DOI: 10.1126/science.1223932, 939 (2012);338 Science

et al.John A. TardunoEvidence for a Dynamo in the Main Group Pallasite Parent Body

This copy is for your personal, non-commercial use only.

clicking here.colleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others

  here.following the guidelines

can be obtained byPermission to republish or repurpose articles or portions of articles

  ): November 16, 2012 www.sciencemag.org (this information is current as of

The following resources related to this article are available online at

http://www.sciencemag.org/content/338/6109/939.full.htmlversion of this article at:

including high-resolution figures, can be found in the onlineUpdated information and services,

http://www.sciencemag.org/content/suppl/2012/11/15/338.6109.939.DC1.html can be found at: Supporting Online Material

http://www.sciencemag.org/content/338/6109/939.full.html#relatedfound at:

can berelated to this article A list of selected additional articles on the Science Web sites

http://www.sciencemag.org/content/338/6109/939.full.html#ref-list-1, 3 of which can be accessed free:cites 54 articlesThis article

http://www.sciencemag.org/content/338/6109/939.full.html#related-urls1 articles hosted by HighWire Press; see:cited by This article has been

registered trademark of AAAS. is aScience2012 by the American Association for the Advancement of Science; all rights reserved. The title

CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience

on

Nov

embe

r 16

, 201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

10. S. Tamura, D. C. Hurley, J. P. Wolfe, Phys. Rev. B 38,1427 (1988).

11. T. Yao, Appl. Phys. Lett. 51, 1798 (1987).12. X. Y. Yu, G. Chen, A. Verma, J. S. Smith, Appl. Phys. Lett.

67, 3554 (1995).13. W. S. Capinski et al., Phys. Rev. B 59, 8105

(1999).14. S.-M. Lee, D. G. Cahill, R. Venkatasubramanian,

Appl. Phys. Lett. 70, 2957 (1997).15. G. Chen, Phys. Rev. B 57, 14958 (1998).16. B. C. Daly, H. J. Maris, K. Imamura, S. Tamura, Phys. Rev. B

66, 024301 (2002).17. Y. K. Koh, Y. Cao, D. G. Cahill, D. Jena, Adv. Funct. Mater.

19, 610 (2009).18. E. T. Swartz, R. O. Pohl, Rev. Mod. Phys. 61, 605

(1989).19. R. Landauer, Philos. Mag. 21, 863 (1970).20. D. Li et al., Appl. Phys. Lett. 83, 2934 (2003).21. P. D. Robb, A. J. Craven, Ultramicroscopy 109, 61

(2008).22. C. A. Paddock, G. L. Eesley, J. Appl. Phys. 60, 285

(1986).23. D. G. Cahill, Rev. Sci. Instrum. 75, 5119 (2004).24. A. J. Schmidt, X. Chen, G. Chen, Rev. Sci. Instrum. 79,

114902 (2008).

25. Materials and methods are available as supplementarymaterials on Science Online.

26. M. A. Afromowitz, J. Appl. Phys. 44, 1292 (1973).27. P. Hyldgaard, G. D. Mahan, Phys. Rev. B 56, 10754

(1997).28. S. Tamura, Y. Tanaka, H. J. Maris, Phys. Rev. B 60, 2627

(1999).29. E. S. Landry, A. J. H. McGaughey, Phys. Rev. B 79,

075316 (2009).30. S. Volz, J. B. Saulnier, G. Chen, P. Beauchamp,

Microelectron. J. 31, 815 (2000).31. S. Baroni, P. Giannozzi, A. Testa, Phys. Rev. Lett. 58,

1861 (1987).32. D. A. Broido, M. Malorny, G. Birner, N. Mingo,

D. A. Stewart, Appl. Phys. Lett. 91, 231922 (2007).33. K. Esfarjani, G. Chen, H. T. Stokes, Phys. Rev. B 84,

085204 (2011).34. J. Garg, N. Bonini, B. Kozinsky, N. Marzari, Phys. Rev.

Lett. 106, 045901 (2011).35. J. Garg, N. Bonini, N. Marzari, Nano Lett. 11, 5135

(2011).36. S. Tamura, Phys. Rev. B 27, 858 (1983).37. P. A. Lee, D. S. Fisher, Phys. Rev. Lett. 47, 882 (1981).38. C. Caroli, R. Combescot, P. Nozieres, D. Saint-James,

J. Phys. C Solid State Phys. 4, 916 (1971).

39. Y. Meir, N. S. Wingreen, Phys. Rev. Lett. 68, 2512 (1992).40. G. Pernot et al., Nat. Mater. 9, 491 (2010).41. R. M. Costescu, D. G. Cahill, F. H. Fabreguette,

Z. A. Sechrist, S. M. George, Science 303, 989(2004).

42. C. Chiritescu et al., Science 315, 351 (2007).

Acknowledgments: We thank A. A. Maznev, K. A. Nelson,K. C. Collins, and J. Johnson for helpful discussions. Thismaterial is based on work supported as part of the Solid StateSolar-Thermal Energy Conversion Center (S3TEC), an EnergyFrontier Research Center funded by the U.S. Department ofEnergy, Office of Science, Office of Basic Energy Sciencesunder award DE-SC0001299/DE-FG02-09ER46577. M.N.L. waspartially supported by the National Science FoundationGraduate Research Fellowship under grant 1122374.

Supplementary Materialswww.sciencemag.org/cgi/content/full/338/6109/936/DC1Materials and MethodsFigs. S1 to S4References (43–57)

4 June 2012; accepted 9 October 201210.1126/science.1225549

Evidence for a Dynamo in the MainGroup Pallasite Parent BodyJohn A. Tarduno,1,2* Rory D. Cottrell,1 Francis Nimmo,3 Julianna Hopkins,2 Julia Voronov,1Austen Erickson,1,2 Eric Blackman,2 Edward R.D. Scott,4 Robert McKinley1

Understanding the origin of pallasites, stony-iron meteorites made mainly of olivine crystalsand FeNi metal, has been a vexing problem since their discovery. Here, we show that pallasiteolivines host minute magnetic inclusions that have favorable magnetic recording properties. Ourpaleointensity measurements indicate strong paleomagnetic fields, suggesting dynamo action inthe pallasite parent body. We use these data and thermal modeling to suggest that some pallasitesformed when liquid FeNi from the core of an impactor was injected as dikes into the shallowmantle of a ~200-kilometer-radius protoplanet. The protoplanet remained intact for at leastseveral tens of millions of years after the olivine-metal mixing event.

Lord Rayleigh (Robert John Strutt) (1)noted the paradox posed by pallasite me-teorites:Olivine andmetal seemingly should

have separated into layers in their parent body.Some models, to avoid segregation, have invokedsmall metal pools throughout a parent body (2),but the putative scenario has remained in forma-tion near a core-mantle boundary (3). There are~50 known pallasite meteorites. Most have iso-topic ratios that fall near the terrestrial massfractionation line and are called “main group”pallasites (4). Olivine ranges from Fa11 to Fa20and often occurs as centimeter-sized (Fig. 1, Aand B) crystals (5–8), with a dislocation density

(9) comparable with those of unshocked terres-trial samples. The metal in main group pallasitesis Ir poor and is thought to have originated fromthe residual melt fraction of a core similar incomposition to IIIAB iron meteorites (3).

Paleomagnetism might help to distinguishbetween models for pallasite formation, but priorattempts have failed to yield interpretable data.The massive FeNi of the pallasite matrix is thelikely culprit. This metal is similar to that com-posing iron meteorites, which carries a highlyanisotropic, soft magnetization; it is notoriouslypoor as a paleomagnetic recorder (10, 11). Paleo-magnetic studies of other meteorites [for example,(12–13)], however, suggest some parent bodieshosted dynamos.Modeling suggests bodies >80 kmin radius could be in the regime of supercriticalmagnetic Reynolds numbers, in which large-scaledynamo action is possible (14, 15).

Rather than studying bulk material, we ap-plied techniques of single-silicate crystal analysis(16, 17) to an investigation of the Imilac andEsquel main group pallasites. We selected gem-like olivine subsamples ≳0.5 cm from the me-

teorite edge and several millimeters from theolivine/metal contact. Prior studies (18, 19) sug-gest that at these distances, heating effects due toatmospheric entry are negligible.

We have observed strings of large inclusions,tens of micrometers in size (Fig. 1C), in someolivines using transmitted light microscopy. Scan-ning electron microscopy (SEM) reveals isolatedand strings of much smaller inclusions (≲10 mm)(Fig. 1D) that are composed of Fe, Ni, S, and Cr(fig S3).Microprobe analyses detail submicrometer-sized, irregularly spaced FeNi particles within thesesmaller inclusions, surrounded by troilite (fig S4).These metal particles are sometimes Ni rich [~51to 58 weight percent (wt %) Ni] and are potentialstable magnetic recorders.

Olivine subsamples lacking inclusions visibleto the naked eye show pseudo-single– to single-domain magnetic hysteresis behavior (Fig. 1, Eand F). In contrast, samples with visible inclu-sions have multidomain behavior. In the formercase, we find only a slight anisotropy (Fig. 1G),and first-order reversal curves (20) fail to showsubstantial magnetic interactions (Fig. 1H). Thus,we further selected olivine subsamples lackingvisible inclusions because they can have optimalproperties for paleointensity determination (21).

Many meteorites have been exposed to mag-netic contamination during collection (13). Wetherefore first used alternating field demagne-tization, which revealed removal of magnetiza-tions after the application of low peak fields (5 to10 mT). Magnetization directions stabilized afterthis pretreatment, and it was here that we startedthermal demagnetization.We used thermal meth-ods because they best replicate the potential mag-netization acquisition process [thermoremanentmagnetization (TRM)] (21). In many meteorites,magnetic mineral alteration accompanying thermaltreatment is severe (11–13). Studies of terrestrialsamples indicate that inclusions in single-silicatecrystals are less susceptible to alteration (16, 17).

1Department of Earth and Environmental Sciences, Universityof Rochester, Rochester, NY 14627, USA. 2Department of Phys-ics and Astronomy, University of Rochester, Rochester, NY14627, USA. 3Department of Earth and Planetary Sciences,University of California, Santa Cruz, CA 95064, USA. 4HawaiiInstitute for Geophysics and Planetology, University of Hawaii,Manoa, HI 96822, USA.

