Extended chondrule formation intervals in distinct physicochemical environments: Evidence from Al-Mg isotope systematics of CR chondrite chondrules with unaltered plagioclase.

Abstract

Al-Mg isotope systematics of twelve FeO-poor (type I) chondrules from CR chondrites Queen Alexandra Range 99177 and Meteorite Hills 00426 were investigated by secondary ion mass spectrometry (SIMS). Five chondrules with Mg#’s of 99.0 to 99.2 and Δ¹⁷O of −4.2‰ to −5.3‰ have resolvable excess ²⁶Mg. Their inferred (²⁶Al/²⁷Al)₀ values range from (3.5 ± 1.3) × 10^‒6 to (6.0 ± 3.9) × 10^‒6. This corresponds to formation times of 2.2 (–0.5/+1.1) Myr to 2.8 (‒0.3/+0.5) Myr after CAIs, using a canonical (²⁶Al/²⁷Al)₀ of 5.23 × 10^˗5, and assuming homogeneously distributed ²⁶Al that yielded a uniform initial ²⁶Al/²⁷Al in the Solar System. Seven chondrules lack resolvable excess ²⁶Mg. They have lower Mg#’s (94.2 to 98.7) and generally higher Δ¹⁷O (−0.9‰ to −4.9‰) than chondrules with resolvable excess ²⁶Mg. Their inferred (²⁶Al/²⁷Al)₀ upper limits range from 1.3 × 10^‒6 to 3.2 × 10^‒6, corresponding to formation >2.9 to >3.7 Myr after CAIs. Al-Mg isochrons depend critically on chondrule plagioclase, and several characteristics indicate the chondrule plagioclase is unaltered: (1) SIMS ²⁷Al/²⁴Mg depth profile patterns match those from anorthite standards, and SEM/EDS of chondrule SIMS pits show no foreign inclusions; (2) transmission electron microscopy (TEM) reveals no nanometer-scale micro-inclusions and no alteration due to thermal metamorphism; (3) oxygen isotopes of chondrule plagioclase match those of coexisting olivine and pyroxene, indicating a low extent of thermal metamorphism; and (4) electron microprobe data show chondrule plagioclase is anorthite-rich, with excess structural silica and high MgO, consistent with such plagioclase from other petrologic type 3.00–3.05 chondrites. We conclude that the resolvable (²⁶Al/²⁷Al)₀ variabilities among chondrules studied are robust, corresponding to a formation interval of at least 1.1 Myr.

Using relationships between chondrule (²⁶Al/²⁷Al)₀, Mg#, and Δ¹⁷O, we interpret spatial and temporal features of dust, gas, and H₂O ice in the FeO-poor chondrule-forming environment. Mg# ≥ 99, Δ¹⁷O ~−5‰ chondrules with resolvable excess ²⁶Mg initially formed in an environment that was relatively anhydrous, with a dust-to-gas ratio of ~100×. After these chondrules formed, we interpret a later influx of ¹⁶O-poor H₂O ice into the environment, and that dust-to-gas ratios expanded (100× to 300×). This led to the later formation of more oxidized Mg# 94–99 chondrules with higher Δ¹⁷O (−5‰ to –1‰), with low (²⁶Al/²⁷Al)₀, and hence no resolvable excess ²⁶Mg.

We refine the mean CR chondrite chondrule formation age via mass balance, by considering that Mg# ≥ 99 chondrules generally have resolved positive (²⁶Al/²⁷Al)₀ and that Mg# < 99 chondrules generally have no resolvable excess ²⁶Mg, implying lower (²⁶Al/²⁷Al)₀. We obtain a mean chondrule formation age of 3.8 ± 0.3 Myr after CAIs, which is consistent with Pb-Pb and Hf-W model ages of CR chondrite chondrule aggregates. Overall, this suggests most CR chondrite chondrules formed immediately before parent body accretion.

1. INTRODUCTION

The presence of ²⁶Al in the early Solar System (e.g. Lee et al., 1976) can be used to infer the relative ages of materials formed during that epoch. Extinct parent ²⁶Al is measured as excess daughter ²⁶Mg in early Solar System materials. As ²⁶Al has a half-life of 0.705 Myr (Norris et al., 1983) it is a well-resolved chronometer, assuming (1) ²⁶Al was homogeneously distributed in the Solar System prior to formation of materials of interest; and (2) there was no later addition of ²⁶Al. Ca- and Al-rich inclusions (CAIs), chondrules, and achondrites have ²⁶Al-²⁶Mg systematics that are generally concordant with other isotope systems, including ²⁰⁷Pb-²⁰⁶Pb, ⁵³Mn-⁵³Cr, and ¹⁸²W-¹⁸²Hf (e.g. Jacobsen et al., 2008; Nyquist et al., 2009; Spivak-Birndorf et al., 2009; Amelin et al., 2002; 2010; Bouvier and Wadhwa, 2010; Schiller et al., 2010; 2015; Bouvier et al., 2011a; 2011b; Kleine et al., 2012; Kita et al., 2013; Wadhwa et al. 2014; Budde et al., 2016; 2018; Schrader et al., 2017 with some exceptions such as ⁵³Mn-⁵³Cr in some CAIs (e.g. Bogdanovski et al., 2002; Papanastassiou et al.; 2002; Nyquist et al., 2009). Among these materials, many CAIs have the highest inferred initial ²⁶Al/²⁷Al, or (²⁶Al/²⁷Al)₀, with a canonical value of ~5.2 × 10^–5 (e.g. Jacobsen et al., 2008; Larsen et al., 2011; MacPherson et al., 2010). As such, the canonical CAI (²⁶Al/²⁷Al)₀ serves to anchor the Al-Mg isotope system, and the agreement among multiple isotope systems from different types of early Solar System materials supports homogeneous ²⁶Al distribution when and after canonical CAIs formed (Kita et al., 2013 and references therein). However, this assumption is actively debated, mainly in two ways. The first considers homogeneous ²⁶Al distribution at the time of canonical CAIs, but that isotopically anomalous and ²⁶Al-poor CAIs, with no evidence of disturbance from secondary processing, formed prior to homogeneous ²⁶Al distribution (e.g. Makide et al., 2013; Kuffmeier et al., 2016; Park et al., 2017). The second considers disk-wide ²⁶Al heterogeneity, based on inconsistencies between Pb-Pb ages and Al-Mg isotope systematics among some CAIs, chondrules, and achondrites (Larsen et al., 2011; Connelly et al., 2012; Schiller et al., 2015; Olsen et al., 2016; Van Kooten et al., 2016 Bollard et al., 2015; 2017). While recognizing the merits of both sides to this debate, we operate under the assumption that ²⁶Al was homogeneously distributed after canonical CAI formation, and use the Al-Mg isotope system to calculate the age of chondrule formation relative to canonical CAIs.

Secondary ion mass spectrometry (SIMS) is a powerful analytical tool for early Solar System chronology, because it provides the ability to measure isotope ratios of small meteoritic components in-situ, with micron-scale resolution. Based on SIMS measurements, chondrules have lower (²⁶Al/²⁷Al)₀ than canonical CAIs, as those from the least metamorphosed chondrites, Acfer 094 (ungr. C3.00), Yamato 81020 (CO3.05), and Semarkona (LL3.01) (classifications: Kimura et al., 2008), have values ranging from 0.24 × 10^–5 to 1.6 × 10^–5 (Nagashima et al., 2018 and references therein; Fig. 1a). Using the aforementioned ²⁶Al half-life and canonical (²⁶Al/²⁷Al)₀, these (²⁶Al/²⁷Al)₀ values correspond to formation ages of 1.2 to 3.1 Myr after CAIs, (Fig. 1b, shaded region). Among CO and Acfer 094 chondrites, FeO-poor chondrules with anorthite-rich plagioclase formed over an even narrower interval (mean relative age: 2.14 ± 0.27 [1SD] Myr after CAIs; Kurahashi et al., 2008a; Ushikubo et al., 2013). SIMS Al-Mg isotope systematics have also been measured in CR chondrite chondrules, and within analytical uncertainties approximately one-third (10 of 35) have resolvable excess ²⁶Mg (Fig. 1b, solid symbols). Their inferred (²⁶Al/²⁷Al)₀ values are generally within the range of chondrules from Acfer 094, CO, and LL3 chondrites. However, roughly two-thirds (25 of 35) of CR chondrite chondrules studied have no resolvable excess ²⁶Mg (e.g. Fig. 1b, open symbols). In particular, SIMS data from Nagashima et al. (2014) and Schrader et al. (2017) indicate that, when applying (²⁶Al/²⁷Al)₀ upper limits, several chondrules formed at least 0.5 Myr after those from Acfer 094, CO, and LL3 chondrites (Fig. 1b).

Fig. 1.

(a) Representative isochrons for canonical CAIs (e.g. Jacobsen et al. 2008; MacPherson et al. 2010) and the ranges from Semarkona (LL3.01), Yamato 81020 (CO3.05), and Acfer 094 (ungr. C3.00) chondrite chondrules (Hutcheon and Hutchison 1989; Kita et al. 2000; Kunihiro et al. 2004; Kurahashi et al. 2008a; Rudraswami et al. 2008; Villeneuve et al. 2009; Hutcheon et al. 2009; Ushikubo et al. 2013). *δ²⁶Mg is defined in Section 2.5.1 of the main text. Slopes of isochrons correspond to inferred (²⁶Al/²⁷Al)₀. (b) CR chondrite chondrule (²⁶Al/²⁷Al)₀ values (symbols) versus the range of chondrule (²⁶Al/²⁷Al)₀ from LL3, CO3, and Acfer 094 chondrites (gray shading and dashed lines). Relative ages are calculated using the canonical (²⁶Al/²⁷Al)₀ shown in (a), and a ²⁶Al half-life of 0.705 Myr (Norris et al. 1983). For CR chondrite chondrule data, references 1–4 correspond to Schrader et al. (2017), Nagashima et al. (2014), Hutcheon et al. (2009), and Kurahashi et al. (2008b), respectively. Closed and open symbols correspond to chondrules with and without resolvable excess ²⁶Mg, respectively. Three chondrules from Hutcheon et al. (2009) were re-measured by Nagashima et al. (2014); only the data from the latter study are shown. (²⁶Al/²⁷Al)₀ values and uncertainties from Hutcheon et al. (2009) were determined using Isoplot. Data from Kurahashi et al. (2008b) with unresolvable excess ²⁶Mg are only reported as (²⁶Al/²⁷Al)₀ upper limits (hence, no symbols are shown).

Among currently available data from CR chondrite chondrules, there appears to be no correlation between their Al-Mg isotope systematics and properties related to the physicochemical environment from which they formed (Schrader et al., 2017). This is an important consideration, because CR chondrite chondrules (1) have mass-independent fractionated oxygen isotope ratios; and (2) become systematically depleted in ¹⁶O as their Mg#’s decrease (where Mg# = mol.% MgO/[MgO + FeO] of chondrule ferromagnesian silicates) (Connolly and Huss, 2010; Tenner et al., 2015; Schrader et al., 2013, 2014a; 2017). As lower chondrule Mg#’s correspond to more oxidized formation environments, based on metal-silicate phase equilibria (e.g. Kring, 1988; Zanda et al., 1994; Ebel and Grossman 2000), Connolly and Huss (2010) suggested this Mg# versus oxygen isotope relationship was caused by variable proportions of ¹⁶O-depleted H₂O ice (e.g. Sakamoto et al., 2007) and relatively ¹⁶O-rich material within chondrule precursors. Similar correlations of decreasing Mg#’s and ¹⁶O-depletion among chondrule populations are observed in other carbonaceous chondrites (e.g. Russell et al., 2010; Rudraswami et al., 2011; Ushikubo et al., 2012; Tenner et al., 2013; 2017; Chaumard et al., 2018; Hertwig et al., 2018; Yamanobe et al., 2018). Therefore, it is worthwhile to investigate further Al-Mg isotope systematics of CR chondrite chondrules with known Mg# and O-isotope signatures, to better understand if excess ²⁶Mg, or the lack thereof, (1) was related to the abundance of H₂O ice within the CR chondrite chondrule-forming environment, and (2) indicates timing similarities and/or differences to other chondrite chondrule-forming environments.

In order to use Al-Mg isotope systematics of early Solar System materials for chronology, well-defined internal isochrons should be established by plotting ²⁷Al/²⁴Mg versus excess ²⁶Mg (i.e. δ²⁶Mg*; defined in Section 2.5) with a large range of ²⁷Al/²⁴Mg ratios, where the slope corresponds to (²⁶Al/²⁷Al)₀ (e.g. Fig. 1a). A caveat of this approach for chondrules is that their isochrons depend almost solely on measurements of glassy mesostasis or plagioclase, which are the only constituent phases with sufficient ²⁷Al/²⁴Mg to resolve excess ²⁶Mg (olivine and pyroxene Al-Mg isotope systematics define the end at the origin). It is uncommon that glass and plagioclase coexist within a single chondrule. Glass and plagioclase often do not have large ranges of ²⁷Al/²⁴Mg ratios within a single chondrule, meaning isochron regressions are made based nearly on a “two-point line”. Therefore, any problems related to the closure of the glass or plagioclase isotope decay system will compromise chondrule isochrons. This is evident among thermally metamorphosed, petrologic type ≥3.2 chondrites, where chondrule plagioclase or glass data with significantly high ²⁷Al/²⁴Mg (>100) and with limited-to-no excess ²⁶Mg (e.g. Huss et al., 2001; Kita et al., 2004a; 2004b) are interpreted as representing a late disturbance of the Al-Mg isotope system. Furthermore, if only considering chondrule plagioclase or glass with excess ²⁶Mg, such data by themselves are often as consistent with a horizontal line as they are with an isochron. As pointed out by Alexander and Ebel (2012), a horizontal line would indicate a late disturbance (after all ²⁶Al decayed) that modified Al/Mg ratios and/or re-equilibrated Mg isotopes in glass or plagioclase. Thus, although Al-Mg isotope systematics of chondrules from petrologic type ~3.0 chondrites are considered robust (e.g. Kita and Ushikubo, 2012), more critical assessments pertaining to the reliability of plagioclase or glass data used for chondrule chronology are beneficial.

