Early aqueous activity on the ordinary and carbonaceous chondrite parent bodies recorded by fayalite

Chronology of aqueous activity on chondrite parent bodies constrains their accretion times and thermal histories. Radiometric 53Mn–53Cr dating has been successfully applied to aqueously formed carbonates in CM carbonaceous chondrites. Owing to the absence of carbonates in ordinary (H, L and LL), and CV and CO carbonaceous chondrites, and the lack of proper standards, there are no reliable ages of aqueous activity on their parent bodies. Here we report the first 53Mn–53Cr ages of aqueously formed fayalite in the L3 chondrite Elephant Moraine 90161 as Myr after calcium–aluminium-rich inclusions (CAIs), the oldest Solar System solids. In addition, measurements using our synthesized fayalite standard show that fayalite in the CV3 chondrite Asuka 881317 and CO3-like chondrite MacAlpine Hills 88107 formed and Myr after CAIs, respectively. Thermal modelling, combined with the inferred conditions (temperature and water/rock ratio) and 53Mn–53Cr ages of aqueous alteration, suggests accretion of the L, CV and CO parent bodies ∼1.8−2.5 Myr after CAIs. The parent bodies of many chondritic meteorites experienced aqueous alteration, the chronology of which helps constrain their histories. Here, the authors synthesize a fayalite standard and report reliable ages of secondary fayalite, from which model accretion ages are determined and the place of accretion is inferred.

M ost chondritic meteorites (chondrites) experienced aqueous alteration resulting in the formation of a diverse suite of secondary minerals, including phyllosilicates, magnetite (FeFe 2 O 4 ), sulfides, carbonates (calcite (CaCO 3 ), dolomite (CaMg(CO 3 ) 2 ), breunnerite (Mg,Fe,Mn)CO 3 ) and siderite (FeCO 3 )), ferromagnesian olivines ((Fe,Mg) 2 SiO 4 ) and Ca-rich pyroxenes (Ca(Fe,Mg)Si 2 O 6 ; ref. 1). Mineralogical observations, isotopic data and thermodynamic analysis suggest that the alteration resulted from interactions between a rock and an aqueous solution in an asteroidal setting 2 . Therefore, dating minerals formed by aqueous alteration provides important constraints on the accretion ages of chondrite parent bodies. The chondrite accretion ages, the conditions of aqueous alteration (temperature and water/rock ratio), and the inferred hydrogen and oxygen isotopic compositions of chondritic water 3,4 can potentially be used to constrain chondrite accretion regions and to test the recently proposed Grand Tack dynamical model of the early Solar System evolution 5 . According to this model, the hydrated P-type, D-type and taxonomic C-complex asteroids, which are commonly associated with carbonaceous chondrites 6 , accreted outside Jupiter's orbit (5-15 AU from the Sun; 1 AU is the distance between the Sun and the Earth), and were scattered and implanted into the main asteroid belt (located between 2 and 4 AU from the Sun) during the migration of Jupiter and Saturn within the first several million years of the Solar System formation.
The short-lived radionuclide 53 Mn, which decays to 53 Cr with a half-life of B3.7 Myr and appears to have been uniformly distributed in the protoplanetary disk 7 , is known to be a useful tool for dating aqueous alteration 1 . As yet, accurate 53 Mn- 53 Cr ages of aqueous alteration are only known for the CM (Migheilike) carbonaceous chondrites 8 containing aqueously formed Mnrich, Cr-poor calcite suitable for in situ radiometric 53 Mn- 53 Cr dating with secondary ion mass spectrometry (SIMS). SIMS measurements of manganese-chromium (Mn-Cr) isotope systematics of a mineral require a proper standard to determine a relative sensitivity factor (RSF ¼ ( 55 Mn þ / 52 Cr þ ) SIMS /( 55 Mn/ 52 Cr) mineral ) to correct for the relative sensitivities between 55 Mn þ and 52 Cr þ ions and to calculate true 55 Mn/ 52 Cr ratios in the mineral. The initial 53 Mn/ 55 Mn ratio, ( 53 Mn/ 55 Mn) 0 , of calcites in CM chondrites has been determined by using an RSF measured on a synthetic Mn-and Cr-doped calcite standard 8 . The inferred initial 53 Mn/ 55 Mn ratios, ranging from B2.7 Â 10 À 6 to B3.4 Â 10 À 6 , correspond to calcite formation ages of B4 À 5 Myr after the formation of calcium-aluminiumrich inclusions (CAIs), the oldest Solar System solids dated 9 . Carbonates, however, are virtually absent in weakly aqueously altered ordinary (H, L and LL) and several carbonaceous chondrite groups, including CV (Vigarano-like) and CO (Ornans-like). Instead, the least metamorphosed meteorites of these groups contain aqueously formed Mn-rich, Cr-poor fayalite (Fa 490 ; Fayalite (Fa) number ¼ atomic Fe/(Fe þ Mg) Â 100 in olivine) suitable for in situ Mn-Cr isotope dating with SIMS [10][11][12] . As for carbonates, SIMS measurements of 53 Mn- 53 Cr systematics in fayalite require a proper standard to determine the RSF and calculate 55 Mn/ 52 Cr ratios. Because natural fayalite contains virtually no chromium, it cannot be used as a standard for Mn-Cr isotope measurements; San Carlos olivine (Fa 10 ) was typically used instead [10][11][12] . However, it has been recently discovered that the RSF in olivine changes as a function of its fayalite content 13,14 . Therefore, all previously published 53 Mn-53 Cr ages of CV and CO chondritic fayalite, acquired with a San Carlos olivine standard [10][11][12] , need to be corrected.
To obtain accurate 53 Mn- 53 Cr ages of fayalite, a Mn-and Cr-doped fayalite (Fa 99 ) standard was synthesized (ref. 14 and Doyle et al., manuscript in preparation). Here we report on the mineralogy, petrography and oxygen-isotope compositions of the fayalite-bearing assemblages and 53 Mn-53 Cr ages of fayalite in type 3 L, LL, CO and CV chondrites. Our data confirm the origin of fayalite during low-temperature aqueous alteration on the L, CV and CO chondrite parent bodies [10][11][12] , and allow us to constrain the accretion ages and the accretion regions of these bodies.
Oxygen isotopic compositions of fayalite and magnetite in A-881317 (CV3) plot along a mass-dependent fractionation line with a slope of B0.5 (Fig. 2a) and have an average D 17 O value of À 0.4 ± 0.9% (Fig. 2b), consistent with the previously published data for fayalite and magnetite from the Kaba and Mokoia CV3 chondrites 17 . The bulk oxygen isotopic compositions of CV chondrites 18 and CV chondrule olivines 19 plot along the carbonaceous chondrite anhydrous mineral (CCAM) line with a slope of B1 (Fig. 2a) and show a range of D 17 O values from À 6 to À 2% (Fig. 2b).
Oxygen isotopic compositions of fayalite and magnetite in MAC 88107 (CO3-like) plot along a mass-dependent fractionation line with a slope of B0.5 (Fig. 2a) and have an average D 17 O value of À 1.6 ± 0.9% (Fig. 2b), whereas chondrule olivines plot along the CCAM line and show a range of D 17 O values from À 9 to À 2% (Fig. 2b). The bulk oxygen isotopic composition of MAC 88107 also plots along the CCAM line and has a D 17 O value of À 4.8% (ref. 18). 53 Mn-53 Cr ages of fayalite formation. To obtain an olivine standard suitable for SIMS measurements of Mn-Cr isotope systematics of chondritic fayalite, we synthesized a suite of ferromagnesian olivines ranging from Fa 31 to Fa 99 (ref. 14; see Methods section and Supplementary Table 3). Our Mn-Cr isotope measurements of the synthesized olivine grains with the University of Hawai'i (UH) ims-1280 SIMS revealed that the RSF strongly depends on the fayalite content in olivine 14 . In olivine with Fa 10-30 , the RSF increases from B0.9 to B1.5 as a function of fayalite content; in more ferroan olivines (Fa 430 ), the RSF is nearly constant, B1.6 ( Fig. 3 and Supplementary Table 4). The observed differences in RSF values between fayalite (Fa 90-100 ) and San Carlos olivine (Fa 10 ) suggest that all previously reported 53 Mn- 53 Cr dating of fayalite in carbonaceous chondrites [10][11][12] are in error and yield systematically young ages 14 .
The 53 Mn- 53 Cr chronometer provides a relative chronology, that is, the 53 Mn-53 Cr age of mineral formation is given relative to a reference material: Dt mineral-reference ¼ 1/l 53 55 Mn ratio in CV CAIs has, however, not been measured owing to a lack of primary CAI minerals having high Mn/Cr ratios and the post-crystallization disturbance of their Mn-Cr isotope systematics 20 . Instead, the initial 53 Mn/ 55 Mn ratio in CV CAIs can be calculated from the difference in the U-corrected Pb-Pb ages of CAIs and the U-corrected Pb-Pb ages angrites (basaltic meteorites) with the measured initial 53   Thermal modelling of L, CO and CV chondrite parent bodies. Mineral thermometry suggests that L, CV and CO chondrite parent bodies reached peak metamorphic temperatures of B950, B600 and B600°C, respectively 23 . Thermal evolution of the L-, CV-and CO-like bodies with different radii heated only by decay of the short-lived radionuclide 26 Al, having a half-life of B0.7 Myr, is modelled with reference to the peak metamorphic temperatures and the ages of fayalite formation in the L, CV and CO chondrites (see Methods section).
Our calculations indicate that L chondrite-like bodies with radii of 20 À 40 km need to accrete within B1.6-1.8 Myr after CV CAIs, respectively, in order to reach a peak metamorphic temperature of 950°C (Fig. 6a). The CV and CO chondrite-like bodies with radii of 20-50 km need to accrete within B2.4-2.6 Myr and B2.1-2.4 Myr after CV CAIs, respectively, to reach a peak metamorphic temperature of 600°C (Fig. 6b). The CO chondrites have lower bulk aluminium contents than the CV chondrites (Supplementary Table 6). As a result, the CO chondrite parent body heated by decay of 26 Al must have accreted earlier than the CV chondrite parent body in order to reach the same peak metamorphic temperature. We note that owing to the balance between the heat generated by decay of 26 Al and heat lost through radiation, the maximum temperature reached by a body larger than 30 km in radius largely depends on its accretion time and not its size (Fig. 6a,b).
Temperature evolution at different depths of the L, CV and CO chondrite-like bodies with different sizes and accretion ages are shown in Fig. 6c-f. Regions B26.5 and B36.5 km from the centre of L chondrite-like bodies with radii of 30 and 40 km, respectively, would have conditions (100 À 200°C) suitable for the formation of fayalite (Fig. 6c,e). In a body 30 km in radius, the fayalite-forming region would not have been heated above 300°C, and would have avoided Fe-Mg diffusion between fayalite and ferromagnesian olivines, consistent with our observations of fayalites in EET 90161. In a body 40 km in radius, the fayaliteforming region would have reached up to 350°C B0.5 Myr after fayalite formation, which would have resulted in some Fe-Mg interdiffusion in fayalite. The CV and CO chondrite-like bodies, each with a 50-km radius, have regions B43 À 47 km from the centre of the bodies, which experienced metamorphic temperatures in the range of 100 À 300°C (Fig. 6d,f). These calculations indicate that secondary fayalite is likely to have formed and survived only near the peripheral parts of the L, CV and CO chondrite parent bodies.

Discussion
The oxygen-isotope systematics of fayalite and magnetite in type 3 ordinary and carbonaceous chondrites indicate that these minerals are in isotopic disequilibrium with the chondrule olivines, precluding a high-temperature origin of fayalite and magnetite during chondrule formation 24    are consistent with a low-temperature origin of these minerals in equilibrium with an aqueous solution 25,26 . The mineralogy, petrography and oxygen isotopic compositions, therefore, support the formation of fayalite-bearing assemblages in type 3 ordinary, CO and CV chondrites during fluid-rock interactions on their respective parent bodies 1,10-12 . Thermodynamic analysis of the gas-solution-rock system 27,28 shows that nearly pure fayalite (Fa 490 ) is stable at low temperatures (B100 À 200°C) and low water/rock mass ratios (B0.1 À 0.2), and, therefore, these conditions are inferred for aqueous alteration on the ordinary, CV and CO chondrite parent bodies. These alteration conditions are different from those recorded by the more extensively aqueously altered but less metamorphosed CM and CI (Ivuna-like) carbonaceous chondrites, which have experienced alteration at temperatures of B20-80°C (refs 18,29) and o150-210°C (refs 18,30), and water/rock mass ratios of B0.3 À 0.6 and 40.8 (ref. 18), respectively. The CI and CM chondrites contain abundant phyllosilicates, carbonates and magnetite, but lack fayalite, which is unstable under these alteration conditions 28 .
The formation ages of fayalite in L, CO and CV chondrites and the conditions of fayalite stability 28 can be linked to the metamorphic histories and the accretion ages of their parent bodies (Fig. 6). The L chondrites define a metamorphic sequence of petrologic types from 3 to 6, and reached peak metamorphic temperatures of B950°C (ref. 23). In contrast, the CV and CO chondrites define a narrower range of petrologic types, 3.1 to 43.6 and 3.0 to 3.8, respectively, and reached peak metamorphic temperatures of B600°C (refs 23,31,32). As it is not known how well the CV and CO chondrite parent bodies are sampled by the recovered meteorites, the estimated peak metamorphic temperatures provide only a lower limit on the maximum temperatures experienced by these bodies.
It is generally accepted that 26 Al was the major heat source responsible for thermal metamorphism and aqueous alteration of early accreted planetesimilas 33 . Assuming 26 Al was homogeneously distributed in the protoplanetary disk with an initial abundance corresponding to the canonical 26 Table 6). Numerical modelling of an L chondrite-like body with a radius of B30-40 km (Fig. 6c,e) suggests that this body accreted with an initial 26 Al/ 27 Al ratio of B9 Â 10 À 6 , corresponding to B1.8 Myr after CV CAIs. Fayalite could have precipitated B0.6 Myr later in the outer portion of this body, which may never have experienced temperatures above 300°C (Fig. 6c,e). Similarly, the CV and CO chondrite-like bodies with radii of 50 km (Fig. 6d,f) accreted with an 26 Al/ 27 Al ratio of B4 Â 10 À 6 corresponding to B2.5 Myr after CV CAIs. Fayalite could have precipitated B1.5-2.5 Myr later in the outer portions of these bodies, which may never have experienced temperatures above 300°C. The modelled accretion age of the CV chondrite parent body is in agreement with the 26 Al- 26  meteorites 41 and the lack of variations in bulk chemical compositions between meteorites with different degrees of aqueous alteration 1,2,42 imply that the alteration occurred in a chemically closed system, with the fluid flow being restricted to 10-100 mm, which is consistent with rare occurrences of short veins in aqueously altered chondrites (Fig. 1d). We suggest that the inferred low abundance of water ices in ordinary, CO and CV chondrite parent bodies resulted from their accretion close to the snow line, possibly slightly inside it, where only the relatively large ice-bearing particles could have avoided instantaneous evaporation and survived to be accreted to the parent body 43 . The location of the snow line in the protoplanetary disk is uncertain; it likely did not reside at a single location, but rather migrated with time as the luminosity of the proto-Sun, the mass transport rate through the disk, and the disk opacity all evolved with time. In viscous disks, where mass accretion occurs throughout the disk, energy from internal dissipation can keep the inner disk warm for much of the lifetime of the disk. As mass accretion rates diminished with time, the viscous dissipation would slow, causing the snow line to migrate inwards with time. While specific details vary with the assumed disk structure and viscosity, models [44][45][46][47] suggest that the snow line would be located beyond 5 AU early in disk evolution, but is likely to have been present at 2-3 AU in the 1.8-2.5-Myr-old disk, which corresponds to the accretion ages of ordinary, CV and CO chondrite parent bodies. Thus, this would suggest that the ordinary, CO and CV chondrite parent bodies most likely accreted in the inner part of the Solar System, close to the current position of the main asteroid belt, rather than experiencing large radial excursions that may have occurred during planet migration 5 . This conclusion is inconsistent with the Grand Tack dynamical model of the Solar System evolution that predicts implantation of hydrated carbonaceous chondritelike planetesimals, formed beyond Jupiter, into the main asteroid belt 5,48 .

Methods
Synthesis experiments. To measure the Mn-Cr RSF in fayalite, ferromagnesian olivines covering a wide compositional range (Fa 31-99 ) were synthesized, as described by Doyle et al. (manuscript in preparation). The methodology is summarized as follows: pre-dried Fe 2 O 3 , MgO, SiO 2 , MnCO 3 , Cr 2 O 3 and NiO were mixed in stoichiometric proportions and ground by hand under ethanol. The respective powders were made into a slurry with a polyvinyl alcohol solution and attached to platinum loops. The loops were suspended from a platinum chandelier within a 1-atm vertical gas mixing furnace in which the temperature was monitored using an S-type thermocouple. Mixtures of H 2 and CO 2 were used to control oxygen fugacity (fO 2 ), which was monitored using a SIRO2 C700 þ solid electrolyte oxygen sensor. The samples were at dwell temperatures ranging from 1,200 to 1,500°C for up to 19 h (Supplementary Table 3), after which they were quenched into water. Oxygen-isotope systematics. Oxygen isotopic compositions of fayalite, magnetite and chondrule olivines were measured in situ by SIMS with the UH Cameca ims-1280 ion microprobe using two measurement protocols, depending on grain size.
For grains 410 mm, an B0.8-1.2 nA Cs þ primary ion beam was focused to a diameter of B5 mm and rastered over a B10 Â 10 mm 2 area for pre-sputtering (120 s). After pre-sputtering, the raster size was reduced to B7 Â 7 mm 2 for automated centring of the secondary ion beam followed by data collection. An energy window of B40 eV was used. A normal-incident electron flood gun was used for charge compensation with homogeneous electron density over a region B70 mm in diameter. Three oxygen isotopes ( 16 O À , 18 O À and 17 O À ) were measured in multicollection mode using multicollection Faraday cups and an axial electron multiplier (EM). The mass-resolving power for 17 O À and for 16 O À and 18 O À were set to B5,500 and B2,000, respectively. 16 OH À signal was monitored in every measured spot and was typically o10 6 counts per second, compared with typical 17 O À count rates of 2 Â 10 5 counts per second. Contribution of 16 OH À onto 17 O À was corrected based on a peak/tail ratio. The correction was typically o0.1% (B0.5% at most). Instrumental mass fractionation effects for fayalite, magnetite and chondrule olivine were corrected by analysing terrestrial fayalite, terrestrial magnetite and San Carlos olivine standards, respectively. The standards were analysed repeatedly before and after each run. Reported errors (2s) include both the internal measurement precision and the external reproducibility (B0.5-1.4% (2s) in both d 17 O and d 18 O) of standard data obtained during a given session.
For grains o10 mm in size, an B20-30 pA primary Cs þ beam was focused to B1-2 mm. The three oxygen isotopes, 16 O À , 17 O À and 18 O À , were measured in multicollection mode using a multicollection Faraday cup, an axial EM and a multicollection EM, respectively. The internal and external reproducibility on the multiple analyses of the standards was B1-3% (2s) for both d 17 O and d 18 O.
Small grains (o10 mm) are generally very difficult to identify under the optical microscope of the ims-1280 ion microprobe. To measure such small grains, we used the UH JEOL-5900LV scanning electron microscope to mark the regions of interest. This is done by focusing the electron beam, centring the grain of interest in the field of view and then increasing the magnification so that the electron beam is effectively a spot on the centre of the grain. The electron beam removes adsorbed water on the carbon coating in such a way that it appears as a dark spot in 16 O À scanning ion image in the ion probe. Scanning 16 O À images are obtained by rastering the B20 pA ion beam over an B10 Â 10 to 50 Â 50 mm 2 square. After the mark is found using 16 O À scanning ion imaging, we centre the point of interest under the ion beam and then turn off the primary beam to prepare for measurement. The sputter rate of the rastered beam is sufficiently low that it does not completely remove the sample's conductive coating during ion imaging. A conductive pathway therefore remains available for the duration of the oxygenisotope measurements.
After analysis, the location of each probe spot was re-imaged by electron microscopy to check for beam overlap between phases and to identify large cracks or impurities that may have affected the result.
Mn-Cr isotope systematic. Mn-Cr isotope data were collected in situ with the UH Cameca ims-1280 ion microprobe using two measurement protocols, depending on grain sizes.
A 16 O À primary beam with a 100 pA current was focused into a spot B5 mm in diameter and used to collect data from terrestrial and meteoritic olivines (Fa 10-34 ), liquidus-phase synthetic olivines (Fa  ) and fayalites (Fa 95-100 ) from the carbonaceous chondrites A-881317 and MAC 88107. Owing to the small size of the fayalite grains (r4-5 mm) in the OC EET 90161, their Mn-Cr isotope data were collected using a 16 O À primary beam with a 65-75 pA current focused into a spot B3 mm in diameter.
The positive secondary ions were accelerated with 10 kV. A 50 eV energy window was used. Three chromium ions, 50 Cr þ , 52 Cr þ and 53 Cr þ , were measured simultaneously using two multicollection EMs and an axial EM, respectively. Subsequently 55 Mn þ was measured on the axial EM by peakjumping. The mass-resolving power was set to B4,500 for 50 Cr þ and 52 Cr þ , and B6,000 for 53 Cr þ and 55 Mn þ . These settings were sufficient to separate the 50 Cr þ , 52 Cr þ , 53 Cr þ and 55 Mn þ ions from interfering species, including 52 CrH þ . Isotopic data were typically collected over 125 cycles (EET 90161) and 100 cycles (A-881317 and MAC 88107). The chromium count rates collected during the first 25 cycles were often higher than during the remaining cycles, so the first 25 cycles were removed due to possible contamination. Corrections were made for both the EM background noises and dead times.
Chromium-isotope and 55 Mn/ 52 Cr ratios were calculated from the total number of counts to suppress systematic bias caused by low count rates of chromium 50 53). As the ion yields changed with time, a time-averaged RSF was used. The measurement durations of the bracketing Fa 99 standards were matched to that of the unknown if (on the rare occasion) the meteoritic fayalite analysis was shortened (for example, owing to the base of the grain being breached). The reported uncertainties (2s) in chromium-isotope ratio and 55 Mn/ 52 Cr ratio include both the internal precision of an individual analysis and the external reproducibility for standard measurements during a given analytical session.
After analysis, the location of each probe spot was re-imaged by electron microscopy to check for beam overlap between phases and to identify large cracks or impurities that may have affected the result.
Age anchors. We used the U-isotope-corrected Pb-Pb absolute age of the CV CAIs (4,567.30 ± 0.16 Ma; ref. 9) as the age of the Solar System. Owing to the presence of isotopically anomalous chromium, absence of primary minerals with Mn/Cr 4 41 and evidence for a late-stage disturbance of 53 Mn- 53 Cr systematics in CV CAIs 20 , the initial 53 Mn/ 55 Mn ratio of the Solar System is not known. In this paper, relative 53 Mn- 53 Cr ages obtained in fayalites as well as those in calcites reported in literature 8 are anchored to the D'Orbigny angrite for which ( 53 Mn/ 55 Mn) 0 and U-corrected Pb-Pb absolute ages are known 21,22 .
Angrites are a small but diverse group of igneous meteorites that are divided into two subgroups: coarse-grained plutonic and fine-grained quenched angrites (ref. 54 and references therein). The mineralogy, bulk chemical and isotopic compositions suggest that both subgroups originated on the same parent body. Crystallization of the quenched angrites predates that of the plutonic angrites by B7 Myr (refs 55,56). The quenched angrites are useful as age anchors for multiple short-lived chronometers (ref. 56) as (1) having crystallized from a melt, they were isotopically homogenized at the time of their original formation; (2) they formed sufficiently early in the history of the Solar System to contain measurable excesses of the daughter products of several short-lived radionuclides; and (3) they crystallized and cooled rapidly at the time of their formation, effectively closing the isotope systems of long-and short-lived chronometers at the same time.
Among the quenched angrites dated, we used D'Orbigny as an age anchor for calculating model ages for the following reasons: (1) it is relatively unmetamorphosed, and therefore preserves the chronological records largely undisturbed 57 ; (2) it has a precise U-isotope-corrected Pb-Pb absolute age The calculated 53 Mn- 53 Cr ages of fayalite were subsequently compared with the U-corrected Pb-Pb absolute age (4,567.30 ± 0.16 Ma) of the CV CAIs 9 . We recalculated the previously reported model ages of secondary calcites in CM carbonaceous chondrites 8 relative to the D'Orbigny angrite anchor (Supplementary Table 5 and Supplementary Fig. 2).
Thermal modelling of chondrite parent bodies. To model the thermal evolution of the L, CV and CO chondrite-like bodies consistent with the peak metamorphic temperatures, the temperature range of fayalite formation and the inferred 53 Mn- 53 Cr ages of fayalite crystallization, we considered spherically symmetric, instantaneously accreting bodies that were heated by decay of 26 Al only. A heat-conduction equation, where r is density, c is specific heat, T is temperature, t is time, r is radius from the centre, K is thermal conductivity, A is initial radiogenic heat generation rate per unit volume and l is the decay constant of 26 Al, is solved numerically using a finite difference method and an explicit integral method 58 . We assumed that (1) 26 Al was the only heat source of a chondrite parent body, and was uniformly distributed in the protoplanetary disk with the canonical 26 Al/ 27 Al ratio of 5.25 Â 10 À 5 ; (2) some regions of these parent bodies reached 100 À 200°C (temperature range of fayalite formation 28 ) at the time of fayalite formation; and (3) that these regions avoided subsequent heating above 300°C (to preclude Fe-Mg diffusion in secondary fayalite). Supplementary Table 6 (part 1) defines parameters specific to the parent bodies from which the L, CV and CO chondrites were derived, such as the fayalite formation ages (Figs 4 and 5) and peak metamorphic temperatures experienced by these parent bodies 23,31,32 . We note that using an initial 26 Al/ 27 Al ratio of 5.25 Â 10 À 5 , rather than 5 Â 10 À 5 , corresponds to 0.05 Myr, which is well within the uncertainty of our isotope measurements (0.4 À 1.8 Myr).
The initial parameters (assumed to be constant) of the L, CV and CO chondrite parent bodies are detailed in Supplementary Table 6 (part 2; refs 28,59). In addition, we used the physical properties (thermal conductivity, specific heat and density) for ice, water 60,61 and rock 62,63 , detailed in Supplementary Table 6 (part 3).
The latent heat of ice melting has been included, and we numerically calculated the effect of ice melting on the thermal evolution. Hydration reactions were not included, as the primary goal is to determine whether or not the parent bodies could reach the formation temperature range of fayalite. The thermal model also does not include the potential influence of an overlying regolith to reduce the thermal diffusivity. Two previous estimates of 26 Al heating of meteorite parent bodies 64,65 reported much higher temperatures than those calculated here, most likely owing to the differences in the assumed physical properties such as the thermal diffusivity (previous studies assumed values of o0.01) and the water/rock ratio (assumed to be zero in the previous studies). We note that increasing the water/rock ratio not only increases the amount of energy required to heat the parent body to a given temperature (ice has a higher heat capacity than rock) but also reduces the amount of 26 Al (per unit mass) available to produce heat as water does not contain any 26 Al.
Parent bodies with radii of 20, 30, 40 and/or 50 km were modelled for the L, CV and CO chondrite parent bodies (Fig. 6). The size of the bodies selected may also represent larger bodies as the maximum temperature reached by a parent body larger than 30 km in radius is largely independent of its size, and there is little difference in the accretion times of the larger bodies (Fig. 6a,b).