*To whom correspondence should be addressed. E-mail:john.tarduno@rochester.edu

www.sciencemag.org SCIENCE VOL 338 16 NOVEMBER 2012 939

REPORTS

on

Nov

embe

r 16

, 201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

Low unblocking temperature magnetiza-tions (<360°C) observed from Esquel olivinelikely have a viscous origin. However, the Esquel

pallasite olivine shows a large decrease in nat-ural remanent magnetization (NRM) and a sta-ble direction between ~360° and 500°C (Fig. 2,

A and B). Only very small NRM changes areseen at higher demagnetization temperatures,between 500° and 750°C. The dominant drop

2 m

m

A

C D

B E F

G H

Fig. 1. Magnetic character of inclusions in pallasite olivine. (A and B) Esquel andImilac meteorite samples, respectively. (C) Large inclusions in olivine (transmitted lightmicroscopy). (D) String of smaller inclusions (between white arrows; SEM). (E and F)

Magnetic hysteresis curves for olivine. (G) Hysteresis parameter versus angle of measurement (16) and (H) First-order reversal curve plot (20) for Esquel olivine.Mr,remanent magnetization; Ms, saturation magnetization; Hcr, coercivity of remanence; Hc, coercivity.

Fig. 2. Paleointensity experiments on pallasite olivine. (A) Demagnetizationof NRM of Esquel olivine (black line). (B) Orthogonal vector plot of (A); red isinclination, blue is declination (orientation relative). (C) Thellier-Coe paleo-intensity data, NRM removed versus TRM gained using a 60-mT applied fieldsuggests a paleofield of 110.7 mT. (D) Demagnetization of a laboratory TotalTRM acquired in a 60-mT field [(A), red curve] suggests a paleofield of 118.8 mT

(calculated by comparing values at three temperature steps highlighted bygray boxes). (E to H) Paleointensity data as discussed above on Imilac olivineindicating paleofields of 64.9 mT (Thellier-Coe technique, 60-mT applied field)and 67.3 mT (Total TRMmethod, 30-mT applied field). (I) An oriented section ofthe Esquel meteorite with metal removed. (J to L) Associated demagnetizationresults.

16 NOVEMBER 2012 VOL 338 SCIENCE www.sciencemag.org940

REPORTS

on

Nov

embe

r 16

, 201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

in NRM suggests a taenite carrier (~50 to 55wt % Ni) (21, 22), which is consistent with ourmicroprobe results. Ordering may be limited invery small taenite particles within troilite inclu-sions (23).

Thellier-Coe (23, 24) paleointensity data (Fig.2C) of a typical sample suggest that a TRM wasimparted in a paleofield of 110.7 T 5.2 mT. Tofurther examine the nature of the NRM, we im-parted a Total TRM to the sample by heating at700°C in the presence of a 60-mT field. The de-magnetization curve of the Total TRM is similarto that of the initial NRM (Fig. 2A); small dif-

ferences may indicate minor thermally inducedalteration. Demagnetization of the Total TRMallows for a second estimation of the paleofield;this yields 118.8 T 5.7 mT. Subsamples from twoadditional crystals from the same Esquel meteor-ite sample yield similar values (116.0 T 5.4 mT,109.6 T 7.0 mT, Thellier-Coe method; 115.0 T 6.9mT, 113.4 T 4.0 mT, Total TRM method). As afurther consistency test, we studied a second Esquelpallasite sample. We observed nearly identicaldemagnetization behavior, with Thellier-Coe andTotal TRM paleointensity estimates of 132.4 T5.7 mT and 134.3 T 6.1 mT, respectively, within

~15% of the values obtained from our first ex-periments (tables S1 to S4).

Olivine subsamples from the Imilac pallasiteshow similar behavior (Fig. 2, E to H). Thellier-Coe experiments on two separate samples yield67.9 T 9.2 mTand 79.3 T 7.2 mT (paleointensitiesbased on Total TRM experiments are 67.7 T 6.2and 77.7 T 2.2 mT, respectively). Total TRM ex-periments using two different applied field valuesyield consistent paleointensities (table S2), sug-gesting no applied field dependence.

The unblocking temperatures we have ob-served, viewed in the context provided by ourmicroprobe results, are inconsistent with terres-trial weathering (23). Also, our experiments dem-onstrate that the dominant magnetization is not anartifact of kamacite-taenite interaction discussedin the study of iron meteorites (10). Our paleo-intensity measurements are on unoriented olivinecrystals. In some meteorites, subsamples havebeen found to have different magnetic directions,precluding the acquisition of a TRM after themeteorite mass had assembled (11–13). In con-trast, at unblocking temperatures >360°C, we ob-served consistent directions from oriented pallasiteolivine crystals (Fig. 2, I to L).

The average field value obtained from theEsquel meteorite (122.3 T 14.4 mT, Thellier-Coemethod; 125.2 T 12.9 mT, Total TRM method) issomewhat larger than those observed on Earth’ssurface but somewhat weaker than Earth’s fieldcalculated at the core-mantle boundary (for ex-ample, the radial component was typically 200 to600 mT in 1990) (25). The average value from theImilac meteorite (73.6 T 8.1 mT, Thellier-Coemethod; 72.7 T 7.1 mT, Total TRM method) iscomparable with Earth’s surface field. These rela-tively high intensities suggest an internally gen-erated magnetic field in the pallasite parent bodybecause other sources create fields orders of mag-nitude weaker (13). We interpret these data asrecording dynamo action after the injection ofmetal into the olivine crystals. The fracture path-ways for the metal injection subsequently healed,and the inclusions cooled below the Curie tem-perature of taenite. This injection probably coin-cided with an impact creating the larger-scaleolivine-metal mixing.

The absolute age of the mixing event is un-known, but Mn-Cr systematics provide an oldestage bound of 4.558 billion years ago (26). Fission-track model ages suggest that the magnetizationwe have measured may have set in as late as 4.4to 4.2 billion years ago (27), values that are con-sistent with an early mixing event followed byslow cooling (23).

Our data thus imply that the parent bodymusthave retained a partially liquid iron core (to permita dynamo) until the pallasites cooled to ~360°C,and therefore they cannot have been too closeto the core-mantle boundary. The magnetic evi-dence is consistent with, and independent of, thediversity of main group pallasite cooling ratesthat previously have been used to argue (28)against a core-mantle boundary origin. A liquid

Fig. 3. Spherically symmetric three-layer conductive asteroid cooling model (23). (Left) Evolution oftemperature as a function of radius and time. The model consists of an insulating regolith, a silicatemantle, and a metallic core. The initial condition is 1600 K everywhere. The core remains isothermal(liquid) until it starts to solidify at 1200 K and thereafter cools conductively. The mantle cools conductivelythroughout. The 800 K and 633 K isotherms correspond to taenite diffusion recording cooling rate andthe lowest paleomagnetic unblocking temperature defining the characteristic magnetization, respectively.The horizontal dashed line indicates the core mantle boundary, and the vertical dashed line indicates thetime at which core solidification is complete. (Right) Cooling rate at 800 K as a function of distance. Thedark shaded box indicates the assumed megaregolith thickness (23). The light shaded box is the 2 to 9 Kper million years cooling rate estimate from pallasite metal experiments (28). The solid and dashed linesrepresent model cooling rates with and without a megaregolith, respectively. The core was still convecting(not solid) when the pallasites reached 633 K. So, the pallasites must be shallower than the depthindicated by the dotted line. For a 200-km-radius body, there is a region at radius (r) = ~160 km at whichboth the cooling rate and the paleomagnetic constraint are satisfied.

www.sciencemag.org SCIENCE VOL 338 16 NOVEMBER 2012 941

REPORTS

on

Nov

embe

r 16

, 201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

core requires a temperature exceeding ~1200 K(29), so assuming conductive cooling (23), thepallasites we have investigated were in the top~60% of the protoplanet mantle. Cooling rates at800 K (the diffusion temperature of taenite) inthis depth range in a 200-km-radius body matchestimated pallasite metal cooling rates (28) of 2 to9 K per million years (Fig. 3). Conversely, in alarger 600-km-radius body the pallasites wouldhave to have resided in the near-surface mega-regolith,which is inconsistentwith their unshockedstate, whereas in a smaller 100-km-radius body,the cooling rate is too fast (Fig. 3). Compositionalconvection in the core (14) can drive the dynamo,and impacts can provide additional short-termstirring (30). For a 200-km-radius body, pressureeffects on the magnetization are likely minor (23).These conclusions on parent body size assume thepallasites were not remagnetized during impactheating subsequent to the olive-metal mixingevent. If such reheating occurred, parent bodiesranging from 100- to 200-km radius could satisfythe data, and the pallasites could have formeddeeper in the parent body, within 10% of thecore-mantle boundary. However, we view this asimprobable because such reheating is inconsist-ent with the low observed pallasite shock state (23).

The factor of ~2 difference between Esqueland Imilac paleointensity estimates could indi-cate different positions within the protoplanet. Forinstance, the Esquel and Imilac meteorites couldhave resided at original depths of 40 km and10 km, respectively, within a 200-km-radius body,assuming a dipolar field. In this case, the Curieisotherm of taenite would be reached at 180 mil-lion and 52 million years after the body formedfor the Esquel and Imilac pallasites, respectively(Fig. 3). The heat fluxes at the core at these timesare 33 and 0.8 mWm−2, respectively; the formerat least is sufficient to drive a dynamo if com-positional convection occurs (14). However, thepaleointensity difference could also be explainedby a smaller difference in original depth com-

bined with a time-dependent dynamo field. Inany event, generation of a strong, magnetic fieldby a dynamo at least several tens of millions ofyears after olivine/metal mixing is required byour data.

