Shape, Metal Abundance, Chemistry, and Origin of Chondrules in the Renazzo (CR) Chondrite

Shape, metal abundance, chemistry, and origin of chondrules in the Renazzo (CR) chondrite

Item Type Article; text

Authors Ebel, D. S.; Weisberg, M. K.; Hertz, J.; Campbell, A. J.

Citation Ebel, D. S., Weisberg, M. K., Hertz, J., & Campbell, A. J. (2008). Shape, metal abundance, chemistry, and origin of chondrules in the Renazzo (CR) chondrite. Meteoritics & Planetary Science, 43(10), 1725-1740.

DOI 10.1111/j.1945-5100.2008.tb00639.x

Publisher The Meteoritical Society

Journal Meteoritics & Planetary Science

Download date 09/10/2021 15:23:24

Item License http://rightsstatements.org/vocab/InC/1.0/

Version Final published version

Link to Item http://hdl.handle.net/10150/656486 Meteoritics & Planetary Science 43, Nr 10, 1725–1740 (2008) Abstract available online at http://meteoritics.org

Shape, metal abundance, chemistry, and origin of chondrules in the Renazzo (CR) chondrite

Denton S. EBEL1*, Michael K. WEISBERG1, 2, Jessica HERTZ3, and Andrew J. CAMPBELL4

1Department of Earth and Planetary Sciences, American Museum of Natural History, New York, New York 10024, USA 2Department of Physical Sciences, Kingsborough College and Graduate School of CUNY, 2001 Oriental Blvd., Brooklyn, New York 11235, USA 3The Columbia Preparatory School, 5 W. 93rd St., New York, New York 10025, USA 4 Department of Geology, University of Maryland, College Park, Maryland 20742, USA *Corresponding author. E-mail: [email protected] (Received 12 January 2006; revision accepted 23 May 2008)

Abstract–We used synchrotron X-ray microtomography to image in 3-dimensions (3D) eight whole chondrules in a ~1 cm3 piece of the Renazzo (CR) chondrite at ~17 µm per volume element (voxel) edge. We report the first volumetric (3D) measurement of metal/silicate ratios in chondrules and quantify indices of chondrule sphericity. Volumetric metal abundances in whole chondrules range from 1 to 37 volume % in 8 measured chondrules and by inspection in tomography data. We show that metal abundances and metal grain locations in individual chondrules cannot be reliably obtained from single random 2D sections. Samples were physically cut to intersect representative chondrules multiple times and to verify 3D data. Detailed 2D chemical analysis combined with 3D data yield highly variable whole-chondrule Mg/Si ratios with a supra-chondritic mean value, yet the chemically diverse, independently formed chondrules are mutually complementary in preserving chondritic (solar) Fe/Si ratios in the aggregate CR chondrite. These results are consistent with localized chondrule formation and rapid accretion resulting in chondrule + matrix aggregates (meteorite parent bodies) that preserve the bulk chondritic composition of source regions.

INTRODUCTION One hypothesis (Zanda et al. 1994, 2002a,b; Kong and Palme 1999; Kong et al. 1999; Campbell et al. 2002a) calls for The CR (Renazzo-type) chondrites are primitive partial evaporation and rapid recondensation of metal to meteorites that although hydrously altered did not suffer explain metal segregation, and textural, chemical and isotopic significant thermal metamorphism, so the properties of characteristics of chondrules, as reviewed by Campbell et al. their chondrules are important records of chondrule (2005b). They argued that Ni and Co concentrations in the formation (Mason and Wiik 1962, Wood 1962, 1963; Lee metal grains of the least circular (in thin section), finest- et al. 1992; Weisberg et al. 1992, 1993, 1995; Krot et al. grained chondrules do not follow a condensation trend, and 2002; DeGregorio et al. 2008). Renazzo is one of only two concluded that the relative amounts of Ni and Co in the CR chondrite falls. In some Renazzo chondrules, metal interior grains were gradually established during multiple occurs in two locations: as large interior metal grains, and chondrule melting events, due to reduction of fayalitic (FeO- as numerous smaller metal grains along the chondrule rim. bearing) olivine and evaporation of S from sulfides. Repeated In other chondrules, metal is in smaller grains and more or extensive heating of some chondrules advanced this evenly dispersed. Rare chondrules have multiple process, coarsened silicate grains, and promoted a spherical concentric silicate layers decorated by metal grains shape of the resulting chondrules. In their scenario, (Weisberg et al. 1993; Ebel and Rivers 2007, their DVD- “primitive” chondrules are FeO-rich, probably also FeS-rich, Fig. 9). Interior metal generally has higher concentrations with a mixture of many small metal and 16O-enriched silicate of the more refractory siderophile elements (e.g., Co, Ni; grains, in a non-spherical, convoluted morphology. More Re, Os, W, Ir, Ru, Mo, Pt) than rim metal, which tends to be “evolved” chondrules have coarser silicate and metal grains, enriched in more volatile elements (e.g., Fe, Cu, Au; volatile-rich metal grains on their rims (Humayun et al. 2002), Weisberg et al. 1993; Connolly et al. 2001; Humayun et al. circular morphologies in 2D, less 16O-rich silicates (Varley 2002; Campbell et al. 2002a, 2002b). et al. 2003), and well-established, volatility-correlated, core-

