Absence of matrix-like chondrule rims in CR2 LAP 02342

Authors

  • John T. Wasson,

    Corresponding author
    1. Institute of Geophysics and Planetary Physics, University of California Los Angeles, Los Angeles, California, USA
    2. Departments of Earth and Space Sciences and Chemistry and Biochemistry, University of California Los Angeles, Los Angeles, California, USA
    Search for more papers by this author
  • Alan E. Rubin

    1. Institute of Geophysics and Planetary Physics, University of California Los Angeles, Los Angeles, California, USA
    Search for more papers by this author

Abstract

In numerous past papers, it was concluded that the fine (<1 μm) matrix immediately adjacent to, and radially symmetric around, chondrules in primitive chondrites consists of compact (low-porosity) rims that were attached in the solar nebula. We present here textural and compositional evidence that no matrix-like (or accretionary) rims around chondrules are present in the well-preserved CR2 chondrite LAP 02342. Fine-grained matrix-rich regions (i.e., candidate “rims”) at the edges of chondrules were studied with an electron-microprobe-based matrix-grid technique; comparison of the “rims” data for matrix regions near these chondrules showed the candidate “rims” to be compositionally heterogeneous, inconsistent with origins as radially symmetric, matrix-like rims formed by gradual accretion. This evidence (together with simulations and laboratory studies indicating that accretionary processes produced highly porous aggregates) strongly suggests that nebular processes did not produce compact matrix-like rims around chondrules in any chondrite group.

Introduction

Chondrites consist of coarse solids (silicate chondrules and chondrule fragments, metal–sulfide assemblages, refractory and mafic inclusions) and fine solids. For practical reasons, the size boundary between these is commonly defined to be around 1 μm, but, in fact, chondrules are commonly >100× larger and nebular fines <10× smaller than this. Most fines are present in the interchondrule matrix. Matrix regions consist mainly of fragments of chondrules and smoke-like materials that probably formed by condensation (e.g., Brearley 1993; Wasson 2008). Several researchers maintain that fine-grained chondrule rims formed in the solar nebula; common terms for these objects are “matrix-like rims” (e.g., Metzler et al. 1992), “dust rims” (Cuzzi 2004), and “fine-grained rims” (e.g., Huss et al. 2005).

Wasson and Rubin (2009) showed that the porosity of fines in the CR2 chondrite LAP 02342 is around 20%. As summarized by Trigo-Rodriguez et al. (2006), most dust researchers (e.g., Blum and Wurm 2008) argue that such low porosities cannot be achieved in the solar nebula. Based partly on this finding and on petrographic studies of primitive chondrites, Trigo-Rodriguez et al. (2006) argued that, like the closely related matrix, most objects identified as matrix-like rims were formed during the compaction of chondrite parent bodies and were not attached to the “host” chondrules in the solar nebula as compact rims.

It is clear that both chondrules and fines were present together in the solar nebula, so it is possible that there were clumps of nebular materials that consisted of mixes of chondrules and fines in approximately the same proportions now observed in chondrites. This is a very different scenario from the view that matrix-like rims gradually accreted (one particle or one cluster of particles at a time) to chondrules when the chondrules were independent objects orbiting the Sun (Metzler et al. 1992; Morfill et al. 1998; Cuzzi 2004).

The commonly quoted evidence for matrix-like rims is a relatively uniform, radially symmetric layer of fine matrix sometimes observed around chondrules. Metzler et al. (1992) claimed that these are present on CM-chondrite chondrules and that the rim thicknesses are proportional to the chondrule radius. However, Trigo-Rodriguez et al. (2006) found that these structures are rarely radially symmetric, do not in fact have thicknesses proportional to chondrule radii, and seem best explained by production during compaction of porous nebular fines (chondrite matrix material) by impacts on asteroids.

The LAP 02342 CR2 chondrite (Fig. 1) is remarkably well preserved (Wasson and Rubin 2009). It shows no signs of brecciation and has many intact chondrules, a small fraction of which contain clear isotropic glass (Harju et al. 2013). Although it is a find recovered from the Antarctic ice, it is relatively unweathered. Like other CR chondrites, LAP 02342 (commonly referred to as LAP in the remainder of this paper) has a relatively high matrix fraction (approximately 32 vol%—Wasson 2008). It shows essentially no textural evidence of matrix-like rims.

Figure 1.

Backscattered electron mosaic of polished thin section LAP 02342,12 showing areas investigated in the study. The distance between lines on the superposed grid is 1.3 mm. The (generally three or four) matrix-grid regions studied are identified in Fig. 3. Chondrules are assigned names based on the grid and 25-part subdivisions of the individual grid squares. Moderately abundant iron-oxide grains, formed by terrestrial weathering, occur on the left side of the section (B4).

Wasson and Rubin (2009) used a new technique to study matrix in LAP; they used an electron microprobe beam focused to 3 μm to analyze 7 × 7 grids in rectangular regions about 50 μm on a side. Details of the technique are given below. The surprising result of this study was that the grid regions differed significantly in composition. Wasson and Rubin (2009) concluded that nebular fines did not exclusively consist of individual particles having sizes ≤1 μm, but that there were also loosely bound clusters of particles that had compositions differing from the mean of nebular fines.