We recall that the pallasite metal is Ir poor,implicating a fractionated source. This require-ment together with the likely position of the pal-lasites in the protoplanet and the time constraintson when the dynamo was active suggest that thepallasite metal was derived from the liquid ironcore of a differentiated asteroid impactor (fig. S7)that struck before the Curie isothermwas reached.The metal could have been introduced into adunite mantle as dike-like intrusions, similar toimpact melt dikes seen in terrestrial impact struc-tures (31). This mechanism provides a solution tothe pallasite paradox because dikes propagatingthrough relatively cold olivine will undergo an ini-tial phase of rapid cooling, freezing in the olivine-metal pallasite structure, before cooling throughthe taenite Curie temperature. The differentiatedpallasite parent body may have been formed inthe terrestrial planet-forming zone (32). If so, thetiming of dynamo action suggests that the pal-lasite protoplanet was one of the few, late survi-vors in this zone before a cataclysmic collisionthat scattered pallasite fragments from a positioncloser to the Sun outward to the asteroid belt.

References and Notes1. L. Rayleigh, Proc. R. Soc. London Ser. A Math. Phys. Sci.

179, 386 (1942).2. H. C. Urey, Mon. Not. R. Astron. Soc. 131, 199 (1966).3. J. T. Wasson, B. G. Choi, Geochim. Cosmochim. Acta 67,

3079 (2003).4. R. N. Clayton, Space Sci. Rev. 106, 19 (2003).5. E. R. D. Scott, Geochim. Cosmochim. Acta 41, 349 (1977).6. P. R. Buseck, Geochim. Cosmochim. Acta 41, 711 (1977).7. D. W. Mittlefehldt, Earth Planet. Sci. Lett. 51, 29 (1980).8. A. M. Davis, E. J. Olsen, Nature 353, 637 (1991).9. T. Matsui, S. Karato, T. Yokokura, Geophys. Res. Lett. 7,

1007 (1980).10. A. Brecher, L. Albright, J. Geomag. Geoelectr. 29, 379 (1977).11. T. Nagata, Mem. Natl. Inst. Polar Res. Spec. Issue 8, 240

(1978).

12. S. M. Cisowski, in Geomagnetism, J. A. Jacobs, Ed.(Academic Press, New York, 1987), vol. 2, pp. 525–560.

13. B. P. Weiss, J. Gattacceca, S. Stanley, P. Rochette,U. R. Christensen, Space Sci. Rev. 152, 341 (2010).

14. F. Nimmo, Geophys. Res. Lett. 36, L10201 (2009).15. B. P. Weiss et al., Science 322, 713 (2008).16. J. A. Tarduno, R. D. Cottrell, A. V. Smirnov, Rev. Geophys.

44, RG1002 (2006).17. J. A. Tarduno, R. D. Cottrell, M. K. Watkeys, D. Bauch,

Nature 446, 657 (2007).18. J. F. Lovering, L. G. Parry, J. C. Jaeger, Geochim.

Cosmochim. Acta 19, 156 (1960).19. T. Nagata, Phys. Earth Planet. Inter. 20, 324 (1979).20. A. P. Roberts, C. R. Pike, K. L. Verosub, J. Geophys. Res.

105, 28461 (2000).21. D. J. Dunlop, Ö. Özdemir, Rock Magnetism,

Fundamentals and Frontiers (Cambridge Univ. Press,Cambridge, 1997).

22. Y.-Y. Chuang, Y. A. Chang, R. Schmid, J.-C. Lin,Metall. Trans. A 17, 1361 (1986).

23. Materials and methods are available as supplementarymaterials on Science Online.

24. R. S. Coe, J. Geomag. Geoelectr. 19, 157 (1967).25. A. Jackson, A. R. T. Jonkers, M. R. Walker, Philos. Trans. R.

Soc. London A 358, 957 (2000).26. G. W. Lugmair, A. Shukolyukov, Geochim. Cosmochim.

Acta 62, 2863 (1998).27. Y. V. Bondar, V. P. Perelygin, Radiat. Meas. 36, 367

(2003).28. J. Yang, J. I. Goldstein, E. R. D. Scott, Geochim.

Cosmochim. Acta 74, 4471 (2010).29. A. Ghosh, H. Y. McSween Jr., Icarus 134, 187 (1998).30. M. Le Bars, M. A. Wieczorek, Ö. Karatekin, D. Cébron,

M. Laneuville, Nature 479, 215 (2011).31. W. U. Reimold, R. L. Gibson, Chem. Erde 66, 1 (2006).32. W. F. Bottke, D. Nesvorný, R. E. Grimm, A. Morbidelli,

D. P. O’Brien, Nature 439, 821 (2006).

Acknowledgments: We thank J. Hunt for assistance withmicroprobe analyses. This work was supported by NASAgrant NNX11AG66G and NSF grants EAR0619467 andEAR1015269 (to J.A.T.) Paleomagnetic data are includedin the supplementary materials.

Supplementary Materialswww.sciencemag.org/cgi/content/full/338/6109/939/DC1Materials and MethodsFigs. S1 to S7Tables S1 to S5References (33–65)

27 April 2012; accepted 5 October 201210.1126/science.1223932

Evidence for Early HaftedHunting TechnologyJayne Wilkins,1* Benjamin J. Schoville,2 Kyle S. Brown,2,3 Michael Chazan1

Hafting stone points to spears was an important advance in weaponry for early humans. Multiple linesof evidence indicate that ~500,000-year-old stone points from the archaeological site of Kathu Pan1 (KP1), South Africa, functioned as spear tips. KP1 points exhibit fracture types diagnostic of impact.Modification near the base of some points is consistent with hafting. Experimental and metric dataindicate that the points could function well as spear tips. Shape analysis demonstrates that the smallerretouched points are as symmetrical as larger retouched points, which fits expectations for spear tips.The distribution of edge damage is similar to that in an experimental sample of spear tips and isinconsistent with expectations for cutting or scraping tools. Thus, early humans were manufacturinghafted multicomponent tools ~200,000 years earlier than previously thought.

Behavioral traits common to both modernhumans and Neandertals could repre-sent shared traits inherited from their

last common ancestor, commonly held to beHomo heidelbergensis (1, 2). The fossil recordfor H. heidelbergensis begins during the early

Middle Pleistocene, and genetic studies situ-ate the divergence of H. sapiens and Neandertallineages at between ~800 and 400 thousand yearsago (ka) (3). Because Middle Stone Age (MSA)hominins and Neandertals probably both hadstone-tipped hunting equipment, it is possible thatH. heidelbergensis also possessed this form oftechnology.

By ~780 ka, hominins were regularly killinglarge game, based on evidence of repeated in situprocessing of complete carcasses of fallow deer atGesher Benot Ya’kov in Israel (4). At the English

1Department of Anthropology, University of Toronto, 19 RussellStreet, Toronto, Ontario M5S 2S2, Canada. 2Institute of HumanOrigins, School of Human Evolution and Social Change, PostOffice Box 872402, Arizona State University, Tempe, AZ85287-4101, USA. 3Department of Archaeology, University ofCape Town, Rondebosch 7701, South Africa.

*To whom correspondence should be addressed. E-mail:jayne.wilkins@utoronto.ca

16 NOVEMBER 2012 VOL 338 SCIENCE www.sciencemag.org942

REPORTS

on

Nov

embe

r 16

, 201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

www.sciencemag.org/cgi/content/full/338/6109/939/DC1

Supplementary Material for

Evidence for a Dynamo in the Main Group Pallasite Parent Body

John A. Tarduno,* Rory D. Cottrell, Francis Nimmo, Julianna Hopkins, Julia Voronov, Austen Erickson, Eric Blackman, Edward R.D. Scott, Robert McKinley

*To whom correspondence should be addressed. E-mail: john.tarduno@rochester.edu

Published 16 November 2012, Science 338, 939 (2012)

DOI: 10.1126/science.1223932

This PDF file includes:

Materials and Methods

Figs. S1 to S7

Tables S1 to S5

References (33–65)

Tarduno et al., Evidence for a dynamo in the main group pallasite parent body

Supporting Online Material

Materials and Methods

Magnetic hysteresis data were collected using the University of Rochester Princeton Measure-

ments Corporation Alternating Gradient Force Magnetometer. Values for the examples shown in

Fig. 1 of the main text are as follows: Hcr, Hc and Mr/Ms are 154.6 Oe, 200.0 Oe and 0.3911

respectively for the Esquel specimen, and 111.1 Oe, 151.9 Oe and 0.3714, respectively, for the

Imilac specimen. For all remanence measurements we select mm-sized gem-like olivine subsam-

ples, lacking any surface discoloration that might be residual contamination from the surrounding

pallasite metal (we note that our initial tests revealed that samples with visible inclusions from

olivine crystal rims altered rapidly when heated). Obtaining suitable samples generally required

cleaning crystals in distilled water. A weak acid (HCl) was used on some crystals to remove surface

contamination. Remanence measurements were made with a 2G Enterprises 3-component 755R

DC SQUID magnetometer and a 2G small (6.3 mm) bore 3-component DC SQUID magnetometer

in the University of Rochester’s magnetically shielded room (ambient field <200 nT). CO2 laser

heating and cooling was conducted (in air) in additional magnetic shields to produce a magnetically

null environment.

Olivine samples 2-3 millimeters in size were mounted on the end of quartz tubes with Omega

cement (both of which are routinely measured to ensure the blank is in the 10−13 to 10−14 A m2

range). The sample holder also served as the target for CO2 laser heating (the 7 mm diameter laser

beam applied at peak temperature for ∼1 minute ensures uniform heating of the crystal; heatings

at each Thellier-Coe paleointensity step were for 3 minutes). The natural remanent magnetization

of approximately 15% of the clean crystal subsamples measured were in the 10−9 to 10−10 A m2

range; these are the focus of our studies as the magnetizations are well within the measuring

range of the DC SQUID magnetometers throughout the demagnetization procedures. The success

rate for crystals having these intensities (yielded interpretable paleointensity results) was ∼50%.

This compares well with paleointensity success rates from Thellier-Coe experiments on whole-rock

terrestrial basalts, which often average 20% (or less).

Thellier-Coe (24) paleointensity data consist of demagnetization of the NRM (field-off step),

followed by the reheating of the sample at the same temperature in a known applied field (field-on

step). We use orthogonal vector plots of the field-off steps to determine the optimal tempera-

ture range to calculate paleointensities. In this study, we typically use a lowermost Thellier-Coe

unblocking temperature for paleointensity calculation that is slightly higher than the lowest un-

blocking temperature where we believe a primary magnetization is held (i.e. 360 oC). This approach

is conservative, and aimed to avoid any influence of magnetizations held at lower unblocking tem-

peratures. For consistency, we use this same temperature range in determining paleointensity from

Total TRM data (see below), although we note that some minor alteration might be expected given

the cumulative time at elevated temperature.