1725 © The Meteoritical Society, 2008. Printed in USA. 1726 D. S. Ebel et al. rim differences in metal composition, particularly higher Co transmission between metal, matrix, and chondrule silicates. and Ni concentrations in interior metal grains. To quantify the Metal grains frequently surround chondrules, setting them degree of melting of chondrules, Zanda et al. measured a apart from matrix in CR chondrites. We used chondrule convolution index (CVI) in thin section (Zanda microtomography to study eight whole Renazzo chondrules et al. 2002a, 2002b; Varley et al. 2003; cf. Nettles et al. 2006). in situ, then sliced and analyzed five of those in 2D sections. The CVI, minimum value 1.0, is defined as the ratio of the Spatial resolution in tomographic images is described by measured perimeter of the chondrule to the perimeter of a the edge length of each cubic volume element (voxel). A circle with the same area: single value for X-ray attenuation, a function of mean atomic CVI = chondrule perimeter/4πchondrulearea . number, is computed for each voxel (or pixel, in a single 2D Correlation of the CVI with chemical data led Zanda virtual slice through the volume). Part of a fragment (~8 × 8 × et al. (2002a) to suggest that the least evolved Renazzo 15 mm, 1.636 g.; Fig. 1a) of Renazzo (AMNH 588) was chondrules formed “by aggregation of numerous droplets in a imaged at 17.11 micron/voxel edge, for a total tomography dust-rich environment.” volume of X = Y = 554, Z = 472 voxels (Fig. 1). Methods of Other hypotheses for the metal segregation in CR tomography and image handling are described in detail by chondrules include condensation of core metal at higher Ebel and Rivers (2007). The technique allows discrimination temperatures, suggesting accretionary growth of the between metal and silicate grains, but not between silicate chondrules as temperature decreased (Weisberg et al. 1992, phases at the chosen X-ray energy and spatial resolution. 1993). Wood and McSween (1977, p. 368) concluded: “the From a 3D tomographic volume describing the X-ray concentric arrangement of metal in these objects suggests that attenuation of every voxel, one can readily produce images of they grew, with successive layers being added by accretion or an infinite number of “virtual”, random sections in any condensation.” Alternatively, additional Fe was incorporated orientation. We chose to examine stacks of sequential virtual into the metal on the chondrule rims due to FeO reduction images, or “slices,” perpendicular to the z-axis (vertical, from adjoining silicates (Lee et al. 1992). Connolly et al. Fig. 1a) of the tomographic volume. (2001) analyzed PGE distributions in CR chondrite metal and Eight chondrules were chosen to be representative of the argued that some rim metal grains, rich in refractory wide range of metal-silicate textures we observed by siderophiles, are original to primary formation of the tomography (Figs. 1 and 2a). From the full data set (e.g., chondrules. They argued that other rim grains, rich in volatile Fig. 2a), a data volume containing each chondrule was siderophiles, were formed by recondensation from extracted, and each image slice was cropped manually using surrounding vapor derived from chondrule reduction and ImageJ (NIH-www 2008) to isolate the chondrule of interest evaporation, and are therefore depleted in refractory in each slice where it appeared. Figure 2b shows number 44 siderophiles. of 99 cropped slices of chondrule 1, this one isolated from Previous studies of CR chondrules have been limited to Fig. 2a. Representative sequential slices for all 8 chondrules two-dimensional (2D) polished thin or thick sections cut at are illustrated in Fig. 3. Each stack of slices was further edited random (e.g., Krot et al. 2002). Here, we use three- to obtain a separate image stack, containing just the metal- dimensional (3D) tomographic imaging (Ebel and Rivers rich rim of the chondrule exposed in each slice. Figure 2c 2007), to obtain sizes, shapes, and volumetric metal/silicate illustrates this rim for slice 44 of chondrule 1. Widths of these ratios of eight representative Renazzo chondrules. We “rim” areas were established as inside the penetration width combine 3D data with 2D petrology and mineral chemistry on of the largest metal grain that was clearly a “rim” grain. surfaces cut through individual chondrules, informed by 3D Metal/(metal + silicate) ratios in rim areas and in each entire images, to assess existing hypotheses for the formation of CR chondrule are reported in Table 1 (columns 13 and 14). chondrules and the accretion history of the CR chondrite The number of voxels in the perimeter and surface area parent body or bodies. This is the first work combining 2D were measured in each slice of each chondrule, and summed and 3D methods to obtain volumetric bulk metal/silicate over all the slices of each chondrule to calculate chondrule ratios of chondrules. We argue that the huge range in metal/ volume and surface area (Table 1, columns 3–6). A CVI was silicate ratio supports theories of chondrule formation in a calculated for each chondrule in each slice (Fig. 4; Table 1 local region with chondritic Fe/Si, and rapid columns 7–10). A three-dimensional CVI (3D-CVI) was also “complementary” accretion of chondrules into parent bodies. calculated for each chondrule (Hertz et al. 2003), defined as the ratio of the measured surface area to the surface area of a METHODS sphere with the same volume as that measured for the chondrule (Table 1, column 11). Tomography and Image Analysis Using identical thresholding techniques for each chondrule, the number of pixels contributing to metal in each Chondrule shape, ratios of metal to silicates, and metal slice was measured (Fig. 4), and summed to the volume fraction grain distribution are well suited to 3D study by of metal in each chondrule (Table 1, column 12). To obtain the microtomography, due to the very high contrast in X-ray data in Table 1, no actual physical cutting was done on the Shape, metal abundance, chemistry, and origin of chondrules in Renazzo 1727

Fig. 1. Renazzo sample. Only part of original rock (A) was imaged. Cross section (B) is tomography slice Z = 380 of 472, illustrating wire saw cuts and chondrule 1 in section B1. Orthogonal tomography slices Y = 300 of 554 in (C) and X = 180 of 554 in (D) illustrate the locations of chondrule 1 and slice Z = 380. Scale bars are in mm. Empty space in (B) has been cropped from the tomography image.

Table 1. Textural characteristics measured and calculated from 3D tomographic data for 8 Renazzo chondrules. The number of virtual slices n yields volume and surface area, from which the maximum, minimum, and standard deviation of CVI observed in all slices of each chondrule are calculated, and an average CVI weighted by chondrule size (area), and a 3D convolution index (3D-CVI). Metal/silicate ratios (as % metal) are for whole chondrule volumes and for portions identified as “rim.” 123 4 5 6 789 10111213 Chon Slices Chondrule volume Surface area Convolution index (CVI) Metal Metal Wtd. n Voxels mm3 Voxel s mm 2 Min Max 1σ av. 3D % total rim% 1 98 359448 1.800 23037 8.26 1.03 1.66 0.08519152 1.155 1.154 12.9 54.8 2 155 903919 4.528 45550 16.33 1.08 1.59 0.071383785 1.188 1.234 4.9 22.3 3 87 218808 1.096 16798 6.02 1.02 1.42 0.086171199 1.157 1.172 4.5 57.5 4 62 88322 0.442 8723 3.13 1.05 1.27 0.04796062 1.106 1.114 1.1 58.7 5 93 307727 1.541 20758 7.44 1.02 1.30 0.067311534 1.145 1.153 10.9 80.4 6 167 1657970 8.305 66342 23.79 1.07 1.95 0.123139994 1.223 1.199 36.9 64.3 7 79 133253 0.667 11982 4.30 1.05 1.23 0.028949385 1.095 1.163 3.4 68.1 8 120 427612 2.142 27989 10.04 1.07 1.33 0.060927306 1.185 1.249 29.2 50.5 1728 D. S. Ebel et al.

Chemical Analysis

Guided by tomographic analysis, we purposely cut the sample (Fig. 1) to expose equatorial surfaces of particular chondrules. The AMNH sections containing these slices are: ch1a = Ren1sB1-S2H (Fig. 2D) , ch1b = Ren1sB1ps4a; ch2a = Ren1sB1ps7a; ch2b = Ren1sB3ps1A; ch3, 4, 5 = Ren1psA. All cuts were made with a 50 or 30 µm tungsten wire saw (Princeton Instruments) with boron carbide abrasive in machine oil slurry. Figure 2b illustrates the tomographic image of chondrule #1 that corresponds as closely as possible to the ~100 µm thick polished section (Fig. 2D). The axes of tomographic data are not exactly orthogonal to the cut and polished surfaces of the irregular prismatic sample (Fig. 1). X-ray mapping, and analyses of silicates and metal grains were performed using the 5-spectrometer Cameca SX100 electron microprobe (EMP) at AMNH. Tables 3 and 4 report averages of n spot analyses, each normalized to 100 wt%, and the average analysis sum. X-ray intensity maps were used to calculate modal abundances of silicate phases in each chondrule (Table 2). Similar modal calculations were done on sub-portions of chondrules 2 (section a) and 3 to obtain modal silicate abundances in the ellipsoid roughly corresponding to the olivine-rich core, and in the ellipsoidal shell corresponding to the pyroxene-rich rim in these igneously layered chondrules (Fig. 5). Technical details are available from the corresponding author. Laser-ablation ion-coupled plasma mass-spectrometry (LA-ICP-MS) analysis of metal grains was performed to confirm the systematic correlations of major and trace siderophile elements in metal grains (e.g., Humayun et al. 2002). LA-ICP-MS was accomplished on chondrule 1, section a (circled, 1–15, Fig. 2D) at the University of Chicago (Campbell et al. 2005a), and on chondrule 1 section b, and chondrules 2–5 on the single-collector Element 2 at the University of Maryland Plasma Mass Spectrometry Lab. No single grain was analyzed in both labs. Spot sizes were 15– Fig. 2. Images of chondrule 1. From original tomographic slice Z = 50 µm depending on grain size. Concentration data for 372 through the bulk sample of Renazzo (A), chondrule #1 is cropped individual analyses and averaged values are reported in online (B), and in (C), the portion considered to be “rim” is isolated. supplemental Tables 1–3. Backscattered electron image (D) at 1 µm/pixel resolution shows a thick section (ch1a) cut and polished almost, but not exactly, parallel to the tomographic slice. Each cubic volume element (voxel) in (A– RESULTS C) has edges 17.11 µm. Brightest pixels are metal. The accretionary material, including the attached chondrule-like object (upper right in Chondrule Shape and Texture by Tomography D), was excluded from analysis. Circled numbers correspond to LA- ICPMS analysis points. The entire imaged portion of Renazzo (Fig. 1) contained original Renazzo sample. Accuracy of 3D data analysis was more than 90 entire chondrules greater than 520 µm in verified by inspection of thin sections through the chondrules. diameter. We measured eight chondrules in detail (Fig. 3). The BSE image in Fig. 2D illustrates the sharp appearance of Measurements on each of the complete tomography the chondrule-matrix boundary in polished section for most sequences are plotted in Fig. 4. Tomographic datasets can be Renazzo chondrules. An accretionary rim surrounds both found in Ebel and Rivers (2007, on DVD). Both metal modal chondrule 1 and also a small attached chondrule, which has a abundances and CVI, measured in 2D for any particular very different metal texture and uniform metal grain chondrule, depend strongly upon which slice is measured, as compositions (c1x in Table 4). This attached chondrule was not evident in Figs. 3 and 4, and the standard deviation in CVI counted as part of chondrule 1 (Fig. 3). (Table 1, column 9). Shape, metal abundance, chemistry, and origin of chondrules in Renazzo 1729

