Oxygen isotope systematics of chondrules in the Murchison CM2 chondrite and implications for the CO-CM relationship

Abstract

High-precision oxygen three-isotope measurements of olivine and pyroxene were performed on 29 chondrules in the Murchison CM2 chondrite by secondary ion mass spectrometry (SIMS). The oxygen isotope ratios of analyzed chondrules all plot very close to the primitive chondrule minerals (PCM) line. In each of 24 chondrules, the olivine and/or pyroxene grains analyzed show indistinguishable oxygen isotope ratios. Exceptions are minor occurrences of isotopically distinguished relict olivine grains, which were found in nine chondrules. The isotope homogeneity of these phenocrysts is consistent with a co-magmatic crystallization of olivine and pyroxene from the final chondrule melts and a significant oxygen isotope exchange between the ambient gas and the melts. Homogeneous type I chondrules with Mg#’s of 98.9–99.5 have host chondrule Δ¹⁷O values ranging from −6.0‰ to −4.1‰, with one exception (Δ¹⁷O: −1.2‰; Mg#: 99.6). Homogeneous chondrules with Mg#’s <96, including four type II chondrules (Mg# ~65–70), have Δ¹⁷O values of around −2.5%. Five type I chondrules (Mg# ≥99) have internally heterogeneous oxygen isotope ratios with Δ¹⁷O values ranging from −6.5% to −4.0%, similar to those of host chondrule values. These heterogeneous chondrules have granular or porphyritic textures, convoluted outlines, and contain numerous metal grains dispersed within fine-grained silicates. This is consistent with a low degree of melting of the chondrule precursors, possibly because of a low temperature of the melting event and/or a shorter duration of melting. The Δ¹⁷O values of relict olivine grains in nine chondrules range from −17.9% to −3.4%, while most of them overlap the range of the host chondrule values.

Similar to those reported from multiple carbonaceous chondrites (Acfer 094, Y-82094, CO, CR, and CV), the Δ¹⁷O ~−5% and high Mg# (≥99) chondrules, which might derive from a reduced reservoir with limited dust enrichments (~50× Solar System), dominate the population of chondrules in Murchison. Other chondrules in Murchison formed in more oxidizing environment (Mg#<96) with higher Δ¹⁷O values of −2.5%, in agreement with the low Mg# chondrules in Acfer 094 and CO chondrites and some chondrules in CV and CR chondrites. They might form in environments containing the same anhydrous precursors as for the Δ¹⁷O ~−5% and Mg# ~99 chondrules, but enriched in ¹⁶O-poor H2O ice (~0.3–0.4× the CI dust; Δ¹⁷0>0%) and at dust enrichments of ~300–2000×.

Regarding the Mg# and oxygen isotope ratios, the chondrule populations sampled by CM and CO chondrites are similar and indistinguishable. The similarity of these ¹⁶O-rich components in CO and CM chondrites is also supported by the common Fe/Mn ratio of olivine in type II chondrules. Although they accreted similar high-temperature silicates, CO chondrites are anhydrous compared to CM chondrites, suggesting they derived from different parent bodies formed inside and outside the snow line, respectively. If chondrules in CO and CM chondrites formed at the same disk locations but the CM parent body accreted later than the CO parent body, the snow line might have crossed the the common chondrule-forming region towards the Sun between the time of the CO and CM parent bodies accretion.

1. INTRODUCTION

CM (Mighei-like) chondrites are the most abundant group of carbonaceous chondrites (CCs) (Weisberg et al., 2006). They are primitive meteorite mostly of petrologic type 2 (e.g., McSween 1979; Kallemeyn and Wasson, 1981; Zolensky et al., 1993; Rubin et al., 2007), which derived from parent bodies that suffered significant aqueous alteration and low to mild thermal metamorphism (e.g., Sears and Dodd, 1988; Brearley and Jones, 1998; Brearley, 2003; and references therein; Busemann et al., 2007; Schrader and Davidson, 2017). CM chondrites, which can be considered as impact breccias, have been recognized as clasts in many other meteorite classes suggesting that CM chondrites may derive from multiple parent bodies, abundant and/or widely dispersed in the main belt (e.g., Zolensky et al., 1996; Gounelle et al., 2003; Bischoff et al., 2006; and references therein). Hence, CM chondrites are one of the most appropriate material to unravel the formation and evolution of the early Solar System.

CM chondrites display similarities to CO chondrites, though they are generally anhydrous, with respect to chemical compositions of anhydrous minerals (Frank et al., 2014; Schrader and Davidson, 2017) and refractory lithophile element abundances of the bulk chondrites (Kallemeyn and Wasson, 1981). Thus, they are sometime referred as CO-CM clan and considered to be closely related groups of chondrites (Kallemeyn and Wasson, 1981; Weisberg et al., 2006). The genetic relationship between CO-CM was also suggested on the basis of oxygen three-isotope systematics. According to oxygen three-isotope studies of CM chondrites by Clayton and Mayeda (1984, 1999), anhydrous minerals in the Murchison CM chondrite (i.e., chondrules) are more ¹⁶O-rich than those in the bulk CM and similar to bulk CO chondrites, while Murchison matrix separates (i.e. phyllosilicates formed in low temperature fluid) are depleted in ¹⁶O. Bulk CM chondrite data are on a mixing line between these two components with a slope of ~0.7 (Fig. 1). These results suggest that materials in CO chondrites would represent components in CM chondrites prior to parent body aqueous alterations.

Fig. 1.

Oxygen three-isotope diagram of bulk and chondritic components from CM and CO chondrites. The CCAM (carbonaceous chondrite anhydrous mineral; Clayton et al., 1977), Y&R (Young and Russell, 1998), and PCM (primitive chondrule minerals; Ushikubo et al., 2012) lines are shown for reference. The terrestrial fractionation line (TFL) is also shown. CM2 bulk data include falls data from Clayton and Mayeda (1999) and the two falls Sayama (Grossman and Zipfel, 2001) and Maribo (Haack et al., 2012). Both altered and less altered lithologies from the CM2 Paris are reported (Hewins et al., 2014). CM2 matrices and CO3 bulk falls are from Clayton and Mayeda (1999) and Greenwood and Franchi (2004), respectively. The Murchison-anhydrous data are from anhydrous mineral separates of the Murchison CM2 chondrite (Clayton and Mayeda, 1984). Y-81020 data are SIMS olivine and pyroxene analyses from chondrules in the Yamato 81020 CO3.0 chondrite (Tenner et al., 2013). The dashed line indicates the linear regression through the oxygen isotope data for the CM2 bulk compositions, Paris, and the anhydrous mineral separates of Murchison.

Recent oxygen three-isotope data of CM falls (Greenwood et al., 2014) and the two lithologies (altered and less altered) of the Paris CM2 chondrite (Hewins et al., 2014) show a better-defined linear trend in δ¹⁸O-δ¹⁷O plot that reinforces the mixing line suggested by Clayton and Mayeda (1999). Hewins et al. (2014) obtained a line with a slope of 0.69 (R² = 0.93) and an intercept of −4.23‰ from a regression of CM falls, Paris, and CO falls that best represent bulk CM chondrite trend. Similar regression lines were reported from bulk and carbonate in CM chondrites (Benedix et al., 2003; Lindgren et al., 2017; Verdier-Paoletti, 2017), as well as from calcite analyses of Sutter’s Mill (Jenniskens et al., 2012; slope of 0.62 and intercept of −4.88‰). As suggested by Jenniskens et al. (2012), the slopes of these regression lines higher than 0.52 strongly suggest mass independent fractionation by a water-rock reaction involving flowing water along a temperature gradient. The end points of these regression lines and the one from Hewins et al. (2014) pass through the field defined by the recent high precision oxygen three-isotope measurements of olivine and pyroxene from chondrules in the Yamato 81020 (Y-81020) CO3.0 chondrite performed by secondary ion mass spectrometers (SIMS; Tenner et al., 2013). These data may indicate that the anhydrous materials (i.e., high temperature silicates in chondrules; ¹⁶O-rich) of CM chondrites are identical to those in CO3 chondrites. Further it has been debated whether CM and CO derived from a single parent body or from distinct parent bodies from common precursors (Greenwood et al. 2014).

To address these questions, we report in situ high precision SIMS oxygen three-isotope analyses on olivine and pyroxene in chondrules from the Murchison CM2 chondrite. In previous SIMS oxygen isotope studies of pristine chondrules, multiple olivine and/or pyroene analyses in a single chondrule are, in most cases, indistinguishable at the precision of SIMS analyses (0.3–0.5‰ in 2SD) except for “relict olivine” (e.g, Ushikubo et al., 2012; Tenner et al., 2013). In Acfer 094 (ungroup carbonaceous chondrite 3.00 with minimal aqueous alteration and metamorphism; Grossman and Brearley, 2005; Kimura et al., 2008), oxygen isotope ratios of glass and minerals in mesostasis are also indistinguishable from olivine and pyroxene phenocrysts in the same chondrules (Ushikubo et al., 2012). Thus, homogeneous oxygen isotope ratios in olivine and pyroxene in chondrules represent those of the chondrule melt during their formation, which has been referred as “host” chondrule oxygen isotope rarios. Using olivine and pyroxene to determine the “host” chondrule oxygen isotope ratios is particularly helpful for chondrules in mildly metamorphosed or aqueously altered chondrites, because olivine and pyroxene have slower oxygen isotope diffusion and are more resistant to hydration compared to glass and plagioclase in chondrule mesostasis. It is true for Murchison, in which chondrule mesostasis are altered to phylosilicates.

We further determine Mg# (= MgO/[MgO+FeO] in mol.%) of olivine and low-Ca pyroxene in individual chondrules that represent oxygen fugacity variations during chondrule formation (Ebel and Grossman, 2000; Tenner et al., 2015). Several studies have shown that oxygen isotope ratios of individual chondrules in each group of CCs vary against Mg# (Nakashima et al., 2010; 2011; Ushikubo et al. 2012; Schrader et al., 2013, 2014; 2017a; Tenner et al., 2013; 2015; 2017; Hertwig et al., 2017; 2018). Tenner et al. (2015) observed the monotonic increase of the Δ¹⁷O (= δ¹⁷0–0.52×δ¹⁸0) values of chondrules in CR chondrites with decreasing Mg#’s, which is interpreted as the result of a mixing between a ¹⁶O-rich anhydrous dust (Δ¹⁷O = −6‰) and ¹⁶O-poor water ice (Δ¹⁷O = +5‰) in the chondrule precursors. For Acfer 094 and CO chondrites, Δ¹⁷O values of chondrules display a bimodal distribution of Δ¹⁷O, with values of ~−5‰ and ~−2.5‰ for Mg#>98 and Mg#~99–30 chondrules, respectively (Ushikubo et al., 2012; Tenner et al., 2013). CV chondrites and the Yamato 82094 (Y-82094) ungrouped CC are dominated by chondrules with lower Δ¹⁷O values of ~−5‰ and high Mg# (>98), though some chondrules show higher Δ¹⁷O and lower Mg#, similar to ~−2.5‰ chondrules in Acfer 094 and CO3 (Hertwig et al., 2017; 2018; Tenner et al. 2017). CH and CB chondrites are dominated by type I (FeO-poor: Mg#>90) chondrules with Δ¹⁷O values of ~−2.5‰, which is very similar to those reported for CB chondrites (Krot et al. 2010; Nakashima et al., 2010; 2011), though CH chondrites contain chondrules with higher Δ¹⁷O values of ~+1.5‰ and a wide range of Mg# (98–60) and lower Δ¹⁷O values of ~−6‰ with high Mg# (≥98). In contrast, most chondrules in H, LL, R, and EH chondrites have Δ¹⁷O values of 0–1% regardless of the Mg# (Kita et al., 2008, 2010, 2015; Weisberg et al. 2010, 2011), while a few chondrules show negative Δ¹⁷O values but rarely below −4%. It thus appears that the chondrules with high Mg# (>~98) and Δ¹⁷O values of around −5% were sampled by all groups of CCs, but not by the ordinary, enstatite, and R chondrites (e.g., Tenner et al. 2017).

