Constraints on the formation environment of two chondrule-like igneous particles from Comet 81P/Wild 2

Abstract

Using chemical and petrologic evidence and modeling, we deduce that two chondrule-like particles named Iris and Callie, from Stardust cometary track C2052,12,74, formed in an environment very similar to that seen for type II chondrules in meteorites. Iris was heated near liquidus, equilibrated, and cooled at ≤ 100 °C/hr and within ≈ 2 log units of the IW buffer with a high partial pressure of Na such as would be present with dust enrichments of ≈ 10³. There was no detectable metamorphic, nebular or aqueous alteration. In previous work Ogliore et al. (2012) reported that Iris formed late, > 3 Myr after CAIs, assuming ²⁶Al was homogenously distributed, and was rich in heavy oxygen. Iris may be similar to assemblages found only in interplanetary dust particles and Stardust cometary samples called Kool particles. Callie is chemically and isotopically very similar but not identical to Iris.

INTRODUCTION

The samples of comet 81P/Wild 2 returned by NASA’s Stardust mission contain abundant igneous rocks that bear striking similarities to igneous objects found in meteorites (Zolensky et al. 2006, Leroux et al. 2008, Nakamura et al. 2008a, Nakamura et al. 2008b, Tomeoka et al. 2008, Bridges et al. 2012, Joswiak et al. 2012, Ogliore et al. 2012, Chi et al. 2009, Matzel et al. 2010, Joswiak et al. 2013, Simon et al. 2008). Many of these objects may have formed in the inner solar system and were subsequently transported to the Kuiper belt. Several investigators (e.g. Ciesla 2007) have attempted to model this transport, though issues still remain, such as the difficulty in maintaining a reasonable crystalline fraction with a total mass flow matching the empirical evidence (Westphal et al. 2009). Oxygen fugacity and isotope measurements may help constrain the mixing processes in the early solar system by tracking the movement of water and mixing of distinct material reservoirs (Shearer et al. 1998).

In addition, olivines, pyroxenes, spinels and sulfides in Stardust cometary samples often exist independent of chondrule, CAI or AOA type assemblages and may be representative of meteoritic matrix-like material, or assemblages found in interplanetary dust particles (IDPs) (Zolensky et al. 2008, Frank et al. 2014b, Joswiak et al. 2012, Stodolna et al. 2012a). Olivine and pyroxene iron compositions also vary widely between and within Stardust tracks, and compare most favorably to anhydrous IDPs (Zolensky et al. 2008, Frank et al. 2014b). Finally, assemblages containing Fe-rich olivine and Na- and Cr-rich clinopyroxene were reported by Joswiak et al. 2009 and given the acrynomn Kool (kosmochloric high-Ca pyroxene and Fe-rich olivine, Joswiak et al. 2009). To date, these Kool assemblages have been found only in Comet Wild 2 and IDPs.

The processes that formed meteorites are generally better understood than those giving rise to IDPs and Wild 2 samples insofar as meteorites are significantly larger and provide better context for petrography. Oxygen fugacity, cooling rates, shock, and mineral formation sequences can often be deduced by comparison between phenocrysts, textures, or averaging properties over large volumes of material. Comet Wild 2 samples must be studied in greater detail using, for example, coordinated synchrotron and TEM analyses to extract comparable information.

METHODS

Sample Preparation

We studied two terminal particles of C2052,12,74 (track 74), a type “B” track (Burchell et al. 2008, Dominguez et al. 2004, Trigo-Rodriguez et al. 2008). Fig. 1A shows the entire track with a bulb on the left side, and three “whisker tracks” which contain the terminal particles. The track was captured in aerogel as described in Tsou et al. (2003). The first terminal particle, named Iris, is one of three terminal particles in track 74. The second terminal particle is named Callie. Iris and Callie are at the end of separate whisker tracks pointing to a common origin within the bulb. They were extracted together in a keystone (Westphal et al. 2004) which was further dissected to separate them into thin aerogel wafers. Both particles were analyzed using synchrotron X-ray micro and nanoprobes. The aerogel wafers were then impregnated with EMBED-812 epoxy, fixed to the end of an epoxy bullet and ultramicrotomed into ≈ 100 nm thick sections. The sections were placed on copper Transmission Electron Microscope (TEM) grids with a 10 nm layer of amorphous carbon. We ultramicrotomed several times and studied the remaining bullets at each interval, typically between 0.5 and 3 μm deeper than the last.

Figure 1.

A) Optical image of Track 74. B) XRF image of Iris showing Fe, Cr and Ni (red, green and blue respectively). C) Optical LRGB image of Iris. The luminance channel corresponds to unpolarized contrast, the red, green and blue channels are acquired through cross polarizers at 0, 30, and 45° and show variation among the orientations of the constituent crystals. D) XRF image of Callie showing Ga (red) Cr (green) and Mn (blue). E) Callie shown as the projection of an optical focus stack in the same orientation as D. The 10 μm scalebar in B applies to B-E.

Synchrotron X-ray Microprobe Analysis

We examined Iris at the Advanced Light Source (ALS) beamline 10.3.2 (Marcus et al. 2004) using hard X-rays between 5 and 15 keV. We did X-ray fluorescence mapping (μXRF, Fig. 1B) of the particle in an aerogel wafer ≈ 100 μm thick before epoxy embedding to study the mineralogy. The FWHM of the beam was ≈ 1.5 μm. The fluorescence spectrum was acquired using an XIA 7-element Ge(Li) energy dispersive detector. The aerogel prevented analysis of elements lighter than potassium because it strongly absorbed photons < 3 keV. We also acquired an 8 keV XRD map on ALS beamline 12.3.2 (Tamura et al. 2009) using a MAR133 CCD. The FWHM was about 1 μm and the energy width was about 1 eV. Images were acquired for 37 seconds/frame, for a total of 132 images in an 11 × 12 pixel raster.

We measured the microtomed sections by Scanning Transmission X-ray Microscopy (STXM) on ALS beamlines 11.0.2 and 5.3.2.1 (Kilcoyne et al. 2003, Warwick et al. 2003, Tyliszczak et al. 2004). STXM provided soft X-ray mapping of oxygen, sodium, quantification of vanadium and manganese in chromite, and iron L-edge X-ray absorption near-edge structure (XANES) spectroscopy. Beam size was > 30 nm FWHM in most cases and the energy resolution was E/ΔE ≥ 3000.

We acquired XRF maps of Callie (Fig 1D) on beamline ID22 NanoImaging endstation at the European Synchrotron Radiation Facility (ESRF) using a ≈ 0.5 μm beam (Bleuet et al. 2008).

Electron Microscopy

We analyzed the bullets using a Tescan Vega 3 scanning electron microscope (SEM) with an Oxford silicon drift detector, an FEI Strata 235 Dual Beam focused ion beam (FIB) at the National Center for Electron Microscopy (NCEM), a Zeiss MA10 SEM, and a JEOL JSM-6490 SEM. The FIB was also used for sample preparation as described in Ogliore et al. (2012).

We did transmission electron microscopy (TEM) at NCEM, and at the University of Washington (UW). At NCEM, we used a Philips CM200 FEG at 200 keV, a Zeiss Libra 200 MC at 200 keV, and a JEOL 3010 LaB6 at 300 keV. We used the CM200 for Energy Dispersive X-ray Spectroscopy (EDS) analysis with an Oxford Si(Li) detector. We did energy filtered imaging on both the CM200 and Libra. At UW, we used an FEI Tecnai TF20 STEM with an EDAX Si(Li) EDS. In some cases we used a LN₂ cooled sample holder to preserve sensitive Na- rich phases, including amorphous silicate and Na-bearing plagioclase feldspar.

EDS Quantification

We carried out SEM EDS quantification using the XPP matrix correction algorithm (Pouchou and Pichoir 1992) as implemented by Oxford Inca software (Oxford Instruments plc 2009). Geller Microanalytical supplied albite, diopside, and bytownite standards for quantifying compositions of plagioclase and glass (Geller and Engle 2002). We used MnS, TiO₂, and VN standards for trace element corrections. We quantified major elements in olivine using the Springwater Fo₈₃ standard from the Smithsonian (Jarosewich et al. 1980) and the Geller standards for Ca and other minor elements. We quantified chromite using Stillwater chromite from the University of California, Berkeley Microprobe facility (UC# 523-9). SEM EDS analyses were done at 6 keV, ≈ 100 pA and 10 keV, ≈ 500 pA. Excitation volumes in olivine allowed imaging resolution < 350 nm for all elements at 6 keV, and < 750 nm for all elements at 10 keV. Plagioclase resolution was < 750 nm for all elements at 6 keV, and < 1 micron for all elements at 10 keV.

We did TEM EDS quantification using the Cliff-Lorimer method of k-factors (Cliff and Lorimer 1975) with a thin film thickness correction assuming a 20 degree takeoff angle using software we developed in MATLAB (The MathWorks, Inc. 2011). K-factors were determined from microtomed Springwater olivine, microtomed chromite (UC# 523-9), and ion milled natural diopside (SAL12) characterized using a microprobe Cameca SX 100 at the Centre Commun de Microscopie in Lille, France. For olivine and chromite, these were exactly the same standards used in SEM (albeit prepared for TEM rather than SEM). We either determined thickness corrections with an accuracy of ±20% from STXM measurements of the oxygen K-shell pre-edge at 520 eV, or we assumed a nominal thickness of 100 nm. TEM measurements showed excess SiO₂ from backscattered electrons striking the nearby copper grid and then aerogel. The TEM EDS quantifications were precise to < 4 atom % for Mg and Fe in olivine as opposed to SEM, which were precise to < 2 atom %. SEM and TEM quantification agreed within uncertainties. Accuracy may be expected to be somewhat worse in some cases, especially TEM measurements where the sample had a complex 3D morphology or was positioned such that backscattered electrons could excite neighboring phases. When possible, SEM measurements were quantified using spectra extracted from low voltage, low current EDS maps using regions that showed no elemental variation over a length scale larger than the excitation volume computed using CASINO (Drouin et al. 2007). This practice reduced inaccuracies from grain morphology and secondary fluorescence. Detection limits were typically on the order of 0.1 atom % or better. Quantifications for elements below 1 wt% are statistically limited on the order of 10-50%.

Because of Na volatility in plagioclase and glassy phases, we determined Na concentrations in the TEM using one of three methods. For long duration spectra we fit the elemental compositions against the plagioclase series of anorthite-orthoclase-albite using software we wrote in MATLAB. For these spectra, the Na was allowed to vary to optimize the fit. In a few cases, several shorter spectra (30 and 60 seconds) were taken and Na was extrapolated back to the value at 0 seconds. For some spectra, we report the results from both approaches and show they are equivalent within confidence limits (Table 3). A third approach was to measure the Na content in STXM which did not volatilize Na. Finally, we did not see Na volatilization in the SEM since the beam voltages and currents were low. Therefore, the Na compositions measured by SEM were considered correct. Corrected TEM, raw SEM and raw STXM measurements all agreed within uncertainties. In EDS analyses of plagioclase, minor detections of Mg and Fe could be due to the proximity of neighboring olivines and chromites.

Table 3.

Compositions of glass and plagioclase in Iris by TEM and SEM EDS.