Fig 3. Sequences of X-ray image reconstructions (CAT-scans) through Renazzo chondrules. High density metal is bright, silicate are dark. Scale bar is 2 mm in each sequence. Surrounding matrix and chondrules have been removed from images. In each sequence, images taken along axis z are aligned to the same (x, y) origin, indicated by an asterisk (*) in the lower left corner of each slice image. Chondrules 4 and 5 are parts of a compound pair.

All of the chondrules studied are FeO-poor type I forsteritic olvine + mesostasis-rich cores, including metal porphyritic chondrules, the dominant type in CR chondrites grains, surrounded by coarse-grained, low-Ca, pyroxene-rich (Weisberg et al. 1995). These chondrules represent ranges of igneous rims. Chondrule 2 is finer grained than 3, with a more grain size, metal abundance, and shape that are typical of the irregular (convoluted) outline (Figs. 3 and 5). Chondrules 4 chondrules in CR chondrites. For example, chondrule 1 is a and 5 appear in tomography to be a classical compound porphyritic olivine chondrule with a 300 µm metal nugget in chondrule pair, in which chondrule 5 was partially molten its core. On its rim are smaller metal grains that are 50–100 µm (Ebel and Rivers 2007, their DVD-Fig. 3B; Ebel-www 2008). in size. Chondrules 2 and 3 are layered (Fig. 5), with This kind of 3D analysis has never been done before in 1730 D. S. Ebel et al.

Fig 4. Modal metal (triangles) and convolution index (CVI, squares) measured in sequential chondrule images. Average values for CVI (dotted) and %metal (dashed) are superposed. Shape, metal abundance, chemistry, and origin of chondrules in Renazzo 1731

Table 2. Modal abundances of minerals, as percentages of total pixels (area) calculated from 2D X-ray maps using criteria given in text. Low-Ca pyroxene is labeled “pyx,” and high-Ca pyroxene “Ca-px.” Two sections (a and b) through chondrules 1 and 2 were analyzed, and a weighted average is given. Independent calculations of silicate modes for inner ellipsoid and outer ellipsoid shell (core, rim) volumes of sections ch2a (0.867 and 1.473 mm3) and ch3 (0.069 and 0.478 mm3) are also given (see Fig. 5). “Total pixels” are the pixels within the masked chondrule in polished section (Fig. 5), and “r” gives the resolution of X-ray mapping (µm/pixel). Label Metal Olivine pyx Ca-px Glass Carbonate Unknown Total pxls r ch1a 13.0 13.6 61.9 2.4 6.7 0.1 2.2 96009 4 ch1b 11.8 22.4 54.4 2.5 6.6 0.1 2.2 496489 2 Wtd. avg. 12.1 22.9 55.7 2.6 6.8 578760 Silicates 24.5 64.9 2.9 7.8 507657 ch2a 12.8 13.8 49.0 2.2 17.7 0.2 4.2 586662 2 ch2b 13.0 16.5 50.5 3.8 12.2 0.3 3.7 416707 3 Wtd. avg. 13.5 17.1 52.6 4.0 12.7 960656 Silicates 18.0 59.9 3.5 18.6 831272 ch3 6.4 16.0 62.9 1.7 10.2 0.5 2.3 2445942 1 Components 6.6 16.5 64.8 1.7 10.5 2376171 Silicates 17.6 69.3 1.8 11.2 2220191 ch4 1.4 7.2 49.5 9.3 27.4 0.1 5.0 1151730 1 Components 1.5 7.6 52.2 9.8 28.9 1092499 Silicates 7.7 53.0 10.0 29.3 1076549 ch5 3.4 38.5 35.9 4.4 13.6 0.2 4.1 1890376 1 Components 3.5 40.3 37.5 4.6 14.2 1809018 Silicates 41.7 38.8 4.8 14.7 1745499 ch2a-core 27.3 36.5 3.2 33.0 236912 2 ch2a-rim 6.7 81.4 1.9 10.0 208937 2 ch2a-total 14.3 64.8 2.4 18.5 445849 ch3-core 39.7 25.5 2.9 31.9 446965 1 ch3-rim 12.0 80.4 1.6 6.0 1773226 1 ch3-total 15.7 73.0 1.7 9.5 2220191

CR chondrites, and allows quantitative comparison of 2D and variation in bulk metal abundance between the chondrules. 3D methods. The images in Fig. 3 suggest that the less (e.g., 2, These inter- and intra-chondrule variations would not be 6, 8) or more coarsened, rounded nature e.g., 1, 3–5, 7) of recognized using only standard 2D techniques. each chondrule can, in most cases, be qualitatively assessed The representative slices in Fig. 3 illustrate, from a single random section through its middle region. qualitatively, the problems inherent in 2D analysis of 3D Quantitative results, however, show significant variations objects that are not spherically symmetric in shape or metal even in these central regions (Fig. 4). The variability in CVI distribution. Metal abundances tend to peak in initial and (Table 1, column 9) is high enough that different random sets final virtual slices that intersect rims, where metal grains are of cross-sections could produce a wide variation in rank order concentrated. Even in the central portions of many of these of “convolution” of these 8 chondrules. In particular, chondrules, large variations are evident in both metal chondrule 1 contains the largest high-Ni “core” metal grain, abundance (Fig. 4) and metal grain shape (Fig. 3). For but has a CVI significantly above unity (i.e., it is rounded but example, slices 45 and 50 of chondrule 1 show a metal-rich not circular). The 3D-CVI are only weakly correlated (r2 = chondrule core, whereas slices 30, 80 and 90 of the same 0.44) with the average 2D CVI, and even more weakly with chondrule show a metal-poor core (Fig. 3). Likewise, slices metal abundance (r2 = 0.29). Although the main variance in 30, 50 and 60 of chondrule 5 appear metal-rich, whereas CVI is between core and rim areas, there is no way in a thin slices 35, 40 and 70 appear metal-poor. section to tell which area one is observing. The high X-ray contrast between metal and silicates allows robust determination of whole chondrule metal/silicate Volumetric Metal/Silicate Ratios by Tomography ratios. Fractional metal abundances in the eight chondrules measured in 3D vary from ~1 vol.% to a remarkably high We report here, after Hertz et al. (2003), the first direct ~37% (Table 1, column 12). These bulk volumetric metal volumetric (3D) measurements of metal abundances in abundances (Table 1, column 12) generally differ from the chondrules. Two results are apparent: (1) both metal apparent abundances acquired by 2-D section analysis abundance and metal grain shape are highly variable among (Table 2), and from the average value across each image 2D slices of a single chondrule, and (2) there is a large stack (Fig. 4). 1732 D. S. Ebel et al.

Table 3. Silicate compositions (wt%). Data are average and standard deviation (parenthetical below) for n electron microprobe analyses. Mean values of oxide totals for n original analyses are also given. Low-Ca pyroxene is labeled “pyx,” and high-Ca grains “Ca-px.” Mineral structures were not determined. K2O, P2O5, and SO2 are below detection limits and omitted from totals. Only analyses with 96 < total < 102 wt% are included.