CM chondrites are the last major group of CCs that are not fully investigated yet for detail Mg#−Δ¹⁷O relationship among chondrules (e.g., Jabeen and Hiyagon, 2003; Chaumard et al., 2016; 2017; Schrader et al., 2017b). We examined a large suite of type I and type II chondrules from Murchison, which is one of the least altered CM chondrite (e.g., Busemann et al., 2007; Schrader and Davidson, 2017) in order to fully characterize oxygen isotope systematics. These data are used to discuss the environment of formation and the relationships between CO and CM chondrites. Such data are also important reference regarding the ongoing near-Earth asteroids sample return missions. Indeed, CM parent bodies accreted significant amount of water/ice and are widely disperse and abundant in the main belt (e.g., Bischoff et al., 2006), so potentially sampled by the Hayabusa2 and OSIRIS-REx missions.

2. ANALYTICAL PROCEDURES

2.1. Sample and chondrule selection

We selected and analyzed chondrules from three thin sections of the Murchison CM2 chondrite allocated by the Smithsonian Institution (USNM 5294–2, USNM 5587–2, and USNM 5587–5). Based on size, texture, and mineralogy, we tried to obtain a chondrule selection as representative as possible, while taking into account the minimum size required for SIMS analyses. Our selection includes 25 type I (Mg#≥90) chondrules (from ~200 μm to 1.2 mm in diameter), three type II (Mg#<90) chondrules (~400–500 μm in diameter), and one large fragment of a type II chondrule (~400 μm in size). Twelve type I chondrules are porphyritic olivine-pyroxene (POP; 20–80% modal olivine), seven are porphyritic olivine (PO; >80% modal olivine), and three are porphyritic pyroxene (PP; <80% modal olivine). Two type I chondrules are granular olivine (GO) and one is granular olivine-pyroxene (GOP). The type II chondrules are all PO in texture.

As in all groups of CCs, chondrules in CM chondrites are predominantly type I (e.g., Jones, 2012). In CM chondrites, it has been reported that ~95% of chondrules are porphyritics, and that 10–40% of these porphyritic chondrules are type II (Jones, 2012; and references therein). Our selection displays ~93% of porphyritic chondrules, ~16% of them being type II and is thus representative of those of CM chondrites. In addition, PO, POP, and PP chondrules in our selection represent approximately 38%, 41%, and 10%, respectively, of the entire population of chondrules analyzed. These values are larger for PO chondrules and slightly lower for POP chondrules compared to CO chondrites (8% for PO and 69% for POP chondrules; Jones, 2012; and references therein).

2.2. Scanning electron microscopy and electron microprobe analysis

Backscattered electron (BSE), secondary electron (SE) imaging and energy dispersive X-ray spectrometry (EDS) analyses were performed using a Hitachi S-3400N scanning electron microscope (SEM) at the University of Wisconsin-Madison using an accelerating voltage of 15 kV. We identified in BSE and SE chondrules containing silicate grains with no cracks, pits, or inclusions of other minerals to be suitable for SIMS analysis.

Quantitative chemical analyses were obtained using a Cameca SXFive FE electron microprobe at the University of Wisconsin-Madison. Olivine and pyroxene analyses were performed with an accelerating voltage of 15 keV and a beam current of 20 nA. Counting times for the peak and background were 10 and 5 sec., respectively. Standards used were natural olivine and synthetic forsterite and enstatite (Mg, Si), jadeite (Na, Al), chromian augite (Ca), microcline (K), TiO₂ (Ti), synthetic Cr₂O₃ (Cr), fayalite (Fe), and synthetic Mn₂SiO₄ (Mn). We used a LTAP crystal for the analyses of Mg, Al, Si, and Na, a LPET crystal for Ca and K, a PET crystal for Cr and Ti, and a LLIF crystal for Mn and Fe. The Probe for EPMA™ (PFE) software was used for data reduction and matrix corrections (ZAF and ϕ_ρ(z)). Based on EPMA measurements, we calculated the Mg# of each analysis. Errors on Mg#’s and chemical compositions reported throughout this manuscript correspond to two standard deviations (2SD).

2.3. SIMS oxygen three-isotope analysis

In situ oxygen three-isotope analyses of olivine and pyroxene in type I and II chondrules were performed in two sessions with the Cameca IMS 1280 at the WiscSIMS laboratory, University of Wisconsin-Madison. Analytical methods and conditions are similar to those reported by Kita et al. (2009, 2010). The primary Cs+ beam was tuned to produce a 15 μm (session #1) and 10 μm (session #2) diameter spot with a primary ion intensity of ~3 nA and ~1 nA, respectively. Using a 15 μm diameter spot, secondary ions of ¹⁶O, ¹⁷O, and ¹⁸O were detected simultaneously with three Faraday cups (FC), with intensities of ~3.5×10⁹, ~1.3×10⁶, and ~7.5×10⁶ counts per second (cps), respectively. FC amplifier resistors were 10¹¹ Ω for ¹⁷O and ¹⁸O, and 10¹⁰ Ω for ¹⁶O as in the previous studies (e.g., Kita et al. 2010). For the analysis session using a 10 μm diameter spot, secondary ion intensities of ¹⁶O, ¹⁷O, and ¹⁸O decreased to ~1.5×10⁹, ~5.5×10⁵, and ~2.9×10⁶ cps, respectively. In order to reduce the noise level of the FC amplifier for ¹⁷O analyses below a 10⁶ cps intensity, we replaced the resistor and capacitor pair on the FC amplifier board by a 10¹² Ω resistor and 1 pF capacitor pair that were on the FC amplifier board from a Finnigan MAT 251 stable mass spectrometer. Typical thermal noise of FC amplifier (measured as 1SD of 4s integration) was reduced from 2,000 cps by using 10¹¹ Ω resistor down to 1,200 cps by using 10¹² Ω resistor. For the two sessions, the mass resolving power (MRP at 10% peak height) was set to ~2200 for ¹⁶O and ¹⁸O using two detectors on the multi-collection array, and 5000 for ¹⁷O using a fixed mono-collector (axial detector). At the end of each analysis, ¹⁶OH was measured to determine its contribution to the ¹⁷O signal following the methods described by Heck et al. (2010). The correction of the ¹⁷O signal from ¹⁶OH was negligible (<0.08‰) for both standards and unknowns, except for a few cases as high as 0.2% in unknowns.

Measured ¹⁸O/¹⁶O and ¹⁷O/¹⁶O ratios were normalized to the VSMOW scale (δ¹⁸O and δ¹⁷O that are expressed as a deviation from standard mean ocean water in the unit of 1/1000; Baertschi, 1976). The external reproducibility has been determined by intermittent measurements of a San Carlos (SC) olivine standard (δ¹⁸O = 5.32%; Kita et al., 2010). We performed 8 SC olivine analyses, 4 before and 4 after, to bracket 9 to 19 unknown chondrule analyses (Kita et al., 2009). External reproducibility is calculated as the 2SD of the SC olivine brackets, with average values during session #1 (15 μm) of 0.2‰, 0.3‰, and 0.3‰ for δ¹⁸O, δ¹⁷O, and Δ¹⁷O, respectively. For session #2 (10 μm), the 2SD average values are 0.3‰, 0.5‰, and 0.5‰ for δ¹⁸O, δ¹⁷O, and Δ¹⁷O, respectively. As discussed by Kita et al. (2009), they represent the spot to spot reproducibility and were thus assigned as the uncertainties of each individual spot analysis.

Corrections for instrumental biases of unknown olivine and pyroxene analyses due to their chemical compositions (EA1) were estimated by measuring olivine (Fo_60–100), low-Ca pyroxene (En_70–97), and diopside (Wo₅₀) standards with known oxygen isotope ratios (e.g., Kita et al., 2010; Tenner et al., 2015). The calibration curves were normalized relative to the bias of the SC olivine bracketing standard (EA1) that were analyzed repeatedly throughout the analysis sessions. The compositional ranges of standards cover those of unknowns measured, with the exception of 2 type II chondrules that contain FeO-rich olivine grains with compositions of ~Fo₅₀. Although these olivine grains may have systematic errors in δ¹⁸O and δ¹⁷O due to extrapolated instrumental bias corrections, those errors are mass-dependent and would not affect the Δ¹⁷O values.

In order to determine host chondrule oxygen isotope ratios, we obtained multiple SIMS analyses per chondrule (n= 6 to 11, average: 8). This number of measurement for each chondrule is necessary to confirm multiple analyses with indsitiguishable Δ¹⁷O values determining the host chondrule values. Many CC chondrules contain olivine grains with heterogeneous oxygen isotope ratios (Kunihiro et al., 2004a, 2005; Wasson et al., 2004; Connolly and Huss, 2010; Rudraswami et al., 2011; Ushikubo et al., 2012; Schrader et al., 2013; 2017a; Tenner et al., 2013). These olivine grains with distinct isotope ratios are commonly considered as “relict” and are likely unmelted crystals that survived during the final high-temperature event of the chondrule formation. They preserved oxygen isotope ratios due to the slow diffusivity of oxygen isotopes (e.g., Chakraborty, 2010). These relict grains were not used to determine the averaged “host” oxygen isotope ratio of each chondrule that is thus considered as representative of the final chondrule melt from which “non-relict” olivine and pyroxene grains crystallized. As defined by Ushikubo et al. (2012) and Tenner et al. (2013, 2015), we define relict olivines those with Δ¹⁷O values exceeding the average 3SD external reproducibility (0.50‰ and 0.75‰, for sessions 1 and 2, respectively) when compared to the average “host” Δ¹⁷O value of the considered chondrule.

Uncertainties of averaged host δ¹⁸O and δ¹⁷O for each chondrule are taken as the propagation of (i) the 2 standard error (2SE) of chondrule analyses (=2SD/√number of analyses), (ii) the 2SE of associated SC olivine bracketing analyses that are used for bias corrections, and (iii) the uncertainty due to the sample topography and/or sample positioning on the SIMS stage as well as uncertainty of instrumental bias corrections, estimated to be 0.3% for δ¹⁸O and 0.15% for δ¹⁷O (Kita et al., 2009). Since (iii) is mass-dependent and does not affect Δ¹⁷O, the propagated uncertainty in Δ¹⁷O only uses (i) and (ii). For relict grains, the uncertainties are the spot-to-spot reproducibility (2SD) as determined by bracketing analyses of SC olivine.

After the two SIMS sessions, we checked each spot analysis using SEM-BSE-SE images. Three of 233 pits either overlapped cracks, imperfections, and/or different phases, or displayed a contribution of ¹⁶OH to the ¹⁷O signal of ≥0.1%, and were rejected from our final dataset.

3. RESULTS

BSE images of representative chondrule textures and chemistries are shown in Fig. 2a-2d. BSE images of all chondrules, in which we reported SIMS pits and analyses numbers, are shown in EA2. Electron microprobe analyses and oxygen isotope measurements are summarized in EA3 and EA4, respectively.

Fig. 2.

Examples of SIMS oxygen three-isotope analyses of chondrules in Murchison. a–d) Backscattered electron images of chondrules B8 (type II), C8 (type I, GOP), C10 (type I, GO), and B1 (type I, PP), respectively. Analysis points are shown by the vertex of the triangles, color-coded for mineral phases (olivine: blue, relict olivine: black, low-Ca pyroxene: green, and high-Ca pyroxene: orange). Scale bars are 150 μm. e-h) Oxygen three-isotope diagrams of individual spot analyses in chondrules B8, C8, C10, and B1, respectively. Except for the heterogeneous chondrule C10 (g), the average values of multiple analyses excluding relict olivine grains are shown as red circles. The four reference lines are the same as in Fig. 1. i-l) Δ¹⁷O values of individual analyses from chondrules B8, C8, C10, and B1, respectively. Data are shown in ascending sequence. Symbols are the same as in oxygen three-isotope plots. Error bars in the oxygen three-isotope and Δ¹⁷O value plots are at 95% confidence level, which are shown in EA4.