Here, we present SIMS Al-Mg isotope systematics of chondrules from Queen Alexandra Range (QUE) 99177 and Meteorite Hills (MET) 00426, which are two of the least altered CR chondrites (petrologic type 2.6 to 3.0; Abreu and Brearley, 2010; Alexander et al., 2013; Harju et al., 2014; Howard et al., 2015). For all chondrules studied, their O-isotope ratios and Mg#’s were previously determined (Tenner et al., 2015), allowing for comparisons to their Al-Mg isotope systematics. We also investigated several attributes of chondrule plagioclase, including elemental relationships, Raman spectra, and characterization by focused ion beam/transmission electron microscopy (FIB/TEM), to determine its reliability with respect to closure of the ²⁶Al-²⁶Mg decay system.

2. SAMPLES AND METHODS

Twelve chondrules, Q7, Q13, Q15, Q16, Q19, Q35, Q36, Q38, Q44, M3, M6, and M23 (as labeled in Tenner et al., 2015) were investigated for SIMS Al-Mg isotope systematics, from thin sections QUE 99177,49 and MET 00426,46. These chondrules were chosen because they have several regions of plagioclase that could be analyzed with 5–10 μm SIMS spot sizes, and cover an appreciable range of O-isotope ratios. Their chondrule Mg#’s range from 94.2 to 99.2, and their Δ¹⁷O values range from –0.9‰ to –5.3‰ (where Δ¹⁷O = δ¹⁷O – 0.52 × δ¹⁸O; δ^17,18O = [(R_sample/R_VSMOW) – 1] × 1000; R = ^17,18O/¹⁶O; and VSMOW = Vienna Standard Mean Ocean Water; Baertschi, 1976). All chondrules studied are within 0.8‰ of the slope-1 δ¹⁸O versus δ¹⁷O primitive chondrule mineral (PCM) line defined by Acfer 094 chondrule phenocrysts (Ushikubo et al., 2012). Chondrules Q13 and Q36 have compound textures (barred olivine interiors surrounded by porphyritic exteriors), and chondrule Q15 is layered, with a porphyritic texture but containing smaller chondrules along the periphery. The other chondrules have porphyritic textures (Fig. 2). In addition to the chondrules described above, plagioclase from a single Acfer 094 chondrule, G15 (as labeled in Ushikubo et al., 2012), was evaluated by transmission electron microscopy (TEM; described below), for comparison to plagioclase from CR chondrite chondrules.

Fig. 2.

Backscattered electron images of QUE 99177 (Q-designation) and MET 00426 (M-designation) chondrules studied for their Al-Mg isotope systematics. Scale bars are 200 μm. All chondrules are FeO-poor; Q7, Q16, Q19, Q35, Q38, Q44, M6, and M23 are type I porphyritic olivine-pyroxene (POP: 20–80% modal olivine, when considering only olivine and pyroxene); M3 is type I porphyritic pyroxene (PP: <20% modal olivine); Q13 and Q36 are type I compound chondrules, each with barred olivine (BO) and POP features; Q15 is a layered chondrule that is type I POP textured with smaller chondrules along the periphery (white arrows).

2.1. Scanning electron microscopy

Backscattered electron (BSE) images of chondrule olivine, pyroxene and plagioclase were collected with a Hitachi S-3400 N SEM equipped with an energy dispersive X-ray spectrometer (EDS). Minerals that were free of cracks, inclusions, and pits, and that could accommodate a 5‒15 μm SIMS spot analysis were identified as regions of interest.

2.2. Electron microprobe analysis

Prior to SIMS, major element oxide compositions (SiO₂, TiO₂, Al₂O₃, Cr₂O₃, FeO, MnO, MgO, CaO, Na₂O, and K₂O) of chondrule silicate minerals were measured by electron probe microanalysis (EPMA) using a Cameca SX-51 instrument at the University of Wisconsin-Madison. Operating conditions included a 15 kV accelerating voltage, a 10 nA beam current, a 1–2 μm diameter beam, and peak and background count times of 10 and 5 seconds, respectively. Standards were olivine (Fo_0, Fo_83, Fo_89, Fo₁₀₀), pyroxene (enstatite, diopside, wollastonite, augite, jadeite, and omphacite), plagioclase (An_0, An_18, An_49, An_67, An_78, An₉₅), rutile, hornblende, chromite, hematite, tephroite, and microcline. Carbon coat thicknesses of standards and thin sections were ~20 nm (indigo red on polished brass; Kerrick et al., 1970). Using a φ(ρZ) matrix correction, data were reduced with Probe for Windows software. Detection limits (99% confidence) were 0.06, 0.05, 0.05, 0.06, 0.09, 0.08, 0.05, 0.04, 0.05, and 0.03 wt.%, for respective oxides listed above. Olivine and pyroxene data are reported in Tenner et al. (2015). For plagioclase, Na count rates were monitored using albite-rich standards and no appreciable decrease was observed. Plagioclase standards were intermittently analyzed as unknowns, to ensure appropriate totals were achieved (99–101 wt.%), and to verify cations were within 0.01 units of their known stoichiometry (Table 1). For all plagioclase data (Table 1; Appendix EA1), endmembers, including Ca(Fe,Mg)Si₃O₈, KAlSi₃O₈, NaAlSi₃O₈, CaAl₂Si₂O₈, (Fe,Mg)Al₂Si₂O₈, and [ ]Si₄O₈, are calculated using methods described in Beaty and Albee (1980).

Table 1.

Plagioclase EPMA data from chondrules and standards analyzed in the same session

	Q7	±(a)	Q13	±	Q15	±	Q16	±	Q19	±	Q35	±	An78 std as unk	±	An78 std value(e)
wt. %
SiO2	46.36	0.74,0.72	47.28	1.32,1.80	47.03	0.85,1.08	46.90	2.24,1.16	49.99	0.57,0.89	49.11	1.88,1.17	49.16	0.36,0.31	49.00
AI2O3	33.29	0.97,0.86	32.10	1.71,1.20	32.62	0.73,0.57	33.02	0.60,1.70	30.69	0.97,0.88	30.80	1.14,1.45	32.49	0.32,0.41	32.90
FeO	0.39	0.66,0.20	0.23	0.38,0.24	0.59	0.37,0.26	0.52	0.48,0.29	0.33	0.26,0.12	0.55	0.58,0.39	0.40	0.13,0.15	0.37
MgO	0.79	0.09,0.14	0.92	0.26,0.31	0.89	0.23,0.20	0.86	0.09,0.10	0.90	0.11,0.12	1.16	0.19,0.29	0.13	0. 04,0.06	0.12
CaO	19.11	0.52,0.44	18.55	0.51,0.66	18.85	0.47,0.35	18.90	0.54,1.38	17.34	0.34,0.68	18.30	0. 58,1.42	15.51	0. 30,0.29	15.75
Na2O	0.07	0.06,0.05	0.42	0.60,0.31	0.14	0.10,0.08	0.24	0.66,0.22	0.83	0.17,0.11	0.12	0.09,0.05	2.49	0.09,0.09	2.53
total	100.01	0.84,1.20	99.50	1.28,0.80	100.22	0.66,0.62	100.44	0.41,1.07	100.07	0.70,0.89	100.14	0.89,1.01	100.35	0.38,0.58	100.74
n	72		46		27		25		8		37		11
cations(b)
Si	2.137	0.028,0.038	2.185	0.054,0.074	2.165	0.026,0.039	2.152	0.090,0.042	2.283	0.038,0.042	2.251	0.091,0.046	2.242	0.015,0.011	2.228
Al	1.807	0.040,0.035	1.747	0.080,0.070	1.768	0.048,0.048	1.785	0.037,0.096	1.650	0.052,0.038	1.662	0.058,0.075	1.745	0.015,0.015	1.761
Fe2+	0.015	0.025,0.007	0.009	0.014,0.009	0.022	0.029,0.010	0.020	0.018,0.011	0.013	0.010,0.005	0.021	0.023,0.015	0.015	0.005,0.006	0.014
Mg	0.054	0.006,0.010	0.063	0.019,0.021	0.061	0.015,0.013	0.059	0.006,0.006	0.061	0.007,0.008	0.079	0.013,0.020	0.009	0. 003,0.004	0.008
Ca	0.943	0.024,0.022	0.918	0.027,0.032	0.929	0.026,0.022	0.929	0.028,0.073	0.848	0.011,0.029	0.898	0.026,0.068	0.757	0.013,0.015	0.766
Na	0.007	0.005,0.005	0.037	0.054,0.027	0.012	0.011,0.008	0.021	0.058,0.020	0.074	0.015,0.010	0.011	0.008,0.004	0.220	0.007,0.007	0.223
total	4.963	0.023,0.017	4.960	0.034,0.034	4.957	0.017,0.020	4.966	0.035,0.029	4.928	0.020,0.013	4.923	0.018,0.052	4.996	0.008,0.009	5.001
endmembers(c)
Ca(Fe,Mg)Si3O8	0.056	0.018,0.010	0.068	0.023,0.020	0.067	0.042,0.011	0.063	0.009,0.011	0.066	0.017,0.010	0.086	0.020,0.032	0.013	0.010,0.010	0.011
NaAISi3O8	0.007	0.005,0.005	0.037	0.054,0.027	0.012	0.011,0.008	0.021	0.058,0.020	0.074	0.015,0.010	0.011	0.008,0.004	0.220	0.007,0.007	0.223
CaAl2Si2O8	0.887	0.018,0.019	0.850	0.044,0.034	0.862	0.035,0.027	0.866	0.026,0.074	0.781	0.019,0.029	0.812	0.017,0.052	0.744	0.010,0.007	0.756
(Fe,Mg)Al2Si2O8	0.013	0.021,0.017	0.005	0.011,0.019	0.016	0.017,0.014	0.016	0.006,0.014	0.007	0.001,0.013	0.014	0.010,0.020	0.011	0.010,0.008	0.012
[ ]Si4O8	0.037	0.020,0.030	0.040	0.044,0.032	0.043	0.023,0.019	0.034	0.033,0.032	0.072	0.018,0.014	0.077	0.061,0.018	0.007	0.008,0.006	−0.005
normalized An(d)	0.993	0.006,0.006	0.958	0.031,0.059	0.986	0.009,0.052	0.976	0.022,0.066	0.914	0.009,0.018	0.987	0.007,0.012	0.768	0.006,0.008	0.769
	Q36	±	Q38	±	Q44	±	M3	±	M6	±	M23	±	An95 std as unk	±	An95 std value(f)
wt. %
SiO2	50.29	1.74,1.00	48.60	0.58,0.66	49.28	1.25,1.25	48.01	0.28,0.36	47.05	1.70,1.12	47.66	0.66,0.81	44.31	0.31,0.42	44.00
Al2O3	30.03	1.10,1.55	31.42	0.60,0.54	31.86	0.61,0.93	32.60	0.20,0.65	32.35	1.07,0.83	31.32	0.97,0.43	35.33	0.42,0.32	36.03
FeO	0.40	0.34,0.16	0.52	0.63,0.31	0.28	0.20,0.10	0.64	0.35,0.29	0.44	0.86,0.32	0.42	0.61,0.20	0.45	0.07,0.10	0.62
MgO	0.90	0.39,0.28	1.08	0.30,0.14	0.89	0.12,0.10	0.56	0.10,0.06	0.90	0.16,0.15	1.01	0.22,0.24	0.06	0.01,0.03	0.02
CaO	16.61	0.74,0.88	18.33	0.40,0.66	17.80	0.63,0.94	16.57	0.30,0.28	18.64	0.65,0.59	18.14	0.65,1.30	19.07	0.16,0.21	19.09
Na2O	1.30	0.79,0.36	0.23	0.24,0.13	0.68	0.43,0.36	2.01	0.09,0.03	0.34	0.26,0.27	1.02	0.81,0.68	0.49	0.12,0.04	0.53
total	99.71	0.76,0.80	100.31	0.85,1.11	100.79	0.15,0.71	100.40	0.29,1.18	99.68	0.84,0.87	99.66	1.15,1.00	99.74	0.49,0.48	100.29
n	30		24		17		7		28		32		6
cations(b)
Si	2.308	0.070,0.042	2.227	0.018,0.030	2.239	0.052,0.044	2.199	0.023,0.014	2.175	0.058,0.034	2.205	0.021,0.039	2.056	0.006,0.012	2.033
Al	1.623	0.045,0.091	1.696	0.033,0.027	1.704	0.041,0.053	1.758	0.013,0.025	1.756	0.052,0.054	1.706	0.052,0.025	1.931	0.014,0.008	1.960
Fe2+	0.015	0.013,0.006	0.020	0.024,0.012	0.011	0.008,0.004	0.025	0.013,0.011	0.017	0.033,0.012	0.016	0.024,0.008	0.017	0.003,0.004	0.023
Mg	0.061	0.026,0.019	0.074	0.020,0.009	0.060	0.008,0.007	0.038	0.007,0.004	0.062	0.011,0.011	0.070	0.014,0.017	0.004	0.001,0.002	0.001
Ca	0.816	0.037,0.044	0.900	0.020,0.036	0.865	0.031,0.048	0.812	0.014,0.007	0.922	0.030,0.037	0.898	0.033,0.069	0.947	0.007,0.016	0.944
Na	0.116	0.070,0.033	0.020	0.021,0.012	0.060	0.037,0.032	0.178	0.008,0.003	0.030	0.023,0.024	0.091	0.072,0.061	0.044	0.011,0.004	0.047
total	4.939	0.031,0.040	4.936	0.016,0.010	4.939	0.013,0.010	5.011	0.009,0.010	4.962	0.026,0.024	4.987	0.034,0.038	5.000	0.006,0.005	5.010
endmembers(c)
Ca(Fe,Mg)Si3O8	0.070	0.035,0.018	0.078	0.019,0.015	0.057	0.017,0.010	0.043	0.013,0.004	0.069	0.024,0.021	0.088	0.017,0.013	0.013	0.006,0.009	0.006
NaAlSi3O8	0.116	0.070,0.033	0.020	0.021,0.012	0.060	0.037,0.032	0.178	0.008,0.003	0.030	0.023,0.024	0.091	0.072,0.061	0.044	0.011,0.004	0.047
CaAl2Si2O8	0.746	0.024,0.058	0.822	0.019,0.032	0.808	0.024,0.049	0.770	0.007,0.012	0.853	0.028,0.041	0.810	0.037,0.061	0.934	0.008,0.007	0.938
(Fe,Mg)Al2Si2O8	0.007	0.004,0.018	0.016	0.012,0.019	0.014	0.012,0.014	0.020	0.003,0.021	0.010	0.021,0.015	−0.003	0.018,0.019	0.008	0.101,0.007	0.019
[ ]Si4O8	0.061	0.047,0.025	0.064	0.015,0.009	0.061	0.018,0.010	−0.011	0.012,0.001	0.038	0.025,0.024	0.013	0.042,0.026	0.001	0.007,0.003	−0.010
normalized An(d)	0.866	0.034,0.073	0.976	0.011,0.026	0.931	0.035,0.044	0.812	0.002,0.010	0.966	0.027,0.026	0.899	0.069,0.080	0.955	0.004,0.012	0.952

^(a):uncertainties represent the range of measured values;

^(b):normalized to 8 oxygens;

^(c):endmembers are calculated following the methods from Beaty and Albee (1980)

^(d):normalized to reported NaAlSi₃O₈ and CaAl₂Si₂O₈ endmembers;

^(e):value reported in Gilbert et al. (2003) with corrected SiO₂;

^(f):value reported in Jarosewich (1980); K₂O (not shown) is 0.07 wt. %

^*:averaged K₂O values for the An₇₈ standard are 0.08 wt. %; all other TiO₂, Cr₂O₃, MnO, and K₂O data are below detection limits.

