Chondrules: Chemical, Petrographic and Chronologic Clues to Their Origin by Impact

Major element contents of chondrite groups were volatility controlled and established in a nebula after formation of Ca-Al-rich inclusions but before chondrules formed. Elemental abundances in chondrules tend to correlate with chemical affinity. Calcium was fractionated from Al by a planetary, not a nebular process. Chondrules were contemporaneous with igneous activity and aqueous and thermal metamorphism. Planetary bodies of varied size and structure co-existed during the first 50-80 million years of the Solar System, when chondrules formed and impact was common. We propose that most chondrules formed by impacts on differentiated bodies. 1. General Introduction Chondrite meteorites are aggregate rocks that escaped melting since they formed in the nascent Solar System some 4565 million years (Ma) ago. They have near-solar elemental proportions that vary somewhat among the three main chondrite classes (see Scott & Krot, this volume): carbonaceous (atomic Mg/Si >1); ordinary (Mg/Si ~0.94), and enstatite (Mg/Si <0.9). The differences among the groups reflect differing reservoirs during accretion, due either to temporal or heliocentric effects. Some chondrites largely escaped the effects of parent body heating and fluid alteration. The constituents of these unequilibrated chondrites thus preserve clues to the process (es) that formed preplanetary matter and to the planetary accretion process itself. The defining constituents of chondrites are chondrules, which occur in all but the CI chondrites. Chondrules, therefore, were widespread within at least the innermost So934 Hutchison, Bridges, and Gilmour lar System when planets were forming, and their origin is the subject of this chapter. Chondrules by definition solidified from partiallyto completely molten droplets. Thus chondrules generally are sub-spherical in shape (or were, prior to fragmentation or deformation), and most range in size from sub-mm to several mm in diameter (much larger and much smaller examples are known; see Connolly & Desch 2004; Krot & Rubin 1996). Chondrule compositions predominantly are rich in ironmagnesium silicates, although some are metaland/or sulfide-rich. Excluded from this definition are calcium-aluminium-rich inclusions (CAIs), many of which also solidified from molten droplets but whose compositions are (as their name suggests) very different. The “chondrule problem” is really two-fold: what process or processes established their bulk compositions, and what process or processes caused melting. Except for a narrow class of models that interpret chondrules as the products of direct melt condensation (as opposed to melting of solid precursors), these two aspects of chondrule petrogenesis are largely (but not completely) separate. The difficulty in the first instance is that the melting process destroys most or all physical traces of the material that was melted (e.g. Dodd 1981, p. 61). Even for the many chondrules that were incompletely melted and contain relict grains, such grains may well derive from an earlier episode of chondrite melting and need not represent the ultimate precursors (Jones 1996). In any case, heating certainly destroyed the record of the physical condition of the precursor material, so there is little direct evidence for the environment in which chondrules formed. Regarding the second part of the chondrule problem, only characteristics of the cooling and solidification part of the melting-solidification cycle are preserved in the final product; inferring the heat source is vastly more difficult. Judging from the papers presented at chondrule conferences in 1982, 1994 and 2004, most workers interpret chondrule formation as occurring over a short timescale within the dusty protosolar nebula or accretion disk. Some argue for early formation with bodies of 20-3000 km diameter as intermediaries (Melosh et al. 2004; Sanders & Taylor, this volume). In contrast, we argue for chondrule formation over extended timescales by disruption of planetary bodies. We review the chemical, physical, and isotopic properties of chondrules and their host chondrites that constrain theories of formation. Although chondrite bulk compositions are related to elemental volatility and presumably resulted from nebular processes, we argue that some chemical differences between chondrules cannot have been achieved by evaporation or condensation. Rather, they arose differently, possibly by crystal-liquid fractionation and core formation in planetary bodies. Chondrules are viewed as one component in an environment that included a range of bodies, and they evolved in a time-frame that also encompassed asteroidal metamorphism, melting, core formation, and massive asteroidal/planetary collisions. Formation of Chondrules by Impact 935 2. Inconsistencies with a “Nebular” Chondrule Origin 2.1 Chondrite vs. Chondrule Chemical Fractionation CI chondrites (the group that lacks chondrules) have bulk chemical compositions closest of all chondrites to that of the volatile-free Sun (Anders & Grevesse 1989). Relative to this “cosmic” composition, the compositions of the various chondrite groups record three kinds of chemical fractionation (Larimer & Anders 1970). The first is known as the refractory lithophile (= rock-forming) element fractionation. Refractory lithophile elements (e.g. Al, Ca, Ti) are enriched in the carbonaceous chondrites, in part by the addition of the early-formed CAIs to solids of “cosmic” composition (Larimer & Wasson 1988). Ordinary and enstatite chondrites, on the other hand, are depleted in refractory lithophiles (Larimer & Anders 1970). The second chemical fractionation established the Mg/Si ratios diagnostic of the chondrite classes. The third chemical fractionation involved the addition or removal of Fe, Ni and Co metal in “cosmic” proportions. Although the mechanisms by which these different chemical fractionations occurred are debated, it is generally agreed that fractionation between chondrite groups was a nebular process. An important observation is that the fractionations exhibited by the bulk chondrite compositions are in some respects decoupled from those exhibited by chondrules. For example, the differing bulk compositions of the different chondrite groups are not the result of differing chondrule abundances. Consider the CM2 and CO3 carbonaceous chondrites, which have chondrule/matrix ratios of ~0.3 and ~1.3 respectively (Scott et al. 1996), and yet have similar major element chemistry (Wasson & Kallemeyn 1988) (Fig. 1a). Consider also the LL group chondrites, which contain abundant volatile-poor porphyritic chondrules (Alexander 1994) and yet are enriched in moderately volatile Mn, K and Na relative to chondrule-poor CM chondrites (Wasson & Kallemeyn 1988). We concur with Huss et al. (this volume) that major element fractionation between chondrite groups occurred before chondrules formed. Another example of chondrule-chondrite chemical decoupling is in their respective Ca/Al ratios (Fig. 1). The solar (CI) Ca/Al value is 0.72 (Fig. 1a). Bulk chondrites exhibit an almost constant mean Ca/Al ratio (Fig. 1a), ranging from 0.76 (atomic) for R chondrites to 0.65 for the EL group. The range among individual meteorites is not much greater: 0.81-0.67 among 59 H, L and LL petrologic types 3-6 ordinary chondrites (Kallemeyn et al. 1989), and 0.82-0.66 among 16 CM, CO and CV chondrites (Kallemeyn & Wasson 1981). The respective individual chondrules, however, have a wider range of Ca/Al ratios (Figs. 1b-d). The Ca/Al ratios of 6 unaltered, FeO-poor chondrules from Murray (CM2) range from 1.23 to 0.31 (Fig. 1b). The range of Ca/Al in 5 chondrules in Allan Hills 85085 (CH3) is 0.99-0.48; that of PO chondrules in Semarkona is 0.94-0.55 (Fig. 1c). The range of Ca/Al, 0.84-0.55, in 9 Type IA and IAB chondrules from CR chondrites is greater than that of H, L, LL, CM, CO and CV chondrite groups. Finally, glass-rich chondrules in unequilibrated ordinary chondrites (UOCs) have extremely low Ca/Al ratios (Fig. 1d), the mean of 12 chondrules being 0.12 (Krot & Rubin 1994). It might be argued that because the data shown in these figures were collected using electron beam microanalysis of thinsections, the random surface areas exposed for analysis are non-representative of the 936 Hutchison, Bridges, and Gilmour whole chondrules in many cases. However, instrumental neutron activation analyses (INAA) of bulk chondrules yield the same results. Twenty-nine chondrules (unclassified) from Semarkona (LL3.0) have atomic Ca/Al ratios in the range 0.900.61 (Grossman & Wasson 1983). Moreover, different chondrule types give systematically different results. For example, in H3 chondrites the mean Ca/Al ratio of radial pyroxene chondrules is 0.96 whereas that of barred olivine chondrules is 0.33 (Fig.1c) Figure 1. Ca/Si vs Al/Si atomic ratios of chondrites and chondrules. (a) Mean ratios of chondrite groups (data of Hutchison 2004). Lines have Ca/Al ratios of 0.82 and 0.66, the upper and lower limits of individual H, L, LL, CM, CO and CV chondrites (Kallemeyn & Wasson 1981; Kallemeyn et al. 1989). (b) Ratios in individual chondrules from carbonaceous chondrites: FeO-poor chondrules in CR chondrites (Connolly et al. 2001); various chondrule types in Allan Hills 85085, CH3 (Weisberg et al. 1988); cryptocrystalline (CC) and skeletal olivine (SO) chondrules in two CBb chondrites (Krot et al. 2001); pristine chondrules in Murray, CM2 (Rubin & Wasson 1986). Dashed reference lines of Ca/Al atomic ratios of 0.82 and 0.66, the limits in chondrites. (c) Ratios in individual FeO-poor (Jones & Scott 1989) and FeO-rich (Jones 1990) PO chondrules from Semarkona, LL3.0; mean ratios in 8 radial pyroxene (RP) chondrules and 7 barred olivine (BO) chondrules from H3 chondrites (Lux et al. 1981); ratios in 4 granular olivine chondrules (Weisberg & Prinz 1996). Separate regression lines are drawn through the FeO-poor and FeO-rich chondrules (d) Individual glass-rich chondrules in type 3 ordinary chondrites (Krot & Rubin 1994). Formation of Chondrules by Impact 937 (Lux et al. 1981). In Semarkona, Type IA (FeO-poor) chondrules define a different slope from type IIA (FeO-rich) chondrules (Fig. 1c; Jones 1990). Differences in Ca/Al ratio are, therefore, not an artifact of sampling. Rather, the differing chondrule varieties formed from different precursors. At least in the case of Semarkona, the observations of Jones & Scott (1989) and Jones (1990) argue that the chondrules were neither aqueously altered nor thermally metamorphosed, so secondary processing was not the cause of Ca/Al fractionation. We conclude that variations in Ca/Al ratios are a primary feature of chondrules, and those variations are different than exist in bulk chondrites. One other striking contrast between chondrites and chondrules exists, namely in how volatile and refractory elements correlate. In many (commonly glassy) chondrules, enrichment in refractory Al paradoxically is accompanied by enrichment in moderately volatile Na rather than refractory Ca. Aluminium is associated with the Na in chondrule mesostases. Similarly, Al is also correlated with more volatile Cr in rare chromite-rich chondrules (Krot et al. 1993), and the elements occur together in spinel minerals (MM2O4; M = divalent Mg or Fe, and trivalent Al, Fe or Cr). Thus, in chondrules, elements tend to be associated on the basis of shared chemical affinity rather than volatility, in contrast to the situation among chondrites. Possible causes of the elemental enrichments or depletions in chondrules are discussed below. 2.2. The Duration of Chondrule Heating is Unconstrained If chondrules formed by melting of solid precursors in a nebular environment, experimental results indicate that a thermal spike is required to produce the range of observed textures and compositions, either a few minutes at 1500−1850°C or hours at perhaps to 1400−1750°C (Hewins et al. this volume). Evaporation of FeO and SiO2 from melts could yield more magnesian compositions in chondrules without the need for the very high temperatures that might otherwise be demanded, so PO chondrules may have formed at 1400−1600°C. On the other hand, if chondrules formed by the break-up of internally heated partly molten bodies, there is no need for a sudden temperature spike. Impact, however, would have caused additional local heating. It is thus essential to include heating rate in discussions of chondrule origins. Consider, for example, how so-called flash heating would affect objects of different sizes. Would rapid heating of finegrained precursors yield the same internal temperature in 0.1 mm and 10 mm objects? Would a small object vaporize while a large object melted at, say, 1750°C? If the peak temperature had been uniform throughout small chondrules, would we not see evidence of temperature gradients in cm-sized, ones? Had heating been rapid, we would expect to find some objects with dusty, unmelted or sintered cores grading outwards into melted rims, or the reverse, objects with melted cores grading outwards to dusty rims, depending on the heating mechanism. Such gradations have not been observed. It has been proposed that rare agglomeratic or granular olivine chondrules, and certain rims around mainly porphyritic chondrules, were flash-heated (see below). The rims, however, formed in secondary events that acted on chondrules that had already undergone primary melting and solidification. Agglomeratic types excepted, throughout their size-range chondrules exhibit no textural evidence of temperature 938 Hutchison, Bridges, and Gilmour gradients incurred during primary melting. Some chondrules have chilled margins consistent with rapid cooling, such as an outer olivine “shell” around many barred olivine chondrules. No chondrules have unmelted or partially melted interiors that grade outwards into more highly melted margins, or the reverse, unmelted or partially melted rims that grade into more highly melted interiors. Rubin & Krot (1996) suggested that survival of partially resorbed relict grains (derived from earlier generations of chondrule melts; Jones 1996) in some porphyritic chondrules requires rapid heating. If, however, the grains are xenocrysts that were introduced into pre-existing melts or partial melts, rapid heating may not be required. In fact, adding crystalline material to chondrule melts has been proposed as a means of promoting crystallization (see Connolly & Desch 2004). Survival of relict grains in chondrules probably requires cooling rates towards the higher end of the postulated range, 10 1000°C/hr (Hewins et al. this volume), rather than rapid heating. Coarse-grained (Rubin 1984) and igneous (Krot & Wasson 1995) chondrule rims and microchondrule-bearing chondrule rims (Krot & Rubin 1996) have been cited as support for chondrule formation by rapid heating. In ordinary chondrites, igneous rims have sharp contacts with their host chondrules; rims on low-FeO chondrules mainly formed by melting the host chondrule, whereas those around high-FeO chondrules “appear to consist largely of melted matrix-like materials” (Krot & Wasson 1995). Rare FeO-poor chondrules in highly unequilibrated L and LL chondrites have rims of “fine-grained FeO-rich matrix-like material” that contain microchondrules (Krot & Rubin 1996). The microchondrules consist mainly of magnesian Ca-poor pyroxene, as in the host chondrules, but some others consist of Fe-rich olivine. Host chondrules have pyroxene-rich surfaces that are “irregular and show evidence of remelting” (Krot & Rubin 1996). Rubin & Krot (1996) recognize that chondrules are the products of multiple heating, but the lack of gradational contacts between chondrule interiors and melted or partially melted rims suggests that these rims formed by secondary heating of “previously formed chondrules or chondrule fragments” (Rubin & Krot 1996). Thus, the primary heating of chondrules was unrelated to rim formation. Chondrule melting was more pervasive, suggesting a longer heating duration than the event that formed the rims. 2.3. Most Chondrules in UOCs are Abraded Remnants of Larger Objects In some UOCs, chondrules with abraded margins outnumber whole solidified droplets by about 3:1 (Dodd 1981, pp. 121-122). Abrasion may have been caused by impact or disturbance during compaction on parent bodies. Regardless, petrofabric analysis has revealed that in some porphyritic olivine chondrules, the long axes of olivine phenocrysts have “a linear preferred orientation ... thought to reflect flowage of a partly crystalline parent magma” (Dodd 1969). Such chondrules were interpreted as solidified pieces from larger semi-molten magma bodies, rather than as melt droplets. 2.4. CAIs, Igneous and Metamorphic Rocks and Low-Temperature Matrix Coexisted with Chondrules It is well known that chondrules occur in chondrites together with clear products of very early nebular processes, such as CAIs (Russell et al., MacPherson et al., and Formation of Chondrules by Impact 939 Jones et al. this volume). There can be no doubt that CAIs were present throughout the chondriteaccreting regions, even though CAIs probably formed in a different location from ferromagnesian chondrules (Krot et al. 2002). Fine-grained matrices in the least metamorphosed carbonaceous, ordinary and enstatite chondrites also contain tiny diamond, silicon carbide, and graphite grains whose isotopic ratios or fractionated noble gases are indicative of a presolar origin (Huss & Lewis 1994, 1995). What is less-well remembered is that chondrules also coexist in chondrites along with products of much later, parent body, processes such as melting and metamorphism. Fragments of igneous rock occur in members of many chondrite groups: CO (Kurat & Kracher 1980), CV (Kennedy & Hutcheon 1992), H/L, L and LL (Hutchison 1992; Kennedy et al. 1992; Bridges et al. 1995). Dodd (1981) cites examples of metamorphosed chondrules or lithic clasts in UOCs. Mezö-Madaras is an L chondrite composed of (1-10) cm equilibrated L chondrite clasts in dark, chondrule-rich material of petrographic type 3. It is a fragmental breccia whose components had different thermal histories (Binns 1968), but see section 2.5, below. Igneous and metamorphosed lithic and mineral clasts, chondrules, matrix, CAIs, sulfide and metal coexisted and accreted together to form chondrite parent bodies. To identify processes that were contemporaneous with chondrule formation, we discuss the times when these components formed. This should help us evaluate potential chondrule-forming mechanisms. 2.5. Chondrule Formation, Igneous Activity, Thermal and Aqueous Metamorphism and Core Formation Overlapped in Time Diverse kinds of radiometric age data have been used to argue that chondrules must have formed in the solar nebula prior to formation of any asteroidor planet-sized bodies. In fact, the situation is not so simple. CAIs are the oldest measured objects that formed in the Solar System, and ages of other primitive objects are commonly compared to those of CAIs. Unfortunately, the precise relative timing of events that were separated in time by only a few Ma in the early Solar System is difficult to achieve using absolute radiometric age dating. Short-lived radionuclides that were present in the solar nebula provide the means to achieve relative (not absolute) age differences as short as 10 years in some cases. One widely applied system, based on the decay of Al to Mg, indicates that many, but by no means all (MacPherson et al. 1995), “normal” CAIs formed ~1.5 Ma before the onset of chondrule formation (Kita et al. and Russell et al. this volume). This is derived from the fact that the inferred initial Al/Al ratios in CAIs, typically ~5×10, are several times higher than ratios measured in chondrules. However, the short half-life of Al, 0.73 Ma, restricts the chronologic usefulness of this system to the earliest ~4 Ma of chondrule history. After ~4 Ma most of the Al had decayed, so the method is incapable of yielding precise ages for objects formed after this time and merely sets upper limits (Huss et al. 2001; Mostefaoui et al. 2002). Additionally, interpretation of the Al-Mg isotopic system in terms of chronology within a chondrule or CAI is unambiguous only when the Mg/Mg ratios (from which Mg excesses, due to the in situ decay of former Al, are measured) in the constituent phases correlate with Al/Mg to give an “isochron” relationship. In CAIs, Mg-rich phases coexist with phases having high Al/Mg ratios, and it is in the latter where in situ decay 940 Hutchison, Bridges, and Gilmour of radiogenic Mg has produced measurably high Mg/Mg ratios. Isochron relationships are reasonably well exhibited by many CAIs. Similarly, in chondrules the Al largely resides in Mg-poor glassy mesostasis or plagioclase, and these are the sites where Mg excesses can readily be measured. However, many chondrules do not show good isochron relationships and instead simply show well-resolved Mg excesses. If (as is commonly believed) chondrule recycling introduced relict olivine grains into later generations (Jones 1996), those grains must have been accompanied by “parentless” radiogenic Mg from mesostases and, possibly, from plagioclase. The relict olivines did not dissolve completely or equilibrate with their new chondrule melts, so Mg from the olivines need not have completely “swamped” excess Mg. Therefore excesses of Mg in such chondrules, without an isochron relationship, do not require that the most recent melting episode occurred when Al was “live”. The Al-Mg method dates the onset of chondrule formation, but not its cessation. Figure 2. A schematic timescale for early Solar System events, calibrated to a Shallowater I-Xe age of 4563.2±0.6 Ma (see Gilmour et al. 2005 for cross-calibration and references). 1. CAI origin. 2. Aqueous alteration (CI, CM, CR carbonates; Mokoia fayalite; Allende dark inclusions; CI magnetite; secondary minerals in CAIs; Monahans halite. 3. Igneous rocks and differentiation (Semarkona CC-1 by Al-Mg; H chondrite clast, Barwell, L6; Serra de Magé; Mars core formation (Hf-W isotope dating, Kleine et al. 2002); Magmatic iron meteorite formation (Hf-W isotope dating; earliest age plotted; Kleine et al. 2005); LEW 86010; Shallowater pyroxene). 4. Chondrule model ages. 5. Chondrule isochrons, Semarkona (LL3.0). 6. Chondrule isochrons, Chainpur (LL3.4). 4540 4545 4550 4555 4560 4565 4570 Age (Ma)