This discovery suggested that we could use the same technique to investigate whether there are matrix-like rims on LAP chondrules. Because nebular rims would have formed one particle (or one small cluster of particles) at a time, jostling and tumbling would lead to isotropic accretion. It is assumed that, at any particular time and nebular location, turbulence would ensure that fine dust grains would be well mixed. Thus, any nebular rims acquired moving through this dust should be spherically symmetric. Rims could become radially stratified in mineralogy or grain size if the orbits of the chondrules were modified or if rims were acquired over long periods of time during which the characteristics of nebular dust underwent secular changes. Nevertheless, the spherical symmetry of the rims around chondrules in the nebula would be preserved (barring collisions). It follows that nonsymmetric, heterogeneous matrix “rims” around individual chondrules would be inconsistent with the idea that, in the nebula, chondrules accreted rims.

We first review the structures observed around LAP chondrules to assess whether they appear to possess matrix-like rims. We then report matrix-grid studies of areas near several chondrules having high matrix contents in the adjacent regions.

Petrographic Techniques

The techniques used in this study are very similar to those we used in our study of LAP matrix (Wasson and Rubin 2009). In Fig. 1, we show a backscattered electron (BSE) mosaic of thin section LAP 02342,12; a grid is superposed on the mosaic and used to label the chondrules and other features. Each square is subdivided into a 5 × 5 subgrid designated by letters. The top row has the letters A through E, with A at the upper left position. Because of human error, the spacing on the grid is 1.3 mm, rather than the intended 1.0 mm. We retain this grid because it was used to assign names to the chondrules in our previous publications.

This composite image formed the basis for locating optimal target areas (i.e., those having some nearby matrix over an extended arc) for detailed study. As discussed in Weisberg et al. (1993), and as is apparent in Fig. 1, most CR chondrules are low-FeO porphyritic types with large metal grains. Many chondrules in LAP are concentric compound chondrules, with the silicate and metal grain sizes generally coarser in the primary chondrule and smaller in the enveloping secondary chondrule and in the igneous rim, if present.

Three kinds of studies using the UCLA JEOL electron microprobe (EMP) were carried out: (1) elemental maps for five elements in regions ranging in size from 1 to 16 mm2; (2) analyses of 10 elements using a 3 μm diameter electron beam over 7 × 7 point grids, each grid being approximately 50 μm across; and (3) analyses of large phases using a finely focused beam.

Maps

Wasson and Rubin (2009) and Wasson (2008) published BSE and five wavelength-dispersive X-ray maps (Mg, Al, S, Fe, Ni) of four areas of LAP 02342,12 having sizes of 1, 4, 4, and 16 mm2; these were produced by moving the sample stage below a finely focused (d ~ 1.5 μm) electron beam. Point spacings were 2 μm and dwell times were 50–55 ms. We also mapped five additional elements (Na, Si, P, K, Ca) in the three smaller areas. These were used in the selection of targets for this study.

Matrix-Grids

Quantitative information about matrix was obtained by analyzing points 3 μm in diameter in rectangular grids about 50 × 50 μm in size; the spot size was chosen to be larger than matrix grains, but small enough to allow recognition of single-phase grains larger than 3 μm. Each grid area and its immediate surroundings were imaged in BSE, then the 3 μm diameter beam was used to determine 10 elements: Na, Mg, Al, Si, S, K, Ca, Cr, Mn, and Fe; this set includes all metals having concentrations >1 mg g−1 except Ni. Natural and synthetic standards were used: chromite for Cr, forsterite for Si and Mg, albite for Na, Mn-rich garnet for Mn, orthoclase for K, millerite for S, grossular for Al and Ca, and magnetite for Fe. The beam current was limited to 15 nA to minimize loss of alkalis.

We used ZAF corrections. Because the matrix material is so fine-grained, it approaches the homogeneity of a single mineral phase; thus, errors introduced by applying this procedure should be minor. In the Wasson and Rubin (2009) study, we carried out duplicate determinations of several grid areas. These showed no evidence of alkali loss by volatilization. That study confirmed that region-to-region compositional variations were reproducible.

We followed the practice of Wasson and Rubin (2009) and discarded anomalous points. In Table 1, we list the limits that were chosen for elements and for totals. The raw data were used to calculate first-draft means and variances; the most deviant points were removed and the means and variances recalculated. After applying this technique to a number of typical matrix-grid areas, we chose a set of uniform criteria that were applied to remove anomalous analyses. These include (1) low totals reflecting holes and (2) compositions implying that the electron beam encountered large mineral grains. The guiding philosophy for choosing the limits was to eliminate points having elemental concentrations that deviated ≥3σ from the mean and to use the same limits for all matrix regions. Many of the discarded points were analyses of coarse mineral grain fragments embedded in the matrix; some low totals occurred when the beam overlapped holes and fractures.

Table 1. Criteria for rejection of grid points. Concentration limits are listed both in wt% oxide and in mg g−1 element.
ElementLower (%)Upper (%)Lower (mg g−1)Upper (mg g−1)
S1.05.51055
CaO3.021
FeO2640202311
MgO9.023.854.3144
Cr2O3 1.0 6.8
SiO220.53896178
Al2O30.72.553.713.5
Na2O 2.2 16.3
K2O 0.22 1.83
MnO 1.5 11.6
Total75102.5  

Because elevated concentrations of the least-abundant elements (Cr and K) had no appreciable effect on the totals of other elements, we set the upper limit high for Cr and did not use the K concentration as a discriminant. The standard deviations varied among the different areas; the limits were therefore adjusted on the basis of trial and error. In the EXCEL files provided in the data compilation in the supporting information, we flag one property that was the basis for discarding the point, but it was not uncommon that two or more properties were outside the limits. All data (separated into those accepted and those rejected) are listed in the supporting information.