3.1. Texture, petrography, and mineralogy of chondrules

In porphyritic type I chondrules, olivine grains are present either as euhedral grains (from ~5 μm to ~100 μm), large anhedral phenocrysts, or both. Olivine is mostly located in the central parts of these chondrules, while low-Ca pyroxene is abundant at the periphery. The low-Ca pyroxene often displays a poikilitic texture and contains numerous pores and cracks, probably formed during the protoenstatite-clinoenstatite transition (Hanowski and Brearley, 2001). Low-Ca pyroxene was also observed as large euhedral grains (~40 μm width and up to ~150 μm length). High-Ca pyroxene grains typically surround low-Ca pyroxene grains, with sizes ranging from a few μm to ~70 μm length. In porphyritic type I chondrules, olivine grains are chemically homogeneous, with an average Mg# of 99.3±0.4 (98.6–99.7). They contain 0.29±0.17 wt% CaO, 0.38±0.25 wt% Cr₂O₃, and up to 0.48 wt% Al₂O₃. The Mg#’s calculated for low-Ca pyroxene grains range from 98.9 to 99.5, with an average composition of low-Ca pyroxene of En_97.5±2.4Fs_0.8±0.2Wo_1.7±2.3. We measured 0.39–1.87 wt% Al₂O₃ and 0.30–0.72 wt% Cr₂O₃. The average composition of high-Ca pyroxene is En_60.3±10.2Fs_0.8±0.4Wo_38.9±10.5, with Mg#’s ranging from 98.4 to 99.1. We measured significant amounts of TiO₂ (0.41–1.20 wt%), Cr₂O₃ (0.59–1.16 wt%), MnO (0.15–0.45 wt%), and Al₂O₃ (0.93–9.21 wt%).

We recognized three granular type I chondrules (chondrules B6, C8, and C10), which are composed of evenly sized grains with sparse mesostasis (Fig. 2b,c), especially near the borders. The Mg#’s of olivine and low-Ca pyroxene in granular type I chondrules range from 96.0 to 99.4 and from 95.4 to 99.3, respectively. Olivine grains contain 0.27±0.03 wt% CaO, 0.44±0.27 wt% Cr₂O₃, and up to 0.47 wt% MnO. The compositions of low-Ca pyroxene range between En_90.0Fs_4.4 and En_97.9Fs_0.7. We measured significant amount of Al₂O₃ (1.15–4.39 wt%), Cr₂O₃ (0.46–1.79 wt%), as well as 0.08–35 wt% TiO₂ and 0.14–1.06 wt% MnO. Although chondrules B6 and C10 contain few grains of high-Ca pyroxene, they were two small to obtain quantitative analyses.

In type I chondrules, small inclusions of metal (<10 μm in size) were observed within euhedral olivine grains. In the center of some type I chondrules, large blebs (larger than ~30 μm in size) display a core of metal surrounded by a mixture between an abundant S-Fe-rich phase and, to a lesser extent, a Fe-rich phyllosilicate. This PCP (Poorly Characterized Phases) -like material (intergrowths of cronstedtite and tochilinite; e.g., Mackinnon and Zolensky, 1984; Tomeoka and Buseck, 1985; Browning et al., 1996) was commonly observed replacing metal in CM type I chondrules. In addition, and based on EDS analyses, the altered mesostasis of type I chondrules mainly consist of phyllosilicates (i.e., “spinach” phase; e.g., Fuchs et al., 1973).

The type II chondrules investigated have a porphyritic texture. Olivine phenocrysts display various degrees of aqueous alteration. They are almost fresh and unaltered in chondrule C12, and partially serpentinized (e.g., Lee and Lindgren, 2016) in chondrule B8. Mesostasis of type II chondrule is composed of phyllosilicates, as well as secondary ferroan olivine (rounded grains, <10 μm in size) in chondrule B9. The Mg# measured for olivine is variable from one grain to another in a single chondrule, as well as from one chondrule to another. The range of Mg# are 53.0–71.3, 63.2–71.7, 53.1–78.3, and 62.6–96.0 for chondrule C12, B9, C11, and B8, respectively. Olivine grains also contain significant amounts of Cr₂O₃ (0.22–0.52 wt%) and MnO (0.12–0.51 wt%).

3.2. Chondrule Mg#

In 19 of the 29 chondrules analyzed, we were able to compare Mg#’s of coexisting olivine and low-Ca pyroxene grains. As shown in Fig. 3, these 19 chondrules contain olivine and low-Ca pyroxene with similar Mg#’s. The 10 other chondrules contain either olivine or pyroxene grains. While the Mg# in olivine and pyroxene from chondrules show a wide range (~50–99), the consistency between olivine and low-Ca pyroxene Mg#’s within a single chondrule indicates they did not suffer significant secondary parent body processes such as thermal metamorphism and aqueous alteration (e.g., Jones, 1994; Tachibana et al., 2003; Ushikubo et al., 2012; Tenner et al., 2013, 2015; Schrader and Davidson, 2017; Schrader et al., 2017a). Because olivine and low-Ca pyroxene are the most abundant phases in chondrules investigated, we calculated the Mg# of individual chondrules defined as the average of its constituent olivine and/or low-Ca pyroxene excluding relict grains (Ushikubo et al., 2012; Tenner et al., 2013, 2015; 2017). Five chondrules investigated (the four type II chondrules and the type I GOP chondrule C8) have Mg#’s ranging between 64.9 and 96.0, while the other 24 chondrules have Mg#’s ranging from 98.9 to 99.6. The chondrule Mg#’s are given in Table 1.

Fig. 3.

Comparison between olivine and low Ca-pyroxene Mg#’s in 19 type I chondrules from Murchison. Each point corresponds to the averaged Mg# of olivine and low-Ca pyroxene coexisting within a single chondrule. Error bars correspond to 2SD of multiple analyses. A 1:1 line is shown for reference.

Table 1.

Mg#’s and oxygen isotope ratios of host chondrules and relict grains.

Chondrulea	Type-texture-relict	Mg#	+/−	n (ol, LPx, HPx)b	δ18O	unc.c	δ17O	unc.c	Δ17O	unc.c	Δ17O 2SD
Session-1:15μm beam size
A5	I-POP	99.5	0.2/0.1	7,0,0	−6.4	0.3	−9.3	0.3	−6.0	0.2	0.5
C2	I-PO	99.4	0.1/0.0	8,0,0	−6.2	0.4	−9.1	0.4	−5.8	0.3	0.7
A2	I-POP	99.4	0.1/0.1	4,2,0	−4.9	0.5	−8.0	0.3	−5.5	0.2	0.5
	relict ol	99.3	n.c.	1,0,0	−4.2	0.3	−7.0	0.4	−4.8	0.4	n.c.
B5	I-POP	99.3	0.2/0.0	4,2,0	−3.5	0.4	−6.9	0.3	−5.1	0.2	0.4
	relict ol	99.3	n.c.	1,0,0	−8.3	0.9†	−11.7	l.l†	−7.3	0.6†	n.c.
	relict ol	99.3	n.c.	1,0,0	−7.5	0.2	−10.6	0.3	−6.7	0.4	n.c.
C9	I-POP	99.3	0.3/0.2	6,2,0	−4.1	0.6	−7.3	0.4	−5.1	0.2	0.6
B2	I-PO	99.3	0.1/0.2	7,0,1	−4.7	0.6	−7.5	0.5	−5.0	0.3	0.6
B1	I-PP	99.2	0.1/0.1	3,4,1	−3.5	0.3	−6.7	0.2	−4.9	0.2	0.3
B7	I-PO	99.3	0.3/0.2	7,0,0	−3.3	0.5	−6.4	0.2	−4.7	0.2	0.6
A1	I-PO	99.0	0.1/0.0	3,1,0	−2.8	0.5	−5.8	0.6	−4.3	0.4	0.7
	relict ol	99.4	n.c.	1,0,0	−7.6	0.3	−9.7	0.4	−5.7	0.4	n.c.
	relict ol	99.4	n.c.	1,0,0	−7.4	0.3	−9.4	0.4	−5.6	0.4	n.c.
	relict ol	99.3	n.c.	1,0,0	−6.3	0.3	−8.7	0.4	−5.5	0.4	n.c.
A6	I-POP	99.0	0.1/0.0	0,2,0	−2.2	0.5	−5.3	0.3	−4.1	0.2	0.0
	relict ol	99.1	n.c.	1,0,0	−5.4	0.1	−8.5	0.2	−5.6	0.2	n.c.
	relict ol	99.1	n.c.	1,0,0	−5.3	0.1	−8.0	0.2	−5.3	0.2	n.c.
	relict ol	99.0	n.c.	1,0,0	−5.4	0.1	−8.6	0.2	−5.7	0.2	n.c.
	relict ol	99.0	n.c.	1,0,0	−4.8	0.1	−7.7	0.2	−5.3	0.2	n.c.
	relict ol	98.9	n.c.	1,0,0	−5.5	0.1	−8.6	0.2	−5.8	0.2	n.c.
C8	I-GOP	96.0	0.2/0.6	2,4,0	0.3	0.3	−2.6	0.3	−2.7	0.3	0.6
	relict ol	96.0	n.c.	1,0,0	−13.2	0.1	−14.7	0.3	−7.8	0.3	n.c.
	relict ol	95.9	n.c.	1,0,0	0.6	0.1	−3.1	0.3	−3.4	0.3	n.c.
Cll	II-fragment	70.3	8.0/17.2	9,0,0	2.0	0.4	−1.5	0.3	−2.6	0.2	0.2
B9	II-PO	67.0	4.7/3.8	8,0,0	1.8	0.3	−1.5	0.2	−2.5	0.2	0.4
B8	II-PO	67.5	3.1/4.9	6,0,0	2.7	0.3	−0.9	0.3	−2.3	0.2	0.1
	relict ol	96.0	n.c.	1,0,0	−5.0	0.2	−8.2	0.4	−5.6	0.4	n.c.
	relict ol	80.5	n.c.	1,0,0	−3.5	0.2	−6.9	0.4	−5.1	0.4	n.c.
Session-2 : 10 μm beam size
B4	I-PO	99.3	0.1/0.2	7,0,0	−5.8	0.6	−9.1	0.6	−6.0	0.3	0.7
	relict ol	99.4	n.c.	1,0,0	−9.8	0.3	−13.0	0.6	−7.9	0.5	n.c.
B6	I-GO	99.1	0.2/0.2	3,2,0	−5.3	0.5	−8.7	0.4	−5.9	0.3	0.3
	relict ol	99.1	n.c.	1,0,0	−34.1	0.3	−35.6	0.6	−17.9	0.5	n.c.
	relict ol	99.0	n.c.	1,0,0	−8.8	0.3	−12.0	0.6	−7.4	0.5	n.c.
C5	I-PO	99.3	0.2/0.1	4,2,1	−5.2	0.3	−8.4	0.3	−5.7	0.2	0.4
C3	I-PO	99.3	0.2/0.2	5,2,0	−4.6	0.3	−7.7	0.3	−5.3	0.2	0.5
	relict ol	99.2	n.c.	1,0,0	−6.9	0.3	−10.0	0.4	−6.4	0.4	n.c.
Cl	I-POP	99.2	0.2/0.1	8,3,0	−3.5	0.6	−6.8	0.4	−5.0	0.2	0.6
C4	I-POP	99.2	0.1/0.1	5,3,0	−3.7	0.3	−6.8	0.3	−4.9	0.2	0.4
A3	I-POP	99.2	0.2/0.1	5,1,0	−1.8	0.4	−5.6	0.4	−4.6	0.4	0.8
C7	I-PP	99.1	0.1/0.1	1,4,3	−3.2	0.5	−6.2	0.3	−4.5	0.2	0.4
C12	II-PO	64.9	6.4/11.9	8,0,0	3.5	0.4	−0.5	0.3	−2.3	0.3	0.2
C6	I-POP	99.6	0.1/0.1	7,1,0	3.3	0.5	0.5	0.4	−1.2	0.2	0.4

^aFor each session, chondrules are sorted following decreasing host Δ¹⁷O values.