1

Heatings were minimized by collecting Thellier-Coe paleointensity data only in the temperature

range where orthogonal vector plots show univectorial decay. An applied field of 60 µT was used for

all Thellier-Coe measurements. After NRM demagnetization and collection of Thellier-Coe data,

a Total TRM was applied. Using a CO2 laser, samples were heated to 700 oC and then cooled

in the presence of a field over a 10 minute time span. The Total TRM was subsequently stepwise

demagnetized using the CO2 laser. An applied field of 60 µT was used in the collection of all initial

Total TRM data. After demagnetization of the first Total TRM, subsample Imilac E3 was given

a second Total TRM in the presence of a 30 µT field (and subsequently demagnetized with a CO2

laser) to check for any potential applied field dependence on paleointensity.

To test for consistency in magnetic directions, an oriented section 1-mm thick was prepared.

Metal was etched away, leaving several mutually oriented gem-like olivine crystals, which we sub-

sequently separated (maintaining orientation) and thermally demagnetized using the CO2 laser.

SOM Text

Paleointensity selection criteria. Examples of accepted results are shown in Figure 2 of the main

text. Two additional examples of accepted results are included here (fig. S1). Results of Thellier-

Coe and Total TRM paleointensity experiments are reported in tables S1-2. Values are judged

acceptable if Thellier-Coe paleointensity and Total TRM paleofield estimates are consistent within

15% (see table S2). The uncertainty in the individual Thellier-Coe and Total TRM paleointensity

estimates must be ≤15%

Here we use demagnetization of a Total TRM to assess alteration because it can readily detect

(and in our case exclude) whole-scale transformations with heating seen in some FeNi magnetic

carriers in meteorites (10). Although our heating times using the CO2 laser are very rapid compared

to those of standard ovens used in paleomagnetism, we note that at the end of our experiments a

specimen has still been exposed to elevated temperatures for a cumulative time exceeding 2 hours.

We forgo pTRM checks (33) which, if applied, would have resulted in even longer cumulative times

at elevated temperature. The Total TRM data also aid in the interpretation of magnetizations

observed at high unblocking temperatures. For example, some Esquel olivine specimens acquire

additional partial TRMs after the temperature at which the NRM appears to have been completely

demagnetized. This is expressed as a flattening of NRM/TRM data (Fig. 2C), which in itself

might suggest that a very low (or null) field is recorded at high unblocking temperatures. However,

demagnetization of a Total TRM reveals only a minor TRM in this same temperature interval (Fig.

2D) suggesting that increases in partial TRM at high temperatures reflect either minor alteration

and/or the influence of minor, and more complex, magnetic phases (see discussion in “Minor high

unblocking temperature magnetizations” below).

Several factors contribute to the cause of unsuccessful experiments. The NRM intensity of

some samples decreased rapidly on AF demagnetization to levels after which measurement with

the SQUID magnetometers through an entire paleointensity run was no longer viable. The main

cause of unsuccessful samples that did not display such AF demagnetization characteristics appears

2

to be thermally-induced alteration. This was manifested by either a scattered NRM demagnetiza-

tion pattern (fig S2A,B) and/or a Total TRM curve that differed markedly from that of the NRM

demagnetization (fig S2B,C).

A B C

D

Inte

nsity x

10

-11 A

m2

Imilac

0

10

20

30

40

50

60

70

80

90

100

0 100 200 300 400 500 600 700 800

Total TRM

NRM

Esquel

410

Inte

nsity x

10

-11 A

m2

Temperature oC

360

410W E

N,Up

S,Down

0

5

10

15

0 5 10

500

410

NR

M (

x 1

0-1

1 A

m2)

TRM (x 10-11 A m2)

109.6 μT

0

20

40

60

80

100

350 400 450 500

Inte

nsity x

10

-11 A

m2

Temperature oC

113.4 μT

0

5

10

0 200 400 600 800

Temperature oC

400 500

NR

M (

x 1

0-1

1 A

m2)

TRM (x 10-11 A m2)

57.9 μT

400

500

0

1

0 1

E F G

H

W E

320

N,Up

S, Down

250

400

Temperature oC

Inte

nsity x

10

-11 A

m2

350 400 450 500

59.9 μT

0

1

2

3

Fig. S1. Additional examples of successful paleointensity experiments on pallasite olivine. (A)

Demagnetization of natural remanent magnetization (NRM) of Esquel olivine (black line). (B)

Orthogonal vector plot of (A), red is inclination, blue is declination (orientation relative). (C)

Thellier-Coe paleointensity data, NRM removed versus thermoremanent magnetization (TRM)

gained using a 60 µT applied field suggests a paleofield of 109.6 µT. (D) Demagnetization of a

laboratory Total TRM acquired in a 60 µT field (red curve in (A)) suggests a paleofield of 113.4

µT (calculated by comparing values at three temperature steps highlighted by grey boxes). (E-H)

Paleointensity data as discussed above on Imilac olivine indicating paleofields of 57.9 µT (Thellier-

Coe technique, 60 µT applied field) and 59.9 µT (Total TRM method, 60 µT applied field).

3

A B

C

0

2

4

6

8

10

12

14

N

S

W E

Temperature oC

Inte

nsity (

x10

−11 A

m2)

0 100 200 300 400 500 600 700

N

S

EW

0

10

20

30

40

50

60

70

80

Inte

nsity (

x10

−11 A

m2)

0 100 200 300 400 500 600 700 800

Temperature oC

0

2

4

6

8

10

12

14

Inte

nsity (

x10

−11 A

m2)

Temperature oC

0 100 200 300 400 500 600 700

N

S

W E

Fig. S2. Examples of paleointensity results that did not meet selection criteria. Intensity versus

temperature plots show natural remanent magnetization (NRM) decay (A,B,C) (black) and Total

Thermoremanent Magnetization decay (B,C) (red). Orthogonal vector plots are shown for NRM

demagnetization (red is inclination, blue is declination of relative orientation).

Paleointensity results and averages. Two pallasite meteorites were sampled (Esquel and Imi-

lac). Two thin slabs from each pallasite were available for study (denoted by 1, 2, respectively in

the tables below). Several consistency tests were performed and the results of these tests were in-

corporated into hierarchial averages (tables S3-4) as follows. For Total TRM, paleofield results from

the same crystal measured at different applied field values were averaged (“applied field average”).

Paleofield results from different subsamples from a single olivine crystal were averaged (“crystal

average”). Results from different crystals from a given meteorite sample were averaged (“meteorite

sample estimate”). “Meteorite averages” were determined by averaging the two meteorite sample

estimates available for each meteorite studied.

4

Table S1. Thellier-Coe paleointensity estimates.

Subsample FThC (µT) T (oC) [N] R2 f g

Esquel 1 (green4) 132.4 ±5.7 400-500 [3] 0.92 0.155 0.278

Esquel 2 (19c) 110.7 ±5.2 400-450 [3] 0.98 0.129 0.493

Esquel 2 (3c) 116.0 ±5.4 410-485 [6] 0.98 0.185 0.788

Esquel 2 (4b) 109.6 ±7.0 410-500 [7] 0.99 0.184 0.724

Imilac 1 (F8) 74.4 ±6.7 400-500 [3] 0.92 0.058 0.392

Imilac 1 (E3) 64.9 ±4.5 400-500 [5] 0.98 0.059 0.742

Imilac 1 (E7) 57.9 ±7.8 400-500 [5] 0.97 0.031 0.707

Imilac 2 (G9)* 82.1 ±6.3 425-520 [5] 0.98 0.038 0.739

Imilac 2 (G12) 79.3 ±7.2 400-520 [6] 0.94 0.081 0.726

Abbreviations: FThC , Thellier-Coe field value with 1σ uncertainty; T , temperature range of fit; N ,

number of temperature steps used in fit; f, g are fraction of NRM fit and gap factor, respectively,

from (33). ∗ Sample omitted from averages because of high Total TRM paleointensity uncertainty

(see table S2).

Table S2. Total TRM paleointensity estimates.

Subsample FTTRM (µT) T (oC) [N] ∆FTTRM−FThC%

Esquel 1 (green4) 134.3 ±6.1 400-500 [3] 1

Esquel 2 (19c) 118.8 ±5.7 400-450 [3] 7

Esquel 2 (3c) 115.9 ±6.8 410-485 [3] <-1

Esquel 2 (4b) 113.4 ±4.0 410-500 [3] 3

Imilac 1 (F8) 72.1 ±1.0 400-500 [3] -3

Imilac 1 (E3)† 65.9 ±4.4 400-500 [3] 2

Imilac 1 (E3)‡ 67.3 ±3.4 400-500 [3] 4

Imilac 1 (E7) 59.9 ±1.0 400-500 [3] 3

Imilac 2 (G9)* 84.5 ±15.5 425-520 [3] 3

Imilac 2 (G12) 77.7 ±2.2 400-520 [3] - 2

Abbreviations: FTTRM , Total TRM field value estimate with 1σ uncertainty; T , temperature range

of fit; N , number of temperature steps used in fit; ∆FTTRM−FThC, difference between Total TRM

and Thellier-Coe paleointensity estimates, expressed as percent of the Thellier-Coe value. † 60 µT

applied field; ‡ 30 µT applied field. ∗ Sample omitted from averages because of high Total TRM

paleointensity uncertainty.

5

Table S3. Thellier-Coe hierarchical paleointensity averages.

Subsample FThC (µT) Crystal Meteorite sample estimate Meteorite

average (µT) (µT) average (µT)

Esquel 1

green4 132.4 ±5.7 132.4 Esquel

Esquel 2 122.3 ±14.4

19c 110.7 ±5.2 112.1 ±3.4 (N=2)

3c 116.0 ±5.4 (N=3)

4b 109.6 ±7.0

Imilac 1

F8 74.4 ±6.7 67.9 ±9.2 Imilac

E3 64.9 ±4.5 61.4 ±4.9 (N=2) 73.6 ±8.1

E7 57.9 ±7.8 (N=2) (N=2)

Imilac 2

G12 79.3 ±7.2 79.3

Abbreviations: FThC , Thellier-Coe field value. All averages shown with 1σ uncertainty.