Individual Phases

Individual phases were analyzed using a focused beam and a beam current of 15 nA. The oxide standards listed in the previous section were used for the analysis of silicate and oxide minerals. For metal and sulfide analyses, the standards were millerite for S, 99.99% pure iron for Fe, and the Fe-Ni-Co alloy NBS 1156 for Ni and Co. Cobalt values were corrected for the interference of the Fe-Kβ X-ray peak with the Co-Kα peak.

The Importance of LAP 02342

Primitive chondrites are rare. The nebular record in chondrites can be degraded by thermal metamorphism, aqueous alteration, impact crushing, shock heating, or various combinations of these processes. Terrestrial weathering can also diminish the quality of a study specimen. The CR chondrites include some of the most primitive meteorites (Weisberg et al. 1993). They have largely avoided the effects of thermal metamorphism, but have experienced variable degrees of aqueous alteration. Only two of them are observed falls, Renazzo and the anomalous, matrix-rich Al Rais.

Several CR chondrites are known from Antarctica and most of these show lower degrees of aqueous alteration than Renazzo (Weisberg et al. 1993; Harju et al. 2013). One of the best preserved, least-altered, minimally weathered CR chondrites is LAP 02342. Recently, Abreu and Brearley (2010) inferred, on the basis of transmission electron microscopic (TEM) study of matrix, that the two small Antarctic CR chondrites MET 00426 and QUE 99177 experienced so little alteration that they could be designated type 3.0. Our recent studies (Choi et al. 2009; Harju et al. 2013) show that LAP is slightly more altered than MET and QUE, but that all three CR chondrites should receive type designations in the range of 2.8–3.0.

LAP 02342 shows a beautifully preserved structure (Fig. 1) with many large, intact chondrules; it has experienced relatively little crushing during asteroidal compaction. Although some parts of LAP show moderate amounts of aqueous alteration, most phases in chondrules appear to have preserved their original compositions and, with the exception of Ca, most matrix regions seem to have experienced little elemental transport (Wasson and Rubin 2009). LAP has suffered only minor weathering; the Meteoritical Bulletin lists it as weathering grade A/B. Thin brown halos surround many metal grains (e.g., Fig. 2) and a minor fraction of the metal grains is partially replaced by oxides. Deposits of iron oxides as veins are common only in the surficial region on the left side of the section (Fig. 1); some veins are visible in Figs. 2a and 2b. In most regions, iron-oxide veins are uncommon and easily avoided in compositional studies. Wasson and Rubin (2009) reported that the total abundance of terrestrial oxide in LAP is only 0.2–0.5 vol%. Because of these positive factors, we conclude that LAP 02342 is well suited for studying compositional properties established in the solar nebula.

Figure 2.

Backscattered electron images of a) complete chondrule E3y and b) detailed view of a smooth rim on the N end of the chondrule. Chondrule E3y contains squiggly metal; it is irregular in shape and may be a chondrule fragment. The detailed image shows that the smooth rim has a ragged interface with the chondrule to the S, but a relatively smooth interface toward the porous matrix to the NW. The smooth rim seems to have replaced an igneous rim; some of its constituents seem to have been leached out of nearby matrix. An iron-oxide vein occurs just beyond the smooth rim. A thin weathering rind occurs around the large metal grain at the top of the image.

Chondrule Selection

We first took steps to identify those chondrules that appeared most likely to possess matrix-like rims. We labeled every chondrule, chondrule fragment, and other object having a long dimension ≥0.4 mm. A total of 101 objects were assigned names; two were identified as aqueously altered “matrix lumps” (or dark inclusions) and the remaining 99 were determined to be chondrules or chondrule fragments. These are briefly discussed by Wasson and Rubin (2009). For the 28 largest chondrules (those with long dimensions ≥0.7 mm), we took notes on several features that are summarized in Table S1, including: (1) an approximate FeO content based on BSE brightness: (high, medium, and low); (2) chondrule shape (round or irregular); (3) degree of fragmentation as measured by the fraction of the original perimeter that is preserved (fragments have original perimeters ≤240°); and (4) presence of an igneous rim (yes, no, or uncertain).

We gathered a moderate amount of compositional data on 15 relatively large chondrules in LAP 02342. One of these (H7c) has a high FeO content (olivine Fa48), one (H2q) an unusually low FeO content (olivine Fa approximately 0.2) and Si-bearing metallic Fe-Ni; the remaining 13 have olivine Fa values in the range 1.3–5.8. All of the low-FeO chondrules contain mesostasis regions with a normative anorthitic plagioclase component. More mineralogic/petrographic details about these chondrules will be published elsewhere. These chondrule features are similar to those described by Weisberg et al. (1993) in other CR chondrites.