Chon. Phase n SiO2 TiO2 Al2O3 Cr2O3 FeO MnO MgO NiO CaO Na2O Total 1 olv 43 42.82 0.04 0.03 0.65 1.74 0.13 55.80 0.01 0.17 0.01 101.38 (0.66) (0.03) (0.04) (0.04) (0.12) (0.03) (0.53) (0.01) (0.02) (0.01) (1.06) 1 pyx 15 59.26 0.11 0.64 0.60 1.78 0.09 38.51 0.02 0.35 0.02 101.34 (1.03) (0.06) (0.19) (0.09) (0.80) (0.03) (0.74) (0.02) (0.08) (0.03) (1.23) 1 Ca-px 7 48.12 1.19 9.65 1.72 1.11 0.14 17.45 0.01 19.97 0.03 99.36 (1.59) (0.21) (0.85) (0.11) (0.48) (0.03) (0.71) (0.01) (0.45) (0.02) (1.79) 1 glass 5 51.18 0.28 21.28 0.46 1.58 0.09 13.92 0.02 13.09 0.59 101.89 (0.49) (0.05) (1.90) (0.08) (0.94) (0.01) (1.48) (0.01) (0.84) (0.06) (0.45) 2 olv 40 44.04 0.04 0.08 0.56 1.09 0.14 55.52 0.02 0.22 0.01 101.71 (0.70) (0.02) (0.16) (0.08) (0.26) (0.05) (0.71) (0.01) (0.08) (0.01) (1.15) 2 pyx 47 59.64 0.24 1.31 0.69 1.66 0.09 37.60 0.04 0.43 0.03 101.69 (1.39) (0.08) (0.54) (0.15) (1.14) (0.04) (0.85) (0.03) (0.13) (0.04) (1.84) 2 glass 14 47.46 0.02 32.86 0.04 1.33 0.03 0.89 0.06 18.37 0.42 101.05 (1.28) (0.01) (0.89) (0.02) (1.29) (0.01) (0.10) (0.06) (0.47) (0.13) (1.04) 3 olv 40 42.68 0.03 0.07 0.62 1.96 0.19 54.43 0.02 0.25 0.04 100.26 (0.31) (0.02) (0.12) (0.12) (0.31) (0.06) (0.43) (0.02) (0.05) (0.03) (0.57) 3 pyx 41 58.58 0.15 1.01 0.72 1.84 0.14 36.95 0.02 0.80 0.06 100.21 (1.20) (0.05) (0.33) (0.14) (0.98) (0.05) (1.21) (0.02) (0.62) (0.09) (1.20) 3 Ca-px 4 49.79 1.14 7.86 1.83 1.05 0.27 18.05 0.03 19.34 0.07 99.35 (0.55) (0.02) (0.19) (0.03) (0.06) (0.00) (0.51) (0.01) (0.42) (0.01) (0.71) 3 glass 5 46.21 0.16 27.98 0.24 1.58 0.08 6.58 0.04 15.68 0.20 98.55 (1.84) (0.09) (5.53) (0.22) (1.00) (0.06) (5.93) (0.04) (2.56) (0.05) (0.87) 4 olv 2 41.97 0.16 0.48 0.66 1.96 0.16 52.14 0.01 0.23 0.66 97.77 (0.12) (0.02) (0.48) (0.01) (0.02) (0.01) (0.31) (0.00) (0.00) (0.66) (0.34) 4 pyx 5 57.70 0.25 1.42 1.01 1.79 0.19 33.51 0.01 1.80 0.00 97.69 (0.22) (0.06) (0.41) (0.09) (0.47) (0.03) (0.44) (0.02) (0.91) (0.00) (0.93) 4 Ca-px 5 53.69 0.97 2.43 1.02 1.39 0.23 20.71 0.01 17.94 0.01 98.39 (0.88) (0.33) (0.51) (0.04) (0.47) (0.06) (1.23) (0.01) (1.42) (0.01) (0.79) 4 glass 4 44.74 0.02 33.04 0.02 0.30 0.00 0.90 0.02 19.16 0.25 98.21 (1.13) (0.02) (0.65) (0.02) (0.13) (0.00) (0.10) (0.02) (0.39) (0.28) (1.42) 5 olv 6 42.25 0.06 0.20 0.44 1.74 0.19 51.77 0.03 0.27 0.16 96.96 (0.26) (0.02) (0.21) (0.17) (0.54) (0.18) (0.65) (0.02) (0.06) (0.35) (0.69) 5 pyx 6 56.34 0.41 4.07 1.17 1.18 0.21 32.86 .03 1.68 0.00 97.96 (2.16) (0.13) (2.78) (0.31) (0.19) (0.09) (1.63) (0.02) (0.83) (0.01) (0.52) 5 Ca-px 2 50.58 1.07 6.49 1.44 1.08 0.26 18.67 0.09 17.62 0.01 97.31 (1.01) (0.10) (0.48) (0.01) (0.11) (0.00) (1.03) (0.05) (0.33) (0.01) (1.26) 5 glass 4 50.97 0.49 21.84 0.37 2.58 0.33 5.71 0.04 14.53 0.74 96.84 (2.18) (0.07) (0.97) (0.06) (2.04) (0.04) (1.07) (0.03) (0.60) (0.23) (0.94)

Silicate Mineralogy and Chemistry 2D (Fig. 3), we approximated the effect of layering on modal calculations by fitting ellipsoids to an outer region (rim), and an Modal abundances of silicates were only accessible by 2D inner region (core) for one section of each of these chondrules methods (Table 2). The results of 2D modal calculation using (Fig. 5). We start with the assumption that the olivine/ custom-written software, output with pixels color-coded by pyroxene/glass ratios in Table 2, and phase compositions in mineralogy as determined by a rule-based algorithm from raw Tables 3 and 4, are representative of rim and core regions each element maps, are indistinguishable from the Mg-Ca-Al entire chondrule. This assumption is justified by the larger mosaic composite X-ray maps shown in Fig. 5. Modal proportion of silicates, relative to metal, exposed in sections; abundances of olivine, pyroxenes, glass, trace carbonate lack of localized concentrations of a single silicate phase; (verified by energy dispersive spectrometry), and smaller grain size of silicates compared to metal; observed unidentifiable pixels in five sectioned chondrules are given in chemical homogeneity of silicate phases; and the fact that we Table 2. could not distinguish among silicates in 3D. Chondrules 2 and 3 are layered, with an inner region that Bulk chemical compositions were calculated for each appears to be nearly spherical, with coarser forsteritic olivine chondrule by modal recombination, using the silicate and large metal grains near its rim (Figs. 3 and 5). Because chemical compositions in Table 3. No systematic differences these two chondrules appear roughly symmetrical in 3D and in major element contents between olivines or pyroxenes in Shape, metal abundance, chemistry, and origin of chondrules in Renazzo 1733

Fig. 5. X-ray composite maps of Renazzo chondrules 1–5. X-ray intensities of Mg, Ca, and Al are overlapped in, respectively, red, green, and blue channels of rgb images. Metal is black. Yellow curves outline inferred rim and core regions in chondrules 2 and 3. 1734 D. S. Ebel et al.

Table 4. Metal grain compositions. Composition ‘c1x” is a small chondrule attached to chondrule 1 (Fig. 1). Data are mean and standard deviation for n electron microprobe analyses. ‘Xmetal” is the volume fraction of chondrule metal in rim and core grains in the whole chondrule (Table 1), except for chondrule 2 with homogeneous metal compositions.