Table S4. Total TRM hierarchical paleointensity averages.

Subsample FTTRM (µT) Applied field Crystal Meteorite sample Meteorite

average (µT) average (µT) estimate (µT) average (µT)

Esquel 1

green4 134.3 ±6.1 134.3 Esquel

Esquel 2 125.2 ±12.9

19c 118.8 ±5.7 116.0 ±2.7 (N=2)

3c 115.9 ±6.8 (N=3)

4b 113.4 ±4.0

Imilac 1

F8 72.1 ±1.0 67.7 ±6.2

E3† 65.9 ±4.4 66.6 ±1.0 (N=2)

E3‡ 67.3 ±3.4 (N=2) 63.3 ±4.7

E7 59.9 ±1.0 (N=2) Imilac

Imilac 2 72.7 ±7.1

G12 77.7 ±2.2 77.7 (N=2)

Abbreviations: FTTRM , Total TRM field value estimate. All averages shown with 1σ uncertainty.† 60 µT applied field; ‡ 30 µT applied field.

6

Minor high unblocking temperature magnetizations. Although the dominant natural re-

manent magnetization is removed by thermal demagnetization between 360 and 500 oC, consistent

with a taenite carrier, we note there is a very small signal (1-5% of the NRM) at demagnetization

temperatures >500 oC in some samples. On the basis of microprobe analyses (discussed below) and

potential unblocking temperatures, we consider these small signals to be carried by a fine-grained

mixture of taenite and kamacite. We further note that some samples show a small NRM and To-

tal TRM remanence increase (and subsequent decrease) at thermal demagnetization temperatures

>500 oC (cf Figure 2). This increase generally occurs over a restricted temperature range (∼100oC), but its exact initiation temperature varies between samples. We interpret this as reflecting ex-

change interaction between fine-grained taenite and kamacite. Because these are very minor phases

compared to the bulk magnetization, this interaction is not apparent in FORC diagrams. We also

note that small amounts of tetrataenite could be recorded at these high unblocking temperatures.

However, the reproducibility of the intensity increase seen in demagnetization of a Total TRM (see

Figure 2e) indicates that tetrataenite cannot be solely responsible for these minor magnetizations

because tetrataenite should not have survived heating to 700 oC (i.e. the temperature at which the

Total TRM was applied).

Terrestrial weathering. Unblocking temperatures similar (but not identical) to those reported

in our study have been reported by Uehara et al. (34) in weathered chondrite meteorites and

interpreted to reflect maghemite and substituted magnetite formed during terrestrial weathering,

resulting in a terrestrial magnetization overprinting an extraterrestrial signal. This was not the

case for chondrites with no or little weathering. Maghemite generally inverts after heating above

250 oC (21), and this results in irreversible magnetic behavior; this was not observed in our thermal

demagnetization experiments. Moreover, evidence for maghemite or a substituted magnetite phase

was not found during our SEM or microprobe analyses (detailed below), whereas clear evidence

for FeNi particles was identified. However, we emphasize that our analyses have been restricted to

gem-like olivine particles. Our meteorite samples were selected to have minimal weathering. Al-

though not studied here, we predict that weathered pallasite olivines do contain magnetic minerals

formed during terrestrial weathering.

SEM and Microprobe analyses of FeNi particles. Scanning electron microscopy (SEM)

analyses were conducted using a Zeiss SUPRA 40VP with EDAX spectrometer at the University

of Rochester. SEM analyses reveal FeNi inclusions that are potential remanence recorders. These

are similar to those reported in some prior studies (35-36) but differ from the tubular symplectic

inclusions studied in the Fukang pallasite (37). We observed some Cr-rich inclusions, but these

are not candidates for the major NRM carrier which demagnetizes between 360 and 500 oC. SEM

analyses of an olivine inclusion that is a candidate remanence carrier from the Esquel meteorite is

shown in fig. S3.

7

Maps of inclusion 7 in crystal D2

FeK FeL

Mg

NiL NiK

OSi

S

C

Esquel - Crystal D2, Inclusion 7

Fig. S3. SEM analyses of an inclusion in olivine of the Esquel pallasite meteorite. EDAX K and

L shell shell composition maps are shown for Fe and Ni.

EDAX spectra show an absence of Si, Mg and O, indicating that the inclusion is distinct from the

olivine matrix. Sulfur-rich regions (darker grey areas of the inclusion in the SEM image) separate

concentrations of FeNi within the inclusion.

Compositions of inclusions were further explored using a JEOL 8900 electron microprobe at

Cornell University with an accelerating voltage of 8 KeV to obtain ∼0.5 micron resolution. Electron

microprobe results reveal FeNi compositions within the inclusion (fig. S4). A pentlandite (Fe,

Ni)9S8 standard from Manibridge, Ontario (weight percentages S: 33.01, Fe: 30.77, Co: 0.10, Ni:

36.12) was used for these analyses. Total weight percentages less than 100% in the analyses plotted

reflect the presence of elements other than Fe and Ni (mostly S). The compositions of Ni-rich

particles overlap with those of the ordered FeNi mineral tetrataenite. However, the dominant

8

changes in NRM intensity do not match the characteristic magnetic decay pattern related to the

∼550 oC Curie temperature of tetrataenite (38) (see also fig. S5). In addition, as noted above, the

reproducibility exhibited by the Total TRM data (Figure 2, tables S2, S4) are inconsistent with

tetrataenite.

Esquel - Crystal D2 Inclusion 7 - Microprobe Analysis

2µm

A

D

F

B

E

G

C

H

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

11000

12000

Coun

ts

0 1 2 3 4 5 6 7 8

keV

Fe

C

S FeNi AlMg

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

11000

Coun

ts

0 1 2 3 4 5 6 7 8

keV

Ni

Fe

C

SAlFeSi

O

0

1000

2000

3000

4000

5000

Coun

ts

0 1 2 3 4 5 6 7 8

keV

Fe

C

S FeNi

0

1000

2000

3000

4000

5000

6000

7000

8000

Coun

ts

0 1 2 3 4 5 6 7 8

keV

Fe

C

S

MgFe

S

S

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

11000

Coun

ts

0 1 2 3 4 5 6 7 8

keV

Ni

Fe

C

SAl FeMg Si

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

Coun

ts

0 1 2 3 4 5 6 7 8

keV

Ni

Fe

C

SAl FeMg Si

Ni

Fe

CS

Al FeO

0

1000

2000

3000

4000

5000

6000

7000

8000

Coun

ts

0 1 2 3 4 5 6 7 8

keV

Fe

C

S Fe

Ni

AlO

Ca0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

11000

12000

Coun

ts

0 1 2 3 4 5 6 7 8

keV

Ni Fe Total Weight 57.022 42.845 99.867Atomic 55.8707 44.1293 100

Ni Fe Total Weight 58.136 42.209 100.345Atomic 56.7148 43.2852 100

Ni Fe Total Weight 52.25 44.877 97.127Atomic 52.5523 47.4477 100

Ni Fe Total Weight 3.237 101.377 104.614Atomic 2.9478 97.0522 100

Ni Fe Total Weight 3.015 99.185 102.2Atomic 2.8103 97.1897 100

Ni Fe Total Weight 52.702 41.762 94.464Atomic 54.5557 45.4443 100

Ni Fe Total Weight 0.126 45.685 45.811Atomic 0.2622 99.7378 100

Fig. S4. Electron microprobe results for Esquel olivine (same inclusion shown in Fig. S4).

Brightest areas (B,F,G,H) are high Ni, surrounded by Fe sulfide (A). Slightly less bright areas are

low Ni, FeNi metal (C,D). Area (E) likely samples region of several adjacent grains, but nonetheless

demonstrates S between region of low and high Ni.

9

0

20

40

60

80

0 200 400 600 800

Inte

nsity

x10

-10

A m

2

Temperature oC

Observed NRM

Observed Total TRM

400

400

Predicted tetrataenitedemagnetization

Fig. S5. Predicted tetrataenite NRM demagnetization behavior versus observed demagnetization

of NRM and total TRM of an Esquel pallasite olivine subsample (cf. Figure 2). Note the NRM

value of the hypothetical tetrataenite component is arbitrary, and the decay qualitative, drawn

after the behavior reported by Wasilewski (38).

We interpret the dominant magnetic recorder in inclusions within the pallasite olivine as

metastable, disordered taenite. We note that this conclusion differs from the result of a thermo-

magnetic analysis of the “silicate” (presumable olivine) part of a pallasite (Yamato-74044) reported

in an early study by Nagata (11) which is consistent with kamacite. We note, however, that our

analyses have been restricted to gem-like olivines lacking large inclusions. It is possible that con-

tamination from the metal surrounding the olivine (especially near the olivine rim) was responsible

for the signal reported by Nagata (11).

Given the slow cooling recorded by the massive FeNi surrounding the olivine (28), we might ex-

pect taenite to have ordered. However, the high-Ni magnetic carriers in olivine have a very different

origin from the tetrataenite rims in the metallic matrix as the latter formed by low temperature

equilibration between FeNi phases whereas the former are minor components in FeS inclusions that

are predominantly composed of FeS. We envisage that the Fe-Ni-S inclusions were injected as a

melt into cracks [perhaps at ∼900 oC, (39)] which were sealed up on cooling. We speculate that

minor Fe exchange between the FeNi and FeS phases during slow cooling may inhibit ordering in

very small particles such as those that characterize the olivine magnetic inclusions.

Pressure effects on magnetization. Hydrostatic pressure can potentially affect magnetizations

(21). Turcotte and Schubert [(40), eq. 2-71] give the pressure as a function of radial position r

inside a homogeneous body of density ρ:

P =2

3πρ2G(R2 − r2) (1)

where R is the radius of the body and G is the gravitational constant. Taking our estimates of the

size of the pallasite parent body, R=200 km, ρ=3 g/cc, we obtain pressures of 9.6 MPa at r =180 km

10

and 22 MPa at r=150 km. Although detailed pressure experiments are not available on sub-micron

sized FeNi particles at these estimated pressures, we note that our estimates of pressures are much

lower than the value at which substantial changes are observed in magnetite and titanomagnetite

[e.g. (41-42)] and FeNi (43). Shock can induce much larger pressures. However, beyond the event

that caused olivine-metal mixing and break-up of the parent body, pallasites do not appear to

have been affected by the massive shocks that characterize other meteorites. For example, pallasite

metal does not show the characteristic hatched pattern seen in shocked IIIAB iron meteorites (44 ),

and dislocation densities are low in both olivine and metal [e.g. (9, 45)].