^bNumbers of mineral phases analyzed (olivine, low-Ca pyroxene, and high-Ca pyroxene).

^cUncertainties of host oxygen isotope ratios are propagated, combining (1) the 2SE of chondrule measurements, (2) 2SE of associated SC olivine bracketing analyses, and (3) influences from sample topography/positioning. Uncertainties of oxygen isotope ratios of relict grains are the spot-to-spot reproducibility (2SD), as determined by bracketing analyses of SC olivine.

^†2SE of the measurement (see EA4). n.c.: not calculated

3.3. Oxygen isotope ratios

We collected a total of 230 spots analyses from 29 chondrules from three sections in two separated sessions (15 and 10 μm beam size; EA4). Regardless of which mineral phase was measured, our data plot between the CCAM (carbonaceous chondrite anhydrous mineral; Clayton et al., 1977) and the Y&R (Young and Russell, 1998) lines, close to the PCM (primitive chondrule minerals; Ushikubo et al., 2012) line, as shown in the examples of Fig. 2e-2h. The δ¹⁸O and δ¹⁷O values of olivine range from −34‰ to +5‰ and from −36‰ to +1‰, respectively, while those of pyroxene range from −7‰ to +3‰ and from −10‰ to +0‰, respectively (Fig. 4). Olivine data extend to lower δ¹⁸O and δ¹⁷O values compared to those of pyroxene because they include ¹⁶O-rich relict olivine data as indicated below. Two analyses of olivine (spots #231 in chondrule B5 and #205 in chondrule A7; EA4) display unusual large internal errors for δ¹⁸O and δ¹⁷O (≥1 ‰). Cycle data plot along a slope 1 line supporting a variation of the oxygen isotope ratios (≥10‰ in δ¹⁸O and δ¹⁷O) with depth, possibly due to the presence of small μm-sized ¹⁶O-rich domains within the analysis spots.

Fig. 4.

Oxygen three-isotope diagram of individual spot analyses of chondrules in Murchison. a) olivine. Relict olivine grains are shown as open symbols. b) pyroxene. Low-Ca pyroxene (LPx): squares, high-Ca pyroxene (HPx): triangles. Uncertainties of individual data, corresponding to the spot-to-spot reproducibility (2SD), are smaller than the size of symbols. The four reference lines are the same as in Fig. 1.

In 19 chondrules, the Δ¹⁷O values of multiple analyses within a chondrule agree within ±3SD external reproducibility (0.50‰ and 0.75‰ for 15μm and 10μm spot analyses, respectively), as shown in Fig. 2k-2l. Of these 19 chondrules, four chondrules (A4, A8, B3, C10) show a significant variability in δ¹⁸O and δ¹⁷O values (2–3%) beyond typical analytical uncertainty (e.g., Fig. 2g, 2k for C10). Their δ¹⁸O and δ¹⁷O values plot along the PCM line with the slope close to unity (0.84–0.91) and coefficients of determination (R²) of 0.74–0.94 (EA4), indicating these three chondrules are internally heterogeneous in ¹⁶O. Therefore, the host chondrules oxygen isotope ratios are calculated from the mean of multiple analyses in each chondrule except for A4, A8, B3, and C10.

In the other 10 chondrules, one or more olivine analyses show Δ¹⁷O value(s) that differ more than 3SD external reproducibility. In most cases, by excluding between one and three olivine data, other coexisting olivine and/or pyroxene measurements are indistinguishable (Fig. 2i-2j). As described in 2.3, these isotopically distinct olivine grains are considered to be “relict” and excluded from the calculation of individual host chondrule values. In chondrule A6, five olivine grains display homogeneous isotope ratios (Δ¹⁷O = −5.5‰), which differ significantly from those in two low-Ca pyroxene grains (Δ¹⁷O = −4.1‰). As in the previous study (chondrule Y22 from Tenner et al. 2013), olivine in A6 are considered to be relict and the mean of pyroxene data is used to estimate the host chondrule data. In chondrule A7, Δ¹⁷O values show a significant range from −6.2‰ to −4.6‰, even excluding two most ¹⁶O-rich olivine data (Δ¹⁷O; −12.8‰ and −8.0‰). This chondrule is considered to be heterogeneous in oxygen isotopes and reliable host value cannot be determined.

In Table 1, oxygen isotope ratios of host chondrules are calculated from 24 chondrules (excluding the five heterogeneous chondrules A4, A7, A8, B3, C10). The texture, Mg#, the number of measurements per mineral, and the relict olivine analyses in the same chondrules are listed together. Individual spot analyses for the five heterogeneous chondrules are listed in Table 2. Host chondrule Δ¹⁷O values along with their relict olivine analyses are summarized in Fig 5, which also shows individual data from five heterogeneous chondrules. Host chondrule Δ¹⁷O values show discrete ranges; most type I chondrules (18 out of 20) range between −6.0‰ to −4.1‰, while five chondrules including all four type II chondrules show a small range of Δ¹⁷O values between −2.6±0.2‰ and −2.3±0.2‰. The highest Δ¹⁷O value of-1.2±0.2‰ is found from the type I POP chondrule C6. These ranges are very similar to those observed from Acfer 094 and CO3 (Ushikubo et al. 2012; Tenner et al. 2013). The Δ¹⁷O values of relict olivine analyses (19 data from 9 chondrules) and individual analyses from five heterogeneous chondrules are mainly distributed between −6.5‰ and −4.0‰, identical to host values from majority of type I chondrules. Several relict olivine analyses are between −6‰ and −8‰, which are within a range of host chondrules in type I chondrules from other CCs, such as Acfer 094, CV, CO, and Y-82094 (Libourel and Chaussidon, 2011; Ushikubo et al., 2012; Tenner et al., 2013, 2017; Hertwig et al. 2018). Only two analyses are significantly below Δ¹⁷O<−8‰; relict olivine in B6 (−17.9‰) and the most ¹⁶O-rich analysis in A7 (−12.8‰). Similar ¹⁶O-rich relict olivine data have been reported in chondrules in CCs (e.g., Ushikubo et al., 2012; Schrader et al., 2013; Tenner et al., 2013, 2017).

Table 2.

Mg# and oxygen isotope ratios of individual grains in heterogeneous chondrules.

Chondrule Type-texture	Mg#	minerals	δ18O	unc.a	δ17O	unc.a	Δ17O	unc.a
Session-1: 15 μm beam size
A7	99.1	ol	−5.5	0.1	−8.5	0.3	−5.7	0.3
I-POP	99.0	ol	−5.8	0.1	−9.3	0.3	−6.3	0.3
	99.1	ol	−6.7	0.1	−9.3	0.3	−5.8	0.3
	99.1	ol	−5.2	0.1	−7.9	0.3	−5.2	0.3
	99.2	ol	−10.8	0.1	−13.7	0.3	−8.0	0.3
	99.1	ol	−21.4	2.0†	−23.9	1.9†	−12.8	0.9†
	99.2	LPx	−4.9	0.1	−8.0	0.3	−5.4	0.3
	98.6	HPx	−3.2	0.1	−6.3	0.3	−4.6	0.3
Session-2: 10 μm beam size
B3	99.1	ol	−2.4	0.3	−5.8	0.6	−4.5	0.7
I-POP	99.0	LPx	−5.2	0.3	−8.0	0.6	−5.3	0.7
	99.2	LPx	−4.5	0.3	−7.0	0.6	−4.7	0.7
	99.2	LPx	−3.2	0.3	−5.6	0.6	−4.0	0.7
	99.2	LPx	−3.4	0.3	−6.5	0.6	−4.7	0.7
	99.2	LPx	−3.2	0.3	−5.9	0.6	−4.2	0.7
	99.2	LPx	−3.0	0.3	−5.8	0.6	−4.2	0.7
	99.3	LPx	−4.7	0.3	−7.9	0.6	−5.4	0.7
	99.1	LPx	−3.7	0.3	−6.8	0.6	−4.9	0.7
A8	98.6	ol	−3.4	0.3	−6.6	0.6	−4.9	0.7
I-POP	98.9	ol	−3.3	0.3	−6.7	0.6	−5.0	0.7
	98.9	ol	−3.5	0.3	−6.7	0.6	−4.9	0.7
	98.9	ol	−2.3	0.3	−5.9	0.6	−4.7	0.7
	98.9	ol	−2.6	0.3	−5.5	0.6	−4.1	0.7
	98.9	ol	−2.1	0.3	−5.5	0.6	−4.4	0.7
	98.9	ol	−2.7	0.3	−5.5	0.6	−4.0	0.7
	99.1	LPx	−3.1	0.3	−6.1	0.6	−4.5	0.7
A4	99.5	ol	−4.7	0.3	−7.9	0.5	−5.5	0.6
I-PP	99.3	ol	−4.7	0.3	−7.6	0.5	−5.1	0.6
	99.2	ol	−7.5	0.3	−10.0	0.5	−6.1	0.6
	99.3	ol	−5.0	0.3	−8.0	0.5	−5.4	0.6
	99.4	ol	−5.3	0.3	−8.3	0.5	−5.6	0.6
	99.2	LPx	−7.4	0.3	−10.3	0.5	−6.4	0.6
	99.4	LPx	−5.3	0.3	−8.6	0.5	−5.8	0.6
	98.9	LPx	−6.2	0.3	−9.2	0.5	−6.0	0.6
	99.3	LPx	−5.5	0.3	−9.3	0.5	−6.5	0.6
CIO	99.3	ol	−3.8	0.3	−7.1	0.6	−5.1	0.6
I-GO	99.3	ol	−6.5	0.3	−9.3	0.6	−6.0	0.6
	99.3	ol	−4.2	0.3	−7.5	0.6	−5.3	0.6
	99.3	ol	−5.7	0.3	−8.9	0.6	−5.9	0.6
	99.3	ol	−5.3	0.3	−7.9	0.6	−5.2	0.6
	99.4	ol	−5.5	0.3	−8.7	0.6	−5.8	0.6
	99.4	ol	−5.8	0.3	−8.8	0.6	−5.7	0.6

^aUncertainties of oxygen isotope ratios are the spot-to-spot reproducibility (2SD), as determined by bracketing analyses of SC olivine.

^†2SE of the measurement (see EA4)

Fig. 5.

Δ¹⁷O values of individual host chondrules in Murchison. Relict olivine grains are shown together. Data are from Tables 1 and 2. The homogeneous chondrule data are sorted following the host Δ¹⁷O values. For heterogeneous chondrules (B3, A8, A7, A4, and C10), individual analyses within each chondrule are shown.

The host Δ¹⁷O values of individual chondrules are correlated to their Mg#, as shown in Fig. 6. The lowest Δ¹⁷O values correspond to the highest Mg#’s. The Δ¹⁷O values range from −6.0‰ to −4.1‰ for Mg#’s ranging from 99.5 to 98.9, and from - 2.7±0.3‰ to −2.3±0.2‰ for Mg#’s ranging from 96 to 65. Only chondrule C6 deviates from this relationship with highest Δ¹⁷O of −1.2% and highest Mg# of 99.6 (Fig. 6).

Fig. 6.

Murchison chondrule Δ¹⁷O values vs. Mg#’s. Each point represents the host Mg# and Δ¹⁷O values of individual chondrules. Chondrule Mg# uncertainties correspond to the range of measured values, while uncertainties in Δ¹⁷O are those of the mean value at 95% confidence level, shown in EA4.