	Normalized Oxide Weight %
Spectrum	SiO2	Al2O3	MgOa	CaO	FeOa	Na2O	K2O	t (nm)	Stat levelb	Na methodc	Crystallinityd
TEM 27	65.94	21.46		4.15		8.45		100	1	Fit	Unknown
TEM 28	68.14	19.74		1.17		10.49	0.47	100	1	Fit	Both
TEM 29	67.50	20.31		1.77		10.42		217	2	Fit	Yes
TEM 30	66.11	19.95		1.74		12.20		217	2	Extrapolated	Yes
TEM 31	64.31	22.39		3.41	0.28	9.61		146	3	Fit	Yes
TEM 32	63.91	22.24		3.38	0.28	10.19		146	3	STXM	Yes
TEM 33	62.59	21.35		5.21		10.65	0.21	150	2	Extrapolated	Yes
SEM 11	62.77	23.06		5.05		8.83	0.30	n/a	3	Measured	Unknown
SEM 12	62.23	22.80	0.45	5.58		8.63	0.32	n/a	3	Measured	Unknown
SEM 13	63.52	22.38	0.33	4.85		8.59	0.32	n/a	3	Measured	Unknown
SEM 14	66.79	23.52	0.83	5.06		3.04	0.76	n/a	3	Measured	Unknown
SEM 15	66.64	17.94	3.29	1.08	0.85	9.60	0.60	n/a	3	Measured	Unknown
	Cations per 4 oxygens
Spectrum	Si	Al	Mg	Ca	Fe	Na	K	Si+Al+Fe	ΣCations	Phase
TEM 27	2.89	1.11		0.19		0.72		4.00	4.91	An21Or00
TEM 28	2.97	1.01		0.05		0.89	0.03	3.98	4.95	An06Or03
TEM 29	2.95	1.05		0.08		0.88		4.00	4.97	An09Or00
TEM 30	2.92	1.04		0.08		1.04		3.96	5.08	An07Or00
TEM 31	2.84	1.16		0.16	0.01	0.82		4.01	4.99	An16Or00
TEM 32	2.83	1.16		0.16	0.01	0.87		4.00	5.03	An16Or00
TEM 33	2.79	1.12		0.25		0.92	0.01	3.91	5.09	An21Or01
SEM 11	2.84	1.23		0.25		0.77	0.02	4.07	5.11	An24Or02
SEM 12	2.77	1.19	0.03	0.27		0.74	0.02	3.96	5.02	An26Or02
SEM 13	2.81	1.17	0.02	0.23		0.74	0.02	3.98	4.98	An23Or02
SEM 14	2.88	1.20	0.05	0.23		0.25	0.04	4.08	4.66	Glass?
SEM 15	2.94	0.93	0.22	0.05		0.82	0.03	3.90	5.02	Glass?

^aDue to proximity of olivines and chromites, minor detections of Fe may be due to secondary fluorescence, backscattered electrons or excitation volume.

^bStatistical uncertainties are bounded by the uncertainty for Si (the most abundant element) and Ca (the least abundant major). They are approximately 1) 1.5% (Si) and 8% (Ca), 2) 1% (Si) and 5% (Ca), 3) < 1% (Si) and 2% (Ca).

^cWe either measured the Na concentration in the SEM under conditions that did not volatilize the Na, extrapolated back to t=0 using a series of acquisitions (TEM), or inferred concentration from a fit against albite-orthoclase-anorthite + SiO₂ (TEM). In one case, we used a measurement from STXM.

^dCrystallinity was not measured (unknown) or was verified with selected area electron diffraction (Yes). In one case, both the presence of crystalline material and amorphous material was verified with brightfield imaging and oxygen XANES

The bulk composition of the particle was computed from SEM EDS maps and spectra using the formula:

$$B_e=\sum _p{C}_{e,p}{A}_p{n}_p$$ (1)

where B_e was the concentration of element e in the bulk measured in At%, C_e,p was the concentration of element e in atomic percent within phase p, A_p was the area (modal abundance) of phase p, n was the atomic density of the phase in atoms/ų. The following atomic densities were used: oligoclase was 0.078 atom/ų, olivine was 0.095 atom/ų, chromite was 0.095 atom/ų, pyroxene was 0.091 atom/ų, and mesostasis was 0.066 atom/ų. We computed the crystalline atomic densities with the CrystalMaker software package (CrystalMaker Software Ltd. 2012) based on unit cells from the American Mineralogist Crystal Structure Database (Downs and Hall-Wallace 2003) or the PDF-4+/Minerals database (International Center For Diffraction Data 2012). We computed the mesostasis atomic density from Fluegel (2007).

Isotope Measurement

The Al-Mg and O isotope systematics in Iris were reported previously in Ogliore et al. (2012). For measurement of O isotopes in Callie olivine, used a similar sample preparation (Westphal et al., 2011) and analytical protocol for the ion microprobe at the University of Hawai’i (Ogliore et al., 2012). Our specially designed mount held the sample at a fixed altitude inside a 1-inch Al round. For Callie, additional set screws allowed sub-degree attitude control for the microtomed surface to minimize topological effects on the isotope measurement. A gold-coated silicon nitride window with a 300-μm ion-milled hole was set over the Callie bullet, which created a flat, conducting surface for the ion probe. A suite of polished isotope standards was set in a circle around Callie, the “buckler.” We used San Carlos olivine as an oxygen isotope standard for Callie (as we did for Iris). No significant instrumental mass fractionation (< 0.5‰) between San Carlos olivine (Fo₈₉) and Iris and Callie olivine (≈ Fo₆₀) was expected (Jogo et al. 2012, Valley and Kita 2009), so we applied no matrix correction to our standardization using San Carlos olivine. We created a standard analog mount to measure any instrumental mass fractionation between the center-mounted grain (similar to Iris and Callie) and the standards in the buckler. For Iris, the center-mounted San Carlos olivine grain was 1.5 per amu lower on the terrestrial fractionation line than the buckler San Carlos olivine. We measured this shift to be the same before and after the measurement of Iris, so we applied this 1.5‰ per amu center-buckler shift to Iris. For Callie, our technique improved with the addition of attitude control, so we prepared a new analogously prepared San Carlos olivine mount. We measured the center-buckler O isotope shift in the new standard mount to be zero, which was also reproducible before and after the measurement. This difference in instrumental mass fractionation between the two analog samples is likely due to topographical variations caused by changes in our sample preparation technique between measurements of Iris and Callie. Measuring contemporaneously prepared standard mounts minimizes the systematic uncertainties associated with mounting, which we showed by obtaining consistent results in a later re-measurement of O isotopes in Iris olivine using the new attitude adjustment mount (Ogliore et al. 2014).

We measured three oxygen isotopes in Callie using the University of Hawai’i Cameca ims 1280 ion microprobe in multicollection mode. A Faraday cup was used to measure ¹⁶O, with ¹⁷O and ¹⁸O measured on electron multipliers. A 25–30 pA Cs⁺ primary ion beam was focused to ≈2 μm to allow for single-grain measurements of separate olivine grains in Callie. The data were corrected for deadtime, background, detector yield, and interference from ¹⁶OH⁻ on ¹⁷O (typically <0.2‰). We measured a total of 51 spots of the San Carlos olivine standard. These standard measurements showed variations in δ¹⁷O and δ¹⁸O about 1.75 times as much as would be expected by their statistical uncertainties, so this systematic uncertainty factor was used to calculate the total uncertainty of the Callie measurements.

RESULTS

Iris contains olivine, chromite, plagioclase, glass, and augite. The olivine ranges from Fo_60-65 with minor Mn, Ca and Cr. Chromite is associated with the olivine and in some cases poikilitically enclosed suggesting coeval evolution. High-Ca pyroxene neighbors olivine. Plagioclase compositions range from An_6-26. Albitic glass is present. An X-Ray Fluorescence (XRF) overview of Iris can been seen in Fig. 1B taken at the ALS. Red shows the presence of Fe tracking olivine and spinel, green shows Cr concentrated in spinels, and blue shows Ni. Fig. 1C shows an optical luminance-red-green-blue (LRGB, Levay et al. 2007) image generated from the combination of transmission optical and cross polarized optical images showing the location of several crystals with different orientations. Fig. 2 shows two different microtomed surfaces of the particle and the porphyritic nature of the grain. The glassy/plagioclase matrix surrounds the euhedral olivines and chromites. Due to aerogel capture at high speed, the periphery of the particle has been ablated away and a thin rim of melted and resolidified material is present. By analogy to the heated exteriors on meteorites we call this a fusion crust: a thin region of altered material produced primarily from the impactor’s mass, but modified by the deceleration in a viscous medium (here, aerogel instead of air).

Figure 2.

A) 15 keV backscattered electron (BSE) image of the Iris bullet remaining after ultramicrotoming. Euhedral olivines (Ol), and a skeletal chromite (Chr) are visible in a plagioclase/mesostasis (M) groundmass. The fusion crust (F) is the shell of melted and altered material due to aerogel capture. B) 5 keV BSE image of the Iris bullet after additional ultramicrotomy. The bright rim (In) is an In, Pb alloy introduced during sample prep for isotopic measurement in the Cameca ims 1280 SIMS.

Callie contains at least four phases: olivine, sodium plagioclase, glass, and an optically opaque, Cr-rich phase that is likely chromite (Figs. 1E and 3). The olivine ranges from Fo_59–64 with minor Mn, Ca and Cr. Olivine and plagioclase were verified with electron diffraction. Fig. 1D shows an XRF image of Callie taken at the European Synchrotron Radiation Facility (ESRF) showing the concentration of Ga, Cr, and Mn. Fig. 1E shows an optical image of Callie produced using Helicon Focus (Helicon Soft Ltd., 2012) showing the presence of opaque phases some of which match the Cr-rich points in the XRF map. Figure 3 shows an SEM/EDS view of the bullet with olivine, plagioclase and glass phases visible.

Figure 3.

A) A 5 keV BSE image of a Callie bullet after ultramicrotoming. Olivine and mesostasis/plagioclase are visible. B) 5 keV EDS map showing Na (red) Al (green) and Mg (blue). The olivine shows blue, the green region at the top is amorphous . C) 5 keV EDS map showing O (red) Mg (green) and Si (blue).

Olivine

Olivine is the modally dominant phase in Iris and Callie and is nearly equilibrated. It varies over Fo_60-65 in Iris, and Fo₅g_₆₄ in Callie. The largest olivine grain in Iris has an apparent size of 6 × 11 μm (Fig. 4) and contains nanophase chromite inclusions. Contact zones between olivine and chromite as well as the chromite inclusions in olivine suggest prior crystallization of spinel or cocrystallization. The typical olivine grain is ≈ 2 μm in diameter (Figs. 4 and 5). Figure 6 shows confirmation of the olivine structure by electron diffraction.

Figure 4.

The Iris bullet after microtoming below the surface shown in Figure 2B. A) 15 keV BSE image. B) EDS map showing Si (red), Mg (green) and Ca (blue). C) EDS map showing Al (red) Na (green) and Mg (blue). Olivine shows as green in B. The middle of the particle is plagioclase/mesostasis. Na is only present with Al (C-yellow) but sometimes Al is present with little Na (C-red) showing that the plagioclase/mesostasis in the middle is inhomogenous. A triangular chromite near the bottom of A is marked with an arrow and can be seen in C as red. The Cr-rich inclusion in the largest olivine is likely also a chromite. A second arrow in the upper left marks a clinopyroxene. The bright rim seen around the periphery in A is an In, Pb alloy introduced during sample preparation for isotopic measurement.

Figure 5.

A) A 10 keV BSE image of an Iris bullet after ultramicrotoming beneath the bullet in Fig. 4. B) 10 keV EDS map showing Al (red) Na (green) and Mg (blue). C) 10 keV EDS map showing Cr (red) Mg (green) and Fe (blue). Olivine, chromite, and mesostasis/plagioclase are visible. An olivine has a chromite inclusion near the bottom. Fig. 8 provides a zoom in of this inclusion. As in Fig. 4, the mesostasis was inhomogenous, and the chromites show brightly in the Al and Fe channels (B-red and C-red). The bright rim around the periphery is an In, Pb alloy introduced for isotopic measurement.

Figure 6.

STEM HAADF image of an Iris olivine (center) bound to amorphous material marked with dotted lines. Some amorphous material overlaps the grain. The diffraction pattern (upper right) is down the 310 zone axis. The direct beam is obscured by a beam block.

Olivine compositions for Iris and Callie measured by TEM and SEM are shown in Table 1. CaO ranges from 0.15 to 0.60 oxide wt%, MnO ranges from 0.38 to 0.90 oxide Wt %, and Cr₂O₃ ranges from below detection to 0.22 oxide wt%. The compositions are similar between and within Iris and Callie.

Table 1.

Olivine compositions from SEM and TEM EDS.

	Normalized Oxide Weight %									Cations per 4 oxygens
ID	SiO2	TiO2	Al2O3	Cr2O3	MgO	CaO	MnO	FeO	Na2O	Si	Ti	Al	Cr	Mg	Ca	Mn	Fe	Na	Fo#	t (nm)
Iris TEM 01	37.78				29.44	0.43	0.60	31.75		1.028				1.194	0.013	0.014	0.723		62.3	100a
Iris TEM 02	36.35				33.03		0.86	29.76		0.985				1.334		0.020	0.675		66.4	100
Iris TEM 03	37.79				30.40	0.60	0.49	30.72		1.032				1.237	0.018	0.011	0.702		63.8	100
Iris TEM 04	36.94				30.95		0.61	31.50		1.006				1.256		0.014	0.718		63.6	100
Iris TEM 05	38.24				28.39	0.38	0.66	32.33		1.042				1.153	0.011	0.015	0.737		61.0	100
Iris TEM 06	37.50				30.63		0.64	31.23		1.019				1.239		0.015	0.709		63.6	100
Iris TEM 07	37.18			0.19	28.63	0.40	0.51	33.09		1.020			0.004	1.170	0.012	0.012	0.759		60.7	100
Iris TEM 08	41.91		2.86		29.38		0.50	24.84	0.51	1.086		0.087		1.134		0.011	0.538	0.025	67.8	100
Iris TEM 09	44.59			0.17	27.41	0.24	0.53	27.06		1.162			0.003	1.064	0.007	0.012	0.589		64.4	150
Iris TEM 10	39.77			0.20	28.42	0.31	0.67	30.63		1.070			0.004	1.140	0.009	0.015	0.689		62.3	150
Iris TEM 11	36.88			0.13	30.11	0.26	0.66	31.96		1.008			0.003	1.226	0.008	0.015	0.731		62.7	150
Iris TEM 12	38.34		0.11	0.10	30.53	0.19	0.64	30.09		1.033		0.003	0.002	1.226	0.006	0.015	0.678		64.4	150
Iris TEM 13	39.89			0.16	28.20	0.19	0.58	30.98		1.073			0.003	1.131	0.006	0.013	0.697		61.9	150
Iris TEM 14	45.88	0.11	0.65	0.10	26.48	0.17	0.44	26.17		1.184	0.002	0.020	0.002	1.018	0.005	0.010	0.565		64.3	75
Iris TEM 15	39.40		0.48		29.25		0.64	30.23		1.057		0.015	0.000	1.169		0.015	0.678		63.3	120
Iris TEM 16	40.64				28.43	0.19	0.58	30.16		1.087				1.133	0.006	0.013	0.675		62.7	150
Iris TEM 17	47.70				26.10	0.17	0.38	25.65		1.221				0.995	0.005	0.008	0.549		64.4	120
Iris TEM 18	45.64				27.04	0.25	0.59	26.48		1.182				1.043	0.007	0.013	0.573		64.5	130
Iris TEM 19	44.88			0.13	27.50	0.20	0.59	26.70		1.166			0.003	1.065	0.005	0.013	0.580		64.7	150
Iris TEM 20	39.14				27.75	0.41	0.69	32.01		1.062				1.122	0.012	0.016	0.726		60.7	90
Iris TEM 21	39.79				30.20	0.15	0.58	29.28		1.063				1.202	0.004	0.013	0.654		64.8	130
Iris SEM 01	36.03				30.10	0.45	0.62	32.80		1.013				1.260	0.013	0.015	0.771		62.0	n/ab
Iris SEM 02	35.86				29.22	0.30	0.89	33.73		1.015				1.232	0.009	0.021	0.798		60.7	n/a
Iris SEM 03	36.14				30.16	0.42	0.56	32.72		1.027				1.277	0.013	0.014	0.778		62.2	n/a
Iris SEM 04	36.47				31.70	0.38	0.85	30.60		0.999				1.294	0.011	0.020	0.701		64.9	n/a
Iris SEM 05	37.21				32.33			30.46		1.024				1.326			0.701		65.4	n/a
Iris SEM 06	35.49				30.48	0.33	0.90	32.80		0.979				1.253	0.010	0.021	0.757		62.3	n/a
Iris SEM 07	35.58			0.19	29.78	0.28	0.51	33.66		0.981			0.004	1.224	0.008	0.012	0.776		61.2	n/a
Iris SEM 08	35.86				31.10	0.24		32.80		0.984				1.272	0.007		0.753		62.8	n/a
Iris SEM 09	36.32				29.99	0.31	0.60	32.78		0.996				1.226	0.009	0.014	0.752		62.0	n/a
Callie TEM 22	36.59			0.22	28.00	0.35	0.66	34.18		1.012			0.005	1.153	0.010	0.015	0.790		59.3	100
Callie TEM 23	36.62			0.13	30.87	0.31	0.82	31.25		0.999			0.003	1.255	0.009	0.019	0.713		63.8	100

^aWe used a thin film correction based on STXM thickness (t) measurements. Where no STXM measurement was available, we assumed a nominal 100 nm correction.

^bWe used the XPP matrix correction on SEM measurements.

We found dislocations in the olivine in both Iris and Callie (Fig. 7). They showed high contrast in dark field imaging with diffraction vector g = 002 and no contrast with g = 200 which suggested a Burgers vector c. The orientation of the lines was parallel to the c axis of the crystal and they were interpreted as screw defects. We found a maximum of about 200 defects/pm, with an average defect length of about 60 nm. Screw defects have been reported for other Wild 2 olivine samples (Tomeoka et al. 2008, Jacob et al. 2009) and are strongly dominant in shock-deformed olivine (e.g., Leroux 2001, and references therein). For comparison, the Tenham meteorite (shock stage 4-5) shows defect densities up to 200 defects/pm (Langenhorst et al. 1995), though many of the defects are longer than 60 nm. On the other hand, Stodolna et al. (2012b) found similar screw defect densities in unshocked olivine that had been shot by light gas gun into aerogel and then microtomed. Therefore, we cannot ascribe them unambiguously to a shock origin.

Figure 7.

TEM dark field image of screw defects in Iris olivine. Two defects are indicated with small arrows. They appear as lines parallel to the c axis with a defect density ≈ 200 defects/μm and an average length ≈ 60 nm. Several defects escape the volume of the shard. The scalebar is 100 nm.

Chromite

Chromites in Iris often shared a compact interface with one or more olivine grains (Figs. 2, 4, 5). They ranged in size from nanocrystalline to about 2 μm in diameter. Fig. 8 shows a euhedral chromite with several corners including one at 118°. Ultramicrotomy has chipped the sample, but the shape remains intact on the lower left side. The fact that is is enveloped by olivine suggests that it formed before the olivine. At 1 micron in diameter it is too large to be a result of exsolution. Fig. 9 shows a TEM image and electron diffraction pattern of a chromite. The smearing of the spots in the diffraction pattern is due to a high defect density in the chromite. We have not characterized the nature or origin of the defect structure.

Figure 8.

A) BSE image of a chromite inclusion (upper left) in olivine in Iris. B) 6 keV EDS map showing Al (red) Si (green) Mg (blue). The chromite is red, the olivine is blue, glass/plagioclase is yellow and aerogel is green.

Figure 9.

Low-angle annular dark field (LAADF) image of an Iris chromite showing diffraction contrast. The diffraction pattern is near the 110 zone axis and shows a significant spreading of the reflections indicating an imperfect crystal. The 002 reflections are dynamical diffraction since the crystal is > 150 nm thick.

In Callie, we saw an opaque phase via optical imaging (Fig. 1E). Using XRF (Fig. 1D) we found that same phase to be very roughly 20 At% Cr. With XRD we found d-spacings consistent with spinel, so the grain was likely chromite.

The chromite compositions are shown in Table 2. Some TEM measurements showed high SiO₂ due to excitation of nearby SiO₂ aerogel, and in these cases we artificially set the Si content to zero which is unlikely to affect any geochemical results.

Table 2.

Iris chromite compositions from TEM and SEM EDS.

	Normalized Oxide Weight %
Spectrum	SiO2	TiO2	Al2O3	V2O3	Cr2O3	Fe2O3	MgO	MnO	FeO	t (nm)
TEM 24	0.00a	1.80	17.90		43.99	2.85	5.69		27.78	100b
TEM 25	0.00a	1.19	19.03	0.48	45.04	2.18	6.15	0.81	25.13	260
TEM 26	1.48	2.48	14.94		47.10	0	5.90		28.10	150
SEM 10	0.73	1.91	14.12	0.57	45.80	3.93	5.75	0.76	26.43	n/a.c
	Cations per 4 oxygens
Spectrum	Si	Ti	Al	V	Cr	Fe3+	Mg	Mn	Fe2+	Fe3+/ΣFe
TEM 24	0.000	0.045	0.695		1.146	0.071	0.279		0.765	0.08
TEM 25	0.000	0.029	0.734	0.006	1.166	0.054	0.300	0.022	0.688	0.07
TEM 26	0.066	0.083	0.789		1.668	0.000	0.394		1.053	0.00
SEM 10	0.024	0.048	0.558	0.008	1.213	0.099	0.287	0.022	0.741	0.12

^aSiO₂ contribution removed. If some glass was present in the spectrum, this could influence other elements.

^bWe used a thin film correction based on STXM thickness (t) measurements. Where no STXM measurement was available, we assumed a nominal 100 nm correction.

^cWe used the XPP matrix correction on SEM measurements.

The concentration of the minor elements V and Mn (Table 2, TEM 25) were determined using a combined TEM EDS + STXM measurement as described in Gainsforth et al. (2010) because interference between Ti - V and Cr - Mn prevented accurate quantification by EDS alone.

We also acquired an Fe L-edge XANES spectrum of the chromite using STXM (see Fig. 10). The spectrum was recorded in 200 meV steps across the fine structure. We identified at least 7 multiplets within the Fe L_2,3 edges, Several of these were cleanly resolved. Fe L-edge XANES is not yet reliably calibrated in the chromite system, but Taftø and Krivanek (1982) used the ALCHEMI technique in TEM/EELS to obtain the Fe L-edge spectra for tetrahedrally and octahedrally coordinated Fe in a single chromite. They found that tetrahedrally coordinated Fe in chromite had a stronger peak at 708 eV than at 710 eV, while for the octahedral Fe this was reversed. Our spectrum clearly favored tetrahedrally coordinated Fe. We also estimated the oxidation state of the iron from the chromite stoichiometry and found Fe³⁺/ΣFe ≤ 12% (Table 2). Thus, we have mostly Fe²⁺, tetrahedrally coordinated.

Figure 10.

Fe-L_2,3 XANES spectrum from the Iris chromite seen in Fig. 8 measured by STXM. The composition of this chromite is in Table 2 TEM 25. Based on chemical stoichiometry, the iron oxidation state is Fe³⁺/ΣFe ≤ 12%. The L₃ edge at 708 eV shows at least 7 multiplets (unique peaks), four of which are well resolved.

Plagioclase

We found plagioclase ranging in composition from An_6–26 in Iris. Oligoclase was one of two phases used for ²⁶Al radioisotope dating, the other being an SiO₂ rich glass, as discussed in Ogliore et al. (2012). Crystallinity was verified with TEM diffraction (Fig. 11). Not all regions with plagioclase composition were crystalline, however. Callie also had albitic plagioclase.

Figure 11.

STEM HAADF image of an ultramicrotomed portion of Iris. Inset shows diffraction from a shard of oligoclase down the 001 zone axis. The diffraction pattern was obtained with a 3 μradian convergence angle and a 10 eV energy filter centered around the zero loss peak. The direct beam is in the exact center of the inset, eclipsed by the beam block.

Na volatilized rapidly in the TEM from some regions with plagioclase composition. The fact that Na volatilized at different rates hinted at a range of crystallinities (Jeanloz and Ahrens 1976).

Fig. 12 shows a 2D-XRD pattern taken at 8 keV using beamline 12.3.2 at the ALS. It shows the average of several images through the center of the Iris grain and shows multiple phases. The feature marked A had d-spacings of 5.6 Å and 5.8 Å consistent with plagioclase and inconsistent with all the other phases we saw in Iris. It showed clear mosaicity spanning 4° in azimuth. The feature marked B also showed 4° mosaicity and had d-spacings consistent with plagioclase (2.1 Å and 2.2 Å). The feature marked C was consistent with olivine without mosaicity. From this we concluded that the plagioclase experienced significant shock or was not well crystallized. The latter is consistent with a glassy mesostasis which had partially crystallized into plagioclase. There is no clear evidence of shock from the SEM and TEM images (see Figs. 2, 4, 5, and 7). In addition, the SEM EDS maps showed variable composition for the plagioclase which should not occur from shock alone (Jeanloz and Ahrens, 1976). The mosaicity is likely related to the crystallization.

Figure 12.

XRD image taken at 8 keV of the middle region of Iris. A) Mosaiced reflections at 5.7 Ådue to plagioclase. B) Another plagioclase reflection. C) A single crystal reflection due to olivine. The plagioclase peaks show a 4° mosaicity. The direct beam is off the figure to the right.

Glass

Iris and Callie contained an amorphous phase (Fig. 13) which we differentiated from plagioclase by TEM diffraction. In the SEM EDS analyses we observed it as regions of a non-stoichiometric phase in proximity to plagioclase (Figs. 4 and 5). The composition of the glass phase was very similar to the neighboring plagioclase except for excess SiO₂ (Table 3).

Figure 13.

A) Brightfield TEM image of an interface between oligoclase (Plag) and glass (Gl). The scale bar is 500 nm. B) O-K XANES spectra acquired using STXM. Vertical lines denote the energies 537.4 and 540.5 eV. Oligoclase and glass are the materials shown in A. Fusion (for fusion crust) is Iris mixed with aerogel from capture. Aerogel outside the fusion crust also contains epoxy. C) A zoomin of the edge of the O-K XANES spectrum shows that the oligoclase and glass have the same edge energy, while the aerogel and fusion crust have an edge energy ≈ 0.5 eV higher.

To distinguish native glass from SiO₂ introduced by aerogel capture we combined TEM and XANES. Fig. 13A shows a brightfield TEM image of an interface between oligoclase and glass. Fig. 13B shows oxygen XANES spectra of the oligoclase, glass, and surrounding aerogel and fusion crust (aerogel mixed with comet and altered by capture) for comparison (Fig. 13B). Fig. 13C shows that oligoclase and glass have the same edge position which is separated from the aerogel and fusion crust by ≈ 0.5 eV.

Callie consists of olivine crystals with plagioclase and glass sandwiched within (Fig. 3). Compare this with Fig. 14 which shows an ultramicrotomed slice of Callie with the glass delineated with a dotted line. Fig. 1D shows a synchrotron XRF image of Callie with Ga (red), Cr (green) and Mn (blue). The Ga trace follows the glass.

Figure 14.

TEM brightfield of a microtomed slice of Callie. The dotted region outlines primarily amorphous material while the rest is crystalline, mostly olivine. Compare with the green region in Fig. 3B, and the purple region in Figure 3C.

Finally, NanoFTIR was applied to the glass/plagioclase interfaces and also found interwoven structures of crystalline and amorphous material of varying composition (Dominguez et al. 2014).

Clinopyroxene

One pyroxene was analyzed from Iris (Fig. 15). The compositions measured by TEM and SEM are given in Table 4. Selected area electron diffraction patterns along the $\stackrel{‒}{1}3\stackrel{‒}{1}$ and 130 zone axes match the C2/c spacegroup expected from high calcium pyroxene. SEM EDS analysis yields the composition En₃₉Fs₁₂Wo₃₈Ko₀₈Ja₀₂ (Table 4). TEM observations indicate that the crystal was monocrystalline and did not contain a high defect density. The composition was enriched in Na and Cr and slightly less enriched in Al and Ti. Stoichiometry calculations show that coupled substitutions involving Na⁺, Cr³⁺, Al³⁺ and Ti⁴⁺ can balance the charge, and that the largest substitution would be a kosmochlor (NaCrSi₂O₆) component. Such compositions are not uncommon in Stardust. Joswiak et al. (2009) reported a large number of kosmochloric pyroxenes in assemblages which bear many similarities to Iris.

Figure 15.

HAADF image of the Na-,Cr-rich clinopyroxene from Iris, and a selected area electron diffraction pattern along the $\stackrel{‒}{1}3\stackrel{‒}{1}$ zone axis. The increased brightness in the HAADF image at the top of the grain is due to overlap with an SiO₂ grain.

Table 4.

Iris clinopyroxene compositions from TEM and SEM EDS.

	Normalized oxide weight %
Spectrum	SiO2	TiO2	Al2O3	Cr2O3	MgO	CaO	FeO	Na2Oa	Phaseb
TEM 34	55.66	0.65	0.55	2.96	13.01	18.79	6.27	2.12	En36Fs10Wo38Ko11Ja05
SEM 16	53.45	0.68	0.39	2.64	14.22	19.12	7.96	1.55	En39Fs12Wo38Ko08Ja02
	Cations per 6 oxygens
Spectrum	Si	Ti	Al	Cr	Mg	Ca	Fe	Na	ΣCations	t (nm)
TEM 34	2.041	0.018	0.024	0.086	0.711	0.738	0.192	0.151	3.961	153
SEM 16	1.985	0.019	0.017	0.077	0.787	0.761	0.246	0.111	4.003	XPP

^aNa concentration extrapolated to t=0 from a sequence of spectra.

^bEn=enstatite, Fs=ferrosilite, Wo=wollastonite, Ko=kosmochlor, Ja=jadeite.

Bulk Composition

To measure the bulk composition of Iris, we acquired SEM backscatter and EDS information from four bullets after each major ultramicrotomy session. In the case of the two bullets shown in Fig. 2, we calculated the modal abundances of the phases from the backscatter images and used compositions from point spectra in each phase. For the two bullets shown in Figs. 4 and 5, we acquired EDS maps and determined modal abundances from phase maps using the Oxford Inca software. The final concentrations are displayed in Table 5.

Table 5.

Bulk composition for Iris derived from EDS data ultramicrotomed bullets shown in Figs. 2, 4, and 5.

At %	O	Na	Mg	Al	Si	K	Ca	Ti	V	Cr	Mn	Fe	Sum	Area (μm)
Fig. 2	58.19	2.12	10.99	4.17	18.25		0.73	0.04	0.02	1.24	0.03	6.84	102.62	232
Fig. 4	57.91	1.16	13.07	2.47	15.73	0.03	0.83	0.04	0.01	0.76	0.19	7.98	100.18	114
Fig. 5	58.06	0.98	13.67	2.15	15.21	0.04	0.46	0.02	0.01	0.60	0.18	8.64	100.01	142
Weighted mean of all bullets	58.09	1.56	12.26	3.19	16.78	0.02	0.67	0.03	0.01	0.94	0.11	7.63	101.29
Weighted Stdev (2 DOF)	0.14	0.65	1.50	1.16	1.73	0.03	0.18	0.01	0.01	0.35	0.09	0.97
Oxide Wt%	-	Na2O	MgO	Al2O3	SiO2	K2O	CaO	TiO2	V2O3	Cr2O3	MnO	FeO	Sum
ΣAll bullets	-	2.03	20.74	6.83	42.32	0.04	1.58	0.10	0.03	3.00	0.33	23.01	100.01

The bulk compositions computed from only the first two microtomed surfaces were compared with the bulk composition using three and four microtomed surfaces. The similarity between the “bulk composition” from each of the surfaces and the sum of them can be seen in Table 5 and implies that the particle is fine-grained on the scale of the measurements and that the bulk composition is probably a reasonable estimate of its true bulk composition. If Iris was a fragment of a larger object, these values should clearly not be interpreted as the bulk composition for the source object, however.

Fine Grained Material

Fine grained primitive material was located next to Iris in microtomed sections (Stodolna et al. 2014). This material contained glass, euhedral nanosulfides, and an enstatite whisker, and is reported elsewhere. Due to the primitive nature of the material, it bears no obvious relationship to Iris, and is likely other cometary material from Wild 2 that was shielded by Iris during capture in much the same fashion as the fine grained material behind the particle Febo (Brownlee et al. 2006).

Isotopes

Ogliore et al. (2012) measured the Al-Mg and O isotope systematics in Iris and did not find evidence for the former presence of ²⁶Al in crystalline oligoclase and amorphous mesostasis from a microtomed grid of Iris. The 2σ upper bound on the initial ²⁶Al/²⁷Al in Iris was found to be 3.0× 10⁻⁶. This implies Iris formed at least 3 Myr after the onset of CAI formation, assuming the ²⁶Al was homogeneously distributed throughout the solar system. The mean O isotopic composition of Iris olivine is similar to terrestrial values: δ¹⁷O=4.6±2.3‰, δ¹⁸O=7.2±1.6‰ (2σ errors). The plagioclase and spinel isotopes plot within 2 of each other, which is within 2σ given the uncertainty of the measurement. Therefore, there is no indication of oxygen fractionation within Iris. Iris is not ¹⁶O-rich and apparently formed from a relatively evolved oxygen isotopic reservoir.

We measured two spots in the largest olivine grain in Callie, and one spot in the next largest olivine grain. All three spots were consistent with each other, giving weighted-mean values of δ¹⁷O=0.7±1.7‰, δ¹⁸O=2.4±1.3‰ (2σ errors), see Table 6. The O compositions of Iris and Callie are shown in comparison to other relevant Stardust olivines in Fig. 16.

Table 6.

Oxygen isotopes for Callie Olivine

Spot	δ18O‰	δ17O‰
Grain 1 Spot 1	3.1 ± 2.2a	0.8 ± 2.9
Grain 1 Spot 2	2.2 ± 2.2	−0.1 ± 2.8
Grain 2	2.0 ± 2.3	1.4 ± 3.0
Weighted Mean	2.4 ± 1.3	0.7 ± 1.7

^aUncertainties are 2σ.

Figure 16.

Oxygen isotopic compositions of olivines in eight comet Wild 2 fragments from five different Stardust tracks: Callie and Iris from Track C2052,12,74 (this work); Torajiro from Track C2054,0,35 (Nakamura et al. 2008a); Gozen-sama from Track C2081,1,108 (Nakamura et al. 2008a); F1 and F4 from Track C2009,20,77 (Nakashima et al. 2012); F7 from Track C2115,24,22 (Nakashima et al. 2012); P0 from Track C2115,1,22 (Mckeegan et al. 2006). The terrestrial fractionation line (TF), carbonaceous chondrite anhydrous mineral line (CCAM, Clayton 2003), and Young & Russell line (Y&R, Young and Russell 1998) are shown for reference. All error bars are 2σ.

Iris and Callie are fractionated from each other along the terrestrial fractionation (TF) line, separated by about 5‰, which is a difference of ≈ 3σ. Iris was measured twice with each measurement separated by one year. The sample was re-microtomed and a fresh olivine standard was used in each case. The second measurement used the newer attitude control mount but resulted in identical numbers within errors (Ogliore et al. 2014). Therefore, we are fairly confident in the reproducibility of our measurement. However, instrumental fractionations along the TF line are notoriously hard to manage, so while we make a tentative conclusion that the oxygen isotopes between Iris and Callie are 3σ different, we will also entertain the possibility that they are the same in subsequent discussion.

DISCUSSION

Different meteorite classes have different distributions of ferromagnesian content in chondrules and it is common to see a bimodal distribution. For example, CO chondrites contain many forsteritic (Type I) chondrules, only a few with ferromagnesian contents near Fo₁₅, but many fayalitic (Type II) chondrules near Fo₃₅ (Brearley and Jones 1998). Type II ferromagnesian chondrules, which contain Fe-rich olivines and pyroxenes, formed in more oxidizing conditions than counterpart type I chondrules reflecting their origin in higher oxygen fugacity environments. High oxygen fugacity can be created in a chondrule forming process, for example, by melting of H₂O ice (Connolly and Huss 2010), which would suggest that type II formation is sampling a region and time near or in the presence of icy bodies (the snow line, Ciesla and Cuzzi 2006). Oxygen isotope measurements of type II chondrules often show heavier O than Type I chondrules (Connolly and Huss 2010, Tenner et al. 2013). In some models, production of heavy oxygen is connected to the presence of water and may even be a tracer for water in the early solar system (Connolly and Huss 2010). ²⁶Al isotopic measurements show that type II chondrules typically formed contemporaneously or later than type I chondrules (Ogliore et al. 2012, Kurahashi et al. 2008, Nagashima et al. 2008). Thus, type II chondrules track the flow of oxygen and possibly water through the early solar nebula.

Na-rich mesostasis and plagioclase are often found in type II chondrules (Jones 1990a, Lauretta et al. 2006). Alexander et al. (2008) showed that many such chondrules formed in a region where chondrules were in chemical equilibrium with their ambient environment in a closed or nearly closed system implying a much higher dust/gas enrichment then type I chondrules (Hewins 1997, Fedkin and Grossman 2013). This is, perhaps, unexpected since nebular gas densities should thin with time. Resolution of this mystery may be related to the formation of the plan-etesimals themselves, if these high densities are produced by the planetesimals in the form of shock waves (Ciesla et al. 2004).

A compilation of grains observed in 16 aerogel tracks suggests that comet Wild 2 contains a greater population of chondrule objects, especially type II chondrule objects, than other igneous refractory assemblages (Joswiak et al. 2012). Therefore, comet Wild 2 is giving us a unique measure of solar system history a few million years into its formation, perhaps during the time when planetesimals are forming. Understanding the characteristics of the chondrule forming process which produced these type II fragments should answer several interesting questions. 1) What is the process by which Wild 2 chondrules formed? 2) Is it similar or identical to processes forming meteoritic chondrules? 3) Is there evidence of nebular or parent body processes between chondrule formation and incorporation into the comet?

The first question, at least in part, can be answered by constraining fugacities, peak temperatures, cooling rates, and dust enrichments. There have been no studies to date which constrain all of these factors for a single object in Wild 2, but in the case of Iris, it should be possible. Here we use geochemical modeling to constrain these factors. We address the second question by comparison to the meteoritic literature below. The third question is related to the first, and requires contemplation of the equilibrium and disequilibirum expected in a 20 μm chondrule/chondrule fragment.

Thermodynamic equilibrium is an idealized concept of a system in a steady state. Of course, no physical system ever achieves perfect equilibrium, but energy gradients can be small enough to allow calculations based on equilibrium to adequately describe an observed system. In many meteoritic type II chondrules, chromite grains are in close association with olivine (Jones 1990b, Johnson and Prinz 1991). Johnson and Prinz (1991) first showed that element partitions between the olivine and chromite could signal the presence or lack of metamorphism in the parent body containing that chondrule. During chondrule formation, rapid cooling freezes in cation concentrations which have partitioned at high temperature. Nebular reprocessing or metamorphism repartitions elements in response to new chemical gradients. At high temperatures ≥ 1000 °C, diffusion becomes so rapid that olivine and chromite could be close to equilibrium during the chondrule forming process for all but the most rapidly cooled objects, and modeling may be justified on the assumption of a closed system and thermodynamic equilibrium. The validity of this model would be dependent on the diffusion time scale and the cooling rate, and we quantify it for our system below.

In the case of zoned olivines or glass, thermodynamic equilibrium would be a poor approximation. The finer olivines in many chondrules from primitive meteorites (3.0) can have fairly uniform composition on the scale of several microns, even near zoned olivines, relict grains and poorly crystallized minerals. For example, figure 4c from Jones (1990b) shows a highly unequilibrated chondrule from Semarkona with extreme zoning and a relict crystal. Figure 5a from the same paper shows a zoning profile measured across the relict crystal (their Figure 4c) showing a difference from Fo₉₀ to Fo₈₀. Assuming the Fo_# is linear with backscatter intensity, then smaller crystals from the same image show smaller gradients. One grain, 10 microns across, shows zoning of 5 Fo_# and another shows no zoning at all. Similar results are easily obtained quantitatively in an SEM, without assumptions about the linearity of published images. This can be understood as a situation where the diffusion length scale is greater than the diameter of the crystal. Depending on the length scales involved, the difference in composition can be small enough to produce useful equilibrium calculation results – for example, with the Johnson and Prinz (1991) geothermometer. It is important to remember that absence of zoning does not imply equilibrium.

Thermodynamic Modeling

At present, the most comprehensive thermodynamic equilibrium model is MELTS (Ghiorso and Sack 1995, Asimow and Ghiorso 1998) which should apply to the refractory components of Iris if they formed in equilibrium within a closed system. We define equilibrium to mean that phases in Iris interdiffused elements sufficiently rapidly to produce compositional gradients were invisible given EDS error bars. We simulated such equilibria using alphaMELTS 1.3 (previously Adiabat_1ph, Smith and Asimow 2005) controlled by MATLAB (The MathWorks, Inc. 2011) and Python (Oliphant 2007, Van Der Walt et al. 2011, Hunter 2007, Perez and Granger 2007). We assumed a pressure of one bar. Pressures < 1 bar should be well represented by computations at 1 bar except for the vapor pressures of volatile elements – which are accomodated by a closed system assumption (Fedkin and Grossman 2013, Alexander et al. 2008).

Type II chondrules formed near the IW buffer (Hewins 1997, Schrader et al. 2012). Therefore, we modeled fugacities from 3 log units below to 2.75 log units above IW (log(f_O2) =−10.3 to −16.1 at 1100 °C). For each fugacity, we computed equilibrium compositions over a temperature range from 1500 °C to 800 °C.At high temperatures (≈1300 °C), equilibrium could be achieved in a few seconds, while at low temperatures (≈500 °C) equilibrium may not occur for millennia. Somewhere between is the point at which the cooling rate of the object matches the equilibrium timescale, and the object falls out of equilibrium.

We quantified the match between each MELTS simulation and Iris by minimizing a fit index over f_O2 and T:

$$F_j(f_{O_2},T)=\sum _i\frac{|C_{i,j}^{\mathrm{mod}el}(f_{O_2},T)-C_{i,j}^{meas}|}{C_{i,j}^{meas}}$$ (2)

where f_O2 and T are oxygen fugacity in bars and temperature in °C for the MELTS calculation. $C_{i,j}^{model}$ is the concentration in oxide weight % for an oxide i in mineral j as calculated by MELTS. $C_{i,j}^{meas}$ is the concentration as determined from SEM EDS. The fit index was calculated for olivine using MgO, FeO, CaO and MnO, for chromite using MgO, FeO, Al₂O₃, Cr₂O₃, and TiO₂, and for feldspar using Na₂O, Al₂O₃, and CaO.

Some fit indexes are shown in Fig. 17 for olivine, spinel and feldspar near the IW buffer. The minima in the fit indexes indicate the temperatures that yield the best fits for the experimental mineral compositions. In this case, olivine and spinel show poor fits to high temperature compositions, but a close fit near about 1000 °C at which point the fit quality degrades again. One can use the average of the two fit indices to determine the temperature at which the olivine/spinel system was closest to equilibrium. In this simulation, feldspar did not begin to crystallize until near 1100 °C, so there is no fit index for higher temperatures. The fact that the feldspar minimum occurs near 900 °C indicates that its closest approach to equilibrium was at a lower temperature than for olivine/spinel.

Figure 17.

Fit index (Equation 2) for olivine, spinel and feldspar as a function of temperature with fugacity held at IW-0.25 log units. The minimum values represent best fits.

The main sources of uncertainty in our computations are the estimation of the bulk composition (Table. 5), the accuracy with which we determine the phase compositions for comparison against the MELTS predictions, and the uncertainty of the MELTS model itself. The latter is difficult to quantify, but as MELTS is a self-consistent model (Ghiorso and Sack 1995), we should expect our predictions to compare favorably to MELTS predictions by other researchers. The confidence limits for the bulk composition are taken from the standard deviation of the three separate bulk composition measurements which we assume to follow a normal distribution (Table 5). We built a monte carlo simulation using Python that varied the abundance of each element in accordance with a normal distribution. Each of these random compositions was fed through the MELTS model and inspected to locate the best fit temperatures for olivine and spinel – chosen to be the miniumum fit index value. Compositions that differed dramatically from the bulk composition also produced systematically poorer fits. To accomodate this correlation, we applied a weight to each monte carlo trial of 1/F(f_O2, T), where F is the fit index from equation 2. Thus, the best fit temperature for the system is the weighted mean of the trial temperatures, and the uncertainty is the weighted standard deviation of the trial temperatures.