Chon n Fe (wt%) Ni P Cr Co Sum Xmetal Co/Fe Ni/Fe 1 Core 124 90.87 (0.28) 8.18 (0.15) 0.32 (0.15) 0.29 (0.05) 0.35 (0.02) 100.03 (0.42) 0.452 0.0039 0.090 1 Rim 194 94.19 (0.45) 5.09 (0.31) 0.42 (0.21) 0.21 (0.05) 0.24 (0.02) 100.38 (1.05) 0.548 0.0025 0.054 2 All 200 93.76 (1.06) 5.25 (0.92) 0.33 (0.19) 0.55 (0.2) 0.25 (0.03) 100.61 (1.14) 1 0.0027 0.056 3 Core 13 87.32 (3.15) 11.96 (3.1) 0.24 (0.03) 0.41 (0.04) 99.42 (0.71) 0.425 0.0047 0.137 3 Rim 27 94.43 (0.27) 5.08 (0.26) 0.18 (0.02) 0.25 (0.01) 100.37 (0.39) 0.575 0.0026 0.054 4 Core 3 89.37 (0.25) 9.91 (0.27) 0.27 (0.01) 0.39 (0.01) 99.11 (0.11) 0.413 0.0044 0.111 4 Rim 5 93.84 (0.05) 5.46 (0.05) 0.40 (0.01) 0.25 (0.01) 99.88 (0.09) 0.587 0.0026 0.058 5 Core 7 89.25 (0.23) 9.88 (0.23) 0.13 (0.01) 0.34 (0.01) 0.40 (0.01) 100.48 (0.49) 0.196 0.0045 0.111 5 Rim 25 93.77 (0.22) 5.22 (0.22) 0.28 (0.03) 0.47 (0.08) 0.25 (0.02) 100.06 (1.09) 0.804 0.0026 0.056 c1x All 10 94.09 (0.09) 5.18 (0.08) 0.26 (0.02) 0.23 (0.02) 0.24 (0.02) 99.55 (0.68) 1 0.0026 0.055 chondrule interiors and chondrule rim regions were detected. 13), and assuming that the chemical compositions of core and A simple calculation based on 2D modal data yielded bulk rim metal grains analyzed in 2D sections (Table 4) are chondrule SiO2, MgO, FeO, CaO, and Al2O3 (Table 6). representative of similarly located grains in the entire Caveats apply to this kind of analysis (Hezel 2007). However, chondrule. We identify the high-Ni grains in chondrule 5 recent work shows that in many cases olivine/pyroxene ratios as “core,” although there are a few other grains not on the are sufficiently constant between serial sections that 2D chondrule rim, but much nearer to it, that have low-Ni modal measurements can be representative of the true 3D compositions typical of rim grains. bulk chondrule composition (Hezel, personal We normalized rim and core mean compositions to only communication). For layered chondrules 2 and 3, a second Fe, Co, and Ni, because these elements dominate metal bulk composition calculation was made, modeling each as a grain compositions (Table 4). We then calculated the bulk combined inner ellipsoid (core) and an outer ellipsoidal shell metal composition in each fractionated chondrule, using the (Fig. 5). These volumes were assumed to be represented by tomographically measured rim/core metal abundances the silicate mineral modes (Table 2) measured in the core and (Table 1, column 13), and also calculated the bulk Fe-Ni-Co % rim areas shown in Fig. 5. Ratios MgO/SiO2(wt ) calculated composition of metal in unfractionated chondrule 2. These this way for chondrule 2 illustrate sensitivity to Ca-pyroxene results are compared in Table 5 with solar abundances. abundance, relative to the simpler 2D calculation. For Part of our goal is to assess the evolution of metal in CR chondrule 3, the second calculation brings the bulk MgO/ chondrules, and trace element chemistry is critical to this end. SiO2 ratio to below solar, due to the thick pyroxene-rich rim. Metal grains were measured by LA-ICP-MS for two sections For comparison with chondritic (solar) ratios, the relative of chondrule 1 (Fig. 2D), and one each of chondrules 2, 3, and mass fractions of the oxides were calculated, preserving their 5. Full results are available in an online supplement. Trace solar proportions relative to the mean FeO of all eight siderophile element compositions are highly variable within chondrules. The composition of this “solar” chondrule differs and between metal grains. Figure 6 illustrates average results from the mean of the measured chondrules in important ways. for chondrules 1 (sections a and b), 2 and 3, normalized to Recalculation of the mean chondrule oxides using these 100 wt% total. Humayun et al. (2002) report Mo, Ru, W, Re, layered compositions for chondrules 2 and 3 brings the mean Os, Ir, Pt, Ni, Co, and sometimes Pd enriched in core metal, chondrule oxide values closer to the solar expectation. and depleted in rims, while Fe, Cu and sometimes Pd are enriched in rims, and Ga and Ge are not measurably core/rim Metal Analysis in Polished Section fractionated (Campbell et al. 2002a). Here, results for Fe, Cu, Mo, Os, Ir, and Pt are similar, but Ni and Pd are always Mean chemical compositions of rim and core metal enriched in core metal grains: core/rim Pd/Fe are 2.33, 2.07, grains are presented in Table 4. Both Cr/Fe and P/Fe are 2.04; in chondrules 1, 3, and 5 respectively. Rim metal grains considerably sub-solar in all metal grains measured. are also enriched in As and Au. Phosphorus is enriched in rim metal, with the ratio (P/Fe) Campbell et al. (2005b) attribute enrichment of volatile, 1.3× and 2.1× higher in rim than in core grains for chondrules but not readily oxidized Pd in core metal grains, and also 1 and 5, respectively. Results for Cr are ambiguous, but Cr is enrichment of Ni, to preferential mobilization of oxidized Fe not as strictly siderophile as the other elements measured in through silicates to evaporate at the rim. In this interpretation, metal. Detailed examination of metal grain microstructures significant Ni/Fe and (especially) Pd/Fe fractionations reveals no subgrains or exsolution. between the core and rim metal are evidence of oxidation/ The absolute Fe, Ni and Co abundances in each reduction (redox) control over silicate solubility of these chondrule were calculated using the relative abundances of elements during formation of these chondrules, in addition to rim and core metal grains observed in 3D (Table 1, column the volatility-based fractionation indicated by the refractory Shape, metal abundance, chemistry, and origin of chondrules in Renazzo 1735