4. DISCUSSION

4.1. Oxygen isotope ratios of host chondrules

4.1.1. Co-origin of olivine and pyroxene

Excluding relict grains, most of the porphyritic type I chondrules analyzed are composed of olivine and low-Ca pyroxene with indistinguishable Δ¹⁷O values. In the previous SIMS oxygen isotope studies of chondrules from various groups of CCs, multiple analyses of olivine and pyroxene phenocrysts within single chondrules show homogeneous oxygen isotope ratios within analytical uncertainties (e.g., Rudraswami et al., 2011; Ushikubo et al., 2012; Tenner et al., 2013, 2015, 2017; Zhang et al., 2014; Schrader et al., 2017a; Hertwig et al. 2018). Such observation suggests that coexisting pyroxene and olivine grains crystallized from isotopically homogeneous melts formed during the final melting event. Consistent oxygen isotope ratios between olivine and pyroxene are clearly demonstrated in Fig. 7 that plot δ¹⁸O, δ¹⁷O, and Δ¹⁷O values of pyroxene against olivine on 1:1 lines, respectively. Similar 1:1 relationship between olivine and pyroxene from 15 individual chondrules has been reported from several groups of CCs (Rudraswami et al. 2011; Ushikubo et al. 2012; Tenner et al. 2012; 2015; 2017; Hertwig et al. 2018). According to these results, pyroxene and most of the olivine grains in these 15 chondrules from Murchison must have crystallized from the same final silicate melt that was homogeneous in oxygen isotopes.

Fig. 7.

Comparison of oxygen isotope ratios between coexisting olivine and pyroxene from 15 individual chondrules in Murchison. The mean low-Ca and high-Ca pyroxene data are plot against the mean olivine data for δ¹⁸O (a), δ¹⁷O (b), and Δ¹⁷O (c). Relict grains are not included in the calculation of mean values. Propagated uncertainties are estimated in the same procedure as host chondrule calculations (see EA4).

Alternatively, several studies suggest that olivine from differentiated planetesimals would have been the precursors of type I chondrules in CV and CR chondrites and experienced gas-melt interactions during chondrule formation (Tissandier et al., 2002; Libourel et al., 2006; Chaussidon et al., 2008; Libourel and Chaussidon, 2011). Chaussidon et al. (2008) reported regression lines with slopes of ~0.8 for δ¹⁸O and δ¹⁷O values between olivine and pyroxene, respectively. Under this scenario, most of olivine in type I chondrules would be relicts and enriched in ¹⁶O, while pyroxene formed as a result of high-temperature reactions between the silicate melt and ¹⁶O-poor SiO molecules in the ambient gas (Marrocchi and Chaussidon, 2015). Co-existing olivine and pyroxene data in type I chondrules from Murchison do not show significant differences (≤0.8‰, ≤0.9‰, and ≤0.7‰ for δ¹⁸O, δ¹⁷O, and Δ¹⁷O, respectively; Fig. 7), thus do not support this scenario. Similarly, previous SIMS studies of chondrules in other CCs including CV and CR chondrites (e.g., Rudraswami et al., 2011; Ushikubo et al., 2012; Schrader et al., 2017a; Tenner et al., 2013, 2015, 2017; Hertwig et al. 2018) all show consistent oxygen isotope ratios between olivine and pyroxene within a single chondrule. Thus, olivine is not systematically ¹⁶O-rich compared to the coexisting pyroxene in most chondrules.

In CR chondrites, Schrader et al. (2013, 2014; 2017a) observed that the range of Δ¹⁷O values measured for olivine phenocrysts in barred olivine (BO) chondrules are similar to those in porphyritic chondrules, while it is generally considered that BO chondrules were completely melted above the liquidus and porphyritic chondrules did not. Schrader et al. (2014) conclude that porphyritic chondrules do not retain abundant relict olivine grains, which is consistent with our observations. This is also an important argument against the origin of olivine as relict fragments of early-formed planetesimals.

4.1.2. Complete oxygen isotope exchange between ambient gas and chondrule melt

As first indicated by Clayton (1983), it has been hypothesized that an incomplete isotope exchange between two distinct isotope reservoirs, ¹⁶O-rich dust and ¹⁶O-poor gas, resulted in the mass independent fractionation of oxygen isotope ratios among chondrules in CCs plotting along a slope ~1 line (e.g., Chaussidon et al. 2008; Schrader et al. 2013; 2014). The rates of the oxygen isotope exchange between the molten silicate and gas (H₂O) were experimentally determined by Yu et al. (1995) and Di Rocco and Pack (2015) in order to evaluate the degree of isotope exchange during a chondrule forming event using the appropriate oxygen fugacity (log fO₂ ~IW-4 to −2 for type I and IW-2 for type II chondrules; e.g., Zanda et al., 1994; Ebel and Grossman 2000). While 90% oxygen isotope exchange is achieved in 30 minutes at a log fO₂value 0.5 units below the IW buffer (experiments performed at 1 atm and 1450°C; Yu et al., 1995), only 50% of oxygen isotope exchange occurred at a lower log fO₂ (from IW-1.3 to IW-3.8) in 1–4 hours (experiments performed at 1 atm and 1500°C; Di Rocco and Pack, 2015). Based on the heating duration at temperatures above the liquidus predicted by the shock models (e.g., Desch and Connolly, 2002; Morris and Desch, 2010), Di Rocco and Pack (2015) estimated the degree of isotope exchange between the chondrule melt and ambient gas to be 50% and 70% for type I and type II chondrule formation conditions, respectively. Thus, the observed internal homogeneity among olivine and pyroxene phenocrysts in chondrules from this work and previous SIMS studies may not be explained unless chondrule precursor and ambient gas had similar Δ¹⁷O values. Furthermore, the oxygen isotope homogeneity between olivine and mesostasis phases (e.g., high-Ca pyroxene, plagioclase, glass) reported from the most pristine chondrites such as Acfer 094 and CR chondrites (Ushikubo et al. 2012; Tenner et al. 2015) is hard to explain if the ambient gas and initial melt had distinct Δ¹⁷O values, because continuous isotope exchange between melt and gas under liquidus temperatures (>10 h) would modify the oxygen isotope ratios of late forming minerals and quenched glass. In the case of chondrules in Murchison, we did not analyze plagioclase or glassy mesostasis that were altered during aqueous alteration. However, we analyzed high-Ca pyroxene grains in several chondrules that are indistinguishable from olivine and low-Ca pyroxene mesostasis in term of oxygen isotope ratios (Fig. 7). Chondrules in Murchison do not show internal oxygen isotope zoning with the crystallization sequence.

As discussed in previous studies (Kita et al. 2010; Ushikubo et al. 2012; Tenner et al. 2013; 2015; 2017), it is likely that the ambient gas during chondrule formation had oxygen isotope ratios close to those of solid precursors. Chondrules likely formed in dust-rich regions (enriched in CI dust by more than ~100×) of the protoplanetary disk (e.g., Ebel and Grossman, 2000; Ozawa and Nagahara, 2001; Alexander, 2004; Alexander et al., 2008; Nagahara et al., 2008; Schrader et al., 2013; Tenner et al., 2015) and in an open-system with significant evaporation and re-condensation (e.g., Cohen et al., 2004; Libourel et al., 2006; Nagahara et al., 2008; Hewins and Zanda, 2012). In consequences, the isotopic ratios of oxygen in the ambient gas would be controlled and dominated by those of the solid chondrules’ precursors (Kita et al., 2010; Ushikubo et al., 2012). Recondensation of oxide further enhances the equilibration of the oxygen isotope ratios between the ambient gas and chondrule melt. Within this framework, the oxygen isotope ratios of the ambient gas and the melt would be similar to the averaged precursor solids that represent local disk regions.

In the series of SIMS oxygen isotope studies of chondrules in CR chondrites, Schrader et al. (2013; 2014) found that BO chondrules show systematically higher and narrower range of Δ¹⁷O than porphyritic chondrule. The authors argued that the isotope exchange between the chondrule melt and ¹⁶O-poor gas was incomplete for all cases, but higher in BO (~90%) than porphyritic chondrules (25–50%). In their model, the ambient gas formed during chondrule formation is derived from a ¹⁶O-poor ice that was evaporated to a H₂O gas. While BO chondrules formed from complete melting and are individually homogeneous in term of oxygen isotope ratios, Schrader et al. (2014) suggested that the significant range of Δ¹⁷O values observed for BO chondrules indicates that even BO chondrules did not experience complete isotope exchange. If true, we should see an isotope zoning in the mesostasis of porphyritic chondrules, in contrast to observations from previous works (e.g., Ushikubo et al. 2012; Tenner et al. 2015). Furthermore, systematically different ranges of Δ¹⁷O values between BO chondrules and porphyritic chondrules in CR chondrites reported by Schrader et al. (2013; 2014) are actually fairly similar; from −3.8‰ to −1.3‰ and from −0.8‰ to +1.4‰ for type I and II BO chondrules, respectively, and from −4.6‰ to −0.3‰ and from −1.8‰ to +0.9‰ for type I and II porphyritic chondrules, respectively. By combining data from Tenner et al. (2015), the lower ends of type I BO chondrules and type I porphyritic chondrules extend to −5.3±0.4‰ and −5.9±0.2‰, respectively, which are marginally different. In the study of Allende chondrules by Rudraswami et al. (2011), the range of Δ¹⁷O values are similar between type I BO and porphyritic chondrules; from −5.6‰ to 0‰ and from −5.8‰ to - 2.2‰, respectively. Alternatively, and as discussed above, a formation of chondrules in an open system and dust-rich environment suggests that oxygen isotope ratios of the chondrule melt and ambient gas would have been very similar and represent the average composition of the solid precursors of localized areas, including ice. As clearly observed, the difference between type I and II chondrules that formed under distinct oxygen fugacities indicates that each chondrite group collected chondrules formed in multiple environments with respect to oxygen isotope ratios. This scenario would explain the similar range of oxygen isotope ratios between BO chondrules and porphyritic chondrules observed in both CR and CV chondrites.

4.1.3. Isotope heterogeneity of chondrules with or without relict grains

Five chondrules were described as heterogeneous regarding the oxygen isotope ratios of their olivine and pyroxene grains (Table 2, Fig. 5). The precursors of these heterogeneous chondrules were thus less melted than the internally homogeneous chondrules. It suggests either a lower temperature of the melting event(s) or a shorter duration of melting, or both.

Interestingly, the four heterogeneous chondrules A7, B3, A8, and A4 have porphyritic textures (Table 2). Although chondrules A7 and A8 display more rounded shapes, the chondrules B3 and A4 have convoluted outlines and contain numerous metal (or altered metal) grains dispersed throughout fine-grained silicates (EA2). All these observations are consistent with an incomplete melting (e.g., Lofgren, 1996; Hewins and Radomsky, 1990; Zanda et al. 2002). Recent experimental works indicate fast dissolution rates of olivine (up to ~22 μm/min at 1531°C) during heating (Soulié et al., 2017). Because of the fast kinetics of dissolution of olivine in type I chondrule-like melts, the preservation of former olivine crystal in porphyritic chondrules required fast cooling rates, in the range of 1000–8000 K/h at 1500°C (Soulié et al., 2017). Although extended to higher cooling rates, these results are in agreement with those from experiments on mineral textures of type I porphyritic chondrules that reported values between 10 and 1000 K/h (e.g., Radomsky and Hewins, 1990; DeHart and Lofgren, 1996). The fifth heterogeneous chondrule displays a granular texture (C10; Fig. 2c). Granular textures also result from the incomplete melting of the precursor (initial temperature before cooling below the liquidus; Hewins and Radomsky, 1990). Both isotope ratios and textural observations support the fact that these five chondrules are internally heterogeneous in oxygen isotope ratios because of the low degree of melting of their isotopically heterogeneous solid precursors. Indeed, the ~50–90% melting experienced by such chondrules allows a high percent of former crystals to survive (e.g., Lofgren, 1996). It should be noted that the range of Δ¹⁷O values in these chondrules significantly overlaps those defined by the host chondrule values (Fig. 5). This observation suggests common isotope reservoirs for the homogeneous chondrules and precursors of these heterogeneous ones.