2.3. Raman spectroscopy of chondrule plagioclase

Structural characteristics of chondrule plagioclase were determined by laser micro-Raman spectroscopy, using a Renishaw inVia spectrometer and a JASCO spectrometer at the National Institute of Polar Research (NIPR) in Japan. A microscope focused the excitation laser beam (532 nm) to a 2 micron spot size, and 200 to 1400 cm^–1 Raman spectra were acquired.

2.4. Transmission electron microscopy of chondrule plagioclase

Regions of chondrule plagioclase were investigated by transmission electron microscopy (TEM) at the Kochi Institute for Core Sample Research, JAMSTEC. Samples were processed into thin films by a Ga⁺-ion beam after deposition of carbon protection layers in a focused ion beam apparatus (FIB: Hitachi SMI4050). The films (~150 nm in thickness) were mounted onto Cu grids by carbon deposition and were positioned with a FIB-equipped micro-manipulator. The TEM instrument (JEOL JEM-ARM200F) operated at an accelerating voltage of 200 kV, and samples were characterized with bright-field transmission electron imaging and by high-angle annular dark field scanning transmission electron imaging (HAADF-STEM). Crystal structures were identified using selected area electron diffraction (SAED). Chemical compositions of the samples were obtained by EDS with a 100 mm² silicon-drift detector in STEM mode.

2.5. Secondary ion mass spectrometry

Among chondrule minerals, the three stable isotopes of Mg, as well as ²⁷Al/²⁴Mg ratios, were measured by SIMS, using the WiscSIMS Cameca IMS 1280. Analytical conditions are similar to those reported by Ushikubo et al. (2013). A primary O^– beam sputtered samples to detect secondary Mg⁺ and Al⁺ ions by applying −13kV and +10KV primary and secondary voltages, respectively. Secondary ions were aligned with a transfer lens magnification of 200×, and with a coupling lens operating in circular mode. The entrance slit width was 90 μm. As detailed below, other settings depended on appropriate conditions for the analysis of mafic minerals (olivine and pyroxene) versus plagioclase, as they have different secondary Mg⁺ ion intensities. Reduction of SIMS Al-Mg isotope data is similar to methods reported in Ushikubo et al. (2017).

2.5.1. Multi-Faraday cup analyses of olivine and low-Ca pyroxene

Olivine and pyroxene were analyzed in a single session. Analyses employed a ~10 μm spot size with a primary beam intensity of ~2.5 nA. Secondary ²⁴Mg⁺, ²⁵Mg⁺, ²⁶Mg⁺, and ²⁷Al⁺ ions were counted simultaneously with four Faraday cups, with an exit slit width of 500 μm. All Faraday cups employed 10¹¹ Ohm feedback resistors, and ²⁴Mg⁺ count rates for olivine and pyroxene ranged from 1.7 to 2.1 × 10⁸ cps. Each analysis duration was approximately 10 minutes. The mass resolving power at 10% peak height was ~2500. Contributions from ⁴⁸Ca²⁺ and ²⁴MgH⁺ to ²⁴Mg⁺ and ²⁵Mg⁺ peaks, respectively, were less than 0.01‰. Terrestrial reference ²⁵Mg/²⁴Mg and ²⁶Mg/²⁴Mg values (0.12663 and 0.13932, respectively; Catanzaro et al., 1966) were used to convert measured (raw) ratios to delta notation (δ²⁵Mg_m and δ²⁶Mg_m). For olivine, San Carlos (Fo₈₉) was used as the bracketing standard, and Fo₁₀₀ and Fo₆₀ standards were also measured; their respective δ²⁵Mg values on the DSM3 scale are –0.072‰ ± 0.089‰ (2SD), –0.366‰ ± 0.086‰, and –0.013‰ ± 0.062‰ (Kita et al., 2016). For low-Ca pyroxene, SP79–11 (En₉₇) was the bracketing standard and IG-Opx (En₈₉) was also measured; their δ²⁵Mg values on the DSM3 scale are –0.058‰ ± 0.055‰ (2SD) and –0.037‰ ± 0.079‰, respectively (Kita et al., 2016).

SIMS instrumental biases of olivine and pyroxene δ²⁵Mg were determined from measurements of standards listed above, and range from –3.6‰ to +0.5‰ (Tables EA2–1 and EA2–2). Repeated analyses of standards were reproducible within 0.1‰, and δ²⁵Mg biases increase with decreasing Fo and En endmembers, respectively (Figs. EA2–1 and EA2–2). For natural and instrumental mass dependent fractionations, we apply a power law function of 0.5128 from Davis et al. (2015). The mass independent term Δ²⁶Mg is defined as Δ²⁶Mg = [(1 + δ²⁶Mg/1000)/(1 + δ²⁵Mg/1000)^1/β ‒ 1] × 1000, where β = 0.5128. SIMS measured Δ²⁶Mg_m values of standards were consistently biased to slight positive values (~0.2‰; Table EA2–1; Table EA2–2), similar to those reported in Ushikubo et al. (2013; 2017). Δ²⁶Mg values of unknowns were corrected for this bias, based on bracketing standard analyses. Δ²⁶Mg uncertainties of unknown data are propagated, combining (1) the maximum of the bracketing standard 2SD versus the internal error of a single analysis (i.e. 2SE of 30 cycles per analysis), and (2) the bracketing standard 2SE. External reproducibilities (2SD) of Δ²⁶Mg_m values for San Carlos olivine and Sp79–11 pyroxene bracketing analyses were 0.08–0.12‰ and 0.06–0.10‰, respectively (n: four brackets consisting of 8 analyses per bracket; Tables EA3–1 and EA3–2). Raw ²⁷Al/²⁴Mg ratios of chondrule olivine and pyroxene were corrected by applying a SIMS relative sensitivity factor RSF= [(²⁷Al/²⁴Mg_m)/ [(²⁷Al/²⁴Mg_true)] of 0.87, which was based on repeated measurements of SP79–11 reference pyroxene (Table EA3–2). We apply a conservative 10% uncertainty to RSF-corrected chondrule olivine and pyroxene ²⁷Al/²⁴Mg values (which is approximately an order of magnitude larger than the 2SE uncertainty of raw ²⁷Al/²⁴Mg values; Tables EA3–1 and EA3–2), particularly because olivine may not have the same RSF as pyroxene; in absolute terms this uncertainty is small, as RSF-corrected chondrule olivine and pyroxene ²⁷Al/²⁴Mg values range from 0.0003 to 0.0742.

For analyses of olivine in chondrules, we observe that their SIMS ²⁴Mg yields (1) are approximately 5 to 10% higher than those of terrestrial standards with similar Fo contents (Fig. EA3–1a); and (2) are systematically higher than terrestrial standards as a function of δ²⁵Mg_m (Fig. EA3–1b). This suggests a matrix effect in chondrule olivine analyses that is not properly corrected by analyses of terrestrial olivine reference materials. In the case of chondrule pyroxenes, their SIMS ²⁴Mg yields are also slightly higher than those of terrestrial standards as a function of En and δ²⁵Mg_m (Figs. EA3–2a and EA3–2b). As such, we do not employ instrumental bias corrected δ²⁵Mg values for chondrule olivine and pyroxene. Instead, we make an assumption that their δ²⁵Mg values are the same as the mean value of terrestrial igneous rocks (–0.13‰ DSM-3 scale; Teng et al., 2010; Hu et al., 2016) and we assign a conservative uncertainty of ±2‰ (Tables EA3–1 and EA3–2); for reference, this uncertainty corresponds to a maximum added error of 0.0005‰ to the δ²⁶Mg* data from chondrule olivine and pyroxene. Under this assumption, the excess δ²⁶Mg* data from chondrule olivine and pyroxene are indistinguishable from their Δ²⁶Mg values because of the formula; δ²⁶Mg* = [(1+ δ²⁶Mg/1000) – (1 + δ²⁵Mg/1000)^1/β] × 1000 = Δ²⁶Mg × (1 + δ²⁵Mg/1000)^1/β.

After analyses, SIMS pits were investigated by SEM. Some show evidence of overlapping small cracks, pinholes, or slivers of an FeO-rich phase (Appendix EA4). However, EDS of the cracks and pinholes gave the same spectra as the olivine or low-Ca pyroxene targeted, suggesting no foreign phase(s) present. Per chondrule, isotope data from SIMS pits with cracks, pinholes or FeO-rich phases are indistinguishable relative to data from pits that appear clean; therefore, such data are not excluded from isochrons.

2.5.2. Plagioclase analyses

Mg-isotopes and ²⁷Al/²⁴Mg ratios of chondrule plagioclase were collected in three sessions. A primary beam intensity of 20–100 pA was employed, corresponding to spot sizes (5–10 microns in diameter) that could accommodate the small regions of chondrule plagioclase. As a result of the low primary beam current, it was necessary to count for ~3.5 hours per analysis for desired analytical precision. A 50 μm diameter mass aperture was used along with a 100 μm beam aperture, to project a Köhler-illuminated 5 μm diameter spot with sharp edges onto the sample surface (i.e. 0.1× magnification). However, spot analysis diameters gradually increased to 10 μm or more within 2–3 days, due to continuous sputter-enlargement of the mass aperture hole. In our final session, we used a custom mass aperture strip that was thicker (75 μm-thick molybdenum, versus the standard 25 μm-thickness), which decelerated sputter-enlargement of the mass aperture hole, keeping spot analysis diameters between 5 and 8 μm over a 7 day period. We note there were no systematic effects on the instrumental bias or RSF due to (1) variability in the primary beam current, and (2) sputter enlargement of the mass aperture hole. The mass resolving power was set to 4000, allowing for separating ²⁴Mg⁺ and ²⁵Mg⁺ peaks from ⁴⁸Ca²⁺ and ²⁴Mg¹H⁺ interferences, respectively. Secondary Mg⁺ ions were detected by an axial electron multiplier (EM) that operated by peak jumping. During measurement of ²⁵Mg⁺, secondary ²⁷Al⁺ ions were simultaneously collected by a Faraday cup located on the high-mass side of the axial EM. All analyses were corrected for a 23 ns EM dead time. Count rates for ²⁴Mg⁺ ranged from 2.0 × 10⁴ to 1.2 × 10⁵ cps. Count durations per cycle were as follows: for ²⁴Mg⁺ it was 3 seconds, and for ²⁵Mg⁺, ²⁶Mg⁺, and ²⁷Al⁺ the durations were each 10 seconds. Cycles per analysis were either 300 (first two sessions) or 400 (the third session).

Data reduction for plagioclase Al-Mg isotopes are similar to those for olivine and pyroxene described above. Instrumental bias was determined by measuring Mg isotope ratios and ²⁷Al/²⁴Mg from three synthetic anorthite glasses with 0.1, 0.6 and 1 wt.% MgO; each have a δ²⁵Mg of –1.77‰ (DSM3 scale; Kita et al., 2012). Based on measurements of the 1 wt. % MgO anorthite glass used as the running standard, the relative sensitivity factor for each session was constant, with a value of 1.04 (Table EA2–3), and external reproducibilities (2SD) of δ²⁵Mg_m and Δ²⁶Mg_m ranged from 0.77‰ to 1.09‰ and 0.70‰ to 0.99‰, respectively, similar to internal errors. The δ²⁵Mg_m and Δ²⁶Mg_m grand averages from running standard measurements were determined for each session; their uncertainties (2SE_std; standard error of the mean), ±0.19‰ to ±0.32‰, and ±0.13‰ to ±0.32‰, respectively, were used to correct unknown data (Table EA2–3). For unknowns, uncertainties are calculated as [(2SE_internal)² + (2SE_std)²]^1/2 for δ²⁵Mg and 2SE_internal for δ²⁶Mg*, where 2SE_internal is the internal error of an individual unknown analysis.

Plagioclase ²⁴Mg signal stability was monitored by plotting ²⁷Al/²⁴Mg versus cycle number (e.g. Fig. 3). For reference glasses, ²⁷Al/²⁴Mg changed systematically over the first 30 cycles and then stabilized for the remainder of cycles, which could reflect implantation of oxygen ions from the primary beam into the sample matrix. In most cases, such depth profile patterns among unknowns matched those produced in anorthite glass running standards (e.g. Figs. 3a–3c), suggesting (1) chondrule plagioclase was the only phase analyzed; and (2) Mg concentrations were homogeneous within the volume analyzed. However, three of the 60 plagioclase analyses, two from chondrule Q36 and one from chondrule M23, had cycles removed because the measurements inadvertently profiled into high-Ca pyroxene grains. This was easily distinguishable, because when such a grain became part of the spot analysis the ²⁷Al/²⁴Mg ratio dropped precipitously over the course of a few cycles; the presence of such grains was also verified by SEM imaging of the sputtered analysis area (e.g. Fig. 3d). We removed cycles with anomalously low ²⁷Al/²⁴Mg, as it is possible the increased ²⁴Mg signal from high-Ca pyroxene could cause inaccurate EM dead time corrections or lead to quasi-simultaneous arrival (QSA) effects that would compromise the accuracy of Mg isotope ratios. We note that a reduced number of cycles may cause changes to the ²⁷Al/²⁴Mg relative sensitivity factor, and we did not account for this because our correction scheme uses the grand averaged ²⁷Al/²⁴Mg values of all running standards for a given session. However, we evaluated this effect by reducing the cycle numbers of standards that bracketed shortened unknown analyses (to the same number of cycles as the shortened unknown analyses), and find that (1) changes to the bracketing standard analysis ²⁷Al/²⁴Mg RSF values are small (–0.002 to +0.012), and that therefore (2) changes to the grand averaged ²⁷Al/²⁴Mg RSF changes are minimal (max: 0.003). We also note that depth profile patterns from SIMS pits containing FeO-rich veins are not systematically different than those from SIMS pits without such veins; e.g., in Fig. 3c, the relatively flat ²⁷Al/²⁴Mg pattern after the first 100 cycles most likely reflects natural variability observed in chondrule M23 plagioclase.