Table 5. Bulk siderophile compositions of chondrules measurement in thin section is unreliable due to the large (wt%) reconstructed from data in Table 1 (col. 13) and standard deviations in CVI (Table 1, column 9) and the range Table 2, compared to solar proportions (AG89: Anders and even among nearby slices (Fig. 4). Nettles et al. (2006) also Grevesse 1989; L03: Lodders 2003). noted that a single criterion like the CVI should be Fe Co Ni supplemented by metal grain size and distribution ch1 93.19 0.291 6.522 information. That is exactly the information which supplies ch2 94.46 0.254 5.286 the visual clues for qualitative assessment of chondrule ch3 91.66 0.317 8.020 melting (Fig. 3; Nettles et al. 2006; see also Ebel and Rivers ch4 92.36 0.309 7.328 2007). ch5 93.54 0.280 6.178 The sphericity of entire chondrules, measured by the 3D- AG89 wt% 94.32 0.249 5.431 CVI, is a more reliable texture indicator than single 2D L03 wt% 94.08 0.275 5.642 measurements. The 3D-CVI correlate with the observed abundance and texture of metal grains, and are lowest in PGEs and the volatile siderophiles. Palladium does not chondrules 4, 5, 1, 7 and 3, in which either the total abundance fractionate significantly from Fe during high-temperature of metal is low (#4 and 7), or metal is concentrated in a few vapor-solid interactions in the solar nebula (Campbell et al. large grains (#1, 3, and 5). The 3D-CVI appears to be weakly 2001), but Pd is less readily oxidized and has a lower correlated to total chondrule volume. As a chondrule”s solubility in the silicate melt than Fe. Enrichment of core surface area increases, it has more opportunity to have a large metal in Pd/Fe (by ~2×) was observed by Humayun et al. 3D-CVI, and surface convolution becomes easier to measure (2002) in only one of 11 chondrules they studied. By contrast, in larger chondrules. We do not observe any correlation we observe core Pd enrichment in all core/rim fractionated between FeO in olivine (Table 3) and the 3D-CVI. chondrules where Pd was measured (Figs. 6 and 7). It has been suggested (Grossman and Wasson 1985; Difference in oxidation and silicate solubility would also have Tsuchiyama et al. 2000) that metal is driven to the rims of affected the abundances of the other siderophile elements, but rapidly spinning, partially molten chondrules by centrifugal their much stronger fractionations (for refractory PGEs) and force. In this scenario, metal should be concentrated on the fractionations in the opposite direction (for volatile elements equatorial plane of a spinning chondrule (Uesugi and Sekiya Cu, As, Au) reveal that volatility effects were more important 2006). Tomographic images in orthogonal directions do not than redox effects on these elements. Finally, we note that the reveal any preferred (equatorial) surface distribution of metal Pd/Fe ratio in the rim grains is subchondritic, whereas it is grains in the CR chondrites we studied. In experimental superchondritic in the core metal. This relationship is analogs of non-spinning chondrules, metal also migrates to consistent with a model in which volatilized Fe (and other outer surfaces (Connolly et al. 1994), apparently to minimize volatile siderophiles, such as Cu and Au) recondensed onto surface energy. Alternatively, metal grains could be accreted rim grains (Connolly et al. 2000; Humayun et al. 2002; directly to the outer surfaces of earlier-formed chondrules. Campbell et al. 2005b). De-coupled accretion of high-temperature metal and silicate grains would explain the high variability in bulk metal/silicate DISCUSSION volumetric ratios in chondrules (Table 1). Separate, sequential accretion of metal and silicates would also explain the Petrology in 2D and 3D existence of chondrules with multiple concentric layers: silicate/metal/silicate/metal. Assigning the perimeter of a chondrule for measurement of the CVI is a subjective exercise. All of our chondrules were Chondrule Layering virtually “cut” from matrix (Fig. 1) by one person, from data at a single spatial resolution and contrast, using consistent In 3D, we have observed a few chondrules in CR criteria. Yet most of our CVI measurements are below 1.45, chondrites that have multiple metal-rich and metal-poor while a majority of those by Zanda et al. (2002a) and Nettles layers. One well-developed example is in sample Acfer 139-2 et al. (2006) are higher. It is likely that perimeters drawn imaged by Ebel and Rivers (2007, their DVD-Fig. 9C; cf. around high-resolution surface maps or photomicrographs are Ebel-www 2008), and consists of a forsterite core decorated generally more convoluted (jagged or embayed) than those with metal, all rimmed by forsteritic silicate, then larger metal drawn around lower-resolution X-ray tomographic images. grains, then a thinner rind of silicate, with an outer rim of Nevertheless, our results allow robust textural comparison smaller metal grains. Imaging in 3D allowed us to precisely within a single dataset. bisect chondrules to reveal, in 2D, not only the metal layering Our results confirm many of the observations of Nettles observed by tomography, but also the igneous layering et al. (2006) concerning the use of the CVI as a single sequence–olivine-rich core, Ca-poor pyroxene-rich rim (Fig. 7 indicator of the degree of chondrule melting. A single CVI inset; Fig. 5)—that is observed in some CR chondrules 1736 D. S. Ebel et al.

Fig. 6. Siderophile trace elements in chondrule core and rim metal grains. Average LA-ICP-MS analyses from U. Chicago (chondrule 1a) and U. Maryland laboratories for data with n > 1 and mean > 1σ. Two sections of chondrule 1 were analyzed independently.

(Weisberg et al. 1993). Our 3D data show that neither type of accreted and annealed, with only partial re-equilibration of layering would be recognized in many 2D sections. Olivine- the inner core minerals. The Pd/Fe ratios of interior and rim pyroxene layering is consistent with high-temperature metal grains (Fig. 6) suggest the later influence of an equilibration with a solar gas, because olivine condenses at oxidizing environment, which would preferentially remove higher temperatures than pyroxene (Ebel and Grossman Fe (Campbell et al. 2005b). Alternative thermodynamic 2000; Ebel 2006). conditions may exist, however, in which core grains could The inner regions of layered chondrules appear to be concentrate Pd at the same temperatures at which Ni and Co nearly spherical, with coarser forsteritic olivine and large would be enriched (Fig. 7). metal grains than the outer regions (e.g., Fig. 5B). Yet the interior metal grains are not always Ni- or Co-rich, for Major Element Abundance and “Complementarity” example in chondrule 2. The outer region (rim) of this chondrule is a more irregular, fine-grained layer (Fig. 5A) that For the 8 chondrules studied here, the standard deviation gives the chondrule an over-all convoluted texture (Table 1, of metal abundance is >95% of its average value (Table 1, Figs. 3 and 4), and the metal grains in this rim are also ~solar column 12; Fig. 3 by inspection). Each chondrule has a in Ni/Fe and Co/Fe ratio (Table 4, Figs. 6 and 7). We have not unique budget of siderophile elements (Table 5). Similarly, found any layered CR chondrules with a convoluted, fine- chondrule bulk silicate compositions show large variability in grained core surrounded by a spherical, coarse-grained rim. chondrule MgO/SiO2 ratios (Table 6). It is clear, however, Metal grains in chondrule 3 (Table 4, Fig. 7) show a core- that bulk samples of CR chondrites (Lodders and Fegley to-rim composition trend consistent with the high- to low- 1998, their Table 16.10), and the Renazzo meteorite temperature trend predicted by condensation calculations (Kellemeyn and Wasson 1982; Weisberg et al. 1993) contain (Fig. 7; Ebel and Grossman 2000, their Fig. 14). It is not at all Mg, Si, Fe, Co, and Ni in solar proportions. The concept of clear how these compositional gradients would be predicted “complementarity” can explain these observations. or explained by evaporation-recondensation models for The complementarity between the chemical compositions bimodal metal compositions (e.g., Zanda et al. 1994; Kong of matrix and chondrules (Palme et al. 1992) has been and Palme 1999; Campbell et al. 2005b). Instead, these addressed in several studies (Kong and Palme 1999; Kong gradients suggest that the core region of chondrule 3 et al. 1999; Klerner and Palme 1999; Bland et al. 2005). equilibrated with vapor at high temperatures, and that Meteorites with chondritic (solar) ratios of major elements successive layers of silicate and metal were subsequently contain chondrules and matrix in varying ratios. Because Shape, metal abundance, chemistry, and origin of chondrules in Renazzo 1737

Table 6. Bulk silicate compositions of chondrules (wt%), calculated from Tables 2, 3, and 4, compared with solar expectation. Cr2O3, MnO, and NiO are omitted. Chondrules 2 and 3 are calculated using both simple modes (above), and the core + rim layer estimate of Table 2 (below). Oxide proportions expected for a solar complement of oxides for comparison (AG89: Anders and Grevesse 1989; L03: Lodders 2003). SiO2 TiO2 Al2O3 FeO MgO CaO Na2O MgO/SiO2 ch1 54.99 0.14 2.41 1.62 38.85 1.93 0.06 0.706 ch2 54.64 0.20 7.32 1.46 31.83 4.45 0.10 0.583 ch3 43.13 0.14 0.72 3.38 51.66 0.79 0.19 1.198 ch4 48.69 0.35 2.96 2.54 40.86 2.48 2.13 0.839 ch5 44.30 0.15 1.64 2.18 50.56 0.72 0.44 1.141 Mean: 49.08 0.20 3.01 2.24 42.82 2.07 0.58 0.893 By layers: ch2 54.37 0.19 7.11 1.49 32.57 4.16 0.10 0.599 ch3 55.33 0.20 4.21 1.52 36.26 2.41 0.06 0.655 New mean: 51.54 0.21 3.67 1.87 39.82 2.34 0.56 0.773 1 : 4.35 0.07 1.92 0.42 6.04 1.11 0.80 0.194 Solar: AG89 52.13 0.17 3.76 1.87 37.56 2.97 1.54 0.720 L03 53.13 0.17 3.79 1.87 36.35 3.12 1.58 0.684

Fig. 7. Co and Ni compositions of rim and core metal grains. Mean core grain (c) compositions for chondrules 1, 4, and 5 are diamond, rectangle and triangle, respectively. Individual analyses of core grains from chondrule 3 are plotted (open circles) with reference to position in the chondrule (inset, BSE image). Rim grains from all chondrules cluster near the canonical solar composition (star, Anders and Grevesse 1989). Solar Co/Ni ratio (dotted) is provided without error estimates. Solid line shows predicted metal composition condensing from a gas of solar composition at a total pressure of 10−3 bar, and dashed line for a gas enriched 100× in CI dust at 10−3 bar (Ebel and Grossman 2000), approximately coincident with the path for a solar gas at 10−8 bar (Ebel 2006). chondrules and matrix differ in mean major element less likely that complementarity would result from exotically abundance, the chondrules must have formed, coexisting with formed chondrules (e.g., near the Sun), transported and mixed the matrix grains, as a closed system in those elements, from with matrix dust local to the region of later parent body a chondritic batch of material in the protoplanetary disk. It is accretion. 1738 D. S. Ebel et al.