Nature and Origin of Smooth Rims

Before we discuss matrix-like rims, we need to note that there is a second material that we call “smooth rims” based on texture. These rims are discussed in some detail by Harju et al. (2013). FeO-rich phyllosilicates, and possibly other hydrated phases, are present. These phases are inferred to have formed by aqueous alteration on the CR parent asteroid; based on the textures, the precursor phases appear to have been igneous rims. A key question is to identify the source of the elements now present in these alteration phases. Our data imply that their constituent atoms originated mostly from igneous rims (especially glass), but also, in part, from fine matrix.

Smooth rims typically form relatively uniform arcs around parts of chondrules. An analyzed arc covering about 90° of the circumference of chondrule E3y is shown in the BSE image in Figs. 2a and 2b. The thickness of the rim varies between 20 and 40 μm. The detailed image (Fig. 2b) shows the presence of grains having different degrees of brightness reflecting mainly differences in Fe and/or Ca. A matrix-grid analysis on this smooth rim shows generally low S contents (approximately 1.9 mg g−1 versus 2.3–4.1 mg g−1 in the “normal” regions), high totals, and high Si contents. The analyzed grid area overlapped the chondrule on the NE and overlapped matrix on the SW. Thus, it was not unexpected that there are wide compositional variations among the different regions. We also carried out additional spot analyses to check totals and elemental concentrations in some of the alteration phases (these are reported in Harju et al. 2013). Another area of smooth rim was accidentally included in a grid area analyzed near (ESE of) chondrule J5l.

In the solar nebula, serpentines only become stable at temperatures <400 K; volume diffusion rates are too low to permit chondrule alteration at these temperatures. However, as discussed by Wasson (2008), hydrated silicates can condense from materials that were flash-evaporated during chondrule formation. This process would produce smoke-size particles, not the larger interlocking grains observed in smooth rims. Harju et al. (2013) therefore inferred that these rims are the result of asteroidal aqueous alteration.

Petrographic Search for Matrix-Like Rims

As discussed in Wasson and Rubin (2009), S appears to be the best elemental marker of nebular fines in CR chondrites. As shown in the color images there and in Wasson (2008), S contents are high and relatively uniform in matrix. This contrasts with the low S contents of chondrules in CR chondrites. Other markers of nebular fines include presolar grains and high C and N contents, none of which we are able to determine using our techniques.

A detailed examination of the mosaic in Fig. 1 shows little petrographic evidence for matrix-like rims. Although modal analysis shows that matrix is abundant (approximately 32 vol%) in LAP, no large chondrules exhibit evidence of rims having discernable borders with the surrounding matrix. Most chondrules border matrix along parts of their perimeters.

The structures that are most rim-like are found on two chondrules, lobate chondrule (with squiggly metal) C4q (Fig. 3a) and spheroidal, armored J5l (Fig. 3f), which are surrounded by relatively uniform-appearing fine-grained “rims” about 20 to 50 μm thick. As discussed below, three grid areas on the rim-like feature on C4q gave relatively uniform compositions. We analyzed two grid areas (one in duplicate) on the J5l “rim” and found more variation.

Figure 3.

Images of six chondrules having moderate amounts of nearby matrix chosen as having the greatest potential of hosting matrix-like rims. The locations of the analyzed regions (typically squares approximately 50 μm on a side) are shown, their locations indicated by arrows: a) C4q, b) E4u, c) F3o, d) I1r, e) I6c, f) J5l.

We studied two other chondrules, E4u (Fig. 3b) and I2r (Fig. 3d), which had analyzable matrix patches separated by large >200° angular extents along their peripheries. We also examined four areas adjacent to the large and beautiful compound chondrule F3o, which has interrupted patches of matrix of variable thickness near about 180° of its perimeter (Fig. 3c). We report data for three matrix patches near chondrule I6c (Fig. 3e).

In Fig. 4, we show details of the individual matrix patches whose locations are marked on Fig. 3. These images show that, in most cases, we were able to find 50 × 50 μm regions that consist entirely of fine matrix. In a few cases, such as F3o-SSE (Fig. 4f) and I2r-SE (Fig. 4k), the areas included clasts. As discussed above, data from these clasts were discarded before determining the means.

Figure 4.

Twenty images showing 21 regions within which we measured matrix or “smooth rims” near six chondrules. Data were gathered on 49 points within each of the rectangular regions. Locations were given the name of the chondrule followed by compass coordinate acronyms. Long dimensions of the rectangles were in the range of 45–54 μm.

Matrix Compositional Data

Because matrix-like rims growing one particle (or one particle cluster) at a time should show the same bulk composition on all sides of the host chondrule, the key question is whether or not matrix patches adjacent to an individual chondrule all show similar compositions.

We studied 20 matrix-grid areas adjacent to 6 chondrules; we reanalyzed one area to confirm its very low Al content. Our mean compositional data in the 21 analyses are listed in Table 2. In each case, 49 points in a 7 × 7 array (about 50 × 50 μm) were analyzed. The number of values that were retained (n) is listed in the last line; we chose not to use the data for grid area I2r-E for which we retained only 10 points. We did use the data for F3o-E with only 18 points and also retained some others with 19 or 20 points (Table 2).

