Earth and Planetary Science

We present evidence for both early-and late-stage magnetic activity on the CV and L/LL parent bodies respectively from chondrules in Vigarano and Bjurböle. Using micro-CT scans to re-orientate chondrules to their in-situ positions, we present a new micron-scale protocol for the paleomagnetic conglomerate test. The paleomagnetic conglomerate test determines at 95% conﬁdence, whether clasts within a conglomerate were magnetized before or after agglomeration, i.e., for a chondritic meteorite whether the chondrules carry a pre-or post-accretionary remanent magnetization. We found both meteorites passed the conglomerate test, i.e., the chondrules had randomly orientated magnetizations. Vigarano’s heterogeneous magnetization is likely of shock origin, due to the 10 to 20 GPa impacts that brecciated its precursor material on the parent body and transported it to re-accrete as the Vigarano breccia. The magnetization was likely acquired during the break-up of the original body, indicating a CV parent body dynamo was active ∼ 9 Ma after Solar System formation. Bjurböle’s magnetization is due to tetrataenite, which transformed from taenite as the parent body cooled to below 320 ◦ C, when an ambient magnetic ﬁeld imparted a remanence. We argue either the high intrinsic anisotropy of tetrataenite or brecciation on the parent body manifests as a randomly orientated distribution, and a L/LL parent body dynamo must have been active at least 80 to 140 Ma after peak metamorphism. Primitive chondrites did not originate from entirely primitive, never molten and/or differentiated parent bodies. Primitive chondrite parent bodies consisted of a differentiated interior sustaining a long-lived magnetic dynamo, encrusted by a layer of incrementally accreted primitive meteoritic material. The different ages of carbonaceous and ordinary chondrite parent bodies might indicate a general difference between carbonaceous and ordinary chondrite parent bodies, and/or formation location in the protoplanetary disk.


Introduction
Chondrites formed mainly by accreting up to millimeter-sized chondrules, micrometer-sized fine-grained dust, millimeter-sized calcium-aluminum rich inclusions (CAIs) and opaque phases approximately 4.6 Ga ago (e.g.Scott, 2007;Connelly et al., 2012).Chondrules are thought to have formed in the presence of magnetic fields before accretion of the parent asteroid, and have the potential to record pre-accretionary magnetic remanences that can be used to estimate the role of magnetic fields in early Solar System momentum transport and chondrule formation (Desch et al., 2012;Fu et al., 2014a).However, many primitive and unequilibrated chondritic meteorites show some evidence of secondary for some meteorites both types of magnetization have been reported, e.g., Butler (1972) found a homogeneous magnetization amongst sub-samples of Allende, then Sugiura et al. (1979) found inhomogeneous magnetization and finally Carporzen et al. (2011) again found homogeneous magnetizations.Some of these inconsistencies could lie in the difficulty of performing conglomerate tests on the sub-millimeter scale: sampling, orienting and measuring smaller than standard samples (∼11 cm 3 in terrestrial studies) leads to greater angular-dispersion errors as well as sampling both chondrule and matrix material together into individual samples (Butler, 1972;Strangway and Sugiura, 1982;Böhnel et al., 2009;Böhnel and Schnepp, 2014).Chondrules are particularly small, with volumes of ∼0.0005 cm 3 .Physical disaggregation and reorientation of individual chondrules from a chondrite by hand is likely to introduce errors that are difficult to quantify and, due to scarcity of material, cannot be reduced by simply measuring more samples as is the case for most terrestrial studies.Gattacceca et al. (2016) were not able to mutually orient their sub-samples of Kaba.Accurate reorientations (<5 • error) of chondrule samples have been achieved by working with cut-sections of material (Carporzen et al., 2011;Fu et al., 2014aFu et al., , 2014b)), however these studies worked with sample sections, and not with whole chondrule samples.
Experimental methods that are non-destructive and minimize human error are particularly important when studying the paleomagnetism of meteorites and their constituents.X-ray microcomputed tomography (micro-CT) has been a successful technique for the 3D volumetric non-destructive characterization of meteorites, and is widely being incorporated into the curation procedure for meteorites (e.g., Hezel et al., 2013a;Smith et al., 2013;Zeigler et al., 2015).In this paper we report a new sub-millimeterscale method for the paleomagnetic conglomerate test that uses x-ray micro-computed tomography (micro-CT) images with the aid of computer software to rotate the ex-situ chondrules to their in-situ positions, allowing their paleodirections to be determined at an accuracy equal to that of routine terrestrial paleomagnetic measurements.We have developed and applied this technique on chondrules from two meteorites: Bjurböle (L/LL4) and Vigarano (CV3).For these two meteorites we have determined the relative orientation of the remanent magnetization of their constituent chondrules.

Samples
The bulk samples from which chondrules were extracted were 686 mg of Bjuböle and 405 mg of Vigarano.To avoid contamination of the magnetization during atmospheric entry of the meteorite, both samples were taken from their interiors, with no fusion crust present.Bjurböle is an ordinary chondrite (L/LL4) that fell in Bjurbdle Borga, Nyland, Finland in 1899, with a 330 kg approximate recovered weight (Grady, 2000).Bjurböle was selected for this study due to its friability, which allows chondrules to be easily disaggregated.Chondrule abundance in Bjurböle is about 66 vol.% and their median diameter is 0.688 ± 0.003 mm (Hughes, 1978).
Vigarano is a carbonaceous chondrite (CV3) that fell in the Emilia-Romagna region of Italy in 1910 with a total recovered mass of 15 kg (Grady, 2000).