Only two olivine grains from the heterogeneous chondrules A7 are significantly ¹⁶O-rich (Δ¹⁷O = −8.0 and −12.8‰; Table 2, Fig. 5) compared to the other coexisting phases (Δ¹⁷O from −4.6 to −6.3‰; Fig. 5) that are similar to the ¹⁶O-rich relict olivine grains in homogeneous chondrules. Some of these ¹⁶O-rich analyses could be caused by a small domain of AOA-like ¹⁶O-rich olivine that might be present as relict grains within these chondrule phenocrysts (Nagashima et al., 2015). Especially, the most ¹⁶O-rich data in A7 with Δ¹⁷O of −12.8±0.9‰ show a large variability in δ¹⁸O and δ¹⁷O along a slope ~1 line within a single analysis. Such ¹⁶O-rich domains could also occur as small μm-sized areas within other olivine analysis spots, which would result in lower δ¹⁷O and δ¹⁸O values along the PCM line compared to true host chondrule values. Although the changes are generally small and marginally resolvable beyond analytical uncertainties, we may not be able to completely rule out the possibility that some chondrules are recognized as “heterogeneous” due to frequent encounters with small ¹⁶O-rich domains during SIMS analyses.

4.1.4. Abundance and origin of relict grains

One type II chondrule among the four analyzed in Murchison contains relict olivine grains (Fig. 5). Because of the small number of type II chondrules analyzed, this abundance of 25% calculated may not be representative of the CM chondrites. Recently, Schrader and Davidson (2017) analyzed 84 type II chondrules in CM chondrites. Using the FeO content of olivine, these authors estimated that ~12% of type II chondrules in CM chondrites contain relict olivine grains. Whatever the method and the number of type II chondrules analyzed, these two values (12% and 25%) are significantly different that the abundance of ~48% of CO type II chondrules that contain relict grains (15/31; Jones, 1992; Wasson and Rubin, 2003; Kunihiro et al., 2004a; Tenner et al., 2013). As discussed by Schrader and Davidson (2017), this difference between CM and CO chondrites does not support a common parent body for these two groups of CCs.

For type I chondrules, approximately 40% of the chondrules investigated contain relict olivine grains, which are either ¹⁶O-poor or ¹⁶O-rich compared to the host chondrules (Fig. 5). The abundance of ~40% is very similar to those in Y-81020 (CO3.05; Tenner et al. 2013), Acfer 094 (ungroup C3.00; Ushikubo et al. 2012), and Kaba (CV3; Hertwig et al. 2018).

The presence of variable oxygen isotope ratios in a single chondrule can be interpreted as the consequence of the mixing process between ¹⁶O-rich precursors and a ¹⁶O-poor nebular gas (e.g., Krot et al., 2005). However, such model cannot explain the presence of ¹⁶O-poor relict olivine grains (in chondrule A2) (Fig. 5). Although amoeboid olivine aggregates (AOAs) may have been the precursors of relict olivine grains with the lowest Δ¹⁷O values (down to −17.9‰ in chondrule B6; Fig. 5), this result suggests that some relict olivine grains did not originated from AOA-like materials or ¹⁶O-rich refractory precursors (e.g., Krot et al., 2006a, and references therein; Ushikubo et al., 2012). Twelve of the 19 relict olivine grains have Δ¹⁷O values ranging between −3.4‰ and −6.0‰. They are thus possibly related to material formed in the major isotope reservoir (¹⁷O~−5%) discussed above, as well as individual analyses from the five heterogeneous chondrules with Δ¹⁷O values ranging between −4.0‰ and −6.5‰ (Table 2). It implies mixing events and material exchanges from one isotope reservoir to another at the time of chondrule formation, i.e., 2–4 Ma after CAIs (e.g., Kita and Ushikubo, 2012; Schrader et al., 2017a). Because approximately 35% of the chondrules investigated (both type I and type II) contain relict olivine grains and thus sampled at least two different oxygen isotope reservoirs, migration of chondrule and/or chondrule precursors within the protoplanetary disk may have been a common process.

The presence of relict olivine grains in some chondrules indicate that they did not completely melt during the heating event that formed the final liquids from which the pyroxene and the non-relict olivine grains crystallized. Due to the low diffusion coefficient of oxygen in olivine (~1×10⁻¹⁸ m/s² at 1700 K; Ryerson et al., 1989; Gerard and Jaoul, 1989, Dohmen et al., 2002), unmelted relict olivine grains can preserve their oxygen isotope signatures. A part of the relict grains could correspond to an earlier generation of chondrule olivine (e.g., Jones et al., 2004; Russell et al., 2005; Kita et al., 2010; Rudraswami et al., 2011; Ushikubo et al. 2012; Schrader et al., 2014; Tenner et al., 2015; Krot and Nagashima, 2017), though some of these relict olivine grains could originate from different disk regions (e.g., Berlin et al., 2011; Weisberg et al., 2011; Schrader and Davidson, 2017; Schrader et al., 2017b; Tenner et al. 2017). Furthermore, refractory precursors, such as AOA-like ¹⁶O-rich olivine, likely existed among chondrule precursors that survived chondrule melting, which suggest that some precursors would derive from oldest inner disk regions (e.g., MacPherson et al., 2012).

4.2. Mg# and oxygen isotope relationship: Implications for the Murchison chondrule formation

As previously expressed by Ushikubo et al. (2012) and Tenner et al. (2013; 2015; 2017), the detailed relationship between chondrule Mg#’s and their host oxygen isotope ratios at sub% precisions from a representative suite of chondrules (n≥30) can provide important information about the origin and formation of chondrules.

4.2.1. Chondrules with Mg# >98.5 and Δ¹⁷O ~−5‰

As shown in Fig. 6, type I chondrules in Murchison become more ¹⁶O-poor with decreasing Mg#, towards the Δ¹⁷O values of type II chondrules at ~−2.5‰. With the exception of chondrule C6 (Δ¹⁷O: −1.2‰; Mg#: 99.6), chondrules with Mg#’s between 99.5 and 98.9 have Δ¹⁷O values ranging from −6.0‰ to −4.1‰. These trends are similar to those obtained for type I chondrules in Acfer 094 and CO chondrites (Ushikubo et al., 2012; Tenner et al., 2013) (Fig. 8), and to a lesser extent for the Y-82094 ungrouped chondrite and CV chondrites (Libourel and Chaussidon, 2011; Rudraswami et al., 2011; Tenner et al., 2015, 2017; Hertwig et al., 2017; 2018; Kita et al., 2016). The majority of chondrules in CR and CH chondrites shows Δ¹⁷O values higher than −4‰, while a small number of chondrules with Δ¹⁷O = −5‰ and Mg#>98 has been reported (Nakashima et al. 2011; Schrader et al. 2013; 2014; Tenner et al. 2015). It indicates that all groups of CCs sampled a common Mg# ~99 chondrule-forming environment (e.g., same dust enrichment relative to the solar gas, similar H₂O content in solid precursors), with a Δ¹⁷O ~−5‰.

Fig. 8.

Comparison of the Mg# and Δ¹⁷O relationships of chondrules from Murchison (this work), Acfer 094 (ungrouped C3.00; Ushikubo et al., 2012), and the Y-81020 (CO3.05; Tenner et al., 2013).

This oxygen isotope reservoir, which can be found in each group of CC, was highly reducing. This is the same reservoir as the −5‰ oxygen isotope reservoir reported in previous works (e.g., Ushikubo et al., 2012) that is also called as the “−5.5‰” (Tenner et al., 2013) or the −5‰ ± 1‰ reservoir (Tenner et al., 2017). Based on thermodynamic equilibrium calculations (e.g., Ebel and Grossman 2000), Mg#’s of ~99 indicate formation at low fO₂ ~−3.5 log units below the IW buffer (e.g., Tenner et al., 2015). As shown in Fig. 6, Δ¹⁷O values increase as chondrule Mg#’s decrease from 99.5. Similar to those in CR chondrite chondrules (Tenner et al. 2015), this relationship can suggest an addition of ¹⁶O-poor H2O ice to the nearly anhydrous chondrule precursors, to form more oxidized chondrules with higher Δ¹⁷O values and lower Mg#’s. According to the mass balance calculations involving a ¹⁶O-rich anhydrous dust and ¹⁶O-poor H2O ice that were given in Tenner et al. (2015) to model Mg#-Δ¹⁷O diagram of CR chondrite chondrules, Mg# 99 chondrules with a similar Δ¹⁷O values in Murchsion might form at dust to gas ratios of ~50× of solar gas and from dust with 0–0.4× the atomic abundance of H2O ice, relative to CI dust. In addition to this mixing process, an increase of the dust to gas ratios can also be invoked for formation of more oxidized chondrules, as well as a combination of these two processes.

We can also note here that one POP chondrule (C6) in Murchison has a high Mg# (99.6) but a higher Δ¹⁷O (−1.2%) that differs from the trend defined by chondrules with Mg#’s > 98.5 (Fig. 6). One type I PP chondrule containing dusty olivine grains in Acfer 094 and one BO chondrule in Y-81020 also display Δ¹⁷O values ranging between −3.3% and −1.2% for Mg# >98.5 (Fig. 8). There are no textural or chemical similarities among these three chondrules.

4.2.2. Chondrules with Mg# <96 and Δ¹⁷O ~−2.5‰

The four type II chondrules analyzed and the type I GOP chondrule C8 have host Δ¹⁷O values of around −2.5% (−2.7±0.3‰ to −2.3±0.2‰) for Mg#’s ranging from 64.9 to 96.0 (Table 1, Fig. 6). These characteristics were reported for many chondrules from Acfer 094 (Ushikubo et al., 2012), the Y-81020 CO3.0 chondrite (Tenner et al., 2013) (Fig. 8), as well as from the Y-82094 ungrouped chondrite (Tenner et al., 2017). Some chondrules in CR (Schrader et al., 2013, 2014; Tenner et al., 2015) and a few in CV3 oxidized (Rudraswami et al., 2011; Hertwig et al., 2018) chondrites also display these characteristics. These observations suggest that in addition to the Δ¹⁷O ~−5% reservoir, another oxygen reservoir (Δ¹⁷O ~−2.5‰) may have been ubiquitous in the region of the protoplanetary disk where CCs formed. However, chondrules in CCs are predominantly type I (Jones, 2012). As noticed by Tenner et al. (2015; 2017), the low modal abundance of type II chondrules relative to type I chondrules in CCs (e.g., Zanda et al., 2006, Jones, points out a probable limited chondrule-forming region for type II chondrules, in term of time and/or volume.

The increase of the oxidation state required by the formation of type II chondrules is often associated with an enhancement of the dust/gas ratios (Ebel and Grossman, 2000; Fedkin and Grossman, 2016) and ice contents in the chondrule precursors (e.g., Fedkin and Grossman, 2006). According to Tenner et al. (2015), the dust to gas ratio may have been variable (~300–2000× solar composition) to account for the range of oxygen fugacities required to form these chondrules (Mg#’s ranging from 65 to 96). Regarding predictions from dynamic models of the protoplanetary disk (e.g., Cuzzi et al., 2001), these high dust enrichments (up to ~2000×) were relatively rare. Thus, it may be the cause of the predominance of type I chondrules compared to type II chondrules in CM chondrites, as in Acfer 094 and Y-81020. Indeed, approximately 20% of chondrules in Acfer 094 are type II (Kunihiro et al., 2005), ~10% and ~10–40% in CO and CM, respectively (Jones, 2012). In Murchison, we only found 10 type II chondrules (excluding fragments and isolated Fe-rich olivine grains) among a total of around 150 chondrules initially identified from three thin sections, suggesting that the abundance of type II chondrules might be lower than 10%, an abundance closer to the value reported for CV chondrites ~5% of chondrules are type II; Jones, 2012).

Interestingly, the type I chondrules with Mg#’s of ~94–96 in CM, CO, and Acfer 094 have host Δ¹⁷O values of ~−2.5‰ (Fig. 8). This ~−2.5‰ value is characteristic of the type II chondrules in CM, CO, and Acfer 094 (Fig. 8). These chondrules have not been observed in CR and CV3 chondrites, in which less than 5% of chondrules are type II. In CR chondrites, some type I chondrules have Mg# ranging between 90 and 96 but their host Δ¹⁷O values are dispersed around ~−3.5‰ and ~−1‰ (Schrader et al., 2013, 2017a; Tenner et al. 2015). Although the Δ¹⁷O ~−2.5‰ isotope reservoir in CM, CO and Acfer 094 is extended towards high Mg# values (up to 96) compared to CV and CR chondrites, it does not seem to support a continuous trend between Mg#’s ~99 and ~65. Compared to chondrules with Mg# > 98.5 (see section 4.2.1), the relatively constant Δ¹⁷O values observed here over a wide range of lower Mg#’s suggest an existence of homogenized isotope reservoirs. Such reservoirs could have been formed as a result of the mixing between ¹⁶O-rich anhydrous dust and ¹⁶O-poor water ice, as in the case of model endmembers by Tenner et al. (2015) with Δ¹⁷O values of −6‰ and +5‰, respectively.