Table 2. Mean compositions of matrix-grid areas in CR2 LAP 02342, chosen as potential matrix-like rims adjacent to six chondrules. The bottom row shows the number of points averaged (in left column) and the Student's t-value for 95% confidence on the mean. Data for individual points are listed in the supporting information.
ElementF3o-EF3o-SSEF3o-SSWF3o-WC4q-NC4q-EC4q-W
MeanStdevMeanStdevMeanStdevMeanStdevMeanStdevMeanStdevMeanStdev
Cr2O30.350.120.290.060.380.250.300.070.290.060.300.060.290.06
SiO227.95.632.71.932.12.131.51.434.31.4234.41.0734.51.5
Na2O0.880.161.000.150.490.060.650.070.800.120.960.120.980.09
MnO0.210.060.190.040.210.070.160.050.560.180.550.160.540.18
K2O0.1180.0400.1470.0270.0930.0210.1850.0230.1510.0190.1470.0200.1520.016
Al2O31.810.251.640.401.580.191.640.171.490.191.500.151.520.18
CaO0.910.530.740.980.280.170.290.120.720.600.410.280.430.26
MgO16.62.114.72.316.02.214.51.415.51.1915.51.3214.71.55
FeO31.11.831.72.731.01.932.42.131.31.930.71.2832.32.03
S3.381.022.720.752.980.582.960.642.520.492.660.412.260.49
Total81.64.684.51.7983.72.3883.11.786.31.3685.71.4886.51.6
n 182.101192.093432017402.021442.016422.018482.012
ElementE4u-NE4u-EE4u-SE4u-WI6c-NI6c-SEI6c-S
MeanStdevMeanStdevMeanStdevMeanStdevMeanStdevMeanStdevMeanStdev
Cr2O30.330.060.380.150.400.100.370.060.350.130.350.100.340.07
SiO229.02.429.71.829.42.727.02.128.23.326.73.627.72.3
Na2O0.510.060.900.110.830.140.800.121.470.280.860.120.890.10
MnO0.220.090.190.060.240.080.150.070.210.120.240.120.230.10
K2O0.0770.0180.1270.0160.1070.0200.0880.0150.1540.0300.0990.0210.1040.019
Al2O32.020.201.770.161.980.222.070.171.690.331.840.301.890.36
CaO0.620.500.370.310.720.521.000.620.590.460.620.520.740.65
MgO16.12.3416.12.017.53.216.92.714.83.216.02.416.12.0
FeO31.82.530.61.830.32.928.11.833.93.432.22.831.42.9
S3.270.713.470.713.860.783.200.983.300.733.480.823.420.62
Total82.22.381.82.783.43.278.03.083.03.480.63.681.42.4
n 312.039322.037292.045242.064262.056282.048362.029
ElementI2r-NEI2r-EI2r-SEI2r-SSWJ5l-ESEJ5l-ESE2J5l-SW
MeanStdevMeanStdevMeanStdevMeanStdevMeanStdevMeanStdevMeanStdev
Cr2O30.340.050.370.060.350.040.340.080.200.060.240.110.330.10
SiO228.82.9829.12.829.02.029.21.934.91.435.11.931.82.7
Na2O0.920.161.110.171.000.130.860.090.860.131.060.231.120.24
MnO0.220.060.240.130.190.030.220.040.500.070.500.080.350.14
K2O0.1590.0390.1760.0360.1520.0290.1230.0140.1590.0220.1670.0310.1470.036
Al2O31.890.241.970.241.860.242.180.291.060.181.160.231.890.40
CaO0.720.610.580.250.360.180.740.580.310.080.360.220.690.25
MgO16.62.316.31.316.02.215.31.813.22.112.81.915.33.7
FeO31.33.030.72.732.62.130.82.632.22.232.21.730.92.2
S3.560.993.600.684.140.643.170.781.760.401.950.652.610.88
Total82.72.082.42.383.62.981.43.084.31.984.52.1584.22.3
n 202.086102.262232.069192.093402.021372.027292.045

In Fig. 5, we plot means and 95% confidence limits for six elements in four scatter diagrams (S-Fe, Si-Fe, K-Al, and K-Na). Individual analytical points are plotted as scatter diagrams in Figs. 6-8; the legends on each of the latter diagrams list mean concentrations of the two plotted elements. The chief value of these diagrams is to examine whether the degree of scatter varies from patch to patch (i.e., from one grid area to another) around a single chondrule. (Wasson and Rubin 2009 reported that there are such compositional variations among matrix areas not associated with specific chondrules.) We will show that, with two exceptions, individual patches differ significantly from the other regions around each chondrule in terms of at least two of the six elements. This is very different from what would be expected from accretionary rims. The identification of individual points can be obtained by noting the mean X and Y compositions given in the legends on Figs. 6-8 or (in a less handy way) in Table 2.

Figure 5.

Means and 95% confidence limits are shown for the 20 studied grid areas. Background dots show the means of the matrix-grid areas studied by Wasson and Rubin (2009). As discussed in the text, if the elemental mean of a matrix-grid area is outside the 95% limits of another grid area, it is <5% probable that they are sampling the same matrix materials.

Figure 6.

Data are compared for six elements in grid areas near chondrules F3o and C4q, chosen because they show maxima and minima in degrees of scatter. If the chondrules were surrounded by matrix-like rims formed by acquisition of one particle (or small cluster of particles) at a time in the nebula, those around each chondrule should show similar compositions. The areas around C4q show much tighter clusters than those around F3o. Note especially the contrast on the K-Al and K-Na diagrams. Legends show the mean concentrations of the plotted elements.