The variation observed here among chondrules proportions of siderophile and lithophile elements (Fig. 6). themselves offers separate confirmation of this theory of These metal grains are preserved in discrete layers inside a complementarity. Renazzo chondrules were formed by a few chondrules, but this texture was obliterated by melting in process that preserved the complementary relationship most chondrules, forming Ni, Co-rich core metal nuggets between silicate and metal fractions in the parent body. That (e.g., ch3, Fig. 5C). A second generation of metal grains then is, all the Renazzo chondrules (and other CR chondrules) formed at lower temperature, with metal grains having near- appear to have formed from a single chondritic (solar) batch solar Ni/Fe and Co/Fe (Fig. 6), along with more SiO2-rich of material that was processed (heated/cooled) selectively and silicate. These later metal grains and silicates adhered to the locally. This produced chondrules with highly variable, non- rims of pre-existing chondrules, or combined to make a next chondritic metal/silicate volumetric ratios, as measured in just generation of chondrules. Perhaps we are observing in these eight chondrules, and observed qualitatively in 3D Renazzo two or more generations of chondrules. Multiple imagery of other chondrules. Yet the independently-formed metal/silicate layers would then result from successive chondrules are themselves “complementary” in that they adherence of generations of metal and silicates onto a single combine with each other, and with matrix, to produce a first generation chondrule. chondritic (solar) bulk abundance of major elements in Renazzo. This complementarity is preserved even though CONCLUSIONS metal and silicate appear to have been fractionated locally, one from another, before, during or following the chondrule Computer-aided X-ray tomography can be used to formation process. quantify the bulk (volumetric) metal/silicate ratios of Klerner and Palme (1999) found that Renazzo matrix chondrules, and to quantify chondrule shape and size. This is % = MgO/SiO2(wt ) 0.496, compared to 0.703 in bulk Renazzo particularly necessary in meteorites in which chondrules vary (Mason and Wiik 1962), and solar 0.684 (Lodders 2003) or greatly in shape and metal abundance, and are not radially 0.720 (Anders and Grevesse 1989). They concluded that symmetric (e.g., CR chondrites). Tomography allows chondrules must, in aggregate, have a high MgO/SiO2 ratio. informed intersection of chondrules for obtaining 2D Our volumetric control on silicate phase abundances is less chemical data, for extrapolation to volumetric chemical strong than for metal/silicate ratios. Nevertheless, our results analysis. (Table 2) illustrate a large spread in modal olivine content The convolution, modal abundances of metal, and just from 2D polished sections, from 8% to 42%. olivine/pyroxene ratios observable in each CR chondrule vary Heterogeneous—and supra-chondritic—bulk chondrule significantly depending upon which random 2D slice of that silicate compositions are evident in calculated MgO/SiO2 chondrule is measured. While the 3D-CVI defined here offers ratios (Table 6). These results support the hypothesis that the an index that agrees with the eye’s intuition of the degree of chondrules are complementary to each other (and to matrix), chondrule melting, perimeter definition is a subjective in the same sense established for metal. exercise in application. Although they remain subjective, Our data do not permit an assessment of whole chondrule metrics including metal grain size and proximity to chondrule Ni/Fe ratios at the level necessary to address chondrule/ rims could be developed to evaluate textures on larger matrix ratios. Zolensky et al. (1993) report Ni/Fe = 0.0787, ensembles of chondrules. At least in CR chondrites, these and Hezel (pers. comm.) 0.0586, very close to the solar values textures can be quantified most accurately by 3D analysis. of 0.0576 (Anders and Grevesse 1989) or 0.0600 (Lodders We present the first 3D determinations of chondrule 2003), and within probable error of bulk Renazzo. Our data metal/silicate volumetric proportions in chondrules in any do not rule out a sub-solar Ni/Fe for Renazzo chondrules, but chondrite. Combined tomographic and chemical study require that such a signature be controlled by the more strongly supports a complementary chemical relationship abundant, high variance rim grains (Fig. 6). between the entire suite of metal-poor and metal-rich chondrules present in the whole Renazzo meteorite. This Origin of Chondrules in CR Chondrites study also suggests a supra-chondritic Mg/Si ratio in the entire chondrule suite. These findings are consistent with Some less spherical chondrules have fine-grained the hypothesis of “complementarity” between components textures, with chondritic metal grains (~solar Ni/Fe and Co/ in meteorites (Palme et al. 1992; Klerner and Palme 1999; Fe) similar to those on the rims of coarse-grained, rounded Bland et al. 2005; Hezel and Palme 2007). Although some chondrules with Ni-, Co-rich interior metal grains. Rare separation of chondrule silicate from metal by thermal multi-layered chondrules preserve multiple generations of maturation of chondrules may have occurred, local metal- metal grains, separated by silicates. A hypothesis that silicate fractionation may, alternatively, have occurred addresses both observations is that high-Ni, high-Co grains in outside of chondrules, before or during chondrule chondrule cores formed at high temperature, able to formation. Local space could then contribute layers of pre- chemically equilibrate with a vapor containing chondritic separated material to accumulate chondrules with layers of Shape, metal abundance, chemistry, and origin of chondrules in Renazzo 1739 different texture, and also metal grains with volatility- or DeGregorio B. T., Stroud R. M., and Ebel D. S. 2008. Pre- and post- accretionary carbonates in the Renazzo CR chondrite (abstract fO2-controlled compositions. These processes would require repeated, very local heating events, and some physical #1887). 2008 Goldschmidt Conference. CD-ROM. Ebel D. S. 2006. Condensation of rocky material in astrophysical mechanism for independent but local formation of metal and environments. In Meteorites and the early solar system II, edited silicate grains, up to sizes of tens of microns. Magnetic by Lauretta D. S. and McSween H. Tucson: The University of phenomena, surface energetics, or nucleation and growth Arizona. pp. 253–277. physics could be involved in such an early separation of Ebel D. S, and Grossman L. 2000. Condensation in dust-enriched metal from silicates. systems. Geochimica et Cosmochimica Acta 64:339–366. Ebel D. S. and Rivers M. L. 2007. Meteorite 3-dimensional synchrotron micro-tomography: Methods and applications. Acknowledgments–Reviews by D. Hezel, R. Jones, and T. Meteoritics & Planetary Science 42:1627–1646. McCoy very much improved this work. Assistance of William Ebel D. S. 2008. Meteorites: Sample images and data. http:// McDonough, Richard Ash, and Mario Luong at the U. research.amnh.org/users/debel/meteorites. Accessed 17 October Maryland Plasma Mass Spectrometry Laboratory is greatly 2008. Grossman J. N., and Wasson J. T. 1985. The origin and history of the appreciated. This research has made use of NASA’s metal and sulfide components of chondrules. Geochimica et Astrophysics Data System Bibliographic Services. Use of the Cosmochimica Acta 49:925–939. APS was supported by the U.S. Department of Energy, Office Hertz J., Ebel D. S., and Weisberg M. K. 2003. Tomographic study of of Science, Office of Basic Energy Sciences, under Contract shapes and metal abundances of Renazzo chondrules (abstract no. W-31-109-ENG-38. Research was supported by the #1059). 34th Lunar and Planetary Science Conference. CD-ROM. Hezel D. C. 2007. A model for calculating the errors of 2D bulk American Museum of Natural History, the University of analysis relative to the true 3D bulk composition of an object, Chicago (AJC), and National Aeronautics and Space with application to chondrules. Computers and Geosciences 33: Administration grants NANG06GD89G (DSE), and 1162–1175. NANG06GG446 (MKW). Hezel D. C. and Palme H. 2007. Constraints for chondrule formation from Ca-Al distribution in carbonaceous chondrites. Earth and Planetary Science Letters 265:716–725. Editorial Handling—Dr. Kevin Righter Humayun M., Campbell A. J., Zanda B., and Bourot-Denise M. 2002. Formation of Renazzo chondrule metal inferred from siderophile REFERENCES elements (abstract #1965). 33rd Lunar and Planetary Science Conference. CD-ROM. Anders E. and Grevesse N. 1989. Abundances of the elements: Klerner S. and Palme H. 1999. Origin of chondrules and matrix in Meteoritic and solar. Geochimica et Cosmochimica Acta 53:197– carbonaceous chondrites (abstract #1272). 30th Lunar and 214. Planetary Science Conference. CD-ROM. Bland P. A., Alard O., Benedix G. K., Kearsley A. T., Menzies O. N., Kong P. and Palme H. 1999. Compositional and genetic relationship Watt L. E., and Rogers N. W. 2005. Volatile fractionation in the between chondrules, chondrule rims, metal, and matrix in the early solar system and chondrule/matrix complementarity. Renazzo chondrites. Geochimica et Cosmochimica Acta 63: Proceedings of the National Academy of Sciences 102:13755– 3673–3682. 13760. Kong P., Ebihara M., and Palme H. 1999. Distribution of siderophile Campbell A. J., Humayun M., Meibom A., Krot A. N., and Keil K. elements in CR chondrites: Evidence for evaporation and 2001. Origin of zoned metal grains in the QUE 94411 chondrite. recondensation during chondrule formation. Geochimica et Geochimica et Cosmochimica Acta 65:163–180. Cosmochimica Acta 63:2637–2652. Campbell A. J., Humayun M., and Zanda B. 2002a. Partial Krot A. N., Meibom A., Weisberg M. K., and Keil K. 2002. The CR condensation of volatile elements in Renazzo chondrules chondrite clan: Implications for early solar system processes. Geochimica et Cosmochimica Acta 66:A117. Meteoritics & Planetary Science 37:1451–1490. Campbell A. J., Humayun M., and Weisberg M. K. 2002b. Lee M. S., Rubin A. E., and Wasson J. T. 1992. Origin of metallic Siderophile element constraints on the formation of metal in the Fe-Ni in Renazzo and related chondrites. Geochimica et metal-rich chondrites Bencubbin, Weatherford, and Gujba. Cosmochimica Acta 56:2521–2533. Geochimica et Cosmochimica Acta 66:647–660. Lodders K. 2003. Solar system abundances and condensation Campbell A. J., Humayun M. and Weisberg M. K. 2005a. temperatures of the elements. The Astrophysical Journal 591: Compositions of unzoned and zoned metal in the CBb chondrites 1220–1247. HH 237 and QUE 94627. Meteoritics & Planetary Science 40: Mason B. and Wiik H. B. 1962. The Renazzo meteorite. American 1131–1148. Museum Novitates 2016:1–11. Campbell A. J., Zanda B., Perron C., Meibom A., and Petaev M. I. Nettles J. W., Lofgren G. E., Carlson W. D., and McSween H. Y. Jr. 2005b. Origin and thermal history of Fe-Ni metal in primitive 2006. Extent of chondrule melting: Evaluation of experimental chondrites. In Chondrites and the protoplanetary disk, edited textures, nominal grain size, and convolution index. Meteoritics by Krot A. N., Scott E. R. D., and Reipurth B. Astronomical & Planetary Science 41:1059–1071. Society of the Pacific Conference Series, vol. 341. pp. 407– National Institutes of Health, Image Processing and Analysis in Java. 431. http://rsb.info.nih.gov/ij/index.html. Accessed 17 October 2008. Connolly H. C. Jr., Hewins R. H., Ash R. D., Zanda B., Lofgren G. E., Palme H., Spettel B., Kurat G., Zinner E. 1992. Origin of Allende and Bourot-Denise J. 1994. Carbon and the formation of reduced chondrules (abstract). 23rd Lunar Planetary Science Conference. chondrules. Nature 371:136–139. pp. 1021–1022. Connolly H. C. Jr., Huss G. R., and Wasserburg G. J. 2001. On the Tsuchiyama A., Kawabata T., Kondo M., Uesugi K., Nakano T., formation of Fe-Ni metal in Renazzo-like carbonaceous Suzuki Y., Yagi M., Umetani K., and Shirono S. 2000. Spinning chondrites. Geochimica et Cosmochimica Acta 65:4567–4588. chondrules deduced from their three-dimensional structures by 1740 D. S. Ebel et al.

X-ray CT method (abstract #1566). 31st Lunar and Planetary Wood J. A. 1963. On the origin of chondrules and chondrites. Icarus Science Conference. CD-ROM. 2:152–180. Uesugi M. and Sekiya M. 2006. Separation of melted iron spheres in Wood J. A. and McSween H. Y. Jr. 1977. Chondrules as chondrules during the chondrule formation (abstract #1502). condensation products. In Comets, asteroids, meteorites, edited 37th Lunar and Planetary Science Conference. CD-ROM. by Delsemme A. H. Toledo, Ohio: University of Toledo Press. Varley L. R., Leshin L. A., Guan Y., Zanda B., and Bourot-Denise M. pp. 365–373. 2003. Oxygen isotopic composition of Renazzo chondrule Zanda B., Bourot-Denise M., Perron C., and Hewins R. H. 1994. olivine and comparison with extent of chondrule melting Origin and metamorphic redistribution of silicon, chromium and (abstract #1899). 34th Lunar and Planetary Science Conference. phosphorus in the metal of chondrites. Science 265:1846–49. CD-ROM. Zanda B., Bourot-Denise M., Hewins R. H., Cohen B. A., Delaney Weisberg M. K., Prinz M., Clayton R. N., and Mayeda T. K. 1992. J. S., Humayun M., and Campbell A. J. 2002a. Accretion Formation of layered chondrules in CR2 chondrites: A petrologic textures, iron evaporation and re-condensation in Renazzo and oxygen isotopic study. Meteoritics 27:306. chondrules (abstract #1852). 33rd Lunar and Planetary Science Weisberg M. K., Prinz M., Clayton R. N., and Mayeda T. K. 1993. Conference. CD-ROM. The CR (Renazzo-type) carbonaceous chondrite group and its Zanda B., Humayun M., Hewins R. H., Bourot-Denise M., and implications. Geochimica et Cosmochimica Acta 57:1567–1586. Campbell A. J. 2002b. The relationship between volatile element Weisberg M. K., Prinz M., Clayton R. N., Mayeda T. K., Grady patterns and chondrule textures in CRs and OCs. Geochimica et M. M., Franchi I., and Pillinger C. T. 1995. The CR chondrite Cosmochimica Acta 66:A869. clan. Proceedings of the NIPR Symposium on Antarctic Zolensky M., Barrett R., and Browning L. 1993. Mineralogy and Meteorites 8:11–32. composition of matrix and chondrule rims in carbonaceous Wood J. A. 1962. Chondrules and the origin of the terrestrial planets. chondrites. Geochimica et Cosmochimica Acta 57:2123–3148. Nature 194:127–130.