Fig. 3.

Example ²⁷Al/²⁴Mg ratios versus cycle number for (a) the anorthite glass running standard with 1 wt. % MgO, and for (b-d) chondrule plagioclase. Increasing cycle numbers represent greater sputtering depth into samples. The shape of the ²⁷Al/²⁴Mg versus cycle number trend is likely related to the SIMS sputtering behavior of the sample, while differences in ²⁷Al/²⁴Mg ranges among samples represent variability in plagioclase compositions. BSE images of SIMS pits associated with (b-d) are also shown, where scale bars are 10 μm. In (b) and (c), chondrule ²⁷Al/²⁴Mg versus cycle number patterns approximate that of the anorthite running standard, suggesting the analyses were not compromised by inadvertent measurement of pyroxene grains. The image affiliated with (c) demonstrates that overlap with thin strands of an Fe-rich phase does not compromise data quality. In (d) the significant ²⁷Al/²⁴Mg decrease after the first 150 cycles indicates partial analysis of an unwanted phase. SEM imaging and EDS of the subsequent SIMS pit confirmed the presence of a small high-Ca pyroxene (HPx) grain. Therefore, only the first 150 cycles of this analysis were used to calculate the Al-Mg isotope systematics.

Following analytical sessions SIMS pits were evaluated by SEM and EDS. Most SIMS pits reveal only plagioclase was analyzed (e.g. image affiliated with Fig. 3b; Appendix EA4). In several cases SIMS pits show that they overlapped thin bands of an FeO-rich phase, but this did not diminish the quality of ²⁷Al/²⁴Mg monitored during cycles (e.g. Fig 3c; Appendix EA4). Therefore, such data were not excluded. In chondrules Q36 and M23, EDS confirmed small grains of high-Ca pyroxene were present in three SIMS pits from analyses with anomalously low ²⁷Al/²⁴Mg cycle values (e.g. image affiliated with Fig. 3d; Appendix EA4).

2.5.3. Chondrule isochrons and relative ages

²⁷Al/²⁴Mg (x-axis) versus δ²⁶Mg* (y-axis) values of chondrules were used to construct ²⁶Al-²⁶Mg isochrons with Isoplot 3.00 (Ludwig, 2003). All isochrons employed a model 1 fit. For each chondrule, (²⁶Al/²⁷Al)₀ values and their uncertainties (95% confidence) are inferred from the slope of the isochron regression. Specifically, (²⁶Al/²⁷Al)₀ = slope × 0.13932/1000, where 0.13932 is the terrestrial reference ²⁶Mg/²⁴Mg value (Catanzaro et al., 1966). Relative ages are calculated as:

$$\Delta t(\text{Myr) }=\text{ln }\left[{\left(\frac{{}^{\text{26}}\text{A}\text{l}}{{}^{\text{27}}\text{A}\text{l}}\right)}_{\text{0; CAI}}/{\left(\frac{{}^{\text{26}}\text{A}\text{l}}{{}^{\text{27}}\text{A}\text{l}}\right)}_{\text{0; chondrule}}\right]\times \frac{0.705}{\text{ln}(2)}$$ (1)

using the canonical CAI (²⁶Al/⁰⁰²⁷Al)₀ of 5.23 × 10^˗5 (Jacobsen et al., 2008; MacPherson et al., 2010), and where 0.705 is the half-life (in Myr) of ²⁶Al (Norris et al., 1983). Relative age uncertainties also employ eqn. 1, using slopes of isochron error envelopes that represent upper and lower age limits.

3. RESULTS

3.1. Plagioclase electron microprobe and Raman spectroscopy results

Chondrule plagioclase electron microprobe data are given in Table 1. Averaged anorthite contents range from 81 to 99. MgO concentrations range from 0.5 to 1.4 wt.%. Oxide totals range from 99.5 to 100.5 wt.%. However, cation totals, normalized to 8 oxygens, are generally low; ten of twelve chondrules have plagioclase with cation values between 4.92 and 4.97, and two chondrules have plagioclase with cation totals of 4.987 and 5.011. These results occur even though measurements of An₇₈ and An₉₅ standards within the same sessions have appropriate oxide totals and have stoichiometric values between 4.994 and 5.001 (Table 1). According to Beaty and Albee (1980), plagioclase EPMA data with element totals at/near 100%, but with too much Si, too little Al, and low amounts of singly and doubly charged cations (e.g. Ca, Na, K), indicate the presence of a [ ]Si₄O₈ endmember component, consisting of a vacancy and excess silica. The “excess silica” component comprises up to 8% of the endmember composition of QUE 99177 and MET 00426 plagioclase studied (Table 1).

Raman spectra from chondrule plagioclase show prominent peaks at 484 and 503 cm^˗1 (Appendix EA5). In many scans peaks at 281, 400, 427, 553, 681, 756, and 974 cm^˗1 are also present. These peaks are consistent with crystalline anorthite (Matson et al., 1986), and are not indicative of polymorphs such as dmisteinbergite. Chondrule Q38 plagioclase has an additional peak near 1020 cm^˗1, consistent with pyroxene (Appendix EA5).

3.2. SIMS Al-Mg isotope systematics

Chondrule Al-Mg isotope data are given in Table 2. Chondrule isochrons are shown in Fig. 4. ²⁷Al/²⁴Mg ratios of olivine and low-Ca pyroxene range from 0.0003 to 0.0742. Their δ²⁶Mg* values range from ‒0.07‰ to 0.13‰, with typical 2SE uncertainties of ~0.1‰. Olivine and low-Ca pyroxene data have no resolvable deficiencies in δ²⁶Mg* relative to terrestrial standard values. Based on these characteristics, olivine and pyroxene data represent origins of chondrule isochrons. The measured δ²⁶Mg* values of chondrule plagioclase range from ‒1.3‰ to 2.0‰, with typical 2SE uncertainties of ~1‰. Their ²⁷Al/²⁴Mg ratios are between 24.8 and 62.2 (Table 2), consistent with CR type I chondrule plagioclase in previous studies, but lower than Al-rich and type II chondrules, with values that can exceed 250 (Hutcheon et al., 2009; Nagashima et al., 2014; Schrader et al., 2017). Per chondrule, measured plagioclase ²⁷Al/²⁴Mg ratios do not significantly vary; in chondrules Q7 and M3 the ranges are ~9 and ~13 units, respectively, and in the other chondrules the ranges are 8 units or less (Table 2). When available, plagioclase Al/Mg ratios by EPMA and SIMS at the same locations generally match (Fig. 5). Two data points with the highest Al/Mg ratios in Fig 5, both from chondrule M3, have higher values by EPMA when compared to those by SIMS. This could reflect variability in the Al/Mg of chondrule M3 plagioclase that was sampled by the EPMA interaction volume, but was not sampled by the shallower depth of SIMS analyses.

Table 2.

Al-Mg istotope systematics of QUE 99177,49 and MET 00426,46 chondrules.

chondrule	spot/phase	27Al/24Mg (2SE)	*δ26Mg (2SE)
Q7.	1. Pl	44.11 ± 0.31	0.92 ± 0.91
Mg#: 99.0 (+0.2,−0.1)(a)	2. Pl	44.05 ± 0.31	1.22 ± 0.87
Δ17O: −5.2 (± 0.5)‰(b)	3. Pl	46.11 ± 0.33	1.15 ± 0.94
isochron MSWD: 0.2	4. Pl	37.80 ± 0.27	1.57 ± 1.24
*δ26Mg0: 0.04 ± 0.04	5. Pl	39.37 ± 0.27	1.06 ± 1.07
(26Al/27Al)0: (3.5 ± 1.3) × 10−6	6. Pl	40.61 ± 0.28	0.87 ± 0.89
Myr after CAIs: 2.8 (−0.3/+0.5)	7. LPx	0.0230 ± 0.00230(c)	0.02 ± 0.07
	8. LPx	0.0298 ± 0.00298	0.05 ± 0.07
	9. Ol	0.0004 ± 0.00004	0.03 ± 0.08
	10. Ol	0.0003 ± 0.00003	0.05 ± 0.08
Q13.	1. Pl	37.44 ±0.14	0.78 ± 1.02
Mg#: 98.4 (+0.2,−0.2)	2. Pl	37.81 ±0.20	0.28 ± 0.93
Δ17O: −1.9 (± 0.6)‰	3. Pl	34.84 ±0.13	0.46 ± 0.84
isochron MSWD: 1.6	4. Pl	31.78 ±0.15	−0.34 ± 1.11
*δ26Mg0: 0.01 ± 0.04	5. Ol	0.0017 ±0.00017	−0.02 ± 0.08
(26Al/27Al)0: (1.3 ± 1.7) × 10−6	6. Ol	0.0027 ±0.00027	−0.07 ± 0.08
Myr after CAIs: >2.9	7. LPx	0.0281 ±0.00281	0.05 ± 0.07
	8. LPx	0.0341 ±0.00341	0.06 ± 0.07
Q15.	1. Pl	35.77 ±0.25	0.04 ± 0.96
Mg#: 98.6 (+0.1,−0.3)	2. Pl	34.48 ±0.14	−0.29 ± 0.93
Δ17O: −2.8 (± 0.5)‰	3. Pl	38.15 ±0.14	0.83 ± 1.03
isochron MSWD: 1.0	4. Pl	36.59 ±0.26	−0.20 ± 0.96
*δ26Mg0: 0.00 ± 0.04	5. Pl	34.36 ±0.24	−0.89 ± 1.05
(26Al/27Al)0: (−0.3 ± 1.7) × 10−6	6. LPx	0.0149 ±0.00149	0.02 ± 0.07
Myr after CAIs: >3.7	7. LPx	0.0276 ±0.00276	−0.01 ± 0.06
	8. Ol	0.0007 ±0.00007	−0.02 ± 0.08
	9. Ol	0.0017 ±0.00017	−0.02 ± 0.08
Q16.	1. Pl	39.47 ±0.27	1.38 ± 1.01
Mg#: 99.0 (+0.2,−0.3)	2. Pl	37.96 ±0.26	1.97 ± 1.16
A17O: −5.0 (± 0.6)‰	3. Pl	36.16 ±0.25	1.28 ± 1.20
isochron MSWD: 0.4	4. Pl	37.45 ±0.26	1.43 ± 0.87
*526Mg0: 0.04 ± 0.01	5. Pl	41.72 ±0.29	1.21 ± 1.05
(26Al/27Al)0: (5.2 ± 1.7) × 10−6	6. Ol	0.0003 ±0.00003	0.00 ± 0.08
Myr after CAIs: 2.4 (−0.3/+0.4)	7. Ol	0.0003 ±0.00003	−0.02 ± 0.08
	8. LPx	0.0286 ±0.00286	0.04 ± 0.06
	9. LPx	0.0166 ±0.00166	0.00 ± 0.06
Q19.	1. Pl	30.13 ±0.31	0.54 ± 1.04
Mg#: 98.7 (+0.7,−0.8)	2. Pl	36.57 ±0.14	0.63 ± 1.16
Δ17O: −4.9 (± 0.9)‰	3. Pl	37.70 ±0.14	0.15 ± 1.17
isochron MSWD: 0.3	4. Pl	33.71 ±0.17	0.37 ± 1.15
*δ26Mg0: 0.00 ± 0.04	5. Pl	34.93 ±0.27	−0.16 ± 1.32
(26Al/27Al)0: (1.2 ± 2.0) × 10−6	6. Pl	35.98 ±0.19	0.03 ± 1.24
Myr after CAIs: >2.9	7. Ol	0.0014 ±0.00014	−0.04 ± 0.13
	8. Ol	0.0014 ±0.00014	−0.06 ± 0.13
	9. LPx	0.0175 ±0.00175	0.01 ± 0.06
	10. LPx	0.0234 ±0.00234	0.00 ± 0.07
Q35.	1. Pl	24.82 ±0.18	−0.37 ± 1.25
Mg#: 98.3 (+0.2,−0.3)	2. Pl	28.11 ±0.23	−0.71 ± 1.28
Δ17O: −2.1 (± 0.8‰	3. Pl	26.57 ±0.19	0.76 ± 1.51
isochron MSWD: 1.0	4. Pl	25.35 ±0.11	0.61 ± 1.11
*δ26Mg0: −0.04 ± 0.05	5. Pl	26.07 ±0.09	−0.58 ± 0.72
(26Al/27Al)0: (−1.0 ± 2.5) × 10−6	6. Ol	0.0003 ±0.00003	−0.06 ± 0.13
Myr after CAIs: >3.6	7. Ol	0.0003 ±0.00003	−0.01 ± 0.13
	8. LPx	0.0742 ±0.00742	−0.04 ± 0.06
Q36.	1. Pl(d)	33.95 ±0.33	−0.52 ± 1.33
Mg#: 97.4 (+0.7,−0.9)	2. Pl	30.31 ±0.22	−0.32 ± 1.78
Δ17O: −1.7 (± 0.3)‰	3. Pl(d)	32.89 ±0.12	−0.01 ± 1.86
isochron MSWD: 0.2	4. Pl	32.45 ±0.23	−0.88 ± 1.62
*δ26Mg0: −0.01 ± 0.04	5. Ol	0.0014 ±0.00014	0.02 ± 0.13
(26Al/27Al)0: (−2.0 ± 3.3) × 10−6	6. LPx	0.0238 ±0.00238	−0.02 ± 0.06
Myr after CAIs: >3.7	7. LPx	0.0139 ±0.00139	0.01 ± 0.06
	8. Ol	0.0023 ±0.00023	−0.03 ± 0.13
Q38.	1. Pl	28.98 ±0.11	1.62 ± 1.11
Mg#: 99.0 (+0.1,−0.2)	2. Pl	31.86 ±0.12	1.40 ± 1.30
Δ17O: −5.3 (± 0.8)‰	3. Pl	30.94 ±0.15	1.88 ± 0.97
isochron MSWD: 2.1	4. Pl	27.36 ±0.10	1.15 ± 1.12
*δ26Mg0: −0.03 ± 0.07	5. Pl	31.08 ±0.13	0.26 ± 1.04
(26Al/27Al)0: (6.0 ± 3.9) × 10−6	6. Ol	0.0006 ±0.00006	−0.01 ± 0.12
Myr after CAIs: 2.2 (−0.5/+1.1)	7. Ol	0.0004 ±0.00004	0.13 ± 0.12
	8. LPx	0.0189 ±0.00189	−0.05 ± 0.07
	9. LPx	0.0317 ±0.00317	−0.06 ± 0.06
Q44.	1. Pl	37.51 ±0.26	−0.27 ± 1.43
Mg#: 98.3 (+0.2,−0.2)	2. Pl	31.65 ±0.12	−0.18 ± 0.82
Δ17O: −2.4 (± 0.5)%	3. Pl	35.69 ±0.15	0.48 ± 1.35
isochron MSWD: 0.7	4. Pl	32.25 ±0.23	0.02 ± 1.54
*δ26Mg0: 0.02 ± 0.04	5. Pl	33.69 ±0.24	−1.04 ± 1.36
(26Al/27Al)0: (−0.9 ± 2.2) × 10−6	6. LPx	0.0230 ±0.00230	0.00 ± 0.06
Myr after CAIs: >3.7	7. LPx	0.0277 ±0.00277	0.05 ± 0.06
	8. LPx	0.0239 ±0.00239	0.01 ± 0.06
M3.	1. Pl	49.10 ±0.25	−0.08 ± 1.01
Mg#: 94.2 (+1.0,−0.7)	2. Pl	51.11 ±0.25	0.74 ± 1.02
Δ17O: −0.9 (± 0.4)‰	3. Pl	56.20 ±0.44	−1.30 ± 1.43
isochron MSWD: 1.2	4. Pl	53.74 ±0.36	0.12 ± 1.25
*δ26Mg0: 0.04 ± 0.06	5. Pl	62.24 ±0.32	0.68 ± 1.42
(26Al/27Al)0: (0.2 ± 1.4) × 10−6	6. LPx	0.0466 ±0.00466	−0.01 ± 0.10
Myr after CAIs: >3.6	7. LPx	0.0287 ±0.00287	0.05 ± 0.10
	8. LPx	0.0439 ±0.00439	0.04 ± 0.10
M6.	1. Pl	38.90 ±0.21	0.69 ± 1.15
Mg#: 99.0 (+0.1,−0.1)	2. Pl	37.16 ±0.21	1.03 ± 1.07
Δ17O: −4.2 (± 0.6)‰	3. Pl	37.65 ±0.20	1.71 ± 1.05
isochron MSWD: 0.5	4. Pl	37.59 ±0.19	1.42 ± 0.94
*δ26Mg0: 0.04 ± 0.04	5. Pl	39.77 ±0.21	0.65 ± 1.08
(26Al/27Al)0: (4.0 ± 1.7) × 10−6	6. Ol	0.0007 ±0.00007	0.05 ± 0.08
Myr after CAIs: 2.6 (−0.4/+0.6)	7. Ol	0.0013 ±0.00013	0.04 ± 0.08
	8. LPx	0.0163 ±0.00163	0.04 ± 0.08
	9. LPx	0.0265 ±0.00265	0.03 ± 0.07
M23.	1. Pl	30.76 ±0.16	0.95 ± 0.90
Mg#: 99.2 (+0.1,−0.1)	2. Pl	28.89 ±0.14	1.07 ± 0.76
Δ17O: −4.7 (± 0.4)%	3. Pl	29.89 ±0.16	1.88 ± 0.91
isochron MSWD: 0.7	4. Pl	30.54 ±0.34	1.36 ± 0.85
*δ26Mg0: 0.00 ± 0.08	5. Pl(d)	31.54 ±0.20	1.23 ± 1.86
(26Al/27Al)0: (6.0 ± 2.0) × 10−6	6. Ol	0.0004 ±0.00004	−0.01 ± 0.08
Myr after CAIs: 2.2 (−0.3/+0.4)