Figure 7.

Matrix-grid data for chondrules E4u and I2 are compared. The areas show similar degrees of scatter, and evidence of compositional variations among different regions around each chondrule, inconsistent with formation as accretionary rims. Note, for example, the large amount of scatter among the matrix fields on the S-Fe and K-Al diagrams for each chondrule. Legends show the mean concentrations of the plotted elements.

Figure 8.

Data are compared for six elements in grid areas near chondrules I6c and J5l. The areas show similar degrees of scatter, and evidence of compositional variations among different regions around each chondrule, inconsistent with formation as accretionary rims. Note how anomalous the J5l data are—very low in S (c) and Al (d), very high in K (d, h) and Si (g) where they appear to reflect aqueous alteration. Region N in I6c is well resolved from the other fields (e.g., in K on b), but, as discussed in the text, the two fields S and SE are not resolved by any of the six elements. Legends show the mean concentrations of the plotted elements.

We introduce the discussion of our matrix-grid data by calling attention (in Fig. 6) to the data sets showing the most and the least scatter; four regions are near the large compound chondrule F3o and three regions are near the smaller, irregularly shaped chondrule C4q. The four grid areas near F3o show large variances and have very different compositions; in contrast, scatter fields for the C4q regions are relatively compact and similar in composition. The areas around the remaining chondrules show properties between these extremes.

The proper way to determine whether two means differ is to use Student's t-test. We carried out this test for some sample pairs and reached the generalization that, if the 95% limits on each mean do not overlap the other mean, then it is <5% probable that the two data sets are samples of the same population. In fact, in most cases the full calculation shows that it is ≪5% probable. It should be noted that each element provides independent evidence regarding possible relationships; thus, it is necessary that the uncertainties on the means for all elements overlap to justify a claim that grid areas are sampling the same population. In Fig. 5, we show the mean compositions and 95% confidence limits for the 20 grid areas as well as the duplicate analysis of J5l-ESE (which is plotted adjacent to the other analysis and is shown without confidence limits). We will refer to information summarized in Fig. 5 as we discuss the detailed scatter diagrams.

Matrix Near F3o and C4q

In Fig. 6, we compare F3o and C4q using six elements on four diagrams (with Fe appearing twice as the independent variable and K twice as the dependent variable). These sets were chosen to emphasize the contrast among grid regions. All data are plotted except those that were discarded because they showed large deviations from the mean composition (as discussed in the matrix-grids section). The four pairs of diagrams for F3o are plotted above those for C4q.

For the F3o regions, the K-Na plots and K-Al diagrams show the greatest contrast. On the K-Na diagram (Fig. 6f), there are very few points that overlap in the four grid areas near F3o; one does not need to compare the means and confidence limits in Fig. 5 to confirm that none of these grid areas is sampling the populations sampled by the other three grid areas. (Although cumbersome, it is straightforward to use the means listed in Figs. 6-8 to identify the points in Fig. 5.) The most closely related are E and SSE, but the means plotted on Fig. 5d show that neither mean is overlapped by the confidence limits attached to the other mean. The situation is similar on the K-Al diagram (Fig. 6b), but there is a very large amount of scatter in the Al value for F3o-SSE with the result that it is not resolved from any of the other grid areas in terms of this parameter. Two F3o grid areas that are not resolved in terms of Fe, S, or Si (SSE and SSW) are strongly resolved in terms of K and Na.

If we compare the two areas in terms of the S-Fe and Si-Fe spaces, we see that there is again much more scatter in the data for the areas near F3o (Figs. 6a and 6e) than in the C4q data (Figs. 6c and 6g), but that one cannot immediately recognize that the F3o grid areas are resolvable. Examination of Fig. 5b shows that the uppermost F3o point (SSE) has so much scatter in Fe that it cannot be resolved from the other areas in terms of this element, but that Si resolves it from two areas (E and W). Figure 5a shows that the same situation prevails in terms of the S-Fe diagram where SSE (the lowest point) is resolved from E and W, but is again not resolved from SSW (the leftmost point). Both Na and K strongly resolve SSE and SSW.

These results are summarized in Fig. 9, a “truth table” that shows open or filled circles for the six elements plotted in Figs. 5-8; a filled circle indicates that the means differ at confidence levels >95%. The results for F3o are summarized in a set of six rectangles. As noted, SSE and SSW are resolved by only two elements, whereas the remaining five rectangles show that the grid areas are resolved by three to five elements. These four grid areas are clearly not parts of a radially symmetric matrix rim formed in the solar nebula.

Figure 9.

This truth table shows the relationship among the different matrix regions around individual chondrules. Each rectangle provides a comparison of two analyzed grid areas near the named chondrule. The rectangles have circles for each of the six elements included in our study. The circles are filled if that element shows no correlation among the compared grid regions.