Martelli and Newton (46) studied shock associated with a hypervelocity projectile (15 km/s)

cratering a terrestrial basalt and proposed that a shock magnetization comparable to a TRM could

be produced. In this case, the shock magnetization was thought to be related to a plasma formed

by the impact. Because the pallasites we have studied lack massive shock features, we believe they

were not near the surface at the time of the impact that disrupted the parent body. In this case, the

more relevant far-field remagnetization process is shock-induced crystallographic transformation.

Dickinson and Wasilewski (47) found that this process normally induces a magnetization much less

than that recorded by our TRM experiments.

Absolute ages of metal-olivine mixing. Lugmair and Shukolyukov (26) report 53Mn-53Cr

systematics on main group pallasites Omolon and Springwater, suggesting ages of 4558.0 ± 1.0 Ma

and ∼4557 Ma, respectively. These ages are consistent with those reported from Re-Os systematics

by Shen et el. (48) (4.60± 0.5 Ga), who interpret the time scale for pallasite fractional crystalization

as spanning 10-20 Myr. These ages provide an upper bound (i.e. oldest age) on the time of metal-

olivine mixing.

In an early K-Ar study, Megrue (49) reported ages of 4.3 Ga for 3 pallasites (Krasnojarsk,

Marjalahti, and Springwater). A much younger 40Ar-39Ar age of 0.86 Ga has been reported for

the non-main group pallasite Eagle Station by Niemeyer (50), and interpreted as recording an

impact of that age. But that study was conducted on weathered olivine [(50), p. 1007] and the age

probably reflects terrestrial processes. Similarly Megrue (49), noted that an apparent young Kr-Ar

age of 0.44 Ga for the Admire pallasite could reflect “diffusion of radiogenic Ar40” or “potassium

contamination of the sample by terrestrial weathering”.

Bondar and Perelygin (27) reported model fission track ages of 4.37 Ga ± 0.01 Ga for the

Marjalahti pallasite, 4.19 ± 0.02 Ga for the Omolon pallasite, 4.18 ± 0.03 Ga for the Bragin pall-

asite and 4.21 ± 0.02 for the Krasnoyarsk pallasite. Given that the temperature for the complete

removal of tracks is 450-500 oC (27), these ages could be close to the time of acquisition of the

magnetizations we have recorded. However, such an assignment depends critically on the accuracy

of the fission track age model. Nevertheless, these ages provide a lower bound (i.e. youngest age)

on the time of metal-olivine mixing.

Thermal modeling. Given the uncertainties in even such basic parameters as its size, we adopt

11

a very simple model to track the thermal evolution of the pallasite parent body. This model re-

sembles those of previous investigations [e.g. (29, 51-53)]; one important complication, however,

is that it includes an insulating, near-surface megaregolith (54-55). Although, especially for the

larger bodies, mantle convection or advection via melt might be important in the early stages, we

simply assume spherically-symmetric conductive transfer of heat:

∂T

∂t=

1

r2∂

∂r

(κr2

∂T

∂r

)(2)

where T (r) is temperature, r is radial position, and κ is thermal diffusivity. We neglect internal

heat production because of the ∼100 Myr timescale of greatest interest to us. Since this timescale

is long compared to the decay of 26Al or 60Fe, the effect of this early heating is incorporated into

our initial conditions. Conversely, over the timescale of interest the decay of long-lived radiogenic

elements like 40K will not qualitatively change our conclusions.

The temperature at the surface (r = R) is kept at a fixed value Ts. The temperature gradient

at r = 0 is zero. We assume an isothermal initial condition (T=1600 K) in which the mantle is

solid but the core is liquid. We solve equation (2) numerically using a centered finite-difference

scheme.

The near-surface megaregolith layer is assumed to have a thermal diffusivity one-tenth that of

the solid rock (54). Based on Fig. 12 of Warren (55), we assume megaregolith thicknesses of 5, 8

and 24 km for bodies of radii 100, 200 and 600 km, respectively.

The megaregolith is important not only as an insulator, but also because it is an unlikely source

of the pallasites (which would have experienced shock and mixing with other silicate phases, neither

of which are observed). We consider large (600 km) objects as unlikely sources of pallasites because

the cooling rate data would require them to have originated within the megaregolith layer (see main

text). We also note that it is possible to ultimately disrupt the parent body by impact without

widespread heating (56).

If core solidification has not yet finished, we only solve equation (2) within the mantle and take

the temperature at the core-mantle boundary (r = Rc) as our boundary condition. The core is

assumed isothermal (due to vigorous convection) and its bulk temperature is updated based on the

heat extracted into the mantle (4πR2ckm

∂T∂r |r=Rc ∆t ) during that timestep ∆t , where km is the

mantle thermal conductivity. We take Rc = R/2.

We assume that the core solidifies at a single temperature Tl, here taken to be 1200 K. If T = Tl

during a particular timestep, the core temperature then remains pinned at that temperature until

the total latent heat of the core (43πR3cρcLc) has been extracted into the mantle via conduction.

Here Lc is the specific latent heat of the core material and ρc its density.

Once the core has completely solidified, we then proceed to solve equation (2) as before, but

now the bottom boundary condition is to set the temperature gradient to zero at r=0. Note that

the thermal diffusivities κm and κc for mantle and core respectively are quite different.

We use a grid spacing ∆r =R/100 and adopt a constant timestep governed by the Courant

criterion: ∆t =0.3∆r2/κc, where we use the core diffusivity because it is more restrictive. We

12

verified that our numerical model reproduces the analytical solutions for a uniform sphere cooling

from an isothermal initial state (57) when the core radius is zero. Parameter values adopted are

given in table S5.

This model ignores many details, such as the fact that core solidification is likely a multi-phase

process which involves a progressive reduction in solidus temperature as solidification proceeds.

Nonetheless, the conclusions reached in the main text are unlikely to be significantly affected by

such uncertainties. This is because the key results - cooling rate and liquid core lifetime - are

both primarily dependent on the size of the parent body, while a low blocking temperature implies

relatively shallow depths, irrespective of the details of the calculations.

In the main text we focus on the 2-9 K/Myr cooling data for main group pallasites obtained by

Yang et al. (28). We do not consider one higher rate reported in that study (18 K/Myr ± 9 K/Myr)

because of its high uncertainty; it also appears to be an outlier in cooling rate, and in cloudy zone

particle size versus tetrataenite bandwidth (Fig. 7 of Yang et al., (28)). Ito and Ganguly (58)

calculate a cooling rate of 20-40 K/Myr at 1000◦C on the basis of a thermochronological model

applied to 53Mn-53Cr age data on olivine from the Omolon pallasite (26). For our 200 km radius

model object, this cooling rate is obtained over a depth range of 6-20 km, which is compatible with

the results shown in Fig. 3 of the main text.

Because we envisage the metallic injections as being relatively narrow, dike-like features, they

will cool rapidly to attain the background temperature of the surrounding material. Their subse-

quent cooling history (including passage through the Curie temperature) will then simply be that

of the surrounding material.

We also considered the possible role of impact heating during the injection event, but concluded

that it was likely to have negligible effect on the overall cooling, as follows. Early in the history of the

asteroid belt, impact velocities were probably comparable to escape velocities. The escape velocity

is roughly 200 m/s (R/100 km) where R is the radius of the body. So for bodies a few hundred

km in radius, impact velocities will be lower than the sound speed of intact rock (a few km/s), i.e.

the impacts are subsonic. As a result, the impact heating (and associated shock pressures) will be

small. If all the kinetic energy of a 200 m/s projectile were contained within one projectile radius,

the temperature increase would only be about 20 K and the actual temperature change would be

even less, because the heat is in reality more broadly distributed.

Table S5. Thermal modeling parameters, based on Table 1 of Sahijpal et al. (52).

Quantity Value Units Quantity Value Units

Ts 250 K Tl 1213 K

ρm 3000 kg m−3 ρc 7800 kg m−3

κm 5× 10−7 m2 s−1 κc 5× 10−6 m2 s−1

Lc 0.27 MJ kg−1 km 3 Wm−1K−1

13

Late impact heating and remagnetization. Many HED meteorites have 39Ar-40Ar ages that

suggest impact-induced heating (59-60) to temperatures capable of partially (or completely) reset-

ting a magnetization that might have been acquired during initial cooling. Some of these impact

heating events occur as late as ∼3.4 Ga, at a time when dynamo action might have ceased on the

HED parent body (Vesta) due to core solidification. If the pallasite meteorites we have studied

were reheated in a similar manner, after dynamo cessation on the pallasite parent body, it would

affect our conclusions. Here, we first consider how our conclusions would be affected, and then

explore whether impact reheating is consistent with available constraints.

If the pallasites were reheated after dynamo action ceased, an ambient magnetic field still must

have been present to account for the magnetizations we have recorded. A potential source of this

ambient field is the entire parent body mantle and crust, which if magnetized during the dynamo

epoch would create a global remanent field. The relevant magnetic carrier of the magnetization

would be kamacite, since its high Curie point (approximately 765 oC, (21)) is consistent with the

idea that it could retain a magnetization even if heated to temperatures sufficient to reset the

magnetizations we have reported from the pallasites.

In this case, the constraints on the pallasite parent body would come from the cooling rates of

the pallasite metal and the inferred magnetization in the parent body mantle and crust (i.e. not

directly from the magnetization of the pallasite olivine, which in this scenario is reset by impact

heating and records only an echo of the dynamo). Given these new constraints, a 600 km radius

body can still be excluded because the pallasites would have to reside in the regolith. However,

we would have less resolution on the position of the pallasites within the parent body and on the

parent body size. Specifically, the pallasites could have formed deeper in the parent body, within

10% of the core-mantle boundary, and parent bodies ranging in size from 100 to 200 km radius

could satisfy the constraints.