^(a)bracketed values represent the range of measured Mg#s, as reported in Tenner et al. (2015).

^(b)bracketed values are the 2SD of multiPle measurements per chondrule used to define the averaged “host” chondrule Δ¹⁷O, as reported in Tenner et al. (2015)

^(c)27Al/²⁴Mg uncertainties are estimated as 10% of the reported value.

See main text for details

^(d)analyses inadvertantly profiled into high-Ca pyroxene grains. As a result, fewer Plagioclase cycles were used to determine ²⁷Al/²⁴Mg and *δ²⁶Mg (215, 151, and 96 cycles for Q36 #’s 1 & 3, and M23 #5, respectively).

Fig. 4.

Al-Mg isochron diagrams from each chondrule studied. Error envelopes are 95% confidence. Inferred initial (²⁶Al/²⁷Al)₀ ratios are determined from slopes of regression lines. Mineral data are from Table 2.

Fig 5.

Plagioclase Al/Mg comparisons of regions analyzed by EPMA and by SIMS. Al/Mg ratios by SIMS are calculated by multiplying the measured ²⁷Al/²⁴Mg by the fraction of Mg measured as ²⁴Mg. EPMA uncertainties represent the maxima of (1) the range of values per region analyzed (1 to 5 EPMA measurements were collected per spot analyzed by SIMS); or (2) the uncertainties of Al and Mg from measurements of plagioclase standards as unknowns (Table 1). SIMS uncertainties are the ²⁷Al/²⁴Mg values given in Table 2, and are smaller than the symbols.

Chondrule mineral isochrons have mean square weighted deviations (MSWD) of 0.2 to 2.1, below the threshold value of 2.5 for rejecting straight-line isochrons, according to Brooks et al. (1972). Five of twelve chondrules, Q7, Q16, Q38, M6, and M23, have resolvable excess ²⁶Mg, with (²⁶Al/²⁷Al)₀ values of (3.5 ± 1.3) × 10^˗6 to (6.0 ± 3.9) × 10^˗6 (Fig. 4; Table 2). Including uncertainties, this corresponds to their formation 1.7 to 3.3 Myr after CAIs (eqn. 1). These chondrules have Mg#’s ranging from 99.0 to 99.2, and have a narrow range of Δ¹⁷O, from –4.2‰ to –5.3‰ (Table 2). The other seven chondrules, Q13, Q15, Q19, Q35, Q36, Q44, and M3, lack resolvable excess ²⁶Mg. Using maxima of isochron uncertainties, their (²⁶Al/²⁷Al)₀ upper limits range from 1.3 × 10^˗6 to 3.2 × 10^˗6, corresponding to their formation >2.9 to >3.7 Myr after CAIs (eqn. 1). These chondrules have lower Mg#’s (94.2 to 98.7) and generally higher Δ¹⁷O values (–0.9 to –4.9‰), relative to chondrules with excess ²⁶Mg (Table 2). The total range of (²⁶Al/²⁷Al)₀ generally agrees with prior CR chondrite chondrule data (e.g. Fig. 1b).

3.3. Plagioclase TEM results

After SIMS, two chondrules, one with resolvable excess ²⁶Mg, M6, and one with no resolvable excess ²⁶Mg, Q44, were selected for analysis by FIB sectioning and TEM (Figs. 6 and 7, respectively). Each chondrule has plagioclase with a resolvable [ ]Si₄O₈ endmember component (Table 1), meaning TEM analysis provides information regarding the nature of excess Si in chondrule plagioclase. For comparison to a sample with minimum thermal/aqueous metamorphism, a third FIB section was made from Acfer 094 chondrule plagioclase (chondrule G15, as labeled in Ushikubo et al., 2012), and was analyzed by TEM (Appendix EA6). This plagioclase also has a resolvable [ ]Si₄O₈ endmember component (Appendix EA1), and has resolvable excess ²⁶Mg (Ushikubo et al. 2013).

Fig. 6.

Electron micrographs of plagioclase from chondrule M6, which has resolvable excess ²⁶Mg (Table 2; Fig. 4). (a) and (b) SEM images of plagioclase sectioned by FIB, where scale bars are 50 μm. (c) Bright-field TEM image of the plagioclase section. The sample appears as a single crystal, largely free of inclusions and defects, except for a small number of dislocations in the bottom right corner. The arrow corresponds to the surface of the meteorite thin section; about 1/3 of the FIB section was lost due to a pre-existing crack. (d) HAADF-STEM (Z-contrast image), showing the sample is chemically homogeneous (elemental EDS maps also show homogeneity in Electronic Appendix EA6). (e) Selected area electron diffraction (SAED) pattern consistent with only an unmodified anorthite structure and showing no superlattice reflections. (f) High resolution TEM (HRTEM) image, showing no inclusions are present, even down to the nm-scale.

Fig. 7.

Electron micrographs of plagioclase from chondrule Q44, which has no resolvable excess ²⁶Mg (Table 2; Fig. 4). (a) and (b) SEM images of plagioclase sectioned by FIB, where scale bars are 50 μm. SIMS pits from plagioclase Al-Mg isotope analyses, as well as a high-Ca pyroxene O-isotope analysis, are visible in (b). (c) thin-foil TEM image of the plagioclase section, consisting of a plagioclase grain and a low-Ca pyroxene grain, separated by a dotted line. The plagioclase is divided into two domains (dashed line; white arrow) by a subgrain boundary consisting of dislocation arrays. Color bands in the image correspond to thickness fringes. The thin gray layer running parallel to the thin section surface is a carbon deposit. Circled regions correspond to free dislocations. (d) HAADF-STEM Z-contrast image. Circled regions depict sub-micron-sized inclusions appearing as tiny bright dots. With the exception of these inclusions, the plagioclase is chemically homogeneous (elemental EDS maps are provided in Electronic Appendix EA6). (e) SAED pattern of plagioclase, showing a diffraction pattern consistent with only an unmodified anorthite structure and showing no superlattice reflections. (f) Bright-field TEM image of a dislocation region in plagioclase domain 1 and associated Fe,Cr,Si-rich inclusions as determined by EDS. (g) HRTEM image of a region analyzed in plagioclase domain 2, which does not have dislocations or inclusions down to the nm-scale.

The FIB section of M6 plagioclase (Figs. 6a & 6b) is a single crystal, based on thin-foil TEM imaging (Fig. 6c). HAADF-STEM Z-contrast imaging (e.g. Fig. 6d) shows chemical homogeneity (as does EDS mapping; Appendix EA6). SAED reveals a pattern consisting only of an unmodified anorthite structure, with no superlattice reflections (Fig, 6e). High-resolution TEM shows no inclusions down to the nanometer-scale (Fig 6f). These features are consistent with those from the Acfer 094 chondrule plagioclase FIB section (Appendix EA6).

The Q44 FIB section (Figs. 7a & 7b) consists of plagioclase and low-Ca pyroxene, based on TEM imaging (Fig. 7c) and EDS analysis. The plagioclase has two domains divided by a subgrain boundary consisting of dislocation arrays (e.g. dashed line with arrow in Fig 7c). Plagioclase domain 1 in Fig. 7c has a small number of free dislocations (circled regions in Fig. 7c), and a weathering vein was identified (Appendix EA6). Plagioclase domain 2 has no defect structures (Fig. 7c). HAADF-STEM Z-contrast imaging shows both plagioclase domains are chemically homogeneous, although domain 1 has a number of sub-micron sized inclusions, appearing as tiny bright dots (circled areas in Fig. 7d). For both plagioclase domains, SAED patterns consist only an of unmodified anorthite structure, with no superlattice reflections (e.g. Fig. 7e). High-resolution TEM imaging shows dislocation-free regions of plagioclase are devoid of inclusions (e.g. Figs. 7f & 7g) down to the nanometer scale. In plagioclase domain 1, 10-to-20 nm-sized inclusions are associated with dislocations (e.g. Fig. 7f). EDS analyses of these inclusions indicate a phase enriched in Fe, Cr, and Si. Due to their small size, EDS spectra of the inclusions partially sample host anorthite (Appendix EA6), making it difficult to properly identify the phase. Excluding the sparse dislocations and inclusions, the near-entirety of plagioclase in the Q44 FIB section has features consistent with M6 (e.g. Fig. 6) and Acfer 094 chondrule plagioclase (e.g. Appendix EA6).

4. DISCUSSION

4.1. Assessing the reliability of plagioclase used for Al-Mg isotope systematics.

The most important consideration of this work involves a critical evaluation of the mineralogical and chemical characteristics of plagioclase used for Al-Mg isochrons, in order to ensure the chronometer is valid. For CR chondrite chondrules this is especially important because the majority of their constituent plagioclase has no resolvable excess ²⁶Mg (Fig 1b; Fig. 4). Therefore, it is necessary to confirm that unresolved plagioclase isochrons do not reflect later secondary processing, but are instead true signatures of later chondrule formation. In this regard, we find that QUE 99177 and MET 00426 chondrule plagioclase show no evidence of secondary processing, and phases other than plagioclase were largely avoided during their analysis. Here, we detail these findings.

4.1.1. Analyzed regions of chondrule plagioclase are free of inclusions and secondary alteration.

Two lines of evidence indicate foreign mineral inclusions were avoided during SIMS analysis of chondrule plagioclase. First, SIMS depth profile patterns of ²⁷Al/²⁴Mg are similar to those collected on the anorthite glass standard (e.g. Figs. 3a–3c). EDS revealed micron-scale high-Ca pyroxene grains in only 3 of 60 plagioclase SIMS pits (e.g. Fig. 3d), and isotope signals from related cycles were eliminated from Al-Mg isotope data. Second, TEM analysis indicates the plagioclase has few-to-no inclusions at the sub-micron level (Figs. 6 and 7). Although plagioclase in the Q44 FIB section has 10–20 nm-sized Fe,Cr,Si rich inclusions (e.g. Fig. 7f), their abundance is low and the inclusions lack Al and Mg, meaning they likely have a negligible influence on plagioclase Al-Mg isotope systematics. Additionally, the plagioclase FIB/TEM samples show no evidence of replacement by other mineral phases during aqueous/thermal metamorphism, down to the nanometer scale. Several plagioclase SIMS pits show evidence of partially sampling weathering veins of an FeO-rich phase (e.g. Appendix EA4: Q19: #1, #3; Q44: #3; M3: #3, #4; M6: #2, #3; M23: #2-#4). However, these data show no systematic differences in ²⁷Al/²⁴Mg and *δ²⁶Mg, relative to plagioclase analyses that did not sample FeO-rich veins (Table 2), suggesting the influence of FeO-rich weathering veins on data quality is minimal. Support for this interpretation comes from the observation that at least some veins are < 100 nm in width (e.g. Appendix EA6), meaning they likely represent a minimal volume of material sputtered during a SIMS analysis, when present.

4.1.2. Chondrule intermineral O-isotope comparisons suggest plagioclase was not disturbed.

Seven of twelve chondrules we investigated have SIMS O-isotope data of coexisting plagioclase, olivine, and pyroxene (from Tenner et al., 2015; Fig. 8). For each chondrule, plagioclase data are within 2‰ of coexisting olivine and pyroxene data on a three-oxygen isotope plot, and often the data overlap. This suggests that, for a given chondrule: (1) plagioclase, pyroxene, and olivine crystallized from a melt that remained nearly-constant in its O-isotope ratio; and (2) the plagioclase O-isotope ratio is unaltered. This second point is important because experimental data extrapolations predict 10 μm² of oxygen diffusion in anorthite occurs per 1 Myr at sustained temperatures of ~500 °C and ~150 °C during dry and hydrous thermal metamorphism, respectively (Fig. 9). Therefore, the per-chondrule O-isotope agreement between plagioclase and coexisting olivine and pyroxene among MET 00426 and QUE 99177 chondrules (e.g. Fig. 8) suggests the plagioclase experienced secondary heating extents of less than 150 °C on the presumably hydrated CR chondrite parent body (e.g. Schrader et al., 2014b). Although the susceptibility of oxygen and Mg isotope systems to disturbance are not related, at temperatures this low, Mg diffusion could also be negligible; however, experimental data for anorthite Mg diffusion in a wet system are currently lacking. In a dry system, Mg and oxygen diffusion rates in anorthite are similar for extrapolations below 1000 °C (Fig. 9), and so it is reasonable to speculate that if oxygen isotopes of chondrule plagioclase appear unaltered in a wet system, then Mg isotopes could also be unaltered. Further support for this hypothesis comes from Kaba (CV3.1) chondrites, as they experienced hydrothermal alteration at ~300 °C, and plagioclase is systematically ¹⁶O-depleted relative to coexisting phases in chondrules and CAIs (Krot et al., 2019). Yet, Al-Mg isotope systematics of Kaba chondrules appear undisturbed (Nagashima et al., 2017).

Fig. 8.

Inter-mineral oxygen isotope ratios of chondrules from this study that have plagioclase data for comparison. Oxygen isotope data are from Tenner et al. (2015). Low-, mid-, and high-Ca pyroxene data have wollastonite endmember contents of 0.8–3, 7–9, and 33–37, respectively. Uncertainties are the 2SD external reproducibility of bracketing standard measurements. Shown are the primitive chondrule mineral (PCM; Ushikubo et al. 2012), carbonaceous chondrite anhydrous mineral (CCAM; Clayton et al. 1977), Young and Russell (1998), and terrestrial fractionation lines.

Fig. 9.

Oxygen and Mg diffusion in anorthite. Solid lines are fits to experimental data, and dashed lines are extrapolations to sustained temperatures required for 10 μm² of diffusion per 1 million years. Dry oxygen diffusion data: Ryerson and McKeegan (1994). Wet oxygen diffusion data: Giletti et al. (1978). Dry Mg diffusion data: LaTourrette and Wasserburg (1998) and Orman et al. (2014).

4.1.3. The importance of excess Si in plagioclase and other compositional relationships for evaluating the Al-Mg isotope systematics of chondrules.

Ten of the twelve chondrules investigated have plagioclase with a resolvable excess silica ([ ]Si₄O₈) component (Table 1). As data from An₇₈ and An₉₅ standards by EPMA are consistent with their reference compositions (Table 1), the measured excess silica in chondrule plagioclase is unlikely due to analytical error. Excess silica is common in anorthite-rich plagioclase from lunar basalts (Dymek et al., 1975; Longhi, 1976; Beaty et al., 1979; Baldridge et al., 1979; Appendix EA1), and was recently found in Pasamonte eucrite plagioclase (Mittlefehldt et al., 2017). According to 1 atm. experiments along the join CaAl₂Si₂O₈˗SiO₂ (e.g. Longhi and Hays, 1979), excess silica is incorporated at high temperatures (~1100 to 1500 °C) and comprises up to several percent of the plagioclase endmember composition in a silica saturated system.

In addition to excess silica, we observe high anorthite contents (0.793 to 0.999, when normalized to anorthite, albite, and K-feldspar components), and appreciable MgO concentrations of 0.5 to 1.4 wt.%, within MET 00426 and QUE 99177 chondrule plagioclase (Table 1). Mg incorporation is most likely achieved by substitution involving the Ca(Fe,Mg)Si₃O₈ endmember (e.g. Longhi et al., 1976; Table 1). Collectively, similar plagioclase characteristics are reported among other FeO-poor CR chondrite chondrules (e.g. Nagashima et al., 2014; Schrader et al., 2017; Appendix EA1), and there are no systematic differences in excess silica, anorthite content or MgO, as a function of excess ²⁶Mg abundance (Figs. 10a and 10b).

Fig. 10.

Plagioclase excess silica (left panels) and MgO (right panels) as a function of anorthite content from various carbonaceous and ordinary chondrite FeO-poor chondrules. For reference, the reported anorthite content is normalized only to the An + Ab + K-feldspar contents. All EPMA data are compiled in Appendix EA1, and plagioclase components are calculated according to methods used in Beaty and Albee (1980). Data sources: CR2 data: Nagashima et al. (2014); Schrader et al. (2017); this study. Acfer 094 data: Appendix EA1, from chondrules studied by Ushikubo et al. (2012). Y-81020 CO3.05 data: Kurahashi et al. (2008); Tenner et al. (2013). L/LL3.01–3.05 data: Hutcheon and Hutchison (1989); Kita et al. (2000); Rudraswami and Goswami (2007). CV3.1–3.6 data: Kimura and Ikeda (1995; 1998); Krot et al. (2002); Jones and Schilk (2009); Nagashima et al. (2017). CO3.2–3.7 data: Tomeoka and Itoh (2004); Wick and Jones (2012). L and LL3.1 data: Mostefaoui et al. (2002); Rudraswami and Goswami (2007). When available, uncertainties represent the range of values from EPMA datasets (Appendix EA1).

When comparing FeO-poor chondrule plagioclase from petrologic type <3.1 chondrites (less thermally metamorphosed) to those from petrologic type ≥3.1 chondrites (more thermally metamorphosed), there are distinct compositional differences. For example, Acfer 094 (ungr. C3.00) and L/LL 3.01–3.05 chondrites have FeO-poor chondrule plagioclase with similar ranges of excess silica, anorthite, and MgO as CR chondrite chondrules (Figs. 10c–10f). FeO-poor chondrule plagioclase from Yamato 81020 (CO3.05) has a slightly lower abundance of excess silica, on average (Fig. 10e), but otherwise has a similar range of MgO and anorthite content as that among CR chondrite chondrules (Fig. 10f). Thus, the combination of excess silica, high MgO and high anorthite content in FeO-poor chondrule plagioclase appears to be characteristic of petrologic type 3.00–3.05 chondrites, and of CR2 chondrites with minimal hydrothermal alteration. In contrast, FeO-poor chondrule plagioclase from type ≥3.1 chondrites, including those from CO, CV, and LL chondrites (Appendix EA1), have reduced proportions of excess silica, anorthite component, and MgO. In particular, data from 31 of 45 of such chondrules are silica deficient ([ ]Si₄O₈ < 0), and the extent of deficiency becomes more pronounced as anorthite components decrease to values of ~0.7 (Fig. 10g). Further, plagioclase MgO concentrations from type ≥3.1 FeO-poor chondrite chondrules are low compared to those from type <3.1 chondrites, mainly ranging from 0 to 0.8 wt.%. (Fig. 10h). This feature could represent diminished Mg solubility in plagioclase during thermal metamorphism. Specifically, plagioclase Mg solubility is enhanced with increasing temperature (see Fig. 2 from Van Orman et al., 2014 and associated discussion), meaning it would have been at a maximum upon crystallization from the final chondrule melt (1150‒1200 °C; Fig. 1 from Ustunisik et al., 2014). During thermal metamorphism of type ≥3.1 chondrites (e.g. 300‒700 °C; Miyamoto, 1991; Huss and Lewis, 1994; Brearley, 1997) (1) plagioclase Mg solubility would have been reduced; and (2) plagioclase Mg diffusion would have been relatively fast (e.g. Fig. 9), meaning chondrule plagioclase may have lost a portion of Mg sequestered during primary crystallization.

Among type ≥3.1 chondrites, some regions of FeO-poor chondrule plagioclase are replaced by nepheline (NaAlSiO₄) (e.g. Fig. 1 from Krot et al. 2002; Fig. 4 from Tomeoka and Itoh 2004; Fig. 2 from Wick and Jones 2012; Fig. 3d from Nagashima et al., 2017). As nepheline is silica-poor relative to albite (NaAlSi₃O₈), we speculate that if EPMA measurements co-sampled plagioclase and nepheline, it could explain a trend of decreasing silica with apparently increasing albite component (i.e. decreasing anorthite component in Fig. 10g) among FeO-poor type ≥3.1 chondrite chondrules. The replacement of chondrule plagioclase by nepheline is commonly interpreted as a reaction product during thermal metamorphism (e.g. Kimura and Ikeda, 1995; 1997; 1998; Ichimura et al., 2017), and could explain disturbed Al-Mg isotope systematics among type ≥3.1 chondrite chondrules with this feature (e.g. Sano et al., 2014; Nagashima et al., 2017). Finally, we note that the extent of thermal metamorphism is often variable throughout a chondrite sample, meaning some FeO-poor chondrules with unaltered plagioclase may exist in type ≥3.1 chondrites. Importantly, electron microprobe analysis could identify such plagioclase based on the presence of excess silica, high MgO, and high anorthite content.

4.2. Chondrule formation in distinct environments I: Timing differences.

A key finding of this study is that chondrules with and without excess ²⁶Mg are distinguishable by Mg# and Δ¹⁷O. Those with excess ²⁶Mg have high Mg#’s (99.0 to 99.2), and a limited Δ¹⁷O range (–4.2‰ to –5.3‰), while chondrules without excess ²⁶Mg have lower Mg#’s (94.2 to 98.7) and generally higher Δ¹⁷O (–0.9 to –4.9‰) (Figs. 11a & 11b). This indicates two distinct physicochemical environments within the CR chondrite accretion region, and we can evaluate the chondrule-formation timing differences between these environments. If considering uncertainties, chondrule M23 has the oldest resolved age (2.2 ‒0.3/+0.4 Myr after CAIs; Table 2). If we compare the youngest possible formation age of chondrule M23 (2.6 Myr after CAIs) to the upper age limits of all chondrules without excess ²⁶Mg (chondrules Q13, Q15, Q19, Q35, Q36, Q44, M3; Table 2), we determine resolved chondrule-formation timing differences of at least 0.3 to 1.1 Myr between the two environments. If we compare the oldest age (including uncertainties) from all chondrules with excess ²⁶Mg (1.7 Myr after CAIs; chondrule Q38; Table 2) to the youngest upper age limit from all chondrules without excess ²⁶Mg (>3.7 Myr after CAIs; chondrules Q15, Q36, and Q44; Table 2), chondrule-formation timing differences of more than 2.0 Myr may have existed between the two environments.

Fig. 11.

(a) Chondrule Mg# versus (²⁶Al/²⁷Al)₀. Relative ages are calculated according to Eqn. 1 in the text, using a canonical (²⁶Al/²⁷Al)₀ of 5.23 × 10⁻⁵. In order to provide a direct comparison, chondrule Mg#’s and uncertainties from Nagashima et al. (2014) (Appendix EA7) and Schrader et al. (2017) are recalculated using definitions from Tenner et al. (2015), as detailed in 4.3.1 of the text. Oxygen fugacities (relative to the Fe-wüstite buffer) during chondrule formation are based on metal-silicate phase equilibria (Eqn. 2 in the text). (b) Corresponding chondrule Mg# versus Δ¹⁷O data from this study and from Schrader et al. (2017). Small cross symbols are additional data from Tenner et al. (2015) that have no current Al-Mg isotope systematics. Δ¹⁷O data from Tenner et al. (2015) and Schrader et al. (2017) are the average and 2SD of all individual measurements per chondrule, excluding relict grains. Shown also is the mass balance model from Tenner et al. (2015), in which dust and gas compositions are based on those from Anders and Grevesse (1989) and Allende Prieto et al. (2001; 2002). Dust to gas ratios are shown as thin semi-vertical lines, and H₂O ice abundances in the dust (thick semi-horizontal lines) are given relative to that in CI dust. Details of the model are given in Section 4.3.2.

Kita and Ushikubo (2012) noted that, while many CR chondrite chondrules have lower (²⁶Al/²⁷Al)₀ than LL, CO, and Acfer 094 chondrules, with most lacking excess ²⁶Mg (e.g. Fig. 1b), it was unknown if such isochrons were potentially influenced by aqueous alteration. Thus, our findings are important because we document that the chondrule plagioclase used for isochrons is unaltered (Section 4.1). This provides confidence that (1) some CR chondrite chondrules have resolvable timing differences exceeding 1 Myr (perhaps greater than 2 Myr), and that (2) at least some CR chondrite chondrules postdate those from LL, CO, and Acfer 094 chondrites by 1 to 2 Myr. Furthermore, the distribution of chondrule formation ages for a given type of chondrite may reflect the degree of local disk turbulence that controlled the time interval of parent body accretion. For example, Cuzzi et al. (2010) calculate that complete parent body accretion within 0.5 Myr would require no turbulence. LL, CO, and Acfer 094 parent bodies are consistent with this scenario based on the narrow standard deviations of their averaged chondrule ages (1SD: 0.4 Myr for each chondrite; Kita and Ushikubo, 2012). Such rapid accretion with no turbulence could explain why chondrules from LL, CO, and Acfer 094 chondrites cluster into discreet Mg# and O-isotope domains, as precursors may have been poorly mixed (e.g. Kita et al., 2010; Ushikubo et al., 2012; Tenner et al. 2013; we note that a small number of Acfer 094 chondrules have unique Mg#, oxygen isotope, and Al-Mg isotope relationships; Hertwig et al., 2019). In contrast, the longer duration of CR chondrite chondrule formation (> 1 million years) suggests a more turbulent environment, which extended the time for complete parent body accretion (e.g. Cuzzi et al., 2010). If true, then at least some CR chondrite chondrules could have formed after establishment of LL, CO, and Acfer 094 chondrite parent bodies. In turn, a more turbulent chondrule-forming environment could have fostered better mixing of solid precursors (e.g. dusts and ices) within the accretion region. This might explain why FeO-poor chondrules from CR chondrites show a continuous Mg# versus O-isotope trend (e.g. Fig 11b), as discussed below.

4.3. Chondrule formation in distinct environments II: Physicochemical characteristics.

4.3.1. Redox conditions during chondrule formation: chondrule Mg# versus Al-Mg isotope systematics.

By relating chondrule Mg#’s to their Al-Mg isotope systematics, redox conditions at the time of chondrule formation can be evaluated. Here, we report chondrule Mg#’s as defined in Tenner et al. (2015). From electron microprobe data per chondrule, average Mg#’s are calculated for olivine and for low-Ca pyroxene. If only one phase is analyzed, its average Mg# is defined as the chondrule Mg#. If both phases are analyzed, the averaged Mg#’s of each phase are weighted equally to define the chondrule Mg#. Chondrule Mg# uncertainties are reported as the range of all measured values. Although this definition does not account for modal abundances of olivine and low-Ca pyroxene within a chondrule (when both phases are present), it is observed among chondrites with petrologic types near 3.0, including QUE 99177 and MET 00426, that Mg#’s of coexisting olivine and low-Ca pyroxene are nearly identical per chondrule, typically differing by one unit or less (see Fig. 4 from Tenner et al., 2015 and associated references). Therefore, any additional uncertainty in chondrule Mg# stemming from olivine and low-Ca pyroxene modal abundances is likely minimal. Using defined chondrule Mg#’s, redox conditions are determined according to a regression constrained by equilibrium thermodynamic models (Ebel and Grossman, 2000; Fedkin and Grossman, 2006; Grossman et al., 2008):

$$\text{chondrule Mg\#}=100-\text{exp}\left((\left[\text{log}f{\text{O}}_2-\text{log IW}\right]+3.444\right)/0.6649$$ (2)

where IW is the Fe-wüstite buffer. Full details of Eqn. 2 are provided in Tenner et al. (2015).

Among the chondrules investigated, those that formed under the most reducing conditions are older than those produced at more oxidizing conditions (Fig. 11a). All chondrules with excess ²⁶Mg (Q7, Q16, Q38, M6, and M23; Table 2; Fig. 4). formed 3.4 to 3.6 log units below IW, according to their Mg#’s (99.0 to 99.2; Eqn. 2). The remaining seven chondrules with no resolvable excess ²⁶Mg formed 2.3 to 3.3 log units below IW, according to their Mg#’s (94.2 to 98.7; Eqn. 2). If one considers the role of early Solar System H₂O ice as an oxidizing agent (e.g. Fedkin and Grossman, 2006; 2016; Connolly and Huss, 2010), these results suggest either (1) ice abundances within the CR chondrite chondrule-forming environment varied on temporal scales; or (2) that separate regions of the CR chondrite chondrule-forming environment coexisted with different H₂O ice abundances, and produced chondrules over different time intervals. Either way, the oldest chondrules are predicted to have formed in a relatively anhydrous environment, while younger chondrules formed in a more ice-enhanced region.

In contrast to our findings, Schrader et al. (2017) report no relationship between redox conditions and Al-Mg isotope systematics among CR chondrite chondrules; they also include data from Nagashima et al. (2014) in their evaluation. However, these datasets have relatively few Al-Mg isotope data from Mg# ≥ 99 chondrules, which, according to our results, are systematically older. In addition, Schrader et al. (2017) calculate chondrule Fe#’s (= 100 ‒ chondrule Mg#) in a manner that slightly differs from our definition described above. Specifically, (1) they use the averaged Fe# of chondrule olivine and the 2SD uncertainty as the equivalent for chondrule Fe#, even if a chondrule has accompanying low-Ca pyroxene EPMA data; and (2) they report the averaged Fe# of low-Ca pyroxene and the 2SD uncertainty as the equivalent for chondrule Fe#, when chondrules have EPMA measurements of only low-Ca pyroxene.

To provide consistency among datasets, we recalculated chondrule Mg#’s from Nagashima et al. (2014) and Schrader et al. (2017) in the same manner as our data. EPMA data related to Nagashima et al. (2014) were kindly provided to us by Kazuhide Nagashima (Appendix EA7). Regarding Mg# > 90 chondrules, all but two of the recalculated chondrule Mg#’s are within 0.4 units of those shown in Fig. 9 from Schrader et al. (2017). However, two chondrules, ED1-MK-#1 and NWA721-#7 from Nagashima et al. (2014) show larger differences. For chondrule ED1-MK-#1, only low-Ca pyroxene EPMA data were collected. Figure 9 from Schrader et al. (2017) shows ED1-MK-#1 has a chondrule Mg# (= 100 ‒ Fe#) of 97.4 with a 2SD of 2.3, while we calculate a higher Mg# of 99.0 +0.3/‒0.2 (Appendix EA7). This difference occurs because we excluded two EPMA point data with low cation totals of 3.95 and 3.96 formula units (assuming 6 oxygens), and with lower Mg#’s of 96.9 and 98.3, respectively. Regarding chondrule NWA721-#7, there are EPMA data from olivine, with an average Mg# of 99.2, and from low-Ca pyroxene, with an average Mg# of 98.0. Schrader et al. (2017) use only olivine data when reporting the chondrule Mg# (i.e. 99.2 ± 0.3; 2SD), while we calculate a slightly lower chondrule Mg# of 98.6 ± 0.7 when using the olivine and low-Ca pyroxene data (Appendix EA7). For type II chondrules (Mg# < 90), recalculated chondrule Mg#’s differ by 0 to 2.8 units relative to those shown by Schrader et al. (2017). These differences are minor within the context of fO₂ variability, as Mg# differences among type II chondrules are relatively insensitive in terms of their corresponding redox conditions (e.g. Eqn. 2; top axis of Fig. 11a).

Recalculated chondrule Mg#’s, and the (²⁶Al/²⁷Al)₀ data from Nagashima et al. (2014) and Schrader et al. (2017) are shown in Fig. 11a. We note one Al-rich chondrule from Schrader et al. (2017), PCA 91082,15 Ch14, is interpreted by the authors to have relict plagioclase, based on heterogeneous O-isotope ratios. This could suggest at least some portion of PCA 91082,15 Ch14 plagioclase did not crystallize from the final chondrule melt, meaning the inferred (²⁶Al/²⁷Al)₀ of PCA91082,15 Ch14 may not be related to its chondrule Mg#. Therefore, we omit PCA 91082,15 Ch14 data in Fig. 12, but note it has a chondrule Mg# of 99.1 +0.3/‒0.7, a (²⁶Al/²⁷Al)₀ of (‒0.3 ± 3.4) × 10^‒6, and a Δ¹⁷O of ‒5.6 ± 1.0‰.

Fig. 12.

Cartoon interpretation of ice migration into the CR chondrite chondrule forming environment, based on (1) the Mg#’s, Δ¹⁷O values, and inferred (²⁶Al/²⁷Al)₀ values of chondrules from this study, and (2) dust enrichment and ice enhancement parameters modeled by Tenner et al. (2015). See also: Fig. 11.

If combining all data shown in Figure 11a, 7 of 8 chondrules with Mg#’s ≥ 99 have resolvable excess ²⁶Mg, only 2 of 14 chondrules with Mg#’s between 98 and 99 have resolvable excess ²⁶Mg, and no chondrules (0 of 7) with Mg#’s between 94 and 98 have resolvable excess ²⁶Mg. As such, chondrules with Mg#’s > 99 appear to more consistently have excess ²⁶Mg, while Mg# 94–99 chondrules do not. We note porphyritic and Al-rich chondrules are not distinguishable with respect to this relationship.

Three type II chondrule data from Nagashima et al. (2014) and Schrader et al. (2017) have Mg#’s ranging from 49.7 to 75.7 (Fig. 11a). Of these, two have resolvable excess ²⁶Mg and one does not. The small sample set and the large internal Mg# variabilities among these chondrules make it difficult to discern if a (²⁶Al/²⁷Al)₀ versus Mg# correlation exists among type II CR chondrite chondrules.

4.3.2. Mg# and oxygen isotope relationships: quantifying the role of ¹⁶O-poor H₂O ice

In addition to the Mg# versus (²⁶Al/²⁷Al)₀ correlation shown in Figure 11a, FeO-poor CR chondrite chondrules produce a related trend in which decreasing chondrule Mg#’s correspond to increased Δ¹⁷O. This is illustrated in Fig. 11b, where chondrules with known O-isotope ratios and Al-Mg isotope systematics are highlighted (large solid symbols; from this study and from Schrader et al., 2017), along with additional data from Tenner et al. (2015) which were not studied for Al-Mg isotope systematics (small x symbols).

As mentioned earlier, it is hypothesized that increased proportions of ¹⁶O-poor H₂O ice in CR chondrite chondrule precursors made the environment more oxidizing, forming lower Mg# chondrules with higher Δ¹⁷O. This seems plausible, as 1 atm. experiments show H₂O vapor readily exchanges its O-isotopes with those from chondrule-like melts (e.g. Yu et al. 1995; Di Rocco and Pack 2015). Dust-to-gas ratios also likely controlled redox conditions of the chondrule-forming environment, according to equilibrium thermodynamic models (e.g. Ebel and Grossman, 2000; Grossman et al., 2008). Tenner et al. (2015) modeled the influence of H₂O ice addition to chondrule precursors in a dust-enriched system by mass balance (Fig. 11b). The model splits chondrule precursors into four components: (1) Solar gas; (2) the silicate portion of dust; (3) H₂O ice in dust; and (4) organic matter in dust. Compositions of each component are based on atomic abundances from Anders and Grevesse (1989) and Allende Prieto et al. (2001; 2002) (see Table 3 from Tenner et al., 2015). Oxygen fractions of each component are constrained at a given dust-to-gas ratio and ice enhancement factor, and H₂O ice proportions are calculated relative to that within CI dust (~18 wt.% H₂O; Anders and Grevesse, 1989). Components are assigned the following Δ¹⁷O values: (1) Solar gas: ‒28.4‰, from Solar wind measurements (McKeegan et al., 2011); (2) silicate in dust: ‒5.9‰, which comes from the lowest Δ¹⁷O chondrule measured by Tenner et al. (2015); (3) H₂O ice in dust: +5.1‰, which satisfies Δ¹⁷O values of the lowest Mg# chondrules studied by Tenner et al. (2015) (~+0.5‰) when employing a mass balance consisting of CI dust; and (4) organic matter in dust: +11.3‰, which is the average value measured in CR chondrites (Hashizume et al., 2011). Chondrule Mg#’s are simultaneously parameterized by metal-silicate phase equilibria (e.g. eqn. 2), as the imposed oxygen fugacity is directly related to the atomic proportions of H, O, and C in the precursor assemblage (e.g. Eqn. 24 from Grossman et al., 2008; Eqn. 3 from Tenner et al., 2015).

Within these constraints, the model predicts most CR chondrite chondrules formed at dust-to-gas ratios between 100× and 200×. The trend of increasing chondrule Δ¹⁷O, from ~–6‰ to ~–1‰, as chondrule Mg#’s decrease from ~99 to ~94, is caused by an increasing proportion of ¹⁶O-poor H₂O ice within the dust, from essentially anhydrous to ~0.8 times the abundance in CI dust (Fig. 12b). Type II chondrules formed in a different environment with higher dust-to-gas ratios (~1000× to ~2500×), and a greater proportion of H₂O ice, corresponding to lower chondrule Mg#’s and higher Δ¹⁷O (Fig. 11b).

4.3.3. Temporal and spatial relationships of chondrule precursors

The links between Al-Mg isotope systematics, Mg#’s, and oxygen isotope ratios of chondrules (Fig. 11) allow for interpreting the temporal and spatial relationships of dust, gas, and H₂O ice within the CR chondrite accretion region. This is shown as a simple cartoon in Fig. 12, which considers ice migration into the CR chondrite accretion region as the protoplanetary disk cooled (e.g. Ciesla and Cuzzi, 2006).

For chondrules with excess ²⁶Mg, Mg#’s ≥ 99, and with Δ¹⁷O values near ‒5‰ (Fig. 12a), the mass balance model from Fig. 11b predicts formation in an environment depleted in H₂O ice, with a dust to gas ratio of ~100×. Based on Al-Mg isotope systematics from this study (Table 2; Fig. 11a), these chondrules formed 2.2 (‒0.5/+1.1) to 2.8 (‒0.3/+0.5) Myr after CAIs.

After formation of the chondrules depicted in Fig. 12a, it is possible that upon cooling of the protoplanetary disk, the CR chondrite accretion region received an influx of ¹⁶O-poor H₂O ice (Δ¹⁷O: ~+5‰; Fig. 12b). This would have increased the oxidation state and the bulk Δ¹⁷O of chondrule precursors. In addition, the mass balance model in Fig. 11b predicts the range of dust to gas ratios expanded, from 100× to 300×, also contributing to more oxidizing conditions. Flash heating events in this environment would have produced Mg# ~94 to ~99 chondrules with Δ¹⁷O values of ~−1‰ to ~−5‰, respectively (e.g. Fig. 11b). Such chondrules formed >2.9 to >3.7 Myr after CAIs, based on our inferred (²⁶Al/²⁷Al)₀ upper limits.

CR chondrite chondrules are predominantly FeO-poor (constituting ~96% of the population; Schrader et al., 2011; 2015), meaning the scenario depicted in Figure 12 accounts for nearly all of the accretion region. Based on limited Al-Mg isotope systematics from Nagashima et al. (2014) and Schrader et al. (2017), the more minor, type II chondrule-forming environment (~4% of the chondrule population) appears to have produced chondrules over a broadly similar timeframe as Mg# < 99 FeO-poor chondrules (Fig. 11a).

4.4. Comparisons to other carbonaceous chondrites and from Wild 2 comet particles.

When comparing Al-Mg isotope systematics, Mg#’s, and Δ¹⁷O to other chondrite chondrules and cometary materials, chondrules from CR chondrites exhibit similarities in some cases, but in other instances there are differences. Comparisons are available for LL3 (Kita et al., 2000; 2010), Acfer 094 (Ushikubo et al., 2012; 2013), and CO3 chondrite chondrules (Kurahashi et al., 2008a; Tenner et al., 2013), as well for Stardust comet particles Iris (Ogliore et al., 2012) and Pyxie (Nakashima et al., 2012; 2015).