At the other extreme in terms of compositional resolution are the three grid areas in C4q; their scatter fields appear much more compact than those of the F3o areas in Fig. 6. As can be seen from the last line of Table 2, we discarded very few points (1, 5, and 7) from these areas; only one F3o area is similar (SSW, 6 points). Note, in particular, how compact and overlapping the distributions are in Fig. 6g. However, despite the impression one gets, the three grid areas are resolved in terms of Fe (Fig. 5). The C4q grid areas plotted in K-Al space (Fig. 6d) are not as compact, but Fig. 5b shows that they cannot be resolved in terms of either element. Grid area C4q-N has a much lower Na content than the other two and is readily resolved on Figs. 6h and 5d. Grid area C4q-W, which could not be resolved from C4q-E in terms of Si, is fully resolved from this area and from C4q-N in terms of S (Figs. 6c and 5a). Figure 9 summarizes the results; although the three grid areas near C4q are relatively close to each other in composition, they are resolvable in terms of two or more elements. As discussed below, the fact that the C4q grid areas have higher Si contents and lower variances than those around other chondrules may reflect a higher degree of aqueous alteration.

Matrix Near E4u and I2r

In Fig. 7, we show data for E4u and I2r; both matrix-grid areas show much scatter. Although we show the data for all four areas near I2r, we do not attach significance to grid area E for which we discarded 80% of the points. Chondrule E4u is a lobate compound chondrule that contains two, and possibly three, generations of “concentric” shells of material. The outermost shell is an igneous rim with a high density of small metal grains. Chondrule I2r is a dumbbell-shaped chondrule with unusually large (up to 300 μm) phenocrysts. In contrast to the other chondrules, it has surficial deposits of FeS and almost no metal in the interior.

The data for the grid areas show moderate scatter and moderate overlap around each of these chondrules. Area E4u-W clearly has a lower Fe content (e.g., Fig. 7e), area E4u-E a higher K content (e.g., Fig. 7b), and E4u-N a lower Al content (e.g., Fig. 7f) than the other E4u areas. Area I2r-SSW has lower K contents (Figs. 7d and 7h) and a lower Na content (Fig. 7h) than the other I2r areas.

Examination of the truth table (Fig. 9) shows that all the areas in E4u are resolved by four to five elements. Among the three areas near I2r (I2r-E was not considered), I2r-SE is well resolved from NE and SSW (by three to five elements), but NE and SSW are resolved by only one element, K.

Matrix Near I6c and J5l

In Fig. 8, we show data for I6c and J5l. Chondrule I6c is round and, based on the distribution of coarse metal, is compound. On the west side is a patch of silicate covering about 45° of arc that contains fine interior metal characteristic of an igneous rim. As can be seen in the images (Figs. 3e and 4p,s,t), there is relatively little matrix adjacent to the perimeter of I6c. Compositionally anomalous grid area I6c-N is located in an embayment just left of the top of the chondrule (Figs. 3e and 4p). The other two regions (S and SE) are located about 400 μm apart in a matrix region showing a relatively uniform texture. Among the 17 areas we used to test for matrix-like rims, this is the only pair that could not be resolved.

Chondrule J5l is round with large exterior metal grains, a classic CR armored chondrule texture. With the exception of the K-Na diagram for I6c (Fig. 8f), the grid areas show much scatter.

Grid area I6c-N has exceptionally high K (Fig. 8b) and Na; the latter values are so high that about 22 of 26 retained points plot off scale on Fig. 8f. An additional five points had Na contents >2.2%, and were discarded. Not surprisingly, Fig. 5 shows that I6c-N can be resolved from the other two grid areas in terms of Na and K; it is also resolvable from these areas in terms of its higher Fe and lower Al. Because alkali elements can be transported in flowing water, we considered the possibility that their enrichments are associated with aqueous alteration rather than nebular processes. We decided that the evidence is inconclusive and included I6c-N in Fig. 9. The other two grid areas (S and SE) could not be distinguished by any of the six elements we included in Figs. 5-8. This pair is the only case where our data are consistent with them being part of a matrix rim. As noted above, the matrix region I6c-N has a very different composition from this pair.

Because of its exceptionally low Al contents (Fig. 8d), we carried out a replicate analysis (offset to the east by about 8 μm) of area J5l-ESE. The second set of Al values was essentially the same (lightly shaded diamonds in Fig. 8d); thus, the Al values are well duplicated. The same two grid areas have the lowest S contents that we observed. After gathering the data, we found that ESE overlapped a smooth, phyllosilicate rim on J5l; the offset to the east of ESE2 reduced this effect, but resulted in even stronger compositional anomalies (e.g., higher mean Na). The ESE and ESE2 means have the two highest Si contents, although they are only marginally higher than those in the three grid areas near C4q. We suggest below that the aqueous alteration responsible for producing phyllosilicate rims is also responsible for the higher SiO2 content and higher totals in J5l-ESE areas and in the three areas near C4q. We therefore chose not to attach importance to a comparison of J5l areas SW and ESE, and did not include a comparison in Fig. 9.

Conclusions: No matrix-Like Rims in LAP 02342 and Probably No Matrix-Like Rims Around Any Chondrules

There is little textural evidence that suggests matrix-like rims in CR2 LAP 02342. Although we chose the six most promising chondrules for study, they do not show uniformly thick rims with well-defined outer boundaries to the interchondrule matrix (Fig. 3). The textural evidence for matrix-like rims was appreciably weaker in the remaining 24 chondrules (in thin section LAP 02342,12) having long axes >0.75 mm.

As shown in Figs. 5 and 9 and discussed above, the compositions of the grid areas near individual chondrules are, with one exception (areas S and SE near chondrule I6c), resolvable in terms of one or more elements. The mean compositions in 10 matrix-grid areas studied by Wasson and Rubin (2009), shown as dots in Fig. 5, cover essentially the same broad composition regions as those from the present study. Although the density of the points in certain regions is higher or lower, this is attributable to stochastic sampling of the continuum. It appears that the two sets of grid areas are sampling the same reservoir of rather diverse matrix compositions.

An unanticipated problem is that grid areas around three of the six investigated chondrules are anomalous. Two of the six chondrules (C4q and J5l) have smooth, phyllosilicate rims. As discussed by Harju et al. (2013), these seem to have formed as a result of aqueous alteration of a pre-existing anhydrous igneous rim. It appears that the matrix areas we chose near these chondrules were in place at the time of the aqueous alteration, and that their compositions were somewhat altered by the transport of elements into and/or out of the analyzed grid areas. Four of the five investigated grid areas around these chondrules showed low Al, high alkali, and high Si values. We suspect that all of these effects are the result of the same aqueous alteration events that produced the phyllosilicate rims. One of the two investigated grid areas near J5l was so anomalous that we discarded it, leaving no area to compare with the other J5l grid area that seems to be “normal.” We retained all three studied areas near chondrule C4q; as shown in Fig. 9, they are resolved from each other by two to three elements.

Although the aqueous alteration effects are patchy, we suggest that they are too common to allow the designation of LAP as subtype 3.0.

Anomalous grid areas were also found near dumbbell-shaped chondrule I2r, the only large chondrule in our set of 30 with large amounts of FeS at the surface. As noted above, we had to discard many points from these grid areas. In the worst case, we discarded 39 of 49 points in grid area I2r-E and in the best case (I2r-SE), we discarded 26 of 49 points. The images of these grid areas (Figs. 3i–3l) show the presence of clasts, but nothing that would imply that the fine, porous parts are anomalous. Because only ten points remained in I2r-E, we did not include it in the Fig. 9 truth table.

The one case where two grid areas could not be resolved at a >95% confidence level are I6c-S and I6c-SE, which are separated by about 400 μm. Grid area I6c-N (which has an anomalously high Na content) is well resolved from both of these. After discarding J5l-ESE made it impossible to compare it with J5l-SW, we are left with 19 grid areas; a check on these shows that the only two candidate regions that cannot be resolved are these two in I6c. Their compositional relationship appears significant; these two areas have compositions that support the idea of matrix-like rims, but the grid area I6c-N on the opposite side of the chondrule conflicts with this view. We conclude that I6c-S and I6c-SE are really sampling the same batch of nebular fines, but that one such instance offers negligible support for the idea that the chondrules in LAP 02342 show evidence of matrix-like rims.

As stated above, the six chondrules were chosen because they showed the strongest textural evidence for matrix-like rims from among the large (D > 0.75 μm) chondrules in thin-section on LAP 02342,12. Examination of Fig. 3 shows that the best-defined (and relatively dark) areas are near chondrules C4q and J5l. We now know that both these chondrules have smooth, phyllosilicate rims adjacent to matrix regions that, in four of five cases, have SiO2 contents about 2% higher than the highest among the other analyzed grid areas, and about 5% higher than the means of the remaining areas. We therefore infer that SiO2 was transported into these matrix areas during the aqueous alteration that affected these two chondrules. This suggests that aqueous alteration was localized, and that the addition of this SiO2 provided cement that helped to keep these matrix regions close to the chondrule during mild impact gardening of the regolith.

As discussed in detail by Blum and Wurm (2008), the physical processes occurring in the solar nebula do not appear to be capable of producing the low porosities observed in chondritic matrix. It was such considerations that led Trigo-Rodriguez et al. (2006) to reinvestigate the rim-like structures observed in CM chondrites, a study that led to the conclusion that these structures (that look like rims in a minority of cases) are the result of compaction produced by impacts on the parent asteroid. Trigo-Rodriguez et al. concluded that the porosity was lowest in matrix regions adjacent to incompressible objects, such as chondrules. We now suggest that aqueous alteration on the CM asteroid may also have led to SiO2 deposition into these regions of compacted matrix and that the SiO2 served as cement that resulted in greater strength and thus facilitated preservation during later impact events.

If compact matrix-like rims were commonly produced in the solar nebula, they should be present in all classes of chondrites and should be especially well preserved in CR chondrites that have experienced negligible thermal alteration and, in the case of LAP 02342, only minor amounts of aqueous alteration and terrestrial weathering. The absence of such rims around LAP chondrules and the absence of definitive evidence of their presence in CM and other chondrites lead to the conclusion that compact rims were not formed around chondrules in the solar nebula. The uncommon chondrules surrounded in two dimensions by roughly concentric distributions of compact matrix are almost certainly stochastic features of asteroid compaction processes.

Acknowledgments

We are grateful to Frank Kyte for help and advice in the use of the electron probe and to Ellen Harju for data and discussions. We thank the Meteorite Working Group and the JSC curators for the loan of the LAP 02342,12 section. Henning Haack provided a most useful detailed review, but disagrees with important details of our conclusions. Mike Zolensky provided valuable suggestions. This research was mainly supported by NASA grants NNG06GG35G and NNX10AG98G (JTW) with additional support from NASA grant NNG06GF95G (AER).

Editorial Handling

Dr. Michael Zolensky

Ancillary