Paleomagnetic directional and paleointensity data from the pallasite olivine are linear between

360 oC and 500 oC (Fig 2). It is unlikely that the ambient magnetic field during a hypothetical late

heating event induced by impact, and the field during initial cooling would be exactly the same.

Hence, following Thelliers’ laws of pTRM additivity (21), if the hypothetical late heating event

reached temperatures greater than 360 oC, but less than 500 oC, a break in slope would be seen in

the paleointensity and/or directional data. Because this is not observed, we conclude that any late

stage heating must have reached a temperature close to 500 oC to be compatible with the magnetic

data.

As noted above, a low velocity impact is insufficient to generate a temperature change great

enough to affect the magnetization. High-velocity impacts create transient high pressures; as the

material subsequently unloads along an adiabat, heating results with higher peak pressures causing

more heating (61). To quantify this effect we used the same approach and parameters as in Barnhart

et al. (62), Appendix A2. We calculate the peak pressure and temperature increase ∆T experienced

directly below the impactor. The results are plotted in Fig S6. As expected, there is an almost

linear relationship between peak pressure and temperature increase. A temperature increase to 500

14

oC is required to reset the magnetization. This implies temperature increases ranging from 300 K

for an impact shortly after core solidification to greater than 400 K for impacts >1 billion years

later. This heating range implies shock pressures ranging from 4.5 GPa to greater than 5.5 GPa

(∼45-55 kbar).

Studies of olivine dislocation density in main group pallasites limit shock pressure to a few

tens of bars (9). These shock levels are incompatible with the much higher shock (i.e. GPa’s)

that would accompany impacts needed to produce heating sufficient to reset the magnetizations we

have recorded. These studies were conducted only on the Admire, Brenham and Dora pallasites.

But metal macro- and microstructure is also sensitive to shock and shock-induced heating. For

example, shock up to 1 GPa can cause Neumann band (mechanical twins) in kamacite, and cloudy

taenite intergrowth can be completely lost upon short term heating (years) at 500 oC (63). Metal

microstructure has been examined in detail for the Esquel and Imilac pallasites (28) and no evidence

for shock has been reported.

Because relatively high shock accompanies impact heating (Fig. S6), and there is no evidence

of shock at these levels in the pallasites following the olivine-metal mixing event, we conclude that

resetting of the pallasite magnetization by impacts is unlikely. We therefore predict that any future

Ar-Ar or fission track dating efforts on the Esquel and Imilac pallasites should yield ages at least

as old as the 4.4-4.2 Ga ages (27) obtained from other main group pallasites.

Figure S6. Shock pressure and temperature increase as a function of impact velocity vi, calcu-

lated using the approach of Barnhart et al. ((62), Appendix A2). We assumed target and projectile

densities of 2.6 g/cc. Other parameters are as given in Barnhart et al. (62).

15

Model for parent body evolution. Our model for the origin of pallasites and the nature and

evolution of their parent body is based on paleomagnetic constraints, cooling rate data (28), thermal

modeling, and previously published geochemical analyses. We envision the impact of a differen-

tiated body with a partially solidified core into a larger differentiated body having a liquid core.

Liquid FeNi from the impactor intrudes into target dunite mantle in the form of dikes, which result

in the formation of the characteristic pallasite compositions. This source of FeNi metal satisfies

Ir constraints (3). The mantle and crust of the impactor are lost and/or incorporated into the

shallowest levels of the target. A similar impact scenario for the generalized case of planetesimal

accretion is presented by Asphaug et al. [(64) Figures 6-7, panels d-f of that paper]. It is also pos-

sible that the impactor had a very thin mantle/core, especially if it had seen prior impacts. There

are more complex scenarios that might meet the available constraints, but our central conclusions

are that the main group pallasites cooled in a body with a core dynamo and that the metal in

the pallasites did not come from that core. Our preferred model (fig. S7) is arguably the simplest

explanation of these findings.

Fig. S7 (next page). Our model for the origin of pallasites and the nature and evolution

of their parent body is based on paleomagnetic constraints and thermal modeling. (A) Liquid

FeNi from an impactor is injected into the mantle of the parent body. (B) Enlargement of boxed

region in (A) shows injection of metal in dike-like intrusions in dunite mantle, which result in the

pallasite texture (C) During the dike intrusion, metal is injected into fractures in the olivine (D);

these fractures subsequently heal upon cooling. Although we illustrate one impact, several similar

impacts might have occurred, and they together might explain the scatter in Ir values seen in main

group pallasites [e.g. (65)]. (E) Heat flux and temperature versus time. At a reference of 30 km

depth, the lower range of taenite Curie temperatures (estimated from the observed paleomagnetic

unblocking temperatures) will be reached at ∼137 million years after initial cooling. Core heat

flux is sufficient at this time to drive a dynamo if compositional convection occurs (14). (F) After

cooling below the lower Curie temperatures of taenite, collision of the pallasite parent body with

a similarly-sized protoplanet occurred, possibly in the terrestrial planet-forming zone [e.g. (32)],

destroying the pallasite parent body. (G) Pallasite meteorites were later scattered by interaction

with a protoplanet.

16

mag

netic

field

magnetic

field

0

50

100

T

ime,

Myr

150

200

250

300

2

0 2

00

4

0 4

00

6

0 6

00

8

0

coo

lin

g r

ate

reco

rded

b

lock

ing

tem

per

atu

re

8

00

-2 Heat flux out of core, mWm

100

C

ore

hea

t fl

ux

core

so

lid

ific

atio

n

star

ts

core

so

lid

ific

atio

n

en

ds

T

emp

erat

ure

(3

0 k

m d

epth

) 1000

120

Temperature, K

1200

~1

00

km

~1

cm

~1

0 μ

m

FG

BA

C

E

Fig

S7

D

17

References and Notes 1. L. Rayleigh, The stone-iron meteorites called pallasites: A synthetic study of their structure

and probable mode of formation. Proc. R. Soc. Lond. A Math. Phys. Sci. 179, 386 (1942). doi:10.1098/rspa.1942.0012

2. H. C. Urey, Chemical evidence relative to the origin of the solar system. Mon. Not. R. Astron. Soc. 131, 199 (1966).

3. J. T. Wasson, B. G. Choi, Main-group pallasites: Chemical composition, relationship to IIIAB irons and origin. Geochim. Cosmochim. Acta 67, 3079 (2003). doi:10.1016/S0016-7037(03)00306-5

4. R. N. Clayton, Oxygen isotopes in the solar system. Space Sci. Rev. 106, 19 (2003). doi:10.1023/A:1024669116828

5. E. R. D. Scott, Pallasites- Metal composition, classification and relationships with iron meteorites. Geochim. Cosmochim. Acta 41, 349 (1977). doi:10.1016/0016-7037(77)90262-9

6. P. R. Buseck, Pallasite meteorites—Mineralogy, petrology and geochemistry. Geochim. Cosmochim. Acta 41, 711 (1977). doi:10.1016/0016-7037(77)90044-8

7. D. W. Mittlefehldt, The composition of mesosiderite olivine clasts and implications for the origin of pallasites. Earth Planet. Sci. Lett. 51, 29 (1980). doi:10.1016/0012-821X(80)90254-X

8. A. M. Davis, E. J. Olsen, Phosphates in pallasite meteorites as probes of mantle processes in small planetary bodies. Nature 353, 637 (1991). doi:10.1038/353637a0

9. T. Matsui, S. Karato, T. Yokokura, Dislocation structures of olivine from pallasite meteorites. Geophys. Res. Lett. 7, 1007 (1980). doi:10.1029/GL007i011p01007

10. A. Brecher, L. Albright, The thermoremance hypothesis and the origin of magnetization in iron meteorites. J. Geomag. Geoelectr. 29, 379 (1977). doi:10.5636/jgg.29.379

11. T. Nagata, Magnetic properties of an iron meteorite (Yamato-75031) and a pallasite (Yamato-74044). Mem. Natl. Inst. Polar Res. Spec. Issue 8, 240 (1978).

12. S. M. Cisowski, in Geomagnetism J.A. Jacobs, Ed. (Academic Press, N.Y., 1987), vol. 2, pp. 525–560.

13. B. P. Weiss, J. Gattacceca, S. Stanley, P. Rochette, U. R. Christensen, Paleomagnetic records of meteorites and early planetesimal differentiation. Space Sci. Rev. 152, 341 (2010). doi:10.1007/s11214-009-9580-z

14. F. Nimmo, Energetics of asteroid dynamos and the role of compositional convection. Geophys. Res. Lett. 36, L10201 (2009). doi:10.1029/2009GL037997

15. B. P. Weiss et al., Magnetism on the angrite parent body and the early differentiation of planetesimals. Science 322, 713 (2008). doi:10.1126/science.1162459 Medline

16. J. A. Tarduno, R. D. Cottrell, A. V. Smirnov, The paleomagnetism of single silicate crystals: Recording the geomagnetic field during mixed polarity intervals. Rev. Geophys. 44, RG1002 (2006). doi:10.1029/2005RG000189

1

17. J. A. Tarduno, R. D. Cottrell, M. K. Watkeys, D. Bauch, Geomagnetic field strength 3.2 billion years ago recorded by single silicate crystals. Nature 446, 657 (2007). doi:10.1038/nature05667 Medline

18. J. F. Lovering, L. G. Parry, J. C. Jaeger, Temperatures and mass losses in iron meteorites during ablation in the earth's atmosphere. Geochim. Cosmochim. Acta 19, 156 (1960). doi:10.1016/0016-7037(60)90002-8

19. T. Nagata, Meteorite magnetism and the early solar system magnetic field. Phys. Earth Planet. Inter. 20, 324 (1979). doi:10.1016/0031-9201(79)90055-4

20. A. P. Roberts, C. R. Pike, K. L. Verosub, FORC diagrams: A new tool for characterizing the magnetic properties of natural samples. J. Geophys. Res. 105, 28461 (2000). doi:10.1029/2000JB900326

21. D. J. Dunlop, Ö. Özdemir, Rock Magnetism, Fundamentals and frontiers (Cambridge Univ. Press, Cambridge, 1997).

22. Y.-Y. Chuang, Y. A. Chang, R. Schmid, J.-C. Lin, Magnetic contributions to the thermodynamic functions of alloys and the phase equilibria of Fe-Ni system below 1200 K. Metall. Trans. A 17, 1361 (1986). doi:10.1007/BF02650117