4.3. The CO-CM relationship

4.3.1. Chondrules in CO, and CM chondrites

In order to investigate the possible CO-CM relationship, we compared our oxygen isotope data obtained for Murchison to those from the CO3.0 chondrite Y-81020, as well as Acfer 094, which can be considered similar to the primitive CM material before aqueous alteration (e.g., Rubin et al., 2007; Hewins et al., 2014) (Fig. 8). Many chondrules from Murchison, Acfer 094, and Y-81020 have Mg#’s higher than 98, with Δ¹⁷O values ranging between approximately −6‰ and −4‰ (Fig. 8). These CCs sampled a same highly reduced and ¹⁶O-rich reservoir (see also section 4.2.1). They also sampled a common ¹⁶O-poor reservoir with Δ¹⁷O values of around −2.5‰, corresponding to chondrules with Mg#’s lower than ~96. As Y-81020, it appears that Murchison did not sample the reservoir observed in CR chondrites and characterized by Δ¹⁷O values of 0% and Mg#’s < 96 (Connolly and Huss, 2010; Schrader et al., 2013, 2014, 2017a; Tenner et al., 2015). The bimodal distribution of the Δ¹⁷O values and the Δ¹⁷O-Mg# relationship observed in chondrules from Murchison is thus similar to those observed in the Y-81020 CO3.0 chondrite by Tenner et al. (2013). Another similarity between Murchison, Y-81020, and Acfer 094, is the apparent lack of chondrules with Δ¹⁷O ~−4‰ to ~−1‰ and Mg#’s of 98.5–97. Recent studies also indicate that this same component is absent in chondrules from CV chondrites (Hertwig et al., 2017; 2018). In contrast, these type I chondrules were commonly observed in CR chondrites (Schrader et al., 2013, 2014, Tenner et al., 2015).

In terms of both Mg#’s and oxygen isotope ratios, our data thus confirm that CM and CO chondrites contain a similar and indistinguishable population of chondrule. It thus supports a common origin for chondrules from these both groups of CCs. This result is in agreement with the chemical similarities reported by Frank et al. (2014) and Schrader and Davidson (2017) for the high-temperature anhydrous silicates in CO and CM chondrites.

As shown in Fig. 9, type II chondrule olivines in Murchison display the same MnO-FeO correlation as other CM chondrites (Frank et al., 2014; and references therein) including the Paris CM2 chondrite (Hewins et al., 2014; without chondrule #3). This correlation is also similar to that for CO chondrites (Berlin et al., 2011; Jones, 2012; Tenner et al., 2013; Schrader and Davidson, 2017) and Acfer 094 (Frank et al., 2014; Ushikubo et al., 2012) (Fig. 9), but is different from what has been observed for OCs and CR chondrites (Berlin et al., 2011; Schrader et al., 2015; Schrader and Davidson, 2017). As discussed by Schrader and Davidson (2017), this similar FeO/MnO correlation in olivine from CO and CM chondrules indicates similar formation conditions (e.g., oxygen fugacity, chondrule precursor compositions). According to data from this study, Ushikubo et al. (2012), and Tenner et al. (2013), type II chondrules in CM, CO, and Acfer 094 have similar FeO/MnO ratios of olivine (~100), similar ranges of Mg#’s (<96), and similar oxygen isotope ratios (Δ¹⁷O ~−2.5‰), which reflects a common formation environment. These data further strengthen the existence of the CO-CM genetic link.

Fig. 9.

MnO (wt%) vs. mol.% Fa for olivine grains in chondrules from Murchison. Data from CO3.0 (Jones, 1992), Acfer 094 (Ushikubo et al., 2012), and Paris (Hewins et al., 2014) are shown for comparison, as well as the fields defined by the CR (Berlin et al., 2011; Frank et al., 2014; Schrader et al., 2015) and unequilibrated ordinary chondrites (UOC) (Frank et al., 2014; and references therein). The CI ratio is represented by the solid line (data from Anders and Grevesse, 1989).

Many olivine grains in type II chondrules from reduced CV chondrites also displays FeO and MnO values that partially overlap the CO-CM field (Frank et al., 2014). In comparison, type II chondrules from oxidized CV chondrites have FeO and MnO contents closer to those measured in olivine in type II chondrules in OCs (Frank et al., 2014; and references therein) and CR chondrites (Berlin et al., 2011; Frank et al., 2014; Schrader et al., 2015). Acfer 094, CO, and CM chondrules thus shared similar formation conditions with at least some CV chondrules, suggesting a common environment in the same region of accretion. Regarding the relative low abundance of type II chondrules in the different groups of CCs compared to type I chondrules (e.g., Scott and Taylor, 1983; Jones, 1992; Weisberg et al., 1993; Kunihiro et al., 2005), this common environment was likely localized in the protoplanetary disk where CCs accreted.

The relict olivine grains from the Murchison and Paris (Hewins et al., 2014) chondrules plot within the fields defined by type I and type II chondrules and some within the gap between them (Fig. 9). As also indicated from oxygen isotope systematics of Murchison chondrules, this observation supports that mechanical exchanges occurred between the type I and type II chondrule-forming reservoirs. Ushikubo et al. (2012) also reported that most of the relict grains in Acfer 094 have Δ¹⁷O values corresponding to the two major reservoirs discussed above (Δ¹⁷O: −5‰; Δ¹⁷O: −2.5‰). Migration of solid precursor may have occurred between these two isotope reservoirs, suggesting they both existed in the protoplanetary disk at the time of chondrule formation (e.g., Ushikubo et al., 2012). Although their spatial relationships are still not known, materials from these Δ¹⁷O “−5‰” and “−2.5‰” reservoirs may have been contemporaneous and near each other (Kurahashi et al., 2008; Tenner et al., 2013; Ushikubo et al., 2013). This oxygen isotope relationship between type I and type II chondrules is in good agreement with petrological and chemical observations. For example, type I forsterite and metal relict grains were reported in type II chondrules in CR, CO, and LL ordinary chondrites (Wasson and Rubin, 2003; Ruzicka et al., 2007; Schrader et al., 2008). Type II chondrules could have been formed from the oxidation of type I chondrules (Connolly et al., 2008; Schrader et al., 2008; Ruzicka et al., 2008; Connolly and Huss, 2010), although a compositionally similar precursor for type I and type II cannot be definitely ruled out. Recently, Villeneuve et al. (2015) experimentally investigated the origin of type II chondrules. The authors also concluded that type II chondrules were mainly formed from type I chondrule material under oxidizing conditions, as supported by the occurrence of fayalitic olivine grains with relict forsterite cores (e.g., Wasson and Rubin, 2003; Ruzicka et al., 2007).

4.3.2. Whole rock and other chondritic components in CO and CM chondrites

The high temperature components in CM and CO chondrites are indistinguishable in term of Δ¹⁷O-Mg# relationship and oxygen isotope ratios. The average oxygen isotope ratios of type I chondrules in Murchison with Mg# > 98.5 and Δ¹⁷O ~−5‰ (except for chondrule C6) is −3.9‰ and −7.1% for δ¹⁸O and δ¹⁷O, respectively. The oxygen isotope ratios of type I chondrules with Mg# > 98.5 and Δ¹⁷O ~−5‰ in the Y-81020 CO3 chondrite (δ¹⁸O: −5.8‰; δ¹⁷O: −8.6‰) are very similar to those measured for type I chondrules in Murchison (Fig. 10). In the same way, the average oxygen isotope ratios of type II chondrules in Murchison is 2.5‰ and −1.1‰ for δ¹⁸O and δ¹⁷O, respectively (Fig. 10). These values are also very close to those of the average oxygen isotope ratios of type II chondrules in the Y-81020 CO3.0 chondrite that are −2.0‰ and −1.4‰ for δ¹⁸O and δ¹⁷O, respectively (Fig. 10).

Fig. 10.

Oxygen three-isotope diagram of CM and CO bulk samples, type I and type II chondrules, CM matrix, and the anhydrous mineral separates in Murchison. The four reference lines (Y&R, PCM, CCAM, and TF lines), regression line (dashed line), and data for the CM2 and CO3 bulks, CM2 matrices, and the anhydrous mineral separates of Murchison are the same as in Fig. 1. For type I chondrule mean values, chondrules with Mg# >97 are used. For Murchison, type I chondrule average values are obtained excluding chondrule C6. For Y-81020, the average value of type I chondrules with Mg# >97 are calculated using data from Tenner et al. (2013). The 2SD uncertainties of the type I and type II chondrules are represented by ovals corresponding to the covariance matrix of δ¹⁸O and δ¹⁷O for each data set.

The average oxygen isotope ratios of type I chondrules in Murchison are very close to those measured by Clayton and Mayeda (1984) (−4.2‰ and −7.4‰ for δ¹⁸O and δ¹⁷O, respectively) for the anhydrous mineral separates from Murchison. In term of oxygen isotope ratios, type I chondrules can thus be considered as representative of the anhydrous part of CM chondrites, as it was calculated for Murchison from seven fractions of olivine+pyroxene with density >3.18 obtained using a density separation method (Clayton and Mayeda, 1984). It is consistent with low abundance of type II chondrules in CM chondrites, possibly less than 10% of the whole chondrule population in Murchison as described earlier. Thus, type II chondrules with higher δ¹⁸O and δ¹⁷O values do not significantly contribute to the oxygen isotope ratios of the anhydrous part of CM chondrites, or are balanced by a small contribution from ¹⁶O-rich refractory inclusions in CM chondrites (e.g., Brearley and Jones, 1998; and reference therein).

As initially proposed by Clayton and Mayeda (1999) and later observed within a single whole rock (the Paris CM2 chondrite; Hewins et al., 2014), CM2 chondrites can be considered in term of oxygen isotope ratios as a mixture of two components between ¹⁶O-rich anhydrous silicates (represented by type I chondrules) and ¹⁶O-poor hydrated phases (matrix). This two component mixing is clearly seen in Fig. 10 in which the average value of type I chondrules in CM plots along the CM mixing line that also pass through the field defined by the bulk compositions of the CO falls (Fig. 10). Regarding our results, it seems that the genetic CO-CM link previously suggested on the basis of their oxygen isotope ratios was due to the similarity between oxygen isotope ratios of anhydrous components of CM and those of bulk CO chondrites. Because of limited degrees of aqueous alterations and thermal metamorphism in CO chondrites, bulk CO chondrites show a narrow range of oxygen isotope ratios, which remained close to those of their ¹⁶O-rich anhydrous components. Indeed, CO chondrites are also dominated by type I chondrules (e.g., Jones, 2012), so that their bulk compositions are very similar to the average compositions of type I chondrules (Fig. 10). By comparison, bulk compositions of CV chondrites display more variability along the CCAM line because of secondary alteration processes (e.g., Clayton and Mayeda, 1999), while pristine type I chondrules in CV chondrites display similarities with type I chondrules from both CM and CO chondrites (Hertwig et al., 2017; 2018). This study thus highlighted the fact that CM chondrites contain high temperature components very similar to other major CCs, not only CO chondrites. The isotope signatures of such components are well preserved in chondrule mafic minerals in Murchison, while bulk Murchison contains a significant amount of low temperature components, such as organic matter, hydrated chondrule mesostasis, and matrix phases.

4.3.3. A single CO-CMparent body?

As discussed above, CM and CO chondrules are similar in terms of oxygen isotope systematics (Tenner et al., 2013; Chaumard et al., 2016; Schrader et al., 2017b) and major and minor element compositions (e.g., Berlin et al., 2011; Schrader and Davidson, 2017). In addition, it has been shown that CO and CM chondrites display similar refractory lithophile element abundances (Kallemeyn and Wasson, 1981). Oxygen isotope studies of bulk CM and CO chondrites show characteristic relationships indicating their genetic relationships (Fig. 10). It has been argued that similar observed properties of CO and CM chondrites may indicate they originated from a single parent body. However, CO chondrites are less hydrated than CM chondrites (e.g., Keller and Buseck, 1990), suggesting that the abundance of water/ice accreted with both chondrites were quite different. As discussed by Greenwood et al. (2014), a possible explanation for water/ice abundances is that CO and CM chondrites might sample the dehydrated inner part and hydrated surface of a single parent body, in which aqueous fluids were generated in the inner part and moved outwards during the heating of the parent body (Fu and Elkins-Tanton, 2014). However, this scenario may not explain the compositional gap in oxygen isotope ratios between the fields defined by the bulk CM and CO falls (Fig. 10), which were recently reduced from the Paris data (Hewins et al., 2014).

Alternatively, CM and CO chondrites may thus derive from two different parent bodies that accreted similar ¹⁶O-rich high temperature and anhydrous components but with different amount of ice/water. Although some CO3 chondrites experienced limited hydrothermal alteration (e.g., Rubin, 1998), i.e., producing secondary magnetite and fayalite (e.g., Doyle et al., 2015), the CO parent body was significantly less hydrated than the CM parent body. This is supported by the observation of many CO3 chondrites with low degrees of thermal metamorphism that still display pristine features such as the presence of Si and Fe-rich amorphous silicate material (i.e., in the CO3.0 Allan Hills A77307; Brearley 1993) and metal (e.g., Keller and Buseck, 1990).

Several properties of CO and CM chondrites argue against a common parent body. There are differences in the chondrule sizes (0.15 and 0.30 mm in CO and CM chondrites, respectively; e.g., Rubin and Wasson, 1986; Rubin, 1989, 2010; May et al., 1999; Weisberg et al., 2006; Friedrich et al., 2015; Schrader and Davidson, 2017), abundance of matrix (30–35 and 70 vol.% in CO and CM chondrites, respectively; Scott et al., 1996; Weisberg et al., 2006; and references therein), cosmic-ray exposure ages (Eugster, 2003), bulk compositions of chondrules from CO and CM chondrites (Rubin and Wasson, 1986), and abundance of FeO-poor relict grains in type II chondrules (see section 4.3; Schrader and Davidson, 2017). Furthermore, lack of breccias composed of CO- and CM-like materials is another argument against a common parent body (Bischoff et al., 2006). As also discussed by Schrader and Davidson (2017), the relationship between the bulk oxygen isotope (Greenwood et al., 2014) and ε⁵⁴Cr compositions (Sanborn et al., 2015) of CO and CM chondrites may not support a common origin from a single parent body.

4.3.4. Accretion time and location relative to the snow line

Constraints on the accretion times of the CO and CM parent bodies were recently provided by the study of their peak temperatures and Mn-Cr ages of aqueously-formed phases. Based on secondary fayalite, Doyle et al. (2015) reported an accretion time of the CO parent body of ~2.1–2.4 Ma after CAI formation. Fujiya et al. (2012) studied secondary calcite and dolomite in the CM chondrites Murchison, Y-791198, Allan Hills 83100, and Sayama. In order to compare results from CO and CM chondrites, Doyle et al. (2015) recalculated the relative ages of CM chondrites analyzed by Fujiya et al.(2012). These corrected ages recalculated by Doyle et al. (2015) and anchored to the same U-corrected Pb-Pb age of the D’Orbigny angrite and relative to the age of CV CAIs give an accretion time of the CM parent body of ~3.7–5.0 Ma after CAIs. Although they use a different CAI age, Sugiura and Fujiya (2014) also calculated a late age of accretion of the CM parent body (~3.5 Ma) by comparison to the CO parent body (~2.7 Ma).

²⁶Al-²⁶Mg systematics of chondrules from the Y-81020 CO3.05 chondrite indicate of formation of type I and type II chondrules 1.7–2.5 and 2.0–3.0 Ma after CAI formation, respectively (Kunihiro et al. 2004b; Kurahashi et al., 2008). Formation ages of type I chondrules are fully consistent with the accretion time of CO3, while type II seem to have been formed slightly later, nearly 1 Ma after the estimated accretion time. The metamorphic temperatures of Y-81020 has been estimated to be 400–550°C using metal chemistry (Shibata and Matsueda 1994; Kimura et al. 2008). At this temperature range, some type II chondrules in Y-81020 may have been disturbed in their Al-Mg system due to fast Mg diffusion in Na-rich plagioclase that commonly occur among type II chondrules (Van Orman et al. 2014). Because of intense secondary processes, formation ages of chondrules in CM chondrites cannot be resolved yet.

Regarding their chemical properties, it is commonly considered that CCs derived from parent bodies formed in the protoplanetary disk at different times and/or locations, as well as various ice/water and dust contents (e.g., Zolensky and McSween, 1988; Brealey, 2006; Chambers, 2006; Krot et al., 2006b; and references therein). With a low water ice/rock mass ratio (~0.1–0.2; Doyle et al., 2015), CO chondrites might be accreted very close to the snow line (e.g., Wood et al, 2005), where the evaporation of ice was not efficient enough to destroy ice-rich particles (e.g., Cyr et al., 1998; Lunine, 2006). Contrary to CO chondrites, CM chondrites have larger water ice/rock mass ratios (~0.3–0.6; Clayton and Mayeda, 1999). It implies that the CM parent body formed beyond the snow line.

Depending on many parameters such as the luminosity of the proto-Sun, the opacity of the disk, and the mass transport throughout the protoplanetary disk, the location of the snow line likely varied with time (e.g., Ciesla and Cuzzi, 2006; Min et al., 2011; Oka et al., 2011; Martin and Livio, 2012; Cieza et al., 2016). As discussed by Doyle et al. (2015), the snow line may have been located around 2–3 AU at the formation of the CO parent body, implying an accretion in the inner part of the Solar System, i.e., close to the current position of the asteroid belt. Because CM and CO chondrites both sampled the same high-temperature ¹⁶O-rich components, it would imply they accreted at the same location across the protoplanetary disk. Indeed, the snow line can move towards the Sun during the early evolution of the disk (e.g., Ciesla and Cuzzi, 2006; Min et al., 2011;Martin and Livio, 2012). Thus, the snow line would have been situated near the 2–3 AU region of the disk when the CO parent body formed (~2.1–2.4 Ma after CAI formation; Doyle et al., 2015), then closer to the Sun at the accretion time of the CM parent body (~3.7–5.0 Ma after CAI formation; ages from Fujiya et al., 2012 corrected by Doyle et al., 2015) that accreted more ice/water than CO chondrites. Within this framework, chondrules in CM could have been formed at the same time as those from CO chondrites, or later. Unfortunately, and as mentioned above, there is no formation ages of CM chondrules because the Al-rich phases in mesostasis are altered and not suitable for ²⁶Al-²⁶Mg chronometers, nor U-Pb and other chronometers. Bouvier et al. (2013) reported bulk Mg isotope measurements of a series of chondrules from the Murchison and Murray CM2 chondrites. No detectable ²⁶Mg* variations was measured in all the chondrules analyzed, even for the one with the highest ²⁷Al/²⁴Mg ratio (though only 0.316; Bouvier et al. 2013). The authors concluded that CM chondrules were formed >1 Ma after CAIs, or more possibly, that the Al-Mg isotope systematics were reset by secondary parent body alteration processes after the decay of ²⁶Al. Although there are numerous similarities between the high-temperature components of chondrules in CO and CM chondrites, we thus cannot argue for a simultaneous formation.

CM-like clasts are ubiquitous within impact breccias of other groups/classes of meteorites (e.g., Zolensky et al., 1996; Gounelle et al., 2003; Bischoff et al., 2006; and references therein) and suggests that CM-like material was widespread in the early Solar System. Although most of the CM2 chondrites likely originated from a parent body in near Earth orbit (Morbidelli et al., 2006), it may support the existence of additional CM parent bodies (e.g., Zolensky et al., 1996). Our data cannot rule out the possibility of multiple CM parent bodies accreted beyond the snow line. However, and as discussed by Hewins et al. (2014), it may be unlikely that CM2 chondrites derived from different parent bodies while they experienced very similar aqueous alteration processes.

5. CONCLUSIONS

The main conclusions obtained by the in-situ oxygen three-isotope analyses of olivine and pyroxene grains in chondrules from the Murchison CM2 chondrite are:

1. Most chondrules consist of olivine grains with oxygen isotope ratios similar to coexisting pyroxene grains suggesting a co-magmatic crystallization. The host values calculated for these internally homogeneous chondrules are considered as representative of the final melt from which olivine and pyroxene crystallized.

2. Some olivine grains in chondrules were considered as relict grains with distict oxygen isotope ratios, which formed earlier than the coexisting minerals and survived during the high-temperature event of the final melt. Since most of the relict olivine grains have oxygen isotope ratios within the range defined by the host chondrule values (Δ¹⁷O from ~−2% to ~−6%), multiple mixing events and material exchanges likely occurred at the time of chondrule formation.

3. Based on the variations of the oxygen isotope ratios of their coexisting olivine and pyroxene grains, five chondrules were described as heterogeneous in oxygen isotope ratios. As supported by their textures, these heterogeneous chondrules might result from the incomplete melting of former chondrule precursors because of a lower degree of the melting event and/or a shorter duration of melting.

4. Oxygen isotope ratios of host chondrules plot close to the PCM line, with Δ¹⁷O values ranging from −6.0‰ to −1.2‰. With one exception, chondrules with high Mg# (>98.5) have host Δ¹⁷O values of~−5‰ consistent with a formation in a reducing environment and dust to gas ratios of ~50×. Chondrules with lower Mg# (~96–65) have host Δ¹⁷O values of ~−2.5‰. These characteristics are consistent with a formation at high dust to gas ratios (up to ~2000×), from the anhydrous precursor of Δ¹⁷O ~ −5‰ that formed Mg# ~99 chondrules, enriched in H2O ice (~0.3–0.4× the CI dust; Tenner et al., 2015). These two reservoirs are similar to those also sampled by other CCs, such as CV (Libourel and Chaussidon, 2011; Rudraswami et al., 2011; Hertwig et al., 2017; 2018), CO (Tenner et al., 2013), Y-82094 (Tenner et al., 2017), and Acfer 094 (Ushikubo et al., 2012).

5. In term of both Mg# and oxygen isotope ratios, the populations of chondrules sampled by CM and CO chondrites are indistinguishable. CO and CM chondrites thus contain similar high-temperature anhydrous silicates, as also supported by the Fe/Mn ratios of the olivine grains in type II chondrules from these two groups of CCs and reduced CV chondrites (Berlin et al., 2011; Tenner et al., 2013; Frank et al., 2014; Hewins et al., 2014; Scharder and Davidson, 2017).

6. Based on chondrule oxygen isotope systematics, we further argued that CM and CO chondrites might derive from distinct parent bodies with different amount of water/ice but accreted similar high-temperature ¹⁶O-rich components, rather than a common parent body. The difference in water/ice abundance between the CO and CM parent bodies might be related to the location of the snow line compared to their accretion regions. The snow line probably moved through the CO and CM chondrule-forming regions towards the Sun between the accretion time of the CO and CM parent bodies, which are considered to be ~2.1–2.4 and ~3.7–5.0 Ma after CAI (Fujiya et al., 2012; Doyle et al., 2015), respectively.

Supplementary Material

EA1

EA1: Instrumental bias correction for SIMS analysis due to variability in endmember composition.

EA2

EA2: Backscattered electron images of studied Murchison chondrules with SIMS pit numbers

EA3

EA3: Electron microprobe analyses of olivine and pyroxene

EA4

EA4: SIMS Oxygen three-isotope data of olivine and pyroxene