corr}}^{* }\right)}_{0}\) and the slope of the line, \(m\), can be used to obtain the initial aluminium isotopic ratio \({\left(\frac{{26\atop}\text{Al}}{{27\atop}\text{Al}}\right)}_{0} = \frac{m}{1000}\times {\left(\frac{{26\atop}\text{Mg}}{{24\atop}\text{Mg}}\right)}_{\text{DSM}-3}\). Using the CAIs as an anchor, we can derive the relative age of the sample from its \({\left(\frac{{26\atop}\text{Al}}{{27\atop}\text{Al}}\right)}_{0}\) value55,

$$\Delta t\,\left(\text{Ma}\right)=\mathrm{ln}\left[\frac{{\left(\frac{{}^{26}\text{Al}}{{27\atop}\text{Al}}\right)}_{0,\text{CAIs}}}{{\left(\frac{{26\atop}\text{Al}}{{27\atop}\text{Al}}\right)}_{0,\text{sample}}}\right]\times \frac{{t}_{\frac{1}{2}}}{\mathrm{ln}\left(2\right)},$$
(6)

where \({\left(\frac{{26\atop}\text{Al}}{{27\atop}\text{Al}}\right)}_{0,\text{CAIs}}=\left(5.23\pm 0.13\right)\times {10}^{-5}\) 28, \({t}_{1/2}=0.717\,\text{Ma}\)58. We then calculate the absolute age of the sample as \({t}_{\text{sample}}(\text{Ma})={t}_{\text{CAIs}}-\Delta t\) using a CAI age of \({t}_{\text{CAIs}}=\,4568.36\pm 0.2\,\text{Ma}\)8. All ages compared to NWA 12264 were recomputed from the \({\left(\frac{{26\atop}\text{Al}}{{27\atop}\text{Al}}\right)}_{0}\) values presented in the literature using the same parameters as stated above for NWA 12264, to enable consistent comparisons across the various samples.