Since the modeled fit is compared against our measured composition for each phase, the confidence limits for our EDS measurements propagate into the term C^meas (equation 2). If we assume that the accuracy of EDS measurements is ± 2%, and since we’re computing 9 elements in total between olivine and chromite, there is a relative uncertainty in F of $\sqrt{9{(0.02)}^2}=0.06$. We added this in quadrature with the uncertainty in the bulk composition.

f_O2 and Sodium During Formation

The MELTS model reproduces the phase assemblage in Iris very well for fugacities near the iron-wÃŒstite buffer (IW) and temperatures > 800 °C by predicting spinel, olivine, feldspar and clinopyroxene with compositions matching our measured bulk composition except for Na₂O. Based on our bulk composition we would expect Na₂O = 2.0 wt% but any open system character may have volatilized a portion of the Na. In this case, the actual bulk composition would have had a higher Na₂O composition than we measured. Fig. 18 shows the results of a computation to determine how well MELTS predicts the phase compositions in Iris for various oxygen fugacities around the IW buffer while also varying the Na₂O content. Fig. 18 shows two minima. The first is a local minimum with Na₂O = 2.1 wt% and log(f_O2) = IW + 0.75. It predicts an olivine/spinel equilibrium temperature of 1020 °C, the presence of orthopyroxene, which we did not see, and fails to predict clinopyroxene. The second minimum is a global minimum, indicating a better fit according to Eq. 2, and occurs with Na₂O = 3.5 wt% at log(f_O2) = IW - 0.25. At the global minimum MELTS produces fit indices for olivine, spinel and feldspar that are significantly lower than at the local minimum and predicts the presence of clinopyroxene instead of orthopyroxene. However, it also predicts nepheline which we do not see. While we cannot be very definitive which of these two solutions is correct, it is more probable based on the fit index metric that the actual bulk composition originally had Na₂O = 3.5 wt% and therefore Iris lost about 40% of its sodium during/after formation. Fig. 18 also provides an error bar for the fugacity prediction: namely IW - 1 < log(f_O2) < IW + 1 produces approximately equivalent fits.

Figure 18.

Average of olivine and spinel fit indices (eq. 2) computed for a range of Na₂O compositions and f_O2 around the IW buffer. The best fit occurs at Na₂O = 3.5 ox wt% and log(f_O2) = IW-0.25.

Therefore, we conclude the optimal fit for fugacity occured at log(f_O2) = −13.3 (IW - 0.25, 1000 °C), so Iris formed near the IW buffer like many type II chondrules.

This fugacity is also reflected in the Fe³⁺/ΣFe in the chromite. We computed Fe₂O₃ content derived from EDS using stoichiometry (Table 2), and measured the Fe-L edge XANES (Fig. 10) from which we concluded that the bulk of the Fe was Fe²⁺ and resides in the tetrahedral site. This is in agreement with MELTS which predicted Fe³⁺/ΣFe = 0.13.

Iris Phases

For the following we will assume our original bulk composition had Na₂O = 3.5 wt%. In this case, MELTS predicted that spinel began to form prior to and concurrently with olivine at temperatures above 1500 °C. At 1130 °C, plagioclase began to crystallize (Fig. 19). The mineral sequence predicted by MELTS agrees with textural evidence that chromite occurs as inclusions in olivine (Fig. 8) and that feldspar surrounds both olivine and chromite (Figs. 4 and 5). We found the best fit equilibration temperature for spinel and olivine was 1040 ± 60 °C (1σ). MELTS predicts the onset of formation of nepheline at 970 °C and then clinopyroxene at 940 °C. This prediction is consistent with the textural evidence because the clinopyroxene (Fig. 4A, top left arrow) is on the edge of the particle between two olivines without affecting their shape. However, we did not see nepheline which probably means that Iris was not in perfect equilibrium at the lower temperatures. Therefore, it is not surprising that measurements taken on feldspar from different bullets and TEM grids yielded both equilibrium and disequilibrium compositions. Horizontal lines plotted in Fig. 20 left frame show the expected oxide wt % for the four primary elements in plagioclase as predicted by MELTS as a function of temperature. Thus, higher temperature feldspar is expected to be richer in Al and Ca, and poorer in Na and Si. Vertical colored lines in Fig. 20 left show EDS measurements of plagioclase from different grids and bullets with compositions matching the MELTS predictions for 880±50 °C. Uncertainties propagated from EDS are also ≈ ±50 °C. Spectra TEM 31 and TEM 32 are actually the same crystal with Na measured by both TEM (9.61 ox wt%) and STXM (10.19 ox wt%). This difference gives a feel for the accuracy of the measurement as well as the precision.

Figure 19.

MELTS predicted phase abundances as a function of temperature for the fugacity IW-0.25, Na₂O = 3.5 wt%. At high temperature, olivine and spinel were co-forming.

Figure 20.

Horizontal traces show the predicted compositions of plagioclase as a function of temperature computed by MELTS for SiO₂, Al₂O₃, Na₂O, and CaO. Colored lines show the measured compositions determined by EDS and STXM. TEM 31 and TEM 32 are the same spectrum except Na₂O has been replaced by a STXM measurement in TEM 32. A perfectly vertical line corresponds to a measurement in perfect agreement with MELTS. An angled line, or one with strong lateral dispersion shows a poor agreement with MELTS. Left) Four compositions are in good agreement with an equilibration temperature of 880 °C. Measurement error bars are on the order of the lateral dispersion of the measurements. Right) Three compositions are out of equilibrium and show several hundred degrees of dispersion from the MELTS compositions. All of the elemental abundances, however, are consistent with a lower temperature than for the equilibrated measurements in the left panel.

The spectra in Fig. 20 right frame are generally consistent with a lower equilibration temperature, but the dispersion between the points increases to > 200 °C and shows a poorer agreement with equilibrium calculations. Furthermore, the spectra show opposing elemental ratios. Na₂O in TEM 28 matches a high temperature prediction while all remaining elements match low temperature predictions, but Na₂O in TEM 33 matches a low temperature prediction while all remaining elements match high temperature predictions. In contrast, all elements in TEM 30 match low temperature predictions except for Si which matches a high temperature prediction. This is further evidence that plagioclase was out of equilibrium. Rapid cooling at temperatures ≤ 900 °C could explain the dispersion in plagioclase compositions because there would be insufficient time to equilibrate across the entire object.

MELTS also predicted the growth of high calcium clinopyroxene (CPX) for temperatures below 940 °C. The predicted composition matched SiO₂ and MgO (< 3% relative error) but had significant errors for CaO and FeO (17% and 30% relative) compared to the clinopyroxene we measured in the SEM. There were also large discrepancies among the minor elements. Since the CPX formed exclusively in the low temperature regime where we have evidence for disequilibrium, we do not try to reconcile the CPX compositions with the MELTS predictions.

The CaO content of olivine is important because CaO only partitions into olivine at concentrations above 0.3 oxide wt % for a specific temperature range centered near 1050 °C given our bulk composition and crystallization sequence. Low-temperature metamorphism would expel it from the crystal in a fashion similar to the loss of Ca observed by Dehart et al. (1992); even slow cooling at a rate much slower than the diffusion timescale for Ca should expel it from the crystal. Fig. 21 shows the predicted composition of the olivine in Iris as a function of temperature. For our system, Fe content increases monotonically as the temperature decreases, and therefore the observed fayalitic content of the olivine will be primarily a function of the partition coefficient of Fe (which in turn derives from bulk composition) and the cooling rate. The MnO content also increases essentially monotonically to ≈ 1300 °C, and is almost contant at 0.5 wt% thereafter. Therefore, how rapidly the crystal cools will not significantly affect the final MnO abundance, and so it is primarily a function of the original bulk composition. Finally, Ca content increases with decreasing temperature down to 940 °C, and then decreases again as clinopyroxene and feldspar begin to form. Therefore, for our assemblage, the Ca content is a very strong function of cooling rate (or metamorphism), and a weaker function of the bulk composition. We measured up to 0.6 wt% CaO in our olivine, while the maximum allowed by MELTS was 0.4 ox wt% which suggests that significant metamorphism did not occur, and that cooling below 1000 °C was relatively rapid compared to the Ca diffusion timescale. Care should be taken not to overinterpret the CaO abundance when the statistical significance of the EDS measurements is limited to 10-50% relative.

Figure 21.

Predicted composition of olivine by MELTS. The Fe content increases monotonically as the temperature decreases. The Mn content also increases essentially monotonically, but is almost constant below 1300 °C. The Ca content increases to a maximum at 940 °C and then decreases again at lower temperatures.

We conclude that the crystallization sequence began with chromite, followed by a chromite/olivine coevolution, and these were able to reach near-equilibrium at temperatures above 1000 °C. High-Ca pyroxene and feldspar began forming around this temperature. Feldspar was in equilibrium until about 900 °C, but fell out of equilibrium at lower temperatures. The remaining melt did not crystallize but formed glass instead. Hence, Iris was probably quenched around 700-800 °C.

Comparison to Other Stardust Particles

For comparison, we also examined published compositions for Torajiro, another chondrule-like object in Stardust, which contained an olivine-chromite pair and enstatite (Nakamura et al. 2008a). We computed a closure temperature of 1090 °C using the olivine-spinel-pyroxene thermobarometer (Sack and Ghiorso 1991, Sack and Ghiorso 1989, Sack and Ghiorso 1994). Their pyroxene composition further constrained the oxygen fugacity to f_O2 = —12 (IW + 1.4). In this case we considered only the olivine and spinel compositions for the closure temperature. Without including the effect of other phases, the temperature is likely somewhat different than we would have computed with a full MELTS model. For comparison, the Iris closure temperature using only the olivine and spinel thermobarometer was 1100 °C, ≈ 100 °C higher than we computed when including the effects of the liquid and feldspar, and very close to Torajiro. An equilibrated Kool grain, Coki-B (Joswiak et al. 2009), contains an olivine poikiolitically enclosing a 50 nm chromite. In this case, the thermobarometer applied to the composition of the olivine and chromite produced a closure temperature of 970 °C, ≈ 50 °C lower than Iris.

Estimation of Cooling Rates from Olivine Diffusion

Using MELTS we determined that the olivine-chromite closure temperature was ≈ 1000 °C, and disequilibrium phases existed at lower temperatures. These results suggest rapid cooling similar to a chondrule forming process. On the other hand, the equilibrated composition of the olivines suggested that this process was not too rapid at the higher temperatures. To put this on a quantitative footing, we did finite difference modeling of the diffusion of Fe and Mg through 3 μm radius spherical olivine grains since our largest grain had a minor radius of 3 μm (Fig 4A). The variation in measured Fo_# was on the order of our experimental precision, so we treated the olivines as unzoned and placed an upper limit on cooling rate as follows.

Morioka (1981) determined that cations in olivine diffuse via a vacancy mechanism where mismatch between cation sizes generates defects that promote diffusion. Thus, the interdiffusion coefficient is concentration dependent and governed by the faster diffusive species (Girifalco 1964, Chapter 3, section 20, Diffusion Couples and the Kirkendall Effect). Dohmen and Chakraborty (2007) later produced a semiempirical relationship (equation 28 in their paper) including the concentration dependence as well as the effect of oxygen fugacity. Using the olivine diffusion coefficient of Dohmen and Chakraborty (2007), we built a diffusion model in MATLAB using spherical coordinates. Fick’s second law then became:

$$\frac{\partial X_{Fe}}{\partial t}=\frac{D_{Fe}}{r^2}\frac{\partial }{\partial r}\left(r^2\frac{\partial X_F{}_e}{\partial r}\right)$$ (3)

where X_Fe = Fe/(Fe+Mg) is the molar fraction of fayalite, r is the distance from the center of the grain, D_Fe is the diffusion coefficient of Fe through olivine. By using spherical coordinates we ignored the influence of crystallographic orientation on the diffusion constant. Buening and Buseck (1973) show that this effect should be small.

To simulate cooling at 1000 °C/hr, we started with homogenous olivine at the composition predicted by MELTS for 1500 °C, with the boundary composition pinned to that predicted by MELTS for 1400 °C. We then allowed diffusion to occur for 6 minutes. Then we reset the boundary condition to the composition predicted for 1300 °C, and continued diffusion for another 6 minutes and so on. We placed an upper limit to the cooling rate by finding the rate that produced zoning as large as or larger than the variation we measured by EDS. Other cooling rates had different time steps and boundary conditions.

Fig. 22 shows the results of the simulations for cooling rates of 1000, 500, 250, 100, and 50 °C/hr. In the top figure, the dotted line shows the compositions predicted for our olivine by MELTS while the solid lines show the simulated cooling profiles. The shaded regions are consistent with our experimentally measured composition of Iris. In the bottom figure, the maximum difference in Fo_# within the crystal is plotted which would be measured experimentally as zoning.

Figure 22.

Cooling rate models for diffusion of Mg and Fe in a 3 μm radius olivine grain. Top: the dashed line shows the Fo_# as a function of temperature predicted by MELTS. The solid lines show cooling rates of 50, 100, 250, 500, and 1000 °C/hr. Slow cooling rates are closest to the MELTS line (dashed). The gray region shows the range of Fo_# measured in Iris. Bottom: profiles show the variation in Fo_# (zoning) at any point in time. The lowest ΔFo_# matches the slower cooling rates. The gray region represents ΔFo_# compatible with Iris measurements.

If this model is correct, cooling rates of ≤ 100 °C/hr could not have produced olivine crystals as homogenous as we found them in Iris. Specifically, we measured an absence of zoning greater than 2-3 Fo_#, whereas the 100 °C/hr model predicted a zoning of 3 Fo_#. Therefore, Iris cooled at ≤ 100 °C/hr. Likewise, the final composition is only close to the predicted composition at the slower cooling rates.

Small grains would reach equilibrium faster and could sustain higher cooling rates. Thus, by forcing the grain size to be static, we perhaps underestimated the cooling rate somewhat. However, the crystal growth would not outpace the diffusion rate – or if it had, we would have seen zoning within each crystal.

Limits on the Thermodynamic Equilibrium Assumption

The degree of supercooling (ΔT) and the cooling rate largely determine whether thermodynamic equilibrium can be assumed. Cooling rate primarily influences the diffusion of elements. However, the degree of supercooling (ΔT) is a better measure of disequilibrium, at least at the onset of crystal formation (Faure et al. 2003). Supercooling is caused by rapid cooling combined with a lack of nucleation centers on which to form crystal grains. In an extreme case, Nagashima et al. (2006) found that microgravity suppressed heterogeneous nucleation with respect to homogenous nucleation and lead to dramatic supercooling (up to ΔT = 1000 ° C in some cases) and radial olivine morphologies. In these experiments, cooling rates were also very high, 100 °C/s, at least three orders of magnitude faster than Iris. These high supercooling differences were made possible only by the elimination of almost all nucleation centers.Donaldson (1976) carried out experiments on the olivine morphologies produced over a wide range of cooling rates, compositions, and supercooling conditions and found that only the slowest cooled melts produced porphyritic olivines. This result was relatively insensitive to melt composition and f_O2. He found that the presence of any supercooling excluded a porphyritic olivine morphology. Faure et al. (2003) later carried out a very systematic study of olivine growth as a function of cooling rate and supercooling. They found two distinct formation regimes. The first was governed primarily by steady state dynamics (i.e. equilibrium crystal growth) and the second was governed by dendritic growth. The competition between these two regimes can be used to explain all observed olivine mophologies. The dendritic regime is diffusion limited, and therefore would be poorly described by thermodynamic equilibrium calculations such as produced by MELTS. However, the porphyritic regime is limited by crystal surface growth dynamics, not diffusion (Faure and Schiano 2005, Faure et al. 2006, Cabane et al. 2005). As a result, porphyritic olivine morphology was only seen for undercooling ΔT< 150 °C, and cooling rates < 100 °C/hr. Faure and Schiano (2005) also found a lack of concentration gradients in the melt neighboring porphyritic olivines up to 100 μm away, further evidence for localized equilibrium. Therefore, in porphyritic olivine assemblages, crystal/melt equilibration calculations should apply.

A key conclusion from these studies is that Iris crystallized at cooling rates in agreement with our diffusion calculations, and was not strongly supercooled or it would not have the observed porphyritic structure. As a corollary, equilibrium calculations are expected to work at the higher temperatures at which diffusion is fast enough to allow for chemical equilibration. On the other hand, we cannot constrain the feldspar equilibrium from the olivine morphologies, since feldspar forms several hundred degrees cooler than olivine.

As a final note about nucleation mechanics, we cannot form a strong conclusion as to whether Iris crystals were seeded by heterogeneous nucleation or homogenous nucleation. Donaldson (1976) noted that at low supercooling, both homogenous and heterogeneous nucleation can form porphyritic crystallites. However, he also observed spinels enclosed in olivines acting as possible nucleation centers. This scenario is compatible with Iris as well.

Kool Grains in Stardust

Joswiak et al. (2009) found that many Stardust tracks contained particles that were assemblages composed of FeO- rich olivines, Na+Cr-bearing high-Ca pyroxenes, sometimes with chromite, albite or albitic glass, and that some of these appeared to have formed as melt objects in a process reminiscent of chondrule formation. Joswiak et al. (2009) called these assemblages Kool grains (Kosmochloric pyroxene and Olivine). Other Kool grains contained similar compositions but the components were clearly not equilibrated and therefore they were not melt objects. They hypothesized that the unequilibrated Kool grains were chondrule precursors, while the equilibrated Kool grains were early-generation chondrules, or mildly heated Kool assemblages. A current definition of Kool particle requires that olivines contain FeO > 10 wt%, Cr₂O₃ < 0.2 wt% and clinopyroxenes contain Na₂O > 0.75 wt%, and Cr₂O₃ > 1.4 wt%.

In Kool particles, kosmochloric high-Ca pyroxenes usually contain Na atomic concentrations close to Cr as one would expect for a pure coupled substitution. However, this is not always the case, since about 1/3 of the measurements done by Joswiak et al. (2009) contained more Na than Cr (up to a factor of two) with additional Al and Ti present. In the Iris high-Ca pyroxene, we also find that the Na is higher than Cr, and that Al and Ti are present. Following the stoichiometry calculations for an ideal pyroxene given by Morimoto et al. (1988), the charges between the four cations balance nicely, with the largest coupled substitution being kosmochlor. In this respect, Iris is similar to the existing Kool population.

However, a first distinction between Iris and Kool particles is that while Kool particles contained Cr spinels solely as inclusions in the olivine, Iris had chromites both internal and external to the olivines. Joswiak et al. (2009) hypothesized that the chromites must have formed entirely before the olivines while our petrographic evidence and the MELTS modeling is also consistent with simultaneous formation.

A second distinction is that Iris contained abundant crystalline plagioclase as well as glassy mesostasis. Although Joswiak et al. (2009) saw some evidence of crystallinity, and even though the authors took care to use low dose acquisitions to preserve delicate crystals, in most cases the plagioclase was amorphous before a diffraction pattern could be recorded. While such delicate crystals were also found in Iris, we additionally found robust plagioclase that held the Na content for many minutes allowing diffraction and EDS quantifications of the Na content. Thus, Iris plagioclase appears to be more crystalline.

A third distinction is the grain size. Joswiak et al. (2009) saw primarily submicron crystals in their finegrained Kool assemblages. Even the Kool melt grains were dominated by grains on the order of ≈ 0.5 μm in diameter, and large crystals were rare. In Iris, the largest crystal was an olivine 11 × 6 μm (Fig. 4), and many crystals, even chromites, were larger than 1 or 2 μm. Thus, Iris was coarser than most Kool particles. The same was true of Callie. As noted by Donaldson (1976), Nagashima et al. (2006), and Faure et al. (2003),- the density of nucleation centers influences– olivine size and morphology. The larger phenocryst grain sizes present in Iris and Callie are consistent with slower cooling and greater equilibration than the Kool grains reported by Joswiak et al. (2009), which provides some context for the cooling rates of other Stardust grains, though to our knowledge, our measurement is the first quantitative cooling rate of an igneous Stardust grain. Iris may also have been more thoroughly melted during liquidus to destroy nucleation centers. Depending on the model of chondrule formation, one could imagine Iris being from a more central portion of a chondrule forming region where peak temperatures were higher, and cooling was slower. Equilibrated Kool assemlages could have come from the periphery of a chondrule forming region, while unequilibrated Kool assemblages were lightly touched or untouched.

Finally, there was one more striking observation. Almost every Kool grain was associated with iron sulfides. In the case of Puki-B (Fig. 8 of Joswiak et al. 2009), sulfides were entirely encased in the body of the object. Therefore, it is probable that sulfides played a significant role in Kool grain formation. Possibly, the original bulk composition was sulfur rich, and the sulfur remained in the liquid until lower temperatures were achieved, and then formed sulfides. In this case, the sulfur would not have significantly altered the phase compositions for the other minerals. If Joswiak et al. (2009) were correct in assuming that the fine grained Kool assemblages were precursors for the igneous Kool assemblages, then one would expect the unequilibrated assemblages to also contain sulfides – and they did.

Future work should examine the presence of sulfur in igneous melts with compositions similar to type II chondrules and Kool particles. Experiments in the vein of Hewins et al. 2005 could see if pyrrhotites form from the liquid after the other phases or show that some other mechanism must be responsible.

Sodium and the Ambient Environment

Excluding formation in a high pressure environment like a differentiated body, the bulk Na concentration (2.0 oxide wt%, Table 5 or 3.5 oxide wt% based on MELTS) and the high concentration of Na in pyroxene, plagioclase and glass require a high partial pressure of Na in the ambient gas during formation (Alexander et al. 2008, Fedkin and Grossman 2013) which in turn argues for a high dust/gas ratio in the environment where Iris and Callie formed. If one can show that the system was closed or nearly closed with respect to Na, it is probable that the system was closed with respect to the other, less volatile elements (Alexander et al. 2008). This would not necessarily always be the case. Libourel et al. (2006) and Hewins et al. (2012) found evidence of open system behavior in type I and II chondrules respectively. However, assuming Iris formed in a closed or nearly closed system from a body of CI composition, a fugacity near the IW buffer, and lost < 50% of its Na, then the dust enrichment was > 10³, and probably closer to 10⁴ (Fedkin and Grossman 2013). With an equilibration temperature of ≈ 1000 °C, and a cooling rate ≤ 100 °C/hr, the Na content of the ambient gas, and consequently the dust enrichment remained high for some hours and behaved as a closed system during the formation of the refractory crystalline phases; only a portion of the plagioclase fell out of equilibrium.

Isotopic Implications

The high dust enrichment (10³) implies that oxygen in the solid and gas phases will equilibrate rapidly (Kita et al. 2010). Because of this, we would not expect dramatic oxygen isotopic heterogeneity within Iris or Callie. Indeed, we found that olivine, plagioclase and glass in Iris were isotopically homogenous, and Callie was also internally homogenous. However the two particles were different from each other in δ¹⁸O by about 5‰ which is a ≈3σ difference. In type II chondrules, differences in isotopic compositions greater than ⪆ 1‰ have only been observed in proximity to a relict grain or as a difference between the olivine and plagioclase/glass within a given chondrule (Kita et al. 2010). Therefore, our evidence suggests that Iris and Callie are not from the same chondrule since the olivine and glass oxygen isotopic compositions are internally uniform, yet different between the two objects. However, as noted previously, since the fractionation is only 3σ along a mass fractionation line, there is still a possibility that this difference is due to unknown instrumental mass fractionation and both Iris and Callie are fragments of a single progenitor chondrule.

Iris and Callie are either two pieces of a single type-II chondrule fragment that broke apart upon aerogel capture, or they are two individual type-II chondrule fragments that formed in similar, but not identical, oxidizing environments. Iris and Callie both have oxygen isotopic composition close to the terrestrial composition. Other Stardust grains also plot in this region (Fig. 16). Torajiro is a type II chondrule/chondrule fragment measured by Nakamura et al. (2008a) and its mean olivine δ¹⁸O composition is ≈10‰ lighter than Callie and Iris, on the Y&R line (Young and Russell 1998). As noted previously, the olivine/spinel geothermometer for Torajiro gives a result (1088 °C) slightly higher than Iris (1020 °C), and a fugacity (f_O2 = —12 at 1088 °C) slightly higher than Iris (f_O2 = —13.4 at 1088 °C). Thus, Torajiro formed in a ¹⁶O rich environment compared to Iris with a slightly higher oxygen fugacity. Joswiak et al. (2012) notes that T22 F7 (Fig. 16) consists of highly unequilibrated, Fe-rich olivine and unequilibrated glass. While it may not be a type II chondrule (there is no hard evidence that it is formed by an igneous process), it nevertheless is similar enough to include in a comparison to Iris. Oxygen isotopes for the olivine plot in the same field as Iris, slightly above the TF line (Nakashima et al. 2012). Two olivine fragments, T77 F1 and F4 (Joswiak et al. 2012, Nakashima et al. 2012; Fig. 16), have fayalitic content similar to Iris olivine. It is possible that these and other lone olivines are chondrule fragments (Frank et al. 2014a). Type II chondrules from CR chondrites have O isotopic compositions dispersed ±5‰ above and below the CCAM line (Clayton et al. 1977), over a range from δ¹⁸O = −2‰ to +12‰ (Krot et al. 2009, Connolly and Huss 2010) which is generally consistent with the FeO-rich Stardust olivines measured so far. Kita et al. (2010) measured LL 3.0-3.1 type II chondrules and found a tighter dispersion with 3.5‰ < δ¹⁸O < 6‰. Thus, Stardust chondrules and Fe-rich olivines are a poorer match for LL type II chondrules compared to CR, but still reasonably close. The primary difference is that LL type II chondrules apparently came from a more homogenized isotopic reservoir as evidenced by their narrow dispersion. Tenner et al. (2013) measured CO 3.0 type II chondrules from Y 81020 and found a dispersion in values −3‰ < δ¹⁸O < 6‰ excluding relict grains. Therefore, CO 3.0 type II chondrules have slightly lighter O isotopic compositions than Stardust. In summary, the oxygen isotope composition for Stardust type II chondrules is virtually indistinguishable from CR, marginally heavier compared to CO, and less homogenized than LL.

Metamorphism and Nebular Processing

The presence of cubanite, magnetite, fluorapatite and brownmillerite in Stardust samples indicates that at least some portion of the material probably experienced aqueous alteration and hints at parent body processing (Berger et al. 2011, Stodolna et al. 2012a). In Iris, however, there are no hydrous phases which may indicate aqueous alteration. The presence of glass also argues against aqueous alteration (Nakamura-Messenger et al. 2011). The question of metamorphic alteration is a little trickier. Grossman and Brearley (2005) showed that even mild levels of metamorphism (≤ petrologic subtype 3.2) cause significant diffusion of Cr in olivines. OC and CO chondrites showed different calibration curves for Cr₂O₃ as a function of petrographic grade (Grossman and Brearley 2005, Fig. 15). Direct application of their curves would classify Iris as petrographic grade 3.15–3.2 (Cr₂O₃ = 0.16 wt%). However, the Cr₂O₃ abundance is also dependent on the specific meteorite, and there is a large dispersion within each meteorite. For example, when we compare against Grossman and Brearley (2005) Figs. 4 and 5, our Cr content is consistent with some ferroan chondrules in Colony, Y81020 (CO 3.0), Acfer 094 (CC ung), and Y793596 (LL 3.0), as well as metamorphosed chondrules. In Iris, the chromite began forming previously and concurrently with the olivine thereby reducing the Cr in the melt for partitioning into the olivine (Libourel 1999). Thus, it is not clear that the Cr calibration will apply to Iris, or Stardust in general where chromian spinel is an early forming phase. Dehart et al. (1992) found that Ca levels in olivine also decrease for increasing metamorphism. In this case, Iris CaO content averages 0.3 wt % which is somewhat higher than even Semarkona’s Fe-rich olivines. This would argue against the presence of metamorphism. The MELTS calculation produced a maximum CaO content of 0.4 wt %, which is consistent with the the EDS measurements considering uncertainty. We conclude that the existing metamorphic indicators serve to restrict the metamorphic grade to ≤ 3.2.

Affinities to Meteorites

On the basis of oxygen isotopes, Iris shows a strong affinity to CR meteorites, some affinity to CO, but poor affinity to OC, as discussed in the isotopic section. Considering ²⁶Al, Iris shows affinity to CR chondrules, many of which also formed > 3 Myr after CAI, but poorer affinity to OC and CO chondrules, most of which had formed by 3 Myr (Ogliore et al. 2012). Frank et al. (2014a) measured > 10³ olivine compositions from matrices and chondrules of major unequilibrated meteorite classes and Stardust. In Iris, we measured an average MnO = 0.63 wt%, Cr₂O₃ = 0.16 wt%, CaO = 0.3 wt% and an average Fo₆₃ (Table 1). Comparing against the compositional fields from chondrules in Frank et al. (2014a), Iris is compatible with CR, CV_ox, or CM. At present, the greatest affinities are with CR, but the oxygen and Al isotopes do not exclude CV_ox or CM meteorites, so future research may show affinities there as well.

CONCLUSIONS

Iris is an igneous object which was first heated to near liquidus and equilibrated. It then cooled at ≤ 100 °C/hr within 2 log units of the IW buffer. The phases found can be modeled by the MELTS code with the exception of the clinopyroxene, though petrography and the approximate composition suggest that the pyroxene also formed from the same igneous process. Iris was not significantly metamorphosed, and exhibits closure temperatures similar to those seen in chondrules. The Na content and MELTS fitting suggests an Na loss of about 40% which ensures that this cooling occured in an Na rich environment, and the ambient partial pressure of Na was near equilibrium with Iris while it cooled giving a dust enrichment of ≥ 10³. Thus, the Iris-forming process exhibited essentially closed-system behavior just as the the type II chondrule forming process has been shown to do in some cases (Alexander et al. 2008). Oxygen isotopes and olivine composition show a closest affinity to CR type II chondrules. Given the absence of ²⁶Al in Iris reported by Ogliore et al. (2012), it probably formed later than many chondrules seen in meteorites. Iris appears as well to be related to the Kool population of particles (Joswiak et al. 2009).

Therefore, we conclude that Iris and Callie formed in a chondrule forming process which is typical of type II chondrules such as found in meteorites, with the closest affinity to CR, but not necessarily the same bulk composition.

Table 7.

Sample numbers

Table	Spectrum	JSC curation #	UC Berkeley ID	Table	Spectrum	JSC curation #	UC Berkeley ID	Table	Spectrum	JSC curation #	UC Berkeley ID
1	TEM 01	C2052,12,74,2,22	B4 1.1.S1	1	TEM 19	C2052,12,74,2,27	B9 7	3	TEM 27	C2052,12,74,2,22	B4 1.2.4
1	TEM 02	C2052,12,74,2,22	B4 1.1.S6	1	TEM 20	C2052,12,74,2,43	D5 10	3	TEM 28	C2052,12,74,2,25	B7 4.2
1	TEM 03	C2052,12,74,2,22	B4 1.2.S3	1	TEM 21	C2052,12,74,2,43	D5 11.3-7	3	TEM 29	C2052,12,74,2,27	B9 A Fit Na
1	TEM 04	C2052,12,74,2,22	B4 1.3.1.S5	1	TEM 22	C2052,12,74,3,12	Callie-B7 5:1	3	TEM 30	C2052,12,74,2,27	B9 A Ext Na
1	TEM 05	C2052,12,74,2,22	B4 1.4.2.S1	1	TEM 23	C2052,12,74,3,25	Callie-D2 1:1-10	3	TEM 31	C2052,12,74,2,27	B9 B Fit Na
1	TEM 06	C2052,12,74,2,22	B4 1.4.2.S2	1	SEM 01	C2052,12,74,2,0	Bullet-D6 1	3	TEM 32	C2052,12,74,2,27	B9 B STXM Na
1	TEM 07	C2052,12,74,2,22	B4 1.4.2.S3	1	SEM 02	C2052,12,74,2,0	Bullet-D6 2	3	TEM 33	C2052,12,74,2,43	D5 1-4 Ext Na
1	TEM 08	C2052,12,74,2,25	B7 1.S4	1	SEM 03	C2052,12,74,2,0	Bullet-D6 3	3	SEM 11	C2052,12,74,2,0	Bullet D6 8
1	TEM 09	C2052,12,74,2,27	B9 11a	1	SEM 04	C2052,12,74,2,0	Bullet-D6 5	3	SEM 12	C2052,12,74,2,0	Bullet D13 3
1	TEM 10	C2052,12,74,2,27	B9 11b	1	SEM 05	C2052,12,74,2,0	Bullet-D6 17	3	SEM 13	C2052,12,74,2,0	Bullet D13 11
1	TEM 11	C2052,12,74,2,27	B9 13	1	SEM 06	C2052,12,74,2,0	Bullet-D13 2	3	SEM 14	C2052,12,74,2,0	Bullet D13 12
1	TEM 12	C2052,12,74,2,27	B9 14a	1	SEM 07	C2052,12,74,2,0	Bullet-D13 6	3	SEM 15	C2052,12,74,2,0	Bullet D13 13
1	TEM 13	C2052,12,74,2,27	B9 14b	1	SEM 08	C2052,12,74,2,0	Bullet-D13 8	4	TEM 34	C2052,12,74,2,43	D5 1
1	TEM 14	C2052,12,74,2,27	B9 10	1	SEM 09	C2052,12,74,2,0	Bullet-D13 10	4	SEM 16	C2052,12,74,2,0	Bullet D6 10
1	TEM 15	C2052,12,74,2,27	B9 12	2	TEM 24	C2052,12,74,2,27	B9 12	Isotopes
1	TEM 16	C2052,12,74,2,27	B9 1	2	TEM 25	C2052,12,74,2,27	B9 1	6	Grain 1 Spot 1	C2052,12,74,3,0	Grain 1 Spot 1
1	TEM 17	C2052,12,74,2,27	B9 2	2	TEM 26	C2052,12,74,2,43	D5 1	6	Grain 1 Spot 2	C2052,12,74,3,0	Grain 1 Spot 2
1	TEM 18	C2052,12,74,2,27	B9 3	2	SEM 10	C2052,12,74,2,0	Bullet D6 4	6	Grain 2	C2052,12,74,3,0	Grain 2