Magnetic carriers of the Bjurböle chondrules
The dominant magnetic mineral in Bjurböle chondrules has been reported as tetrataenite, a magnetically hard mineral, with a high coercivity of up to 600 mT (Wasilewski, 1988).Coercivity can be directly related to paleomagnetic stability; therefore, Bjurböle's tetrataenite phase is unlikely to have acquired a magnetic overprint since its formation.Tetrataenite transformed from taenite during cooling to below 320 • C on the parent body, so any remanent magnetization it carries will have originated from cooling to below 320 • C (Gattacceca et al., 2014).Transmission electron microscope (TEM) observations of equilibrated ordinary chondrites found the tetrataenite occurring in plessite grains consisting of >1 µm tetrataenite in a kamacite matrix and a high-coercivity cloudy zone of 25-250 nm tetrataenite in zoned taenite, with a rim of 1-14 µm tetrataenite grains (Uehara et al., 2011).The disordering of tetrataenite to taenite upon heating above 320 • C has been thoroughly investigated experimentally by Dos Santos et al. (2015); our sample was not thermally demagnetized to prevent this.The presence of tetrataenite also indicates the meteorite did not undergo heating to greater than 320 • C after peak metamorphism (Collinson, 1989).Uehara et al. (2011) reported tetrataenite in the cloudy zone of individual taenite grains have homogeneous crystallographic orientations; the crystallographic orientation of tetrataenite is significant, as tetrataenite has such a strong magnetocrystalline anisotropy that can result in remanence directions that can diverge by up to 90 • from the paleofield direction (Collinson, 1989).In a general conglomerate test, divergence of greater than 90 • from a paleofield direction would usually suggest post-metamorphic brecciation of Bjurböle.It is expected that tetrataenite would carry no remanent magnetization if the transformation occurred under zero-field conditions (Uehara et al., 2011).If the transformation occurred in the presence of a magnetic field, the tetrataenite may carry a phase transformation remanent magnetization (Ph-TRM) due to the bias field (Bryson et al., 2014).The dynamics and length-scales of PhTRM acquisition in tetrataenite are uncertain, however, and alignment of the remanence to the crystallographic axes may result in scatter of the recorded paleodirections (Gattacceca et al., 2014;Bryson et al., 2014).

Magnetic carriers of the Vigarano chondrules
The magnetic mineralogy of the chondrules in Vigarano is largely Fe-Ni metal in the form of kamacite, and the matrix is magnetically dominated by magnetite (Brecher and Arrhenius, 1974).Vigarano is believed to be a regolith breccia of the CV chondrite parent body (Kojima et al., 1993;Bischoff et al., 2006) having originated from an aqueously altered precursor chondrite that was subject to impacts of 10 to 20 GPa and transported to re-accrete as Vigarano (Jogo et al., 2009).Vigarano likely underwent its secondary accretion in an anhydrous environment, so a second phase of aqueous alteration is ruled out (Krot et al., 2000;Tomeoka and Tanimura, 2000).The aqueous alteration product fayalite has been 53 Mn/ 53 Cr dated by Jogo et al. (2009) and recalibrated by Doyle et al. (2016) to date the alteration as 4563 Ma ago and brecciation of the precursor chondrite approximately 5 Ma later.There is petrographic evidence for peak metamorphic temperatures of 400-500 • C (Lee et al., 1996), which could have resulted in a partial thermoremanent magnetization (pTRM) in the presence of a bias field.These secondary alterations occurred prior to the final accretion of the Vigarano breccia (Jogo et al., 2009), so would not be evident in a conglomerate test as a homogeneous magnetization.As Vigarano underwent at least two accretions on the parent body 4565 and 4558 Ma ago (see Fig. 7 of Jogo et al., 2009), both, parent body alteration prior to final accretion and a retained Solar Nebula remanence would be manifested as a random distribution.A shock remanent magnetization (SRM) is most likely the origin of remanence in Vigarano, considering its 10 to 20 GPa impact history during brecciation.SRM can be identified by stable and efficient AF (alternating field) demagnetization (Funaki and Syono, 2010;Gattacceca et al., 2010;Tikoo et al., 2015), and if induced during brecciation of the precursor chondrite, it would likely be observed as a random distribution by the conglomerate test due to the re-accretion of Vigarano.For a Solar Nebula remanence to be retained through the brecciation process, the magnetic carriers need to have a high coercivity (e.g.dusty olivine (Lappe et al., 2013;Fu et al., 2014a;Einsle et al., 2016)).Strangway and Sugiura (1982) briefly report a conglomerate test on Bjurböle, in which they find inhomogeneous magnetizations amongst chondrule and matrix sub-samples; however the directions were not entirely random as they are on the same hemisphere as the natural remanent magnetization (NRM) of the whole meteorite, i.e., the chondrules and matrix.Another conglomerate test on 83 mutually oriented sub-samples of chondrules, matrix, metal, and bulk material from Bjurböle found directions completely scattered on the equal-area projection, with no hemispheric trend, although the sub-sample type is not labeled on the projection (Wasilewski et al., 2002).The random scatter found by Strangway and Sugiura (1982) does not preclude scatter introduced due to anisotropy, however the findings reported by Wasilewski et al. (2002) and chemical anomalies amongst Bjurböle chondrules reported by Scott et al. (1985) suggest Bjurböle underwent brecciation post-metamorphism.Weiss et al. (2010) briefly report a conglomerate test on three bulk samples of Vigarano, finding an unstable non-unidirectional high-coercivity component, however, they note that their sample has an IRM contamination due to a hand magnet.

Methods
By micro-CT scanning the bulk meteorite prior to disaggregation, and micro-CT scanning disaggregated chondrules, it is possible to relate the ex-situ and in-situ volumes to each other, and computationally determine the rotation matrix necessary to compare the magnetic measurements as if they are mutually in-situ.To test for internal consistency and completeness we also conducted rock magnetic experiments and determined paleointensity (ancient field intensity) estimates.

Meteorite disaggregation
In order to measure the magnetization of individual chondrules, they must be disaggregated from the bulk meteorite.Bjurböle's chondrules were easily separated by hand from the matrix due to its high friability.However, Vigarano is significantly more lithified, and freeze-thaw disaggregation following Butterworth et al. (2004) was necessary to separate the chondrules.The Vigarano sample was submerged in 1 ml of distilled water in a 5 ml volume Pyrex container.Following initial degassing using a vacuum pump, we put the sample through 138 cycles of freeze-thaw disaggregation: submerging the base of the container in liquid nitrogen until the water surrounding the sample is frozen, and then melting using an ultrasound bath.The permeating and expanding frozen water slowly disaggregates the sample, which we extracted chondrules from using a stereoscopic microscope.We used a thermocouple to ensure the sample's temperature remained above 122 K, the Verwey transition for magnetite (Verwey, 1939).

X-ray microtomography
X-ray micro-computed tomography (micro-CT) is a non-destructive technique allowing the 3D visualization of a scanned object by reconstructing the x-ray attenuation of the object (Ebel and Rivers, 2007;Elliott and Dover, 2011;Hezel et al. 2013aHezel et al. , 2013b)).
The magnetic field inside the CT scanner is <2 mT, which is not strong enough to contaminate the NRM of the meteorite.The volume can be imaged as a result of variable attenuation intensities due to density contrasts (Ketcham, 2005;Elangovan et al., 2012;Griffin et al., 2012).The bulk meteorite sample and the individual chondrules extracted from it were measured using the X-Tek HMX ST 225 CT System at the Natural History Museum, London.A 0.25 mm copper filter was used to remove low-energy x-rays.Once the chondrules were extracted, they were mounted on carbon stubs of 11 mm height and 6 mm diameter in the center.A fine etch was made into the stubs so that the position of the chondrules maintained the same during micro-CT and magnetic measurements.Scanning in batches of four, 40 chondrules from Bjurböle and 19 chondrules from Vigarano were imaged.An additional benefit of volumetric characterization of the sample prior to disaggregation is that the disaggregation process can be targeted towards constituents of interest, such as those with denser and unique identifying features, and thus preserving more material from the destructive disaggregation process.
The raw micro-CT data were converted to bitmap image stacks using the VGStudio MAX software, and the image stacks were viewed in ImageJ (Schneider et al., 2012) to match ex-situ chondrules to in-situ chondrules from the scan of the bulk specimen.Once chondrules were identified, the ex-situ and in-situ volumes were loaded into AvizoFire ® (VSG Inc., USA) to determine the reorientation.Unique and dense features such as barred olivine, Fe-Ni and sulphides within the chondrules and the shape of the chondrules were used to determine the rotation for a mutual in-situ alignment applied to the remanent magnetization directions of the chondrules.The 'Register Images function' on the AvizoFire ® software can in theory be used to automatically align a chondrule to the bulk-reference orientation; however, this function was found to be unreliable and it was only successful with one sample (BJB37).The other re-orientations were achieved by manually rotating the 3D volume, with accuracy greater than 1 • .

Magnetic remanence measurements
To determine the magnetic remanence direction, the chondrules were mounted on carbon stubs of 11 mm height and 6 mm diameter and placed in paleomagnetic sample boxes of size 22 × 22 × 22 mm, such that the chondrule was in the center of the box.The chondrules from Bjurböle were alternating-field (AF) demagnetized up to 120 mT using a 2G Enterprises SQUID Magnetometer, with an in-line AF demagnetizer at The National Oceanography Centre, University of Southampton, and then further AF demagnetized up to 200 mT field using an ASC Scientific D-2000 AF Demagnetizer at Imperial College London.We followed the same procedure for Vigarano, except using the 2G Enterprises SQUID Magnetometer at the University of Oxford.
After AF demagnetization of the NRM, an isothermal remanent magnetization (IRM) of 900 mT was applied using a pulse magnetizer and then subsequently AF demagnetized and measured.These data were used to determine the paleointensity (see section 3.5).

Rock magnetic measurements
Hysteresis loops, backfield curves and first-order reversal curves (FORCs) (Roberts et al., 2000) were measured using a Princeton Measurements Alternating Gradient Magnetometer (AGM) at Imperial College London for characterization and to produce an input distribution to determine the paleointensity of the chondrules with the Preisach paleointensity method (see section 3.5).

Paleointensity estimation
Standard paleointensity estimation protocols such as the Thellier and IZZI methods require heating of the specimens to repli-cate the remanence acquisition process under laboratory conditions (Yu and Tauxe, 2005).The magnetic carriers in meteorites are typically heavily reduced, and laboratory heating results in significant thermal alteration of the remanence carriers, and a loss of the paleomagnetic information stored, e.g.troilite oxidizing to magnetite and tetrataenite disordering to taenite (Herndon et al., 1976;Dos Santos et al., 2015).In order to estimate the paleointensity without altering the specimens, non-heating methods such as the Ratio of Equivalent Magnetization (REM) methods (Kletetschka and Kohout, 2003;Gattacceca and Rochette, 2004), and the Preisach method (Muxworthy and Heslop, 2011) have been developed.The REM method estimates the paleointensity by normalizing the NRM of the sample by its saturation isothermal magnetization (SIRM), and by multiplying by an experimentally determined calibration factor f , i.e., the paleointensity is determined such that paleofield B = f *(NRM/SIRM).To allow for multicomponent magnetizations the REM' (Gattacceca and Rochette, 2004) method compares the magnetization of the NRM and SIRM after AF demagnetization isolates the characteristic remanent magnetization (ChRM).The Preisach protocol also compares the AF demagnetization data of the NRM and SIRM, however, it uses the FORC distribution of the sample to generate a Preisach distribution to more accurately determine f for each sample (Muxworthy and Heslop, 2011).The determine f in this study, we made a minor modification to the outline described in Muxworthy and Heslop (2011): f is only determined for the part of the AF demagnetization spectrum for which the paleointensity estimate is relatively constant (to with variation <30%); this is similar to the REM' approach.Of these nonheating paleointensity protocols, the Preisach method has been demonstrated to be the most accurate (Emmerton et al., 2011;Muxworthy and Heslop, 2011).Lappe et al. (2013) note it may not be accurate when studying a sample dominated by single vortex (SV) state grains, however, no methods accommodate such behavior.

Micro-CT scanning and in-situ alignment
The bulk volume of Bjurböle was micro-CT scanned at a resolution of 6.2 µm (Fig. 1), and subsequently 40 chondrules ranging in mass from 0.1 mg to 7.7 mg were extracted.Due to the friability and ease of disaggregation of Bjurböle, the chondrules from Bjurböle had no matrix material attached to them.Individual chondrules were imaged at resolutions between 6 and 12 µm, which was sufficient to identify individual chondrules and their positions in-situ.Of the initial 40 chondrules, 16 chondrules were too weakly magnetic to constrain magnetic directions, and were discounted from the alignment process.Magnetic data is presented on 19 chondrules, of which seven could be re-oriented.
The bulk volume of Vigarano was micro-CT scanned at a resolution of 6.4 µm.Using the freeze-thaw disaggregation method we extracted 19 chondrules from Vigarano with masses between 0.1 and 5.3 mg.The chondrules were imaged at resolutions of 7 to 12 µm, which was sufficient to identify individual chondrules and their positions in-situ.Of the 19 chondrules, ten were identified in the bulk volume, and eight had sufficient distinctive constituents to accurately rotate to in-situ positions.

Rock magnetic properties
It was possible to saturate the magnetization of the chondrules from Vigarano in a field of 1 T to accurately record hysteresis loops and FORC diagrams.However, for Bjurböle it was not possible to achieve true saturation.A Day plot (Day et al., 1977) of the hysteresis parameters for measured samples is presented in Fig. 2, with regions associated with single domain (SD), pseudo-single domain (PSD) and multi-domain (MD) behavior labeled.Singledomain grains are small (<100 nm for magnetite), uniformly magnetized and considered stable recorders of magnetic fields over geologic timescales.Between the SD and MD size range are the PSD grains, which are typically the size of grains found in rocks (∼100-1000 nm), and have recently been shown (in magnetite) to be reliable recorders of remanence over geological timescales (Almeida et al., 2016).MD grains are large, non-uniformly magnetized and considered to be poor magnetic recorders.
An ideal sample will have an assemblage of SD or PSD grains.The majority of the chondrules, and all of those that were reoriented plot within the region associated with PSD behavior, except for VG02, which plots in the region associated with MD behavior (Fig. 2).The chondrules are grouped into FORC diagram groups A, B and C (Fig. 3).Group A represents chondrules with a central peak on the FORC diagram that extends from 20 to 40 mT (Fig. 3a).The FORC diagrams in group A indicate a SD grain assemblage similar to that seen in Roberts et al. (2000).The average coercivity of these chondrules is 26 mT.Group B represents chondrules with a central peak near the origin that extends up to 60 mT in some samples (Fig. 3b), and also includes BJB17.The FORC diagrams in group B indicate a PSD grain assemblage similar to that seen in Roberts et al. (2000).Group C represents chondrule VG02 and nine of the ten Bjurböle chondrules chosen for FORC analysis.These chondrules have no defined peak in their FORC diagrams (Fig. 3c).

AF demagnetization
In a maximum AF field of 200 mT, the majority of the Bjurböle chondrules were only demagnetized to 40 to 70% of their NRM, likely due to the tetrataenite component of magnetization (Fig. 4).Some chondrules increased in remanence during demagnetization (e.g.Fig. 4b), which is likely due to low-coercivity remanence carriers being demagnetized, and allowing the highcoercivity tetrataenite to be isolated.Of the 40 extracted chon-  drules, we were able to determine a ChRM in 24 chondrules using principal component analysis (PCA) (Kirschvink, 1980) (Fig. 4a  and b).The remaining chondrules were magnetically weak, and no clear direction could be identified.
Ten of the Vigarano chondrules behaved in a similar way to Bjurböle, in that approximately only ∼50% of the remanence could be demagnetized by AF fields up to 200 mT (e.g.Fig. 4c).The other nine, including most of the chondrules for which a reorientation was possible (except VG02 and VG15) had two magnetic components; a soft component that was demagnetized by AF peak-fields of 6 mT, and a ChRM that was demagnetized in peak fields of 50 mT (Fig. 4d).The soft component is likely a (low-coercivity)  A overprint due to the terrestrial magnetic field, referred to as a viscous remanent magnetization (VRM).
For the ChRM, 17 chondrules had identifiable stable paleomagnetic directions, an example of which is shown in Fig. 4 (c and  d), which shows the in-situ NRM directions for VG02 and VG18 on orthogonal-projection plots and equal area projections.VG02 retained ∼44% of its magnetization (Fig. 4c), VG15 retained ∼26% of its magnetization; the rest of the reoriented chondrules were demagnetized to <10% of NRM (e.g.Fig. 4d).The median destructive field (MDF) of the NRM AF demagnetization spectra ranged from 6.5 to 90 mT with a median of 16 mT, and for the SIRM the MDF was 16 to 61 mT with a median of 28 mT.

Paleointensity estimates
Preisach and REM' paleointensity values for 19 chondrules each from Bjurböle and Vigarano are tabulated in Table 1.A calibration factor of 1600 µT was used for the REM' method as determined for synthetic meteoritic samples by Lappe et al. (2013); usually a value of 3000 µT is used (Gattacceca and Rochette, 2004).For Bjurböle, the range of paleointensities determined by the REM' method is 1.9-54 µT with a median of 9.2 µT and mean weighted by the variance of 8.7 ± 1.0 µT.The majority of these estimates are larger than the 3.2 ± 0.2 µT for Bjurböle chondrules reported by Acton et al. (2007), who used the REMc method.It is possible that the paleofield has been overestimated, as the IRM acquisition of 900 mT would not have saturated the tetrataenite mag-netization of the chondrules, so the REM' ratio would be greater than expected.The FORC diagrams of Bjurböle were not sufficient to determine Preisach paleointensities; it was not possible to saturate the tetrataenite phase.For Vigarano, the range of paleointensities determined by the REM' and Preisach methods is 1.7-150 µT and 1.1-91 µT respectively, with medians of 14 µT and 6.7 µT.The weighted means of the REM' and Preisach estimates are 4.4 ± 0.04 µT and 3.6 ± 1.0 µT.The 10 samples were selected for the Preisach method based on the quality of their orthogonalprojection plots and FORC diagrams.
Currently all paleointensity protocols, including the non-heating REM' and Preisach protocols, assume a thermoremanent origin of magnetization (TRM).All theories used for paleointensity protocols also assume magnetically uniform single domain (SD) assemblages.These assumptions are, however, rarely fulfilled.This is also true for meteoritic materials containing FeNi, as the SD threshold size is very small (<30 nm) (Muxworthy and Williams, 2015).Therefore, all paleointensity estimates need to be interpreted very cautiously.The phase transformation of the tetrataenite in Bjurböle is poorly understood, introducing an additional error to consider when calibrating between PhTRM and TRM acquisition mechanisms.Hence, the paleointensity estimates of the Bjurböle chondrules may have an understated error.A shock origin of remanence (SRM) has been shown to be 10% of the recording efficiency of a TRM for pressures up to 1.8 GPa (Cisowski and Fuller, 1978;Tikoo et al., 2015).Hence, the paleointensity estimation could un- derestimate the paleofield strength -if a SRM -by up to an order of magnitude.Hypervelocity impacts on planetary surfaces in the presence of an ambient field have been shown to result in significant crustal remanence recordings (Gattacceca et al., 2008).The Vigarano chondrules may have recorded remanence at a greater efficiency than 10% given they were subjected to 10 to 20 GPa brecciating impacts, and recording efficiency increases with greater shock pressure (Gattacceca et al., 2008).

Discussion
We disaggregated, micro-CT scanned (Fig. 1) and measured the demagnetization spectra (Fig. 4) of 40 chondrules from Bjurböle and 19 chondrules from Vigarano.Of these, we were able to reorient and determine ChRM directions for seven Bjurböle chondrules and eight Vigarano chondrules (Table 1, Fig. 5).Magnetic carriers of the chondrules were characterized by their rock magnetic properties, presented on a Day plot (Day et al., 1977) and FORC diagrams (Roberts et al., 2000) in Figs. 2 and 3 and Table 1.Paleointensity estimations were made for chondrules from both meteorites using the REM' and Preisach methods (Table 1).

Conglomerate test
The reorientation of the ChRM directions (Fig. 5) to their relative in-situ positions allow us to statistically determine if they were aligned or not using the Watson test for randomness (95% confidence) (Watson, 1956).
For Bjurböle, the randomness of the ChRM directions cannot be disproved at the 95% confidence interval, i.e., Bjurböle is said to have passed the conglomerate test, and indicates that the chondrules have random magnetic directions.This is in agreement with the previous conglomerate test on Bjurböle, which also found scattered paleodirections from seven chondrules (Strangway and Sugiura, 1982).
For Vigarano, the randomness of the ChRM directions cannot be disproved at the 95% confidence interval, i.e., Vigarano has also passed the conglomerate test, and chondrules carry random magnetic directions.This is in agreement with previous findings by Weiss et al. (2010) on three oriented bulk samples of Vigarano.
The low-coercivity (LC) components of the Vigarano chondrules' magnetizations (<6 mT peak AF fields) fail the conglomerate test, indicating that they are not from a random distribution.It is very likely the LC components are a terrestrial magnetization (a viscous remanent magnetization) (Brecher and Arrhenius, 1974), and thus would be expected to be aligned.The alignment of the Vigarano LC components validates the reorientation procedure.

Origin of Bjurböle's remanent magnetization
The demagnetization data presented in Fig. 4a and b indicate that the chondrules in Bjurböle carry a remanent magnetization, and the inability to demagnetize the magnetization with peak AF fields of 200 mT or saturate in fields of 1 T during hysteresis suggests it is carried by high-coercivity tetrataenite.We present a schematic to illustrate the likely history of the magnetization of Bjurböle's chondrules in Fig. 6.Pre-accretionary remanence carrying chondrules (Fig. 6a) are heated on the L/LL parent body and cooled, undergoing phase transformation to the final ferromagnetic tetrataenite, with ferromagnetic taenite as its precursor (Uehara et al., 2011) (Fig. 6b).The remanence of the tetrataenite in Bjurböle is unlikely to have been overprinted since formation due to its very high coercivity, and probably represents an ambient magnetic field at the time of cooling on the parent body to below 320 • C (Uehara et al., 2011;Bryson et al., 2015) (Fig. 6c).Given the 1 to 2 cm length scale of the samples the chondrules originated from, an ambient field on the parent body at the time of cooling would result in a mutual alignment of the chondrules' magnetic moments within 180 • (Collinson, 1989) (Fig. 6c).The paleodirections of Bjurböle fall within 180 • of each other (Fig. 5a), so an anisotropy scattered record of an ambient field cannot be precluded (Fig. 6c).However, the paleointensity estimates range from 1.9-54 µT, which is a comparatively high-level of variability, thereby supporting inhomogeneity.The inhomogeneity found in this study, by Strangway and Sugiura (1982) and by Wasilewski et al. (2002), suggest that Bjurböle was brecciated postmetamorphism (Fig. 6c).Bjurböle has been previously suggested as a breccia due to chemical anomalies (Scott et al., 1985).Shock typically affects the low-coercivity spectrum of grains (Yu et al., 2011;Tikoo et al., 2015), so the brecciation process is unlikely to alter the remanence carried by the high-coercivity tetrataenite grains.
The paleointensities and tetrataenite remanence require the existence of an external magnetic field at the time of tetrataenite formation (Fig. 6c).Peak metamorphism on the L/LL parent body has been estimated to have prolonged for 0.5 to 5 Ma (Seitz et al., 2016).Thermochronometry suggests that Bjurböle would have cooled at a rate of 2 • C/Ma, which would require up to 80 to 140 Ma to cool to 320 • C (Willis and Goldstein, 1981) (Fig. 6b).Fig. 6.A schematic formation scenario for Bjurböle.Initially chondrules acquire a pre-accretionary thermoremanent magnetization (TRM) and are accreted as a random distribution on the Bjurböle parent body (a).The TRM is subsequently erased due to thermal metamorphism (b).Cooling of the body over 80 to 140 Ma (Willis and Goldstein, 1981) to 320 • C results in the formation of ferromagnetic tetrataenite as a phase-transformation remanent magnetization (PhTRM) (c).It is unknown whether the remanence of the ferromagnetic taenite (b), the precursor to tetrataenite, retains any of its remanence.The random magnetization distribution of chondrules in Bjurböle (see Fig. 5a) is either due to PhTRM of an ambient field by the anisotropic tetrataenite when Bjurböle cooled to below 320 • C, or a PhTRM of the ambient field by the chondrules that are then brecciated and re-accreted (c).Phase-transformation remanent magnetization (PhTRM) is not well understood, and it may be possible that tetrataenite inherits remanence from its precursor taenite.Solar wind is unlikely to have been the ambient magnetizing field, as it would not have been stable relative to the rotating L/LL parent body on the Ma timescale to acquire a PhTRM.The possibility of a planetesimal sustaining magnetic activity more than 50 Ma after accretion has been observed in the tetrataenite cloudy zones found within pallasites (Bryson et al., 2015), and it may be possible that the L/LL parent body began core solidification to sustain a magnetic dynamo, recorded as a PhTRM by the Bjurböle chondrules.

Origin of Vigarano's remanent magnetization
The demagnetization spectra of the chondrules from Vigarano (Fig. 4) indicate that they each carry a low-coercivity (LC) as well as a medium-coercivity (MC) component of remanence.The di-rections of the LC components align and are demagnetized with peak-AF fields of up to 6 mT, suggesting they are an overprint acquired on Earth due to the geodynamo.The MC components are different to the LC terrestrial components and have a random magnetization distribution, indicating that they originated prior to the accretion of the Vigarano breccia.
We present a schematic after Jogo et al. (2009) to illustrate the likely history of the magnetization in Vigarano's chondrules in Fig. 7. Pre-accretionary remanence carrying chondrules were accreted onto the CV parent body (Fig. 7a), which underwent heterogeneous aqueous alteration (Fig. 7b).The impacts that brecciated the precursor chondritic material at the chondrule scale (Fig. 7c) to then reaccrete as the Vigarano breccia may be the origin of the remanence observed in Vigarano's chondrules (Kojima et al., 1993;Jogo et al., 2009) (Fig. 7d).The efficient AF demagnetization of the NRM MC component compared to the demagnetization of the IRM (Fig. 4c, d), evident in the comparison between their MDFs (the peak AF required to remove 50% of remanence; see section 4.3.1)suggests that the NRM is likely of shock origin (SRM) (Tikoo et al., 2015).A previous conglomerate test on basalts imparted with an SRM suggests that a remanence proportional to the ambient field and homogeneous in intensity and direction on the mm 3 scale would be acquired (Gattacceca et al., 2010).Subsequent transport and re-accretion of shocked clasts would then result in a heterogeneous remanence distribution (Fig. 7).
The paleointensity estimates for Vigarano determined by the REM' and Preisach methods range from 1.7-150 µT and 1.1-91 µT respectively, with medians of 14 µT and 6.7 µT, and weighted means of 4.4 ± 0.04 µT and 3.6 ± 1.0 µT.There are no previously published paleointensity estimates for Vigarano that we are aware of.The paleointensity estimates display variability, particularly when plotted on the equal-area projection of ChRMs for oriented chondrules, suggesting that the inducing field was not the same for chondrules with similar directions (for example VG13 and VG18: Table 1, Fig. 5).This supports the positive conglomerate test, and suggests the remanence was induced prior to final accretion.If the ChRM is an SRM in origin, the reliability of the paleointensity estimates is reduced; the estimates assume a TRM origin.Tikoo et al. (2015) note that the magnetization intensity recording efficiency of shock remanence is at least 10% of a TRM.An ambient field up to ten times as intense as estimated is required, however, Tikoo et al. (2015) only tested pressure up to 1.8 GPa, so this may not correlate for Vigarano's more intensely shocked chondrules.
The SRM is unlikely to be a recording of an ambient field due to solar wind at the time of impact (Tarduno et al., 2016).The present-day field due to the solar wind is on the nanotesla-scale at 1 AU (Tarduno et al., 2014).The surface magnetic field, rotational period and mass loss rate of the young Sun has been modeled and observed in comparable stars and suggests the solar wind's magnetic field to have been not significantly greater than the present day value (Wood et al., 2014).The ∼50 µT field present during the first 2 to 3 Ma of the Solar System (Fu et al., 2014a) has been observed in angrites and Kaba to have decayed to <0.3 µT by 4 Ma after CAI formation (Wang et al., 2015;Gattacceca et al., 2016).The low recording efficiency of SRM compared to TRM means the paleointensities for Vigarano chondrules would have been underestimated by up to an order of magnitude (Tikoo et al., 2015).The paleofield would have been approximately ∼40 µT, which is of the order expected for planetary dynamos (Selkin and Tauxe, 2000).The brecciation and associated SRM acquisition occurred between 4563 to 4558 Ma ago (Jogo et al., 2009), which chronologically agrees with the dynamo identified on the CV parent body of Kaba (Gattacceca et al., 2016).

Implications for parent body magnetism
Asteroid bodies can feasibly sustain magnetic dynamos for the first 50 Ma of the Solar System due to their thermal structure, and much later once they have cooled enough to initiate core solidification (Elkins-Tanton et al., 2011;Laneuville et al., 2014;Scheinberg et al., 2015).
Vigarano originates from the CV parent body, and recent investigations of the CV3 chondrite Kaba have interpreted its magnetic remanence as due to an Earth-like magnetic dynamo on the CV parent body that was active >6 Ma after CV CAIs (Gattacceca et al., 2016).
However, Nagashima et al. (2017) dispute this interpretation; they claim that the initial abundance of 26 Al that has been measured in CV chondrules is insufficient to have melted the interior of their parent body, thereby suggesting dynamo could not have existed on chondrite parent bodies.However, Nagashima et al. (2017) did not consider the late injection of 60 Fe that likely acted as a heat source for chondrite parent bodies (Mostefaoui et al., 2005).The lack of sufficient 26 Al to melt the interior of a CV parent body does not preclude the possibility of an incrementally accreted parent body, with undifferentiated CV chondritic material accreting onto an already existing differentiated planetesimal (Nagashima et al., 2017).The Mo isotope relationship between certain groups of magmatic iron meteorites and carbonaceous chondrites suggests they formed in the same region (Budde et al., 2016).The incremental accretion models proposed by Elkins-Tanton et al. (2011) and Sahijpal and Gupta (2011) may describe the formation mechanism of the CV parent body, and how it could have had a differentiated interior with an undifferentiated crust.
The remanence in the Bjurböle chondrules is carried by highcoercivity tetrataenite, so regardless of whether it is a breccia or not, it acquired a PhTRM at least 80 to 140 Ma after peak metamorphism.A parent body dynamo active for greater than 80 Ma would require a radius of 800 to 1000 km, or smaller if the recorded dynamo activity is due to core solidification, which can extend the lifetime of the planetesimal dynamo by up to 10 Ma (Sterenborg and Crowley, 2013;Scheinberg et al., 2015).The chondrules in Bjurböle are the first non-pallasite indication of late-stage parent body magnetism that may be due to core solidification.
The magnetization of the chondrules in this study demonstrate the different mechanisms for magnetism to be sustained on parent body asteroids due to their evolution and thermal histories.Asteroids can, therefore, sustain at least three periods of magnetic dynamo activity: (i) one early stage dynamo active <50 Ma (Elkins-Tanton et al., 2011;Gattacceca et al., 2016; this study), (ii) a period of quiescence or potentially impact-driven magnetism (Nichols et al., 2016) and (iii) a later dynamo once the body has cooled sufficiently (Tarduno et al., 2012;Bryson et al., 2015; this study).

Conclusion
We have developed a new micron-scale method for the paleomagnetic conglomerate test that utilizes micro-CT scanning to generate a rotation matrix for disaggregated sub-millimeter clasts from a conglomerate.We applied this method to two chondritic meteorites (Bjurböle (L/LL4) and Vigarano (CV3)) to understand the origin of the magnetization of the chondrules.The alignment of the reoriented low-coercivity components of Vigarano's chondrules' magnetizations clearly demonstrates that the reorientation protocol was successful.We estimate that the reorientation is accurate to <1 • , that is, the directional error in this protocol is the same as that of a standard paleomagnetic measurement (e.g. a 1-inch core).
Bjurböle has a random magnetization distribution amongst the chondrules.We argue that the chondrules in Bjurböle, initially carrying a pre-accretionary magnetization, recorded a phasetransformation remanent magnetization (PhTRM) on the parent body after cooling from peak-metamorphism to 320 • C over 80 to 140 Ma.After this PhTRM acquisition, we cannot preclude that the scattered paleodirections are due to an ambient field being recorded by the highly anisotropic tetrataenite.However, the variable paleointensities, scatter in paleodirections and previously found chemical anomalies suggest Bjurböle formed as a breccia from a brecciated precursor chondrite.
Vigarano has a random magnetization distribution amongst the chondrules.Given that Vigarano formed by re-accreting chondrules and matrix material that had been brecciated by 10 to 20 GPa impacts on the parent body and its demagnetization behavior; it is most likely that its NRM is a shock remanence overprint due to impacts.The harder, non-viscous component of remanence (MC), suggests these brecciating impacts occurred in the presence of an ambient field prior to the accretion of the Vigarano breccia 4558 Ma ago.
Vigarano chondrules evidence the early-stage magnetic activity (ca. 10 Ma after Solar System formation) on asteroid bodies driven by their thermal structure, and the remanence in the Bjurböle chondrules is an indication of late-stage parent body magnetism (>80 to 140 Ma after Solar System formation) likely due to core solidification.The results of our study support earlier findings of Elkins-Tanton et al. (2011) and Gattacceca et al. (2016) that primitive chondrites originate from asteroids that must have sustained a magnetic field.It seems ever more evident that primitive chondrites need not to have originated from entirely primitive, never molten and/or differentiated parent bodies.Rather, it seems parent bodies of primitive meteorites consisted of a differentiated interior, encrusted by a layer of incrementally accreted primitive meteoritic material.Asteroids can, therefore, sustain at least three periods of magnetic dynamo activity: (i) one early stage dynamo active <50 Ma, (ii) a period of quiescence or potentially impact-driven magnetism and (iii) a later dynamo once the body has cooled sufficiently.
The difference in magnetic activity between the carbonaceous chondrite Vigarano and the ordinary chondrite Bjurböle might indicate a general difference between carbonaceous and ordinary chondrite parent bodies, and/or formation location in the Solar Nebula (e.g.Budde et al., 2016).However, this needs to be confirmed by additional studies.

Fig. 1 .
Fig. 1.(a) A micro-CT scan slice of Bjurböle prior to disaggregation.(b) A slice of the micro-CT scan of the individual chondrule, circled in-situ in (a), after disaggregation.Image is in a grayscale to represent material density, with denser features such as Fe-metal and sulphides appearing brighter.Size and dense features within chondrules are used to locate ex-situ chondrules in their original position prior to disaggregation in the bulk meteorite before loading the 3D volume into AvizoFire (VSG Inc., USA) to determine the reorientation rotation matrix.

Fig. 2 .
Fig. 2. A Day Plot(Day et al., 1977) with hysteresis parameters coercivity (B C ), coercivity of remanence (B CR ), saturation magnetization (M S ) and saturation remanence (M RS ) labeled and plotted as M RS /M S versus (B CR /B C ). Regions associated with single domain (SD), pseudo-single domain (PSD) and multi-domain (MD) are labeled.

Fig. 3 .
Fig. 3. Representative first-order reversal curve (FORC) diagrams for the sample set (a, b, c).The chondrules are grouped into A, B and C type FORC diagrams.FORCs were measured over 50 to 60 min, and a smoothing factor (SF) of 5 was used when processing the FORCS.

Fig. 4 .
Fig. 4. Representative Zijderveld plots and equal area projections of the sample set for the AF demagnetization of (a, b) Bjurböle, and (c, d) Vigarano.Both natural remanent magnetization (NRM) and isothermal remanent magnetization (IRM) are plotted on the demagnetization plot insets, which are normalized to the initial NRM and IRM.Low-coercivity (LC) overprint and characteristic remanent magnetization (ChRM) components of Vigarano chondrules' remanence have been highlighted and labeled.

Fig. 5 .
Fig. 5. Equal-area projections of the magnetic directions for the chondrules that were reoriented from Bjurböle (a) and Vigarano (b).Directions representing individual samples were determined by principle component analysis of their demagnetization spectra.

Fig. 7 .
Fig. 7.A schematic formation scenario for Vigarano afterJogo et al. (2009) with the ages recalibrated byDoyle et al. (2016).Annotations indicate the proposed remanent magnetization carried by the chondrules of Vigarano during the formation process of the breccia.The chondrules initially carry a pre-accretionary remanent magnetization (a), which is overprinted by a chemical remanent magnetization (CRM) if the chondrules were subsequently oxidized (b).These pre-accretionary and chemical remanences are later overprinted by a shock-remanent magnetization (SRM) due to the breakup of the precursor chondrite (c).The final assembly of the Vigarano breccia results in this SRM have an inhomogeneous distribution at the millimeter to centimeter scale (d).