LL3, Acfer 094, and CO3 chondrite chondrules have distinct Δ¹⁷O and Mg# groups, even though they have similar ranges of (²⁶Al/²⁷Al)₀ (Fig. 13a). LL3 chondrite chondrules have uniform Δ¹⁷O (~0‒1‰) over a wide chondrule Mg# interval (75‒96.3), indicating constant precursor O-isotope ratios, even though redox conditions varied (e.g. Eqn. 2). Acfer 094 and CO3 chondrite chondrules mainly cluster into two groups: those with Mg#’s ≥ 97 largely have Δ¹⁷O values of ~‒5‰, while those with Mg#’s < 97 have Δ¹⁷O values near ‒2.5‰ (Fig. 13a). Like CR chondrites, the link between decreasing chondrule Mg# and increasing Δ¹⁷O among CO and Acfer 094 chondrites is attributed to increased ¹⁶O-poor H₂O ice in chondrule precursors, and elevated dust-to-gas ratios (Ushikubo et al., 2012; Tenner et al., 2013). As chondrules from LL3, CO3, and Acfer 094 chondrites have similar ²⁶Al age ranges, their distinct oxygen isotope signatures (Δ¹⁷O: ~0‒1‰ for LL3; Δ¹⁷O: ~‒2.5‰ and ~‒5‰ for CO3 and Acfer 094) suggest the parent bodies were spatially separated. If true, this would support a hypothesis for physically separated ordinary and carbonaceous chondrite accretion regions, which is based on their distinct Cr, Ti, Mg and O-isotope relationships (Warren, 2011; Van Kooten et al., 2016; Gerber et al., 2017). This division is interpreted to have been caused by Jupiter’s formation and its influence on dynamic properties of the protoplanetary disk, including pressure regimes, aerodynamic drag, and turbulence (Walsh et al., 2011; 2012; Desch et al., 2018).

Fig. 13.

(²⁶Al/²⁷Al)₀ vs. Δ¹⁷O for (a) LL3 (Kita et al. 2000; 2010), Acfer 094 (Ushikubo et al., 2012; 2013), and CO3 (Kurahashi et al., 2008; Tenner et al. 2013) chondrite chondrules, and for (b) CR2 chondrite chondrules (this study; Schrader et al., 2017: S17) and Stardust comet particles Iris (Ogliore et al., 2012) and Pyxie (Nakashima et al., 2012; 2015). Mg# ranges and uncertainties are as reported in each study, with the exception of chondrule Mg#’s from Schrader et al., 2017, which are recalculated as described in section 4.3.1.. Relative ages are calculated according to Eqn. 1 and use a canonical CAI (²⁶Al/²⁷Al)₀ of 5.23 × 10^˗5 (Jacobsen et al., 2008; MacPherson et al., 2010).

CR chondrite chondrules with Mg#’s ≥ 99 and with Δ¹⁷O values of ~‒5‰ have similar (²⁶Al/²⁷Al)₀ as those from Mg# ≥ 97, Δ¹⁷O ~‒5‰ CO3 and Acfer 094 chondrite chondrules (e.g. Fig. 13b). This suggests an anhydrous and low dust density chondrule-forming environment that was common to multiple carbonaceous chondrites over a similar timeframe. In contrast, the lack of resolvable excess ²⁶Mg in most Mg# < 99 CR chondrite chondrules with Δ¹⁷O higher than ‒5‰ (Fig. 14a) differs from lower Mg# CO and Acfer 094 chondrite chondrules, which also have higher Δ¹⁷O (predominantly near ‒2.5‰), but have excess ²⁶Mg (Fig. 13a). Thus, while H₂O ice abundances and dust to-gas ratios in the CO and Acfer 094 chondrule-forming environments varied spatially, these characteristics of the CR chondrite chondrule-forming environment likely varied temporally.

Mafic minerals from comet Wild 2 have Mg#-Δ¹⁷O characteristics similar to those of CR chondrite chondrules (Nakashima et al., 2012; Defouilloy et al., 2017 and included references). Two comet particles, Iris (Ogliore et al., 2012) and Pyxie (Nakashima et al., 2012; 2015), have Mg#’s of 64 and 95, respectively, Δ¹⁷O values of −0.3‰ and −1.1‰, respectively, and neither have resolvable excess ²⁶Mg (Fig. 13b). These features are similar to Mg# < 99 CR chondrite chondrules that are relatively ¹⁶O-poor and lack resolvable excess ²⁶Mg (Fig. 13b). One possibility for these similarities is that at least some CR chondrite-associated materials were transported to comet-forming regions, possibly involving disk wind (Cuzzi and Hogan, 2003). Evidence for such long-distance transport comes from the presence of CAI-like materials within Stardust particles (Brownlee et al., 2006; McKeegan et al., 2006; Zolensky et al., 2006, Ishii et al., 2010). Ciesla (2007a; 2007b) also show models of a viscously evolving protoplanetary disk induce large-scale outward flows about the midplane, with the ability to carry chondrule and CAI-sized materials to 10 AU.

4.5. Comparisons to other isotope systematics of CR chondrite chondrules

Budde et al. (2018) report a concordant mean age of CR chondrite chondrule formation when comparing Pb-Pb (3.66 ± 0.63 Myr after CAIs; Amelin et al., 2002), Al-Mg (3.7 +0.3/–0.2 Myr after CAIs; Nagashima et al., 2014; Schrader et al., 2017), and Hf-W isotope systematics (3.63 ± 0.62 Myr after CAIs; Budde et al., 2018). Although these isotope systems are concordant, the relationship between the mean age of CR chondrite chondrule formation and the true time interval over which chondrules formed requires further attention. For example, the Hf-W data from Budde et al. (2018) and the Pb-Pb data from Amelin et al. (2002) rely on chondrule aggregates/sets, meaning a bulk chondrule formation age is established, but the actual range of chondrule formation is not constrained. Regarding Al-Mg isotope systematics, although individual chondrule ages are routinely generated, many CR chondrite chondrules have unresolvable excess ²⁶Mg, meaning only upper age limits are determined. Thus, although Al-Mg isotope systematics from this study indicate CR chondrite chondrules formed over a duration of at least 1.2 million years, the true interval is not fully constrained. Considering these limitations, estimates of the CR chondrite parent body accretion age, 3.5 ± 0.5 Myr after CAIs (Sugiura and Fujiya, 2014), as well as the age of CR chondrite parent body alteration, 4–5 Myr after CAIs (Jilly-Rehak et al., 2017), are useful, because they anchor the latest point CR chondrite chondrules could have formed. Nearly half of the chondrules studied here (5 of 12) have inferred ages of >3.6 Myr after CAIs, indicating a significant proportion formed immediately prior to parent body accretion. If there is a contingent of relatively old chondrules, as suggested by Al-Mg isotope systematics (this study, Nagashima et al., 2014; Schrader et al., 2017), and as suggested by Pb-Pb ages of some individual chondrules (Bollard et al., 2017), they likely represent a minor population, as hypothesized by Schrader et al. (2017).

With this in mind the mean CR chondrite chondrule formation age by Al-Mg isotope systematics can be refined, so that it reflects the proportions of chondrules with and without resolvable excess ²⁶Mg. To do so, we use the relationship where Mg# ≥ 99 chondrules predominantly have resolvable, positive (²⁶Al/²⁷Al)₀ (e.g. 7 of 8; Fig. 11a), and where most Mg# < 99 chondrules do not (e.g. 4 of 24; Fig. 11a). Next, we estimate the modes of Mg# > 99 and Mg# ≤ 99 chondrules using data from Tenner et al. (2015), who determined Mg#’s of every chondrule in the MET 00426,46 thin section, constituting 29 chondrules/chondrule fragments. Of these, five have Mg#’s ≥ 99 (17.2%), and 24 have Mg#’s below 99 (82.8%). We then refine the mean chondrule formation age by mass balance, using (1) the respective weighted mean Al-Mg isotope systematics of Mg# ≥ 99 and Mg# < 99 CR chondrite chondrules; and (2) the corresponding chondrule Mg# modes (Appendix EA8). In the following, we provide calculations using (1) only data from this study, and (2) the combined data shown in Fig. 11a, from this study, Nagashima et al. (2014) and Schrader et al. (2017).

In terms of (²⁶Al/²⁷Al)₀, the mass balance generates a value of (0.9 ± 0.6) × 10^‒6, using only data from this work, and a value of (1.6 ± 0.3) × 10^‒6, using the combined data shown in Fig. 11a (Appendix EA8). Uncertainties are propagated errors coming from the weighted mean (²⁶Al/²⁷Al)₀ of each Mg# mode. These (²⁶Al/²⁷Al)₀ values correspond to refined mean chondrule formation ages of 4.2 (+1.2/‒0.6) Myr, if using only data from this work, and 3.5 (+0.2/–0.1) Myr, if using the combined dataset shown in Fig. 11a. However, we stress that (²⁶Al/²⁷Al)₀ values change exponentially with relative ²⁶Al ages, meaning that an age calculated from a mean (²⁶Al/²⁷Al)₀ is biased toward an older value when the collective duration of chondrule formation is longer than the ²⁶Al half-life, which is the case for CR chondrite chondrules. Therefore, we refine the mass balance even further, by employing the mean relative ages of each chondrule Mg# mode, rather than their mean (²⁶Al/²⁷Al)₀ values (Appendix EA8). By doing do, the mass balance generates a refined mean age of 3.8 ± 0.3 Myr after CAIs, if using the combined data shown in Fig. 11a. Here, the uncertainty is the propagated error coming from the mean ages of each Mg# mode. If considering only the data from this work, the mass balance cannot produce a fully constrained age in this manner, because the mean relative age of Mg# < 99 chondrules is only an upper limit. Thus, our preferred mean chondrule formation age for all CR chondrite chondrules is 3.8 ± 0.3 Myr after CAIs (if only considering Mg# ≥ 99 chondrules their mean formation age is 2.4 ± 0.1 Myr after CAIs, and for Mg# < 99 chondrules their mean age is 4.1 ± 0.3 Myr after CAIs) (Appendix EA8). This value is similar to the estimate from Schrader et al. (2017), but benefits from the added clarity of the chondrule Mg# mode/chondrule age relationship (e.g. Fig. 11a), and also accounts for the non-linearity between chondrule (²⁶Al/²⁷Al)₀ and the corresponding relative age. As such, the mean Al-Mg age of CR chondrite chondrules is consistent with Pb-Pb and Hf-W isotope systematics of chondrule aggregates/sets (e.g. Amelin et al., 2002; Budde et al., 2018). Finally, the relationship between chondrule ages, Mg#’s, and oxygen isotopes offers a pathway to improve the significance of Hf-W and Pb-Pb isotope systematics of CR chondrite chondrule aggregates, particularly if Mg# > 99 and Mg# ≤ 99 are measured separately. The latter would better constrain the mean age for CR chondrite chondrules that are too young to resolve by Al-Mg isotope systematics. However, care must be taken to identify samples that avoided disturbances related to the isotope system of interest.

CONCLUSIONS

We investigated Al-Mg isotope systematics of twelve FeO-poor CR chondrite chondrules by SIMS, and assessed the quality of plagioclase that control slopes of mineral isochrons. We found that:

1. Five chondrules have resolvable excess ²⁶Mg, with (²⁶Al/²⁷Al)₀ values of (3.5 ± 1.3) × 10^˗6 to (6.0 ± 3.9) × 10^˗6. Including isochron uncertainties, and using a canonical (²⁶Al/²⁷Al)₀ value of 5.23× 10^˗5, this corresponds to their formation 1.7 to 3.3 Myr after CAIs. These chondrules have high Mg#’s (99.0 to 99.2), and a narrow Δ¹⁷O range (–4.2‰ to –5.3‰).

2. Seven chondrules lack resolvable excess ²⁶Mg. Their (²⁶Al/²⁷Al)₀ upper limits range from 1.3 × 10^˗6 to 3.2 × 10^˗6, corresponding to formation >2.9 to >3.7 Myr after CAIs. They have lower Mg#’s (94.2 to 98.7) and generally higher Δ¹⁷O (–0.9 to –4.9‰), relative to the five chondrules with excess ²⁶Mg.

3. When comparing chondrules with and without excess ²⁶Mg, the estimated chondrule formation duration was at least 1.1 Myr, and if considering uncertainties, may have extended beyond 2 Myr.

4. Several characteristics indicate the plagioclase used to establish chondrule isochrons have an undisturbed Al-Mg isotope system. First, SIMS depth profile patterns of ²⁷Al/²⁴Mg match those collected from anorthite glass standards. Second, TEM analysis shows little-to-no inclusions, and no replacement by other mineral phases during thermal metamorphism, even at the nanometer scale. Third, when comparisons are available, oxygen isotope ratios of chondrule plagioclase match those from coexisting olivine and pyroxene, indicating a low extent of thermal metamorphism (< 150°C in a wet system). Fourth, nearly all chondrule plagioclase studied has a resolvable “excess silica”, [ ]Si₄O₈ component, which is an anhydrous high-temperature primary crystallization feature. In addition, the chondrule plagioclase studied have high anorthite contents (0.793 to 0.999, when normalized to anorthite, albite, and K-feldspar components), and have relatively high concentrations of MgO (0.6 to 1.2 wt.%). These features are consistent with chondrule plagioclase from the least thermally altered 3.00 to 3.05 petrologic type chondrites.

5. Using relationships between chondrule (²⁶Al/²⁷Al)₀, Mg#, and Δ¹⁷O, we interpret spatial and temporal features related to dust, gas, and H₂O ice abundances within the FeO-poor CR chondrite chondrule-forming environment. We estimate that from 2.2 (‒0.5/+1.1) to 2.8 (‒0.3/+0.5) Myr after CAIs the environment was relatively anhydrous, with a dust-to-gas ratio of ~100× that produced Mg# ≥ 99, Δ¹⁷O ~−5‰ chondrules with resolvable excess ²⁶Mg. After formation of these chondrules, we interpret that ¹⁶O-poor H₂O ice migrated into this chondrule-forming region, and that the dust-to-gas ratio increased (100× to 300×), leading to the production of more oxidized Mg# 94–99 chondrules with higher Δ¹⁷O (−5‰ to –1‰), and with no resolvable excess ²⁶Mg.

6. Using data from this study and from previous work (Nagashima et al., 2014; Schrader et al., 2017), we establish a refined mean CR chondrite chondrule formation age of 3.8 ± 0.3 Myr after CAIs. This is accomplished via mass balance, employing (1) modal proportions of Mg# ≥ 99 and Mg# < 99 chondrules in CR chondrites; and (2) mean relative ages from each Mg# mode. This age is consistent with those from Pb-Pb and Hf-W isotope systematics of CR chondrite chondrule aggregates (3.66 ± 0.63 Myr after CAIs and 3.63 ± 0.62 Myr after CAIs, respectively; Amelin et al., 2002 and Budde et al., 2018, respectively).