23. Materials and methods are available as supplementary materials on Science Online.

24. R. S. Coe, The determination of paleointensities of the Earth’s magnetic field with emphasis on mechanisms which could cause non-ideal behavior in Thellier’s method. J. Geomag. Geoelectr. 19, 157 (1967). doi:10.5636/jgg.19.157

25. A. Jackson, A. R. T. Jonkers, M. R. Walker, Four centuries of geomagnetic secular variation from historic records. Philos. Trans. R. Soc. Lond. A 358, 957 (2000). doi:10.1098/rsta.2000.0569

26. G. W. Lugmair, A. Shukolyukov, Early solar system timescales according to 53Mn-53Cr systematics. Geochim. Cosmochim. Acta 62, 2863 (1998). doi:10.1016/S0016-7037(98)00189-6

27. Y. V. Bondar, V. P. Perelygin, Cosmic history of some pallasites based on fossil track studies. Radiat. Meas. 36, 367 (2003). doi:10.1016/S1350-4487(03)00153-7

28. J. Yang, J. I. Goldstein, E. R. D. Scott, Main-group pallasites: History, relationship to IIIAB irons, and origin. Geochim. Cosmochim. Acta 74, 4471 (2010). doi:10.1016/j.gca.2010.04.016

29. A. Ghosh, H. Y. McSween, Jr., A thermal model for the differentiation of asteroid 4 Vesta based on radiogenic heating. Icarus 134, 187 (1998). doi:10.1006/icar.1998.5956

30. M. Le Bars, M. A. Wieczorek, Ö. Karatekin, D. Cébron, M. Laneuville, An impact-driven dynamo for the early Moon. Nature 479, 215 (2011). doi:10.1038/nature10565 Medline

31. W. U. Reimold, R. L. Gibson, The melt rocks of the Vredefort impact structure—Vredefort granophyre and pseudotachylitic breccias: Implications for impact cratering and the evolution of the Witwatersrand Basin. Chem. Erde 66, 1 (2006). doi:10.1016/j.chemer.2005.07.003

2

32. W. F. Bottke, D. Nesvorný, R. E. Grimm, A. Morbidelli, D. P. O’Brien, Iron meteorites as remnants of planetesimals formed in the terrestrial planet region. Nature 439, 821 (2006). doi:10.1038/nature04536 Medline

33. R. S. Coe, C.S. Gromm_e, E.A. Mankinen, Geomagnetic paleointensities by the Thelliers’ method from radiocarbon dated lava ows on Hawaii: The Pacific non-dipole low. J. Geophys. Res. 83, 740 (1978).

34. M. Uehara, J. Gattacceca, P. Rochette, F. Demory, E. M. Valenzuela, Magnetic study of meteorites recovered in the Atacama desert (Chile): Implications for meteorite paleomagnetism and the stability of hot desert surfaces. Phys. Earth Planet. Inter. 200-201, 113 (2012). doi:10.1016/j.pepi.2012.04.007

35. V. V. Sharygin, S.V., Kovyazin, N.M. Podgornykh, Mineralogy of olivine-hosted inclusions from the Omolon pallasite. Lunar Planet. Sci. Conf. XXXVII, abstr. 1235 (2006); available at www.lpi.usra.edu/meetings/lpsc2006/pdf/1235.pdf.

36. Y. Sonzogni, B. Devouard, A. Provost, J.-L. Devidal, Olivine-hosted melt inclusions in the Brahin pallasite. 72nd Ann. Meteoritical Soc. Meet. abstr. 5070 (2009); available at www.lpi.usra.edu/meetings/metsoc2009/pdf/5070.pdf.

37. M. R. Stevens, D. R. Bell, P. R. Buseck, Tubular symplectic inclusions in olivine from the Fukang pallasite. Meteorit. Planet. Sci. 45, 899 (2010). doi:10.1111/j.1945-5100.2010.01054.x

38. P. Wasilewski, Magnetic characterization of the new magnetic mineral tetrataenite and its contrast with isochemical taenite. Phys. Earth Planet. Inter. 52, 150 (1988). doi:10.1016/0031-9201(88)90063-5

39. G. Kullerud, The Fe-Ni-S System, Year Book. Carnegie Inst. Washington 62, 175 (1962/63).

40. D. L. Turcotte, G. Schubert, Geodynamics (Cambridge Univ. Press, Cambridge, ed. 2, 2002).

41. S. A. Gilder, M. LeGoff, J. C. Chervin, J. Peyronneau, Magnetic properties of single and multi-domain magnetite under pressures from 0 to 6 GPa. Geophys. Res. Lett. 31, L10612 (2004). doi:10.1029/2004GL019844

42. S. A. Gilder, M. LeGoff, Systematic pressure enhancement of titanomagnetite magnetization. Geophys. Res. Lett. 35, L10302 (2008). doi:10.1029/2008GL033325

43. N. S. Bezaeva, J. Gattacceca, P. Rochette, R. A. Sadykov, V. I. Trukhin, Demagnetization of terrestrial and extraterrestrial rocks under hydrostatic pressure up to 1.2 GPa. Phys. Earth Planet. Inter. 179, 7 (2010). doi:10.1016/j.pepi.2010.01.004

44. V. F. Buchwald, Handbook of Iron Meteorites: Their History, Distribution, Composition and Structure (Univ. California Press, Berkeley, CA, 1975).

45. A. Desrousseaux, J. C. Diukhan, H. Leroux, J. C. van Duyen, An analytical electron microscope investigation of some pallasites. Phys. Earth Planet. Inter. 103, 101 (1997). doi:10.1016/S0031-9201(97)00025-3

46. G. Martelli, G. Newton, Hypervelocity cratering and impact magnetization of basalt. Nature 269, 478 (1977). doi:10.1038/269478a0

3

47. T. L. Dickinson, P. Wasilewski, Shock magnetism in fine particle iron. Meteorit. Planet. Sci. 35, 65 (2000). doi:10.1111/j.1945-5100.2000.tb01974.x

48. J. J. Shen, D. A. Papanastassiou, G. Wasserburg, Re-Os systematics in pallasite and mesosiderite metal. Geochim. Cosmochim. Acta 62, 2715 (1998). doi:10.1016/S0016-7037(98)00187-2

49. G. M. Megrue, Rare gas chronology of hypersthene achondrites and pallasites. J. Geophys. Res. 73, 2027 (1968). doi:10.1029/JB073i006p02027

50. S. Niemeyer, I-Xe 40Ar-39Ar analyses of silicate from the Eagle Station pallasite and the anomalous iron meteorite Enon. Geochim. Cosmochim. Acta 47, 1007 (1983). doi:10.1016/0016-7037(83)90230-2

51. P. J. Hevey, I. S. Sanders, A model for planetesimal meltdown by 26Al and its implications for meteorite parent bodies. Meteorit. Planet. Sci. 41, 95 (2006). doi:10.1111/j.1945-5100.2006.tb00195.x

52. S. Sahijpal, P. Soni, G. Gupta, Numerical simulations of the differentiation of accreting planetesimals with Al-26 and Fe-60 as the heat sources. Meteorit. Planet. Sci. 42, 1529 (2007). doi:10.1111/j.1945-5100.2007.tb00589.x

53. L. T. Elkins-Tanton, B. P. Weiss, M. T. Zuber, Chondrites as samples of differentiated planetesimals. Earth Planet. Sci. Lett. 305, 1 (2011). doi:10.1016/j.epsl.2011.03.010

54. H. Haack, K. L. Rasmussen, P. H. Warren, Effects of regolith megaregolith insulation on the cooling histories of differentiated asteroids. J. Geophys. Res. 95, (B4), 5111 (1990). doi:10.1029/JB095iB04p05111

55. P. H. Warren, Ejecta-megaregolith accumulation on planetesimals and large asteroids. Meteorit. Planet. Sci. 46, 53 (2011).

56. T. M. Davison, G. S. Collins, F. J. Ciesla, Numerical modeling of heating in porous planetesimal collisions. Icarus 208, 468 (2010). doi:10.1016/j.icarus.2010.01.034

57. H. S. Carslaw, J. C. Jaeger, Conduction of Heat in Solids (Oxford Univ. Press, Oxford, ed. 2, 1986).

58. M. Ito, J. Ganguly, Diffusion kinetics of Cr in olivine and 53Mn-53Cr thermochronology of early solar system objects. Geochim. Cosmochim. Acta 70, 799 (2006). doi:10.1016/j.gca.2005.09.020

59. D. D. Bogard, G. H. Garrison, 39Ar-40Ar ages of eucrites and thermal history of asteroid 4 Vesta. Meteorit. Planet. Sci. 38, 669 (2003). doi:10.1111/j.1945-5100.2003.tb00035.x

60. D. D. Bogard, G. H. Garrison, Ar-Ar Impact Heating Ages of Eucrites and Timing of the LHB. Lunar Planet. Sci. Conf. XL, abstr. 1131 (2009); available at www.lpi.usra.edu/meetings/lpsc2009/pdf/1131.pdf.

61. H. J. Melosh, Impact Cratering: A Geologic Process (Oxford Univ. Press, Oxford, 1989).

62. C. J. Barnhart, F. Nimmo, B. J. Travis, Martian post-impact hydrothermal systems incorporating freezing. Icarus 208, 101 (2010). doi:10.1016/j.icarus.2010.01.013

4

5

63. J. Yang et al., Thermal and impact histories of reheated group IVA, IVB, and ungrouped iron meteorites and their parent asteroids. Meteorit. Planet. Sci. 46, 1227 (2011). doi:10.1111/j.1945-5100.2011.01210.x

64. E. Asphaug, M. Jutzi, N. Movshovitz, Chondrule formation during planetesimal accretion. Earth Planet. Sci. Lett. 308, 369 (2011). doi:10.1016/j.epsl.2011.06.007

65. D. W. Mittlefehldt, T. J. McCoy, C. A. Goodrich, A. Kracher, in Planetary Materials, J.J. Papike, Ed. (Mineralogical Society of America, Washington, DC, 1998), chap. 4.