Petrographic constraints on the formation of silica‐rich igneous rims around chondrules in CR chondrites

In the CR (Renazzo‐like) chondrite group, many chondrules have successive igneous rim (IR) layers, with an outer layer that contains a silica mineral and/or silica‐rich glass (silica‐rich igneous rims, SIRs). Models for SIR formation include (1) accretion of Si‐rich dust onto solid chondrule surfaces, followed by heating and cooling and (2) condensation of SiO(gas) onto the surface of partially molten chondrules. We evaluate these models, based on a petrographic study of five Antarctic CR chondrites that have undergone minimal secondary alteration. We obtained electron microprobe analyses of minerals and glass with quantitative wavelength‐dispersive spectroscopy mapping, and identified silica polymorphs with Raman spectroscopy. Common SIRs contain silica, low‐Ca pyroxene, Ca‐rich pyroxene, Fe,Ni metal, ± glass ± plagioclase ± rare olivine. We also describe near‐monomineralic SIRs where a narrow zone of cristobalite occurs at the outer edge of the chondrule. All crystalline silica is cristobalite, except for one SIR that consists of tridymite. Some rims contain silica‐rich glass (>80 wt% SiO2) but no silica mineral. Features such as sharp interfaces and compositional boundaries between chondrules and SIRs indicate that SIRs were formed from solid precursors. Consideration of the stability fields of silica polymorphs and computed liquidus temperatures indicates that SIRs were heated to >1500°C for limited time periods, followed by rapid cooling, similar to conditions for chondrule formation. We infer that in the CR chondrule formation region, the same heating mechanism was repeated multiple times while the chemical composition of the nebular gas evolved to highly fractionated silica‐rich compositions.


INTRODUCTION
Chondrules are the most abundant component of chondritic meteorites, and yet there is still debate about what process was responsible for chondrule heating.Models for chondrule formation include those that propose melting of free-floating dusty aggregates in the protoplanetary disk, for example, as a result of heating in a shock front, as well as those that propose dispersal of (partially) molten material in collisions between planetesimals (Connolly & Jones, 2016;Desch et al., 2012;Johnson et al., 2018;Morris & Boley, 2018).Any viable chondrule formation model must account for observed physical and chemical properties of chondrules, for example, the thermal histories they record and their chemical and oxygen isotope compositions (Jones et al., 2018;Tenner, Ushikubo, et al., 2019).Chondrule thermal histories include heating to 1400-1700°C, and cooling at rates of tens to hundreds of degrees per hour.An important feature of chondrules is that they record multiple heating events, in the form of relict grains and igneous rims (e.g., Hewins, 1997;Jones, 1996;Jones et al., 2018;Rubin & Krot, 1996).The nature of successive heating events can be a key parameter in assessing the viability of a given chondrule formation model.Interpreting this record of multiple heating is, therefore, an important aspect of understanding chondrule formation, particularly because different groups of chondrites contain chondrules with different characteristics, including the record of multiple heating (Jones, 2012;Kita & Ushikubo, 2012;Scott & Krot, 2014).
The CR chondrites show characteristic differences compared with other carbonaceous chondrite (CC) groups such as CM, CO, and CV chondrites.They have distinctive isotopic properties (D 17 O, e 50 Ti, e 54 Cr) (e.g., Marrocchi et al., 2022 and references therein), and chondrules in CR chondrites formed 1-2 million years later than chondrules in other CC groups (Nagashima et al., 2018;Schrader et al., 2017;Tenner, Nakashima, et al., 2019).CR chondrites are also characterized by the presence of chondrules that have multiple igneous rim layers, including silica-bearing layers that are unique to this group (Krot et al., 2002(Krot et al., , 2004;;Noguchi, 1995;Prinz et al., 1986;Weisberg et al., 1993;Weisberg & Prinz, 1996).These layered chondrules consist of porphyritic chondrule cores (which are olivine and/or pyroxene-rich) surrounded by olivine and/or pyroxenerich igneous rims (IRs) AE an Fe,Ni metal layer AE a phyllosilicate-rich layer AE a carbonate layer AE a silicarich igneous rim (SIR) (Krot et al., 2004;Noguchi, 1995;Weisberg et al., 1993).Phyllosilicate and carbonate layers are secondary alteration features.Layered chondrules do not always consist of this specific layering sequence and can be overlain in different orders: This includes chondrules that are overlain directly by an SIR.The layers have differences in mineralogy, texture, and composition compared to their host chondrules, and rims have more 16 O-poor oxygen isotope compositions than their host chondrules (Weisberg et al., 1993;Weisberg & Prinz, 1996).
SIRs are common in CR chondrites: Noguchi (1995) determined that 40% of chondrules in the CR chondrite Pecora Escarpment (PCA) 91082 contain silica at their margins.In a detailed study of SIRs in 30 CR chondrites, Krot et al. (2004) described SIRs as containing zoned low and Ca-rich pyroxenes, Fe,Ni metal, glassy mesostasis, and silica (Krot et al., 2004).Silica minerals were inferred to be cristobalite, although this was not confirmed.Komatsu et al. (2019) identified cristobalite as the dominant polymorph in SIRs.Krot et al. (2004) also showed that mesostasis increases in Si, Na, K, and Mn and decreases in Ca, Mg, Al, and Cr contents from the center of the chondrule to the margins and surrounding rims, including SIRs.Most recently, Mart ınez and Brearley (2022) have described SIRs in Queen Alexandra Range (QUE) 99177, in the context of alteration processes in CR chondrites, and observe similar compositional trends as well as identifying silica in SIRs as cristobalite.
The nature of the boundary between SIRs and the layers that they overlie (host chondrule, IR, or metal mantle) is an important feature that informs formation models.Unlike the boundary between IRs and host chondrules, SIR boundaries are not always sharp (Krot et al., 2004;Noguchi, 1995).This may in part be because aqueous alteration can obscure features of the interface region (Komatsu et al., 2019).The most highly altered CR chondrites, such as Renazzo and Al Rais, do not contain SIRs because silica has been altered (Krot et al., 2002).Other studies (e.g., Harju et al., 2014;Krot et al., 2000Krot et al., , 2004) also note the replacement of SIRs via alteration and refer to the initial stages of alteration as phyllosilicates.However, Brearley and Jones (2016) and Mart ınez and Brearley (2022) have shown that these "phyllosilicates" are actually composed of amorphous material rich in SiO 2 and FeO, that replaces cristobalite.
As for IRs, accretion followed by melting is a likely formation mechanism for SIRs (Krot et al., 2004;Noguchi, 1995;Rubin, 2018).However, Krot et al. (2004) suggested that in some cases, there is evidence for SiO (gas) condensing directly into chondrule melts.A similar process has also been proposed to explain mineralogical zonation of porphyritic type I chondrules in C and O chondrites, in which olivine is concentrated in the center of chondrules and low-Ca pyroxene is concentrated at the edges (Barosch et al., 2019;Chaussidon et al., 2008;Friend et al., 2016;Jacquet et al., 2012;Libourel et al., 2006;Libourel & Portail, 2018;Nagahara et al., 1999Nagahara et al., , 2008;;Tissandier et al., 2002).The fundamental difference between these models is that formation of a chondrule/IR/ SIR sequence in an accretion and melting model requires three successive episodes of solid deposition and heating, whereas in a condensation model, a single heating event could be invoked.Therefore, understanding the constraints on IR and SIR formation is important for distinguishing the dominant chondrule formation model.The presence of SIRs is also evidence for chemical evolution of gas and condensate material in the CR chondrule formation region of the disk.The formation of highly fractionated SiO 2 -rich solid precursor material (Mg/Si <1) for an accretion and melting model has been proposed as resulting from either isolation of forsterite or partial evaporation during chondrule formation (Krot et al., 2000(Krot et al., , 2002(Krot et al., , 2004) ) as well as equilibrium condensation to pyroxene-dominated condensates (Rubin, 2018).
Chondrule formation models emphasize the need to reproduce the thermal histories of chondrule formation.However, for IRs and SIRs, thermal histories are currently poorly constrained.An experimental study conducted by Connolly and Hewins (1991) determined that IRs could be formed by heating fayalitic dust that was glued onto the surfaces of synthetic chondrules and heated to peak temperatures ≥1000°C for a few minutes to an hour.Krot and Wasson (1995) suggested that the peak temperatures for melting igneous rims in OCs was ≥1126°C, based on calculations using MELTS, but did not estimate a maximum peak temperature or heating duration.Matsuda et al. (2019) used a consideration of oxygen isotope diffusion to suggest that an IR in the CV chondrite Northwest Africa (NWA) 3118 could have been heated for a maximum of several hours to several days at near liquidus temperatures (~1720°C).Interpretations of IR and SIR formation conditions are therefore limited.More robust constraints on peak temperatures, the duration of heating, and cooling rates would help to interpret their formation environment and the relationship of IR heating events to chondrule-heating processes.
In this paper, we revisit the petrography of SIRs.We have characterized the mineralogy and petrology of SIRs from some of the least-altered CR chondrites, along with determining the silica polymorphs present.Our aim is to better understand SIR formation conditions and thermal histories.Understanding the formation of SIRs provides insight into the chemical, physical, and temporal changes that took place in the CR chondrite chondrule-forming region of the protoplanetary disk, and helps to interpret what distinguishes this region from chondrules observed in other chondrite groups.

CR Chondrite Samples
All CR chondrite samples used for this study are Antarctic meteorites, from the Antarctic Meteorite Collection at Johnson Space Center.We studied five chondrites (Table 1) that have petrologic subtypes ranging from 2.4 to 2.6 on the Alexander et al. (2013) scale and 2.7-2.8 on the Harju et al. (2014) scale (2.8 being the least aqueously altered): Queen Alexandra Range (QUE) 99177, Elephant Moraine (EET) 92042, EET 92062, Meteorite Hills (MET) 00426, and Graves Nunataks (GRA) 95229.There is disagreement on classification of individual meteorites between these two studies, probably due to the complex nature of the processes involved after parent body accretion.The Harju et al. (2014) scale is based on petrological parameters and thus is a more useful scale for the purposes of this study.For this study, it was important to have relatively unaltered samples to ensure that the Si-rich phases are primary, and the effects of secondary alteration were minimized, so that formation conditions could be studied without the overprint of alteration.Values listed in Table 1 for the number of chondrules, IRs, SIRs, and percentages of IRs and SIRs are likely minimum values.Krot et al. (2004) also studied some of the CR chondrites in this study, but different thin sections (QUE 99177 7, 8;MET 00426 5, 6;EET 92042 22, 23;GRA 95449 17, 18, 21, 29).

Analytical Methods
The FEI QUANTA 650 field emission gun (FEG) scanning electron microscope (SEM) at the University of Manchester (UoM) was used to collect high-resolution backscattered electron (BSE) images, x-ray maps, and energy-dispersive x-ray spectroscopy (EDS) spectra.We used an accelerating voltage of 15 kV and variable beam current (e.g., 3-7 nA) depending on the requirements of the analyses.
Mineral and glass compositions in chondrules and their SIRs were analyzed by electron probe microanalysis (EPMA).Because many SIRs are fine grained, a traditional point analysis method for EPMA analysis proved challenging.Instead, we used a quantitative mapping approach, outlined by Donovan et al. (2021).Wavelength-dispersive x-ray spectroscopy (WDS) x-ray mapping was done on the JEOL JXA-8530F electron microprobe at UoM. Maps were collected at 15 kV with beam currents ranging from 10 to 30 nA, and dwell times between 80 and 200 ms depending on what was most appropriate for a given sample.Quartz was used as the Si standard for silica, and wollastonite was used as the Si standard for the silicate and glass analysis.For each map, the same 12 elements (Na, Si, Al, Ca, Mg, Ti, Cr, Mn, K, P, Fe, Ni) were measured over three spectrometers.Full analytical details for EPMA, including standards, analytical conditions, map conditions, and spectrometer setup, are given in the Supporting Information, Appendix B. φ(qz) matrix corrections were applied to all analyses.Mean atomic number (MAN) corrections were applied to calculate background positions.We accepted analyses with analytical totals between 98.5 and 102 wt%, using a filter as described below.We also used stoichiometry limits for acceptable analyses which were 2.98-3.02for olivine (on the basis of 4 oxygens), 3.97-4.02for pyroxene (6 oxygens), and 4.98-5.02for plagioclase (8 oxygens).
The CalcImage programme© was used to extract quantitative compositions from WDS x-ray maps.Data were extracted from a mapped area by drawing a 10 9 10 pixel box on a grain of interest.The average wt% oxide value for each element within this box was calculated based on the combined compositions of the pixels inside the extracted area.Prior to extraction, the phase of interest (e.g., pyroxene) was isolated from the surrounding phases by "Pixel Filtering" based on a defined range of wt% values found suitable for each phase in each map.For example, low-Ca pyroxene in the SIR QUE 99177 Ch6 had an MgO wt% filter of >30<40 to isolate it from the surrounding Ca-rich pyroxene, plagioclase, and Si-glass phases.This was particularly useful since Fe-rich alteration affects many of the SIRs studied, so an FeO filter could be enabled to limit the number of pixels that were affected by beam interaction with the surrounding Fe-rich alteration.Numerous extractions were made from multiple phases within each map area and each extraction was treated like a point analysis.Full details are given in the Supporting Information, Appendix A. Average compositions shown in data tables below are averages of multiple extracted analyses.In these tables, "No. of analysis areas" refers to the number of 10 9 10 pixel boxes that were combined to give the reported analysis.Quantitative WDS x-ray maps were processed using ImageJ to show the spatial distribution of element oxides (Supporting Information, Appendix C).
Average detection limits (in wt%) for quantitative WDS maps of each phase are as follows: For the SIRs, for low-Ca pyroxene, 0.21 for Na 2 O, 0.19 for MnO, 0.10 for K 2 O, and 0.12 for TiO 2 ; for Ca-rich pyroxene, 0.21 for Na 2 O and 0.09 for K 2 O; for plagioclase, 0.26 for P 2 O 5 , 0.14 for Cr 2 O 3 , 0.20 for MnO, 0.09 for K 2 O, and 0.13 for TiO 2 ; for glass, 0.27 for P 2 O 5 , 0.14 for Cr 2 O 3 , and 0.19 for MnO; for silica, 0.20 for Na 2 O, 0.24 for P 2 O 5 , 0.11 for MgO, 0.13 for Cr 2 O 3 , 0.07 for CaO, 0.18 for MnO, 0.10 for K 2 O, 0.12 for TiO 2 , and 0.22 for NiO.For chondrules, for olivine, 0.37 for Na 2 O, 0.17 for Al 2 O 3 , 0.23 for MnO, and 0.17 for K 2 O; for low-Ca pyroxene, 0.31 for Na 2 O, 0.22 for MnO, 0.15 for K 2 O, and 0.14 for TiO 2 ; for high-Ca pyroxene, 0.40 for Na 2 O and 0.16 for K 2 O; for plagioclase, 0.32 for P 2 O 5 , 0.17 for Cr 2 O 3 , 0.24 for MnO, 0.16 for K 2 O, and 0.16 for TiO 2 ; for glass, 0.27 for P 2 O 5 , 0.17 for Cr 2 O 3 , and 0.20 for MnO.However, the detection limits are variable among individual analyses.As an example, Table 3 below gives five olivine analyses from WDS maps, in which the detection limit for Na 2 O varies from 0.30 to 0.52 wt%.In these tables, where an element is below detection, we indicate the value of the detection limit for that individual analysis.These detection limits are themselves averages of the detection limits of the individual 10 9 10 pixel areas.
In addition to the WDS mapping, we conducted a few point analyses.These were also obtained on the JEOL JXA-8530F electron microprobe at UoM, using an accelerating voltage of 15 kV and a 10 nA beam current.For point analyses, typical detection limits (in wt%) were: for low-Ca pyroxene, 0.02 for Na 2 O, 0.016, 0.02 for MnO, 0.01 for K 2 O, and 0.02 for TiO 2 ; for olivine, 0.02 for Na 2 O, 0.01 for K 2 O, and 0.02 for TiO 2 .Detection limits for point analyses are lower than detection limits for quantitative WDS x-ray maps (e.g., typical detection limits for Na 2 O in olivine from quantitative WDS maps were 0.370 compared to 0.017 for point analyses) due to different dwell times of the analyses.
We conducted Raman spectroscopy to determine silica polymorphs, using the Horiba XploRA instrument at UoM.We used a 532-nm laser and the following settings: 1009 objective lens, 10%-25% laser power, 1800 diffraction grating (g mm À1 ), 50 lm entrance slit, and 300 lm confocal hole.With the laser wavelength and objective lens used during these analyses, the spot size is ~0.7 lm.Raman spectra collected from samples were compared to reference Raman spectra available from the Rruff database (Lafuente et al., 2015).Cristobalite has characteristic peaks at ~110, 220, and 420 cm À1 and tridymite has a characteristic double peak at ~400 cm À1 with additional minor peaks (Figure 1).Si-rich glass is also present in the samples studied: Glass has no characteristic peaks (Figure 1).We did not observe any quartz.

Overview of Silica-Rich Rims and their Host Chondrules
We studied five CR chondrites, with petrologic subtypes ranging from 2.4 to 2.6 on the Alexander et al. (2013) scales and 2.7-2.8 on the Harju et al. (2014) scale and weathering grades ranging from A to B (Table 1).The number of chondrules in each sample ranges from 9 to 50.Both igneous rims (IRs) lacking Sirich phases, and SIRs, are present around chondrules in all of the samples studied, and IRs are more abundant than SIRs.IRs surround 18%-51% and SIRs surround 10%-27% of the chondrules (Table 1).However, these are minimum values because not all chondrules were examined in detail, and some SIRs might have been overlooked.
We studied a total of 27 chondrules that apparently have SIRs, based on the observation that a high-Si region was observed in EDS x-ray maps of the section, and the presence of an Si-rich phase.Closer inspection showed that some of these chondrules do not actually contain a silica mineral phase, but rather contain a glass with high SiO 2 content (Table 2), which we return to below.All the chondrules with SIRs that we identified are porphyritic, predominantly with porphyritic olivine and pyroxene (POP) textures, and some with porphyritic pyroxene (PP) textures.We did not observe any non-porphyritic chondrules, which are rare in CR chondrites.They range in diameter from ~150 lm to 3 mm and are typically subrounded to rounded.Several chondrules are mineralogically zoned, with olivine concentrated in the center of the chondrules and low-Ca pyroxene concentrated toward the edge.Olivine is commonly poikilitically enclosed in low-Ca pyroxene.Other phases typically present in chondrules include Ca-rich pyroxene, glass, Fe,Ni metal, AE plagioclase.Some chondrules also  contain grains of crystalline silica, which are always cristobalite.Sulfides were not observed.Metal in most chondrules occurs as rounded to subrounded grains ranging in diameter from 5 to 500 lm.The distribution of Fe,Ni metal varies: in some chondrules, metal is concentrated in the central region of the chondrule, and in others, it is found toward the edge.Some chondrules also have an Fe,Ni "mantle" surrounding the chondrule, separating the host chondrule from the IR or SIR.
Where SIRs are present, they surround the chondrule either completely or partially.Widths of SIRs vary among chondrules, from around 10 to 325 lm (Table 2) and the width of a single SIR can either be uniform or variable.SIRs are typically finer grained than their host chondrules.In the majority of SIRs, the dominant phases are silica, low-Ca pyroxene, Ca-rich pyroxene, Fe,Ni metal AE glass AE plagioclase AE olivine (Table 2).Olivine is rare, typically only 1-3 grains in a given rim.The boundary between the chondrule or IR and the surrounding SIR varies in sharpness.The presence of an Fe,Ni mantle can mean that the boundary between the chondrule and the rim can be seen more distinctly.Alteration features and x-ray maps can also highlight the boundary more clearly, but this is not always the case.

Petrography and Silica Polymorphs of Silica-Rich Rims
For the 27 chondrules studied, we observed different types of SIRs.Below we describe these different rim types in detail in terms of texture, mineralogy, silica polymorphs, and mineral chemistries.We term the most abundant rim type, present in 14 of the SIRs, as "common" (Table 2).These rims contain cristobalite, low-Ca pyroxene, Ca-rich pyroxene, AE glass AE plagioclase AE Fe,Ni metal.In 10 of the SIR chondrules, the silica-rich phase is glass: We name these Si-glass rims (Si-gl).The presence of a narrow layer of silica on the edge of chondrules (Si-e) occurs in seven SIRs.Multiple rim types can be found within a single SIR: Six SIRs in this study contain more than one of the different rim types, for example, QUE 99177 chondrule 1 (Ch1) has regions of common rim texture as well as regions of Siglass texture (Table 2).A multilayered rim with a layer of monomineralic silica in QUE 99177 Ch7 and a tridymite rim in EET 92042 Ch11 are each found in only one SIR.

Common Silica-Rich Igneous Rims
The most abundant type of silica-rich igneous rim (denoted "common" (C) in Table 2) occurs on 14 of the 27 chondrules.Examples of some common rims are shown in Figures 2-4. Figure 2a-c shows the POP chondrule EET 92062 Ch8.The SIR is clearly visible in the combined elemental x-ray map, Figure 2b: It has a sharp contact with the host chondrule.The SIR shows the typical texture of SIRs: silica (identified as cristobalite) is predominantly enclosed in euhedral grains of low-Ca pyroxene (Figure 2c).Euhedral grains of Ca-rich pyroxene are also present, some of which overgrow the low-Ca pyroxene.Glass is present interstitial to silica and pyroxene, and there are some rounded grains of Fe,Ni metal.
EET 92042 Ch14 is a subrounded POP chondrule with a fractured edge (Figure 2d-f).The chondrule is partially surrounded by an SIR, which is clearly visible in a combined elemental EDS x-ray map (Figure 2e).The chondrule/rim boundary is sharp, and silica is concentrated at the chondrule/rim boundary (Figure 2e).The rim contains cristobalite, low-Ca pyroxene, zoned Carich pyroxene, Fe,Ni metal, and an interstitial mesostasis that is composed of a fine-grained mixture of Si-rich glass and elongate needles of plagioclase (Figure 2f; see CaO, Na 2 O, and Al 2 O 3 WDS maps in Supplementary Materials Appendix C, Figure S1, hereafter Figure AppC_S1).Lathshaped Fe-rich grains that occur at the margins of the large pyroxene grain in this rim appear to consist of fayalite, which we consider likely to result from aqueous alteration.
In some cases, an SIR can be separated from the host chondrule by an igneous rim (IR).An example is POP chondrule QUE 99177 Ch5 (Figure 3).The host chondrule is mineralogically zoned, consisting of an olivine-rich core and low-Ca pyroxene concentrated in the outer region (Figure 3a,b), in addition to interstitial Ca-rich pyroxene, plagioclase, and Si-rich glass.There is little Fe,Ni metal in the host chondrule, but a well-defined Fe,Ni mantle with a thickness of ~90 lm partially surrounds the chondrule (Figure 3a,b).A combined elemental EDS x-ray map (Figure 3b) clearly shows successive layers of an IR and then an SIR which are overlain on either the Fe,Ni mantle or directly onto the chondrule.The IR is predominantly composed of olivine and low-Ca pyroxene (Figure 3c,d), with some Fe,Ni metal.The contact between the IR and the SIR is sharp (Figure 3d).In the SIR, cristobalite is intergrown with low-Ca pyroxene and Ca-rich pyroxene (Figures 3e,f and AppC_S2).Fe,Ni metal, plagioclase, and mesostasis are also present, the mesostasis being composed of a fine-grained intergrowth of Si-rich glass and elongated plagioclase needles.Where there is no Fe,Ni mantle, the boundary between the IR and the chondrule is not always clear, but the SIR remains well defined.In places, the SIR consists of only a narrow band of silica (Figure 3g,h), a texture that we describe as silica-edge (Si-e) below and in Table 2.The SIR of QUE 99177 Ch5 shows secondary alteration and weathering in the form of iron veining, Fe-, K-, and P-rich alteration of silica, and oxidation of metal (e.g., Figures 3f,h and AppC_S2).
In five of the chondrules, grains of silica occur within the host chondrule as well as in the rim, with or without an Si-rich glass (Table 2).In POP chondrule EET 92062 Silica-rich igneous rims in CR chondrites Ch2 (Figure 4), the host chondrule has a very similar texture and mineralogy to its associated SIR.The chondrule is mineralogically zoned and contains small (~50 lm) olivine grains in the center in association with small, rounded Fe,Ni blebs (Figure 4a,b).Pyroxene grains are elongated in the center of the chondrule, and pyroxene also forms a shell around the perimeter that is overlain by a partial mantle of Fe,Ni metal (Figure 4a,b).Abundant plagioclase occurs both as laths (~25 lm in width), and in an interstitial intergrowth with cristobalite and Ca-rich pyroxene in the chondrule mesostasis (Figures 4c and AppC_S3).The SIR is not obvious in elemental x-ray maps because silica is very fine-grained (Figure 4b).In the SIR, cristobalite grains, up to ~10 lm across, occur in a fine-grained intergrowth with plagioclase and Ca-rich pyroxene, as a mesostasis interstitial to larger grains of low-Ca pyroxene and Fe,Ni metal (Figures 4d and AppC_S4): Metal grains are smaller than those in the host chondrule.
We can make several general statements about common SIRs.Common rims can either completely or partially surround chondrules and typically have sharp boundaries with the host chondrule or the IR that they overlie.The widths of common rims range from ~10 to  2).The abundance and size of Fe,Ni metal particles in the host chondrule and associated SIR are variable.Some chondrules (3 of 14) with a common SIR have an Fe,Ni mantle, which helps define the chondrule/ rim boundary (Figures 3 and 4) although Fe,Ni mantles are not always continuous around the entire chondrule.Common rims contain silica grains, low-Ca pyroxene, Ca-rich pyroxene, AE glass AE plagioclase AE Fe,Ni metal.Olivine can also be present in low abundance.The silica polymorph in common SIRs is always cristobalite.

Silica on Edge of Rim
Seven POP chondrules in this study have a thin layer on their outer edge, where there is a high abundance of silica.We term these silica-edge (Si-e) rims; examples are shown in Figure 5.A section of the rim on QUE 99177 Ch5 also has an Si-e texture as mentioned above (Figure 3g,h).
In MET 00426 Ch1 (Figure 5a), the silica-rich zone is <20 lm in width (Table 2).It consists of subangular cristobalite in association with minor Ca-rich pyroxene and an SiO 2 -rich glass: Ca-rich pyroxene appears to occur as an overgrowth on the pyroxene at the edge of the chondrule.A similar texture occurs as a narrow (<20 lm) rim on EET 92062 Ch3, where there are also abundant submicron blebs of Fe,Ni metal (Figure 5b).This rim also contains glass although it is not illustrated.The texture in these zones, where the exact boundary between the SIR and the underlying pyroxene is not sharp, and where Carich pyroxene mingles with silica, indicates that melting of Ca-rich pyroxene at the edge of the chondrule likely occurred along with formation of the silica-rich overgrowth zone.
Figure 5c,d shows the POP chondrule MET 00426 Ch5 which is partially surrounded by an IR and then an SIR.One section of the IR/SIR appears itself to be overgrown by a second lobe of IR then SIR (Figure 5c).The IR (width ~300 lm) consists of olivine poikilitically enclosed in low-Ca pyroxene, and minor mesostasis.Fe,Ni metal is much more abundant in the IR than in the host chondrule, and metal is to some extent concentrated at the chondrule/IR boundary (Figure 5c).Metal has undergone considerable alteration to Fe-oxides.The Si-e SIR consists predominantly of a narrow layer of cristobalite (Figure 5d).It has an irregular, lobate outer surface, and the Si-e rim decorates all the re-entrant surfaces (Figure 5d).Cristobalite has undergone extensive alteration to an Fe-rich phase, making the boundaries between SIR and the surrounding fine-grained rim (FGR) material hard to identify.
Mineralogically zoned POP chondrule MET 00426 Ch3 also has an igneous rim (Figure 5e-g).The IR consists predominantly of pyroxene and glass with little olivine.It is clearly defined from the host chondrule by its smaller grain size, and abundant small Fe,Ni metal grains (metal is heavily altered in this chondrule).A zone of larger metal grains outlines the chondrule/IR boundary (Figure 5e).The chondrule/IR boundary is not very clearly defined in BSE images which show a continuous texture between the host chondrule and IR layers (Figure 5f), suggesting that melting during IR formation has obscured this boundary.A narrow (~25 lm) Si-e SIR occurs at the edge of the chondrule, and it is itself overlain with a finegrained rim (FGR).The Si-e SIR is easily recognizable in an SiO 2 WDS x-ray map (Figure 5g).In the SIR, cristobalite is the dominant phase, with minor low-Ca pyroxene, Ca-rich pyroxene, and less abundant glass.Some parts of the SIR also contain a feldspathic glass (Figure AppC_S5).In addition to oxidation of metal, alteration of silica has occurred at the contact with the FGR and there is a zone of smooth alteration between the SIR and the FGR.
We can make several general statements about Si-e rims.Si-e rims have very clearly defined boundaries with their host chondrules or IRs, but a small amount of melting of pyroxene may have occurred at the interface between the two.Si-e rims mostly consist of cristobalite with minor low-Ca pyroxene, Ca-rich pyroxene, and glass.The majority of chondrules that exhibit this type of SIRs show evidence of Fe-rich alteration at the FGR interface.

Silica-Rich Glass (Si-gl) Rims
In a significant population of chondrules (10 of the 27 studied), glass with high silica contents is observed.We refer to these as Si-glass (Si-gl) rims.In most Si-glass rims, fractional crystallization of plagioclase results in very high silica contents of the residual glass, for example, >80 wt% SiO 2 (see below).In several cases, Raman spectroscopy was needed to distinguish the amorphous glass phase from crystalline silica (Figure 1).
Examples of Si-glass rims are shown in Figure 6.PP chondrule EET 92062 Ch5 has a wide (up to 270 lm, Table 2) igneous rim, which we describe as an Si-gl SIR.Areas of mesostasis, consisting of smooth (in BSE images) silica-rich glass and plagioclase, occur as a mesostasis interstitial to low-Ca pyroxene, Ca-rich pyroxene, and Fe,Ni metal (Figure 6a).A similar texture occurs in the Si-gl SIR of POP chondrule QUE 99177 Ch6 (Figures 6b and AppC_S6).A wide (200 lm, Table 2) Si-gl rim occurs on PP chondrule, MET 00426 Ch4 (Figure 6c).The Si-glass component of this rim occurs with elongate plagioclase in a fine-grained mesostasis, interstitial to pyroxene.Olivine is present in low abundance (2 grains within the whole rim).In an Sigl rim surrounding PP chondrule QUE 99177 Ch10, Siglass, plagioclase, and Ca-rich pyroxene make up a finegrained mesostasis interstitial to low-Ca pyroxene (Figure 6d).In some parts of the SIR, scattered small (<10 lm) grains of cristobalite are also present in low abundance (Figure 6d), which means that these areas could be described as having a common SIR texture.
We can make several general statements about Siglass rims.Fe,Ni metal tends to be in low abundance in chondrules surrounded by Si-glass rims and is concentrated at the edge of the chondrule or within the Siglass rim itself.Si-glass rims either completely or partially surround chondrules, with variable widths.The rims may arguably be referred to as IRs rather than SIRs, but we define these as SIRs because the glass is significantly higher in Si than in typical IRs (see compositions discussed below) and because they can easily be misidentified as containing crystalline silica in an x-ray mapping survey.The typical texture of an Si-gl rim consists of grains of pyroxene and Fe,Ni metal, with an interstitial mesostasis of plagioclase, Ca-rich pyroxene, and Si-rich glass.

QUE 99177 Ch7: A Composite Rim with a MultiLayered (ML) SIR Region
Chondrule QUE 99177 Ch7 is a POP chondrule that has plagioclase and cristobalite in its mesostasis (Figure 7).It contains little Fe,Ni metal, but it is partially surrounded by an Fe,Ni mantle (Figure 7a).The SIR is complex as different parts of it consist of different types of SIR, including parts with a common texture, containing cristobalite (Figure 7b), and parts with an Si-gl texture in which plagioclase and Si-rich glass form a fine-grained honeycomb texture interstitial to pyroxene (Figure 7c).One section of the SIR is multilayered (Figures 7d,e and  AppC_S7): Adjacent to the chondrule, there is a layer consisting entirely of fine-grained (10-25 lm in length) low-Ca pyroxene.This is overlain by a well-defined monomineralic layer of subrounded cristobalite, grain size <10 lm, and the cristobalite is in turn overlain by a wider (~20 lm) SIR with a fine-grained Si-gl texture that consists of Si-rich glass, low-Ca pyroxene, Ca-rich pyroxene, and plagioclase.
EET 92042 Ch11: Tridymite Rim EET 92042 Ch11 is an elongate, irregularly shaped PP chondrule with a maximum diameter of ~900 lm (Figure 7f).Fe,Ni metal is present in the outer zone of the host chondrule, but does not form an Fe,Ni mantle.The SIR around this chondrule is unique among the chondrules studied, in that it contains tridymite rather than cristobalite (Figure 1).The SIR has a sharp boundary with the chondrule (Figure AppC_S8).It has a maximum thickness of ~80 lm and only partially surrounds the chondrule.The SIR consists almost entirely of tridymite, which occurs as subrounded to subangular grains and which are notably larger (up to 50 lm in diameter) than cristobalite in other rim types (Figure 7g).Very minor amounts of Ca-rich pyroxene and glass are also present (Figure AppC_S8).There is extensive Fe-veining in the tridymite, and a "smooth" Fe-and K-rich alteration rim at the boundary with matrix (Figures 7g and AppC_S8).

Mineral and Glass Compositions of SIRs and Host Chondrules
We measured mineral compositions in selected chondrules that represent the different types of SIRs, including three common rims (QUE 99177 Ch5, EET 92042 Ch14, EET 92062 Ch2), one with silica on the edge (MET 00426 Ch3), the multilayered monomineralic rim QUE 99177 Ch7, the tridymite rim (EET 92042 Ch11), and two silica glass rims (QUE 99177 Ch7 and QUE 99177 Ch6).Compositions of olivine, pyroxene, plagioclase, glass/mesostasis, and silica from chondrules and their SIRs are given in Tables 3-8, and shown in Figures 8-14.Individual analyses are provided in the supporting files (Supplementary Material, Appendix D).Most compositions were extracted from quantitative WDS x-ray maps as described above (see Appendix A and  Figures AppC_S1-S8), with the exception of a few olivine and pyroxene point analyses.

Olivine
Olivine analyses were only obtained from chondrules as olivine is rare in SIRs.Olivine grains in chondrules are magnesium-rich (Fo 95.9-98.2 ) and contain low contents of MnO (<0.5 wt%) and CaO (<0.3 wt%) (Table 3, Figure 8).These data are similar to analyses of olivine from chondrules with SIRs reported by Krot et al. (2004).

Low-Ca Pyroxene
Low-Ca pyroxene compositions are close to enstatite endmember compositions (Table 4, Figure 9).Generally, low-Ca pyroxene in SIRs has higher contents of CaO, MnO, and Cr 2 O 3 compared to chondrules (Table 4, Figure 10).In plagioclase-rich chondrule EET 92062 Ch2 (Figure 4), low-Ca pyroxene compositions are different from most other chondrules (e.g., lower FeO and higher CaO: Figure 10a), and host chondrule and SIR compositions are quite similar.Pyroxene in SIRs may have higher or lower Al 2 O 3 contents compared with pyroxene in the host chondrules.CaO, MnO, Cr 2 O 3 , and Al 2 O 3 contents are similar among the common SIRs,  Note: For mixed point and WDS map analyses, detection limits are based on the WDS maps.<n indicates the detection limit for each oxide.P 2 O 5 and NiO were also measured but values were below detection.a No. of analysis areas refers to the number of 10 9 10 pixel boxes in WDS maps that were combined to give the reported analysis.Silica-rich igneous rims in CR chondrites whereas the Si-gl rims show a wider range in CaO and MnO contents.
A chondrule with a composite SIR, QUE 99177 Ch7, contains a multilayered (ML) sequence including monomineralic layers of pyroxene and cristobalite (Figure 7d).Low-Ca pyroxene in the pyroxene rim (PR) layer is more similar to the host chondrule than low-Ca pyroxene in the SIR, although low-Ca pyroxene in the PR has lower Al 2 O 3 contents than pyroxene in the chondrule.

Ca-Rich Pyroxene
We only obtained a limited number of analyses for Ca-rich pyroxene in SIRs, because they are typically very fine-grained.For Ca-rich pyroxene, average Wo contents are 34.8-40.5,consistent with augite (Table 5, Figure 9).There is a range of minor element compositions, and heterogeneity within individual SIRs and chondrules (Figure 11).Typical MnO and Cr 2 O 3 contents up to 3 wt % are visible in the quantitative WDS x-ray maps of the common rims in QUE 99177 Ch5 (Figures 3 and  AppC_S2), QUE 99177 Ch6 (Figures 6b and AppC_S6), and EET 92062 Ch2 (Figures 4 and AppC_S4), as well as in the host chondrule of EET 92062 Ch2 (Figure AppC_S3).WDS maps show that MnO, Cr 2 O 3 , and TiO 2 contents at the outer edge of Ca-rich pyroxene grains can be very high, for example, up to 8 wt% MnO (Figure AppC_S6).An unusual, Wo-rich pyroxene (Wo 50.5 ) from the Si-e rim in MET 00426 Ch3 (Figure 5e-g) is very MnO-rich (average 8.6 wt%) and Cr 2 O 3 -rich (average 4.1 wt%, Table 5).Intermediate pyroxene compositions, presumably pigeonite, were observed in zoned pyroxene grains in the SIRs of EET 92042 Ch14 (Figures 2f, 9, 11, and AppC_S1) and QUE 99177 Ch6 (Figures 6b and  AppC_S6).In the three chondrules for which we have data from both chondrule and SIR (EET 92062 Ch2, QUE 99177 Ch5, and MET 00426 Ch3), no consistent trends between the two are observed.

Plagioclase
Plagioclase in chondrules and SIRs shows a wide range of compositions, An 60-91 (Table 6, Figure 12).K 2 O is below detection limits in most analyses, but plagioclase in Si-gl chondrule QUE 99177 Ch6 has measurable K 2 O (Or 1.1 ).Plagioclase in each chondrule or SIR is quite homogeneous, apart from in the plagioclase-rich chondrule EET 92062 Ch2, which has compositions An 87- 93 (Figure 12a): Plagioclase in the host chondrule has narrow (<5 lm) Na 2 O-rich rims (Figure AppC_S3), but these are not observed in the SIR (Figure AppC_S4).For two chondrules in which we analyzed both SIR and host chondrule plagioclase (QUE 99177 Ch6 and EET 92062 Ch2), host chondrules are more anorthitic than their corresponding SIRs (Figure 12b).

Silica
In all but one chondrules, silica is cristobalite.It is nearly pure SiO 2 , although with some variation: SiO 2 contents are 93.7-99.6 wt% (Table 7) and it invariably contains small amounts of Al 2 O 3 (up to 2 wt%), CaO (up to 0.9 wt%), Na 2 O (up to 0.6 wt%), and MgO (up to Note: <n indicates the detection limit for each oxide.NiO was also measured but values were below detection.Abbreviation: Pig, pigeonite.a No. of analysis areas refers to the number of 10 9 10 pixel boxes in WDS maps that were combined to give the reported analysis.
b Average composition includes pyroxene from both the ML and the Si-gl SIRs.
0.8 wt%) (Figure 13).Na 2 O and Al 2 O 3 are positively correlated (Figure 13b).It is possible that some minor elements are attributable to analytical overlap with adjacent phases (glass, plagioclase), but we made considerable effort to select areas from the maps that were centered on silica grains, away from grain edges where possible.We also isolated silica grains from surrounding phases by pixel filtering as described in the Methods section and outlined in Appendix A. FeO contents vary from 0.6 to 1.4.Although FeO filters were placed on WDS maps when extracting data, in order to avoid Fe-rich alteration (see Appendix A), alteration is extensive in some rim areas and difficult to avoid completely.Therefore, it is possible that the FeO contents measured in silica grains are from beam interaction with the surrounding alteration and veining and not actually a component of the silica grains.Alternatively, it is possible that some of the FeO may be indigenous since there are significant contents of other elements in silica grains.
For EET 92062 Ch2, we analyzed cristobalite in both the SIR and the host chondrule.Both have the lowest SiO 2 content (93.7-95.9wt%) of the silica grains analyzed (Table 7, Figure 13).Raman spectra show sharp peaks at ~110, 220, and 420 cm À1 , diagnostic of cristobalite, for both occurrences, so it is clearly not glass.The SIR cristobalite has higher contents of CaO, Na 2 O, Al 2 O 3 , and MgO than the host chondrule.
The SIR of EET 92042 Ch11 contains tridymite (Figure 7f,g).The tridymite composition is close to pure SiO 2 (99.6 wt%, Table 7).It has measurable Al 2 O 3 (0.2 wt %) which is lower than most cristobalite analyses, and most other elements are below detection.The measured FeO content of 0.74 wt% could be the result of contamination since tridymite is extensively altered (Figures 7g and AppC_S8).

Glass and Mesostasis
Glass analyses were extracted from WDS maps of the Si-gl rims QUE 99177 Ch7 and QUE 99177 Ch6, the Si-e rim and host chondrule MET 00426 Ch3, and mesostasis analyses were extracted for the common rim QUE 99177 Ch5.We define mesostasis as fine-grained interstitial material representing the final residual liquid: Mesostasis typically contains glass with fine-grained crystalline material.Average glass and mesostasis compositions in SIRs and their host chondrules have high contents of SiO 2 (66-90 wt%), Al 2 O 3 (5-19 wt%), CaO (up to ~9 wt%), and Na 2 O (up to ~6 wt%) (Table 8, Figure 14).Each SIR has a well-defined glass or mesostasis composition, but overall SiO 2 and Al 2 O 3 are negatively correlated (Figure 14a).Glass in two Si-gl rims (QUE 99177 Ch6 and Ch7) has SiO 2 contents >80 wt%, but quite different Al 2 O 3 and TiO 2 contents.Because both of these Si-gl rims contain plagioclase, we attribute the high SiO 2 contents to  Note: <n indicates the detection limit for each oxide.NiO was also measured but values were below detection.Abbreviations: Gl, glass; Mes, mesostasis.a No. of analysis areas refers to the number of 10 9 10 pixel boxes in WDS maps that were combined to give the reported analysis.
fractional crystallization of plagioclase resulting in a very silica-rich residual liquid.In comparison, the chondrule and the Si-e SIR of MET 00426 Ch3 do not contain plagioclase, and in both the glass is less Si-rich and more feldspathic (total cation values of 4.7-4.8 on the basis of 8 oxygens: Table 8).Glass in the Si-e SIR has lower Al 2 O 3 and CaO contents, and higher SiO 2 and K 2 O contents, compared to the host chondrule, and glass in the SIR is more homogeneous than glass in the chondrule (Figure 14).

Bulk Compositions
We calculated bulk compositions for two SIRs using a modal recombination analysis (MRA) method, taking into account the densities of the individual phases as described by Berlin (2009).We determined vol% abundances via point counting on grids overlain on BSE images.Mineral densities were obtained from mindat.org.Densities of glass and mesostasis were computed according to Fluegel (2007) and Fluegel et al. (2008).Compositions of individual phases were taken from Tables 3 to 8. The computed bulk compositions, determined for the silicate portion of the rims only (i.e., excluding metal), are given in Table 9.
The two SIRs were selected as representative of their rim types: (1) the common part of SIR QUE 99177 Ch5 (Figure 3) and (2) the Si-gl rim QUE 99177 Ch6 (Figure 6b).The common rim has a high SiO 2 content, 72 wt%, as well as significant MgO (17 wt%), CaO (5 wt %), and Al 2 O 3 (3.1 wt%) (Table 9).In contrast, the Si-gl rim has significantly lower SiO 2 (58.4 wt%) as well as higher Al 2 O 3 and MgO.Equilibrium liquidus temperatures were computed for these bulk compositions, as well as for an SIR bulk composition from Krot et al. (2004), using the MELTS for Excel software which implements the rhyolite-MELTS thermodynamic model (Asimow & Ghiorso, 1998;Ghiorso & Sack, 1995) (Table 9).Note: <n indicates the detection limit for each oxide.P 2 O 5 and NiO were also measured, but values were below detection.a No. of analysis areas refers to the number of 10 9 10 pixel boxes in WDS maps that were combined to give the reported analysis.

DISCUSSION
We studied 27 SIRs and their host chondrules from five CR chondrites, in order to better understand SIR origins, crystallization history, and their relationship to the host chondrules.We describe different types of SIRs which all occur around porphyritic type I chondrules: common (14/27 chondrules) and silica on the edge (Si-e) of the chondrules (7/27 chondrules).Some chondrules have sections of more than one type of SIR.Two unique rim types are a multilayered (ML) rim that includes a layer of monomineralic silica, and a tridymite-bearing rim (Trid).Cristobalite is the dominant polymorph in SIRs, with the exception of the one tridymite rim observed.Rims in which silica has not crystallized, but in which there is a silica-rich glass, are also common (10/27 chondrules).These can be mistaken for silica mineral-bearing assemblages in mapping surveys.We describe these as SIRs, within the definition that the acronym SIR means silica-rich igneous rims (the term coined by Krot et al., 2004).If the acronym is interpreted as silica mineral-bearing igneous rims, these would not be classified as SIRs.

Models for Formation of SIRs
There are currently two main models for SIR formation, proposed by Noguchi (1995), Krot et al. (2004), andRubin (2018).Noguchi (1995) and Krot et al. (2004) presented arguments against several models, including reduction of ferromagnesian silicates in chondrule peripheries, partial melting of the outer parts of chondrules, and fractional crystallization of chondrules.We concur with these conclusions.Krot et al. (2004) argued that most evidence points to accretion of silica-rich materials onto chondrule surfaces, followed by melting and near-equilibrium crystallization: This is illustrated as Model 1 in Figure 15a.Addition of solid precursors onto solidified chondrule surfaces is similar to that proposed for IR formation (e.g., Weisberg et al., 1993).However, Krot et al. (2004) also argued that there is evidence for direct condensation of SiO (g) into chondrule melt (Tissandier et al., 2002) for some chondrules and their SIRs (Model 2, Figure 15b).For Model 1, the silica-rich material was proposed to originate from a highly fractionated nebular gas (Si/Mg >1), with fractionation resulting either from condensation and partial isolation of forsterite (Petaev & Wood, 1998), or from partial evaporation during chondrule formation.Condensation of silica from a fractionated nebular gas has also been invoked to account for the presence of silica in an amoeboid olivine aggregate (Komatsu et al., 2018).Rubin (2018) proposed an alternative model for producing solid silica-rich material (Model 3, Figure 15c) which involves equilibrium fractionation of the nebular gas.Krot et al. (2004) argued that the following lines of petrologic and mineralogical evidence require the need of Si-rich solid precursor material (Model 1): The presence of SIRs around layered chondrules; irregular outlines of the rimmed chondrules; distinct compositional boundaries (particularly as observed in x-ray maps) between host chondrules and their rims in elements such as Si, Mn, and Na; and the presence of silica-rich chondrules that are not layered.Our similar observations lead us to agree with this interpretation.A similar model involving gas-solid condensation has also been suggested by Hezel et al. (2006) who studied SiO 2 -rich fragments and chondrules in ordinary chondrites, and determined that they formed by condensation of an Si-rich phase from a fractionated nebular gas, followed by melting.
In Model 2 (Krot et al., 2004), gas-melt condensation of SiO (gas) occurs directly into a partially solidified chondrule.This condensation model is supported by the experimental work conducted by Tissandier et al. (2002) who were investigating the formation of mineralogically zoned type I chondrules.The experimental study produced silica on the margins of a melted chip of CM chondrite Murchison, in one experiment that was held at a peak temperature of ~1400°C for 150 s in an SiO-rich vapor and then quenched.The resulting texture of silica, pyroxene, and glass is similar to that of some SIRs.Model 2 was proposed by Krot et al. (2004) to explain observations such as the presence of nonlayered porphyritic chondrules, for example, chondrules which do not contain multiple layers such as an IR and an Fe,Ni mantle, some of which contain silica in an SIR in direct contact with olivine, and others in which pyroxene separates olivine from the SIR material.Krot et al. (2004) also argued that this model explains the compositional trends in host chondrules and associated SIR mesostasis  (Libourel et al., 2006;Matsunami et al., 1993;Nagahara et al., 1999).
In the Rubin ( 2018) model (Model 3), equilibrium condensation of a nebular gas of solar composition (pressures of 10 À3 to 10 À4 atm: Grossman (1972) and Grossman and Larimer (1974)) produces pyroxene-rich dust which aggregates onto the solid surfaces of chondrules.These dust aggregates then become enriched in volatile elements Na, K, and Mn which evaporated from nearby molten chondrules, and condensed onto the dust aggregates before they were accreted onto chondrule surfaces.Following melting, silica crystallizes along with enstatite.Despite arguing for an accretion model which does not require high partial pressures of SiO (g) in the nebula, Rubin (2018) states that high partial pressures of SiO (g) cannot be ruled out.However, Rubin (2018) also points out that the observations of zoned mesostasis by Nagahara et al. (1999), which Krot et al. (2004) used to support their condensation model, can be attributed to parent body aqueous alteration rather than condensation.

Formation Conditions of SIRs
We now discuss the conditions of formation for each of the rim types we describe, based on petrography, mineral/glass chemistry, and silica polymorphs observed.We assess the extent to which our observations support or argue against the different models proposed for SIR formation in CR chondrites.

Abundance of SIRs and Effects of Alteration
CR chondrites experienced varying degrees of aqueous alteration but little thermal metamorphism (Alexander et al., 2013;Harju et al., 2014;Krot et al., 2002Krot et al., , 2004;;Noguchi, 1995;Schrader et al., 2011;Wasson & Rubin, 2010;Weisberg et al., 1993).Most CR chondrites are thus classified as petrologic type 2. Subtypes 2.1-2.8,where 2.8 is the least altered, have been proposed by Alexander et al. (2013), based on bulk H contents in water/OH, and by Harju et al. (2014) based on the presence of alteration features and whole rock oxygen isotope compositions.Our study focused on the least-altered CR chondrites (subtype 2.8) because they should best preserve pristine chondrule properties.However, alteration of silica to an Fe-rich amorphous phase (Mart ınez & Brearley, 2022), which may also be Kand P-rich, is pervasive in all the chondrites studied (e.g., Figures 3f,h, 5b, and 7f).
The susceptibility of cristobalite to alteration, coupled with terrestrial alteration that has resulted in Fe-veining and oxidation of metal (e.g., Figures 3e,f, 4c,d, 5a,e,h, 6b, and 7b,f), makes it difficult to determine the abundance of SIRs.Krot et al. (2002Krot et al. ( , 2004) also argued that the absence of SIRs in more altered CR chondrites such as Renazzo and Al Rais is attributable to the replacement of silica.Our survey indicates that SIRs surround 10%-27% of chondrules present in this study (Table 1).However, this is likely a lower limit as it is possible that some SIRs were overlooked, in addition to the possibility that some silica may be obscured by alteration.

Interface between Host Chondrules and Rims
The presence of a sharp chondrule/rim boundary is important for interpreting formation models of SIRs.A sharp boundary strongly favors models in which the host chondrule is solidified prior to accretion of the rim (Models 1 and 3, Figure 15), and would not be expected in a gas/melt condensation model (Model 2).Krot et al. (2004) invoke Model 2 to explain textures where a chondrule has an outer, silica-rich zone in a continuous igneous texture with the chondrule interior (their Figures 5 and 6).We suggest that an alternative explanation for this texture is that it represents more extensive melting of an SIR produced by melting a solid, accreted layer of Si-rich dust, resulting in mingling of silica-rich material with the outer zone of the chondrule.We did not observe a comparable texture in any of the chondrules studied.Our observations of melting at the interface are limited, for example, we infer that incipient melting of pyroxene in contact with an SIR occurred in Figure 5a,b.Noguchi (1995) also suggested that the absence of sharp chondrule/rim boundaries indicates that chondrules must have been partially molten when the SIR accreted.However, Komatsu et al. (2019) indicated that there may be a disagreement on whether boundaries between host chondrules and their associated rims are sharp, as aqueous alteration of the CR parent body may have affected the interface.We therefore need to be aware that variability in the presence of sharp boundaries between host chondrules and rims may be linked to the extent of aqueous alteration experienced by the chondrite.
For most chondrules in this study, we observe sharp boundaries between host chondrules and their associated SIRs, as well as between IRs and SIRs.This favors an accretion and melting model, that is, a similar mechanism that is proposed to form IRs. We see no correlation between rim type and the sharpness of the boundary.The preservation of a sharp interface between SIRs and their host chondrules indicates that in most cases, the host chondrule did not experience significant reheating during rim formation.
Fe,Ni metal mantles further support these arguments.Several chondrules in this study have Fe,Ni mantles that separate the host chondrule from either the IR or SIR (e.g., Figures 3a,b, 4a,b, and 5c-e).The rounded to subrounded morphology of Fe,Ni metal grains within the mantles (e.g., Connolly et al., 2001;Krot et al., 2002;Wasson & Rubin, 2010) indicates that the metal was initially an immiscible melt within the molten chondrule.Wasson and Rubin (2010), following Wood (1963), proposed that during chondrule formation, metal from the chondrule interior migrated outward and wetted the chondrule surface.Based on trace element geochemistry, Jacquet et al. (2013) extended this and argued for a two-stage cooling model, in which some of the metal originated external to the host chondrule (i.e., accretion of rim material containing Fe,Ni metal grains).Whatever the cooling process, it is clear that if an IR or an SIR overlays a metal mantle, the chondrule plus the Fe,Ni mantle must have solidified prior to the formation of the IR/SIR.Since the metal surrounds the host chondrule, condensation of SiO (g) directly onto a silicate melt at the chondrule surface (Model 2) would not be a viable explanation for SIR formation.

Petrographic Constraints on SIR Formation Conditions
In previous studies of SIRs, formation conditions have not been constrained clearly.Petrography and mineral compositions can be used to interpret parameters such as peak temperature and cooling rates, as well as the chemical evolution that took place between formation of chondrules and formation of SIRs.
Common SIRs contain silica, low-Ca pyroxene, Carich pyroxene, AE glass/mesostasis, AE plagioclase, AE minor olivine, AE Fe,Ni metal (Table 2).They are consistently much finer grained than their host chondrules, which likely indicates a high number of nucleation sites and hence a short period of time at nearliquidus temperatures (see e.g., Lofgren, 1996).Igneous textures of common SIRs, including the presence of glass, indicate crystallization from a rim that was essentially entirely molten.Based on texture, it is hard to determine whether silica or low-Ca pyroxene crystallized first: They potentially co-crystallized.Low-Ca pyroxene crystallized before Ca-rich pyroxene, followed by plagioclase, and glass represents the residual liquid.Rare olivine grains could be relict grains, or the first phase to crystallize.Estimated liquidus temperatures of SIRs are variable, because the proportions of minerals are variable.The liquidus temperatures computed for the bulk Note: For our data, bulk compositions are for the silicate-only portion of the rims (i.e., metal free).a Krot et al. (2004) bulk composition is for the SIR, PCA 91082 #4.Liquidus temperatures were computed with "MELTS for Excel." compositions of a common SIR and Si-gl SIR from our study are 1557 and 1672°C, respectively, and an SIR bulk composition from Krot et al. ( 2004) is 1486°C (Table 9).Compositional data from the chondrules we studied generally support previous observations on compositional differences of minerals between host chondrules and SIRs (e.g., Krot et al., 2004).For example, low-Ca pyroxene in several SIRs generally has higher CaO, MnO, and Cr 2 O 3 compared to the host chondrules (Figure 10).However, there are exceptions: for example, in plagioclase-rich chondrule EET 92062 Ch2 (Figure 4), in which the host chondrule and SIR have very similar textures, compositions of low-Ca pyroxene in the host chondrule and SIR are similar.Unlike Krot et al. (2004) and Mart ınez and Brearley (2022), our data have similar FeO contents for low-Ca pyroxene in SIRs and their host chondrules.This difference may be because EPMA point analyses can overlap with Fe-rich alteration, which is hard to avoid in fine-grained SIRs, whereas with our WDS map analyses, we could filter out Fe-rich alteration.
Data from Krot et al. (2004) show that Ca-rich pyroxene in SIRs contains higher Cr 2 O 3 and MnO contents and lower Al 2 O 3 and TiO 2 contents compared to chondrules.However, our data do not show such a clear relationship (Figure 11).We also measured lower Al 2 O 3 than in the Krot et al.'s data, overall.Since Ca-rich pyroxene crystallizes at a late stage, its chemistry will depend on what other phases have crystallized before.It is also strongly zoned and can have extreme enrichments in MnO and TiO 2 at the margins of grains (Figures AppC_S1,S2,S6).Therefore, we can expect measured Ca-rich pyroxene compositions to be variable, in which case they are probably not as reliable as low-Ca pyroxene for interpreting the evolution of bulk compositions from chondrule to SIR (which we discuss below).
In most SIRs, plagioclase occurs typically as finegrained intergrowths in mesostasis.Krot et al. (2004) did not observe plagioclase, but we find it to be a common mineral in SIRs (Table 2).One chondrule we studied, EET 92062 Ch2, contains coarse grains of plagioclase in both the host chondrule and the SIR (Figure 4).In two chondrules, plagioclase in the host chondrule is more anorthitic than in the SIR (Figure 12), consistent with the SIR precursor material being more Na 2 O-rich than the chondrule.Wick and Jones (2012) showed that in order for plagioclase to nucleate and grow in type I POP chondrules from CO chondrites, slow cooling rates are needed at low temperatures (e.g., 1-10°C h À1 at 800-1000°C).Although SIRs have different bulk chemistries to POP chondrules, we suggest that nucleation of plagioclase will still be inhibited, and cooling rates need to be slow enough to account for its presence.FIGURE 15.Literature models for SIR formation.(a) Model 1: Gas-solid condensation of Si-rich material followed by accretion onto solidified chondrules which is subsequently melted and cooled (Krot et al., 2004).(b) Model 2: Gas-melt condensation into chondrule melts followed by cooling (Krot et al., 2004).(c) Model 3: Pyroxene-rich dust, enriched in Na, K, and Mn evaporated from chondrules, accretes onto solidified chondrules prior to melting and cooling (Rubin, 2018).(Color figure can be viewed at wileyonlinelibrary.com) Our glass and mesostasis analyses show similar variation trends as those of Krot et al. (2004) although our data extend to higher values of SiO 2 (Figure 14).Krot et al. (2004) saw a trend of higher Si, Mn, Na, and K and lower Ca, Mg, Al, and Cr in SIR glass/mesostasis than in chondrule glass/mesostasis.Unlike Mart ınez and Brearley (2022), our glass analyses have totals close to 100 wt%.These authors suggested that their low EPMA totals in glass from SIRs (87-95 wt%) could be partly attributable to the glass being hydrated, but we see no evidence for this, and prefer their alternative explanation that it was an analytical problem in their study.We do not have enough quantitative data to demonstrate the host chondrule/SIR relationship, but in general, SIR glass and mesostasis have low CaO and Al 2 O 3 contents that lie on negatively correlated trends on oxide versus wt% SiO 2 diagrams, along with data from other studies (Figure 14a,b).We only analyzed one chondrule for comparison of glass compositions in both, MET 00426 Ch3, in which Na 2 O and TiO 2 are similar in the glass of the host chondrule and the SIR, but K 2 O is higher in the SIR glass.The Na 2 O, K 2 O, and TiO 2 contents of glass and mesostasis in SIRs are quite variable, which might be expected following crystallization of plagioclase, and Carich pyroxene with variable compositions, as well as variability in the fraction of liquid remaining that is now represented as glass/mesostasis.
SiO 2 -rich glass in SIRs represents the residual liquid in a system that has undergone fractional crystallization.We would expect a silica mineral to crystallize from this melt, and attribute its absence to sluggish kinetics.The importance of identifying Si-glass rims is that they represent a different bulk composition to common SIRs, and hence a different crystallization sequence: Silica would crystallize last in an Si-gl rim, as opposed to being the first or second phase to crystallize in common rims.From examination of Si x-ray maps alone, it is not always easy to identify glass, because SIRs are very fine-grained.We include Si-glass rims under the definition of SIR, with the explicit statement that SIR means silica-rich igneous rim, rather than silica-bearing igneous rim.The bulk composition we determined for an Si-gl rim (Table 9) has 58 wt% SiO 2 , intermediate between the bulk composition of a pyroxene-rich IR determined by Krot et al. (2004), 53 wt% SiO 2 , and common SIRs, which have >70 wt% SiO 2 .
Our analyses of cristobalite in SIRs have high contents of CaO, Na 2 O, Al 2 O 3 , and MgO compared to cristobalite in the host chondrule.Krot et al. (2004) reported two analyses of silica, both with lower minor element contents (Figure 13).Our data are likely more representative as they include more analyses and more diverse rim textures.Measurable amounts of minor elements are not unusual in cristobalite in other planetary settings.In Figure 17, we compare our data with literature data for cristobalite from ordinary and enstatite chondrites, as well as igneous compositions from eucrites, lunar, Martian, and terrestrial sources.The data include cristobalite that has formed during chondrule crystallization, chondrite metamorphism, chondrite partial melting, planetary igneous processes, and in impact processes.Minor element contents are variable with no clear process-related trends, and it is difficult to use the chemical compositions to interpret the compositions measured.
Silica has several stable polymorphs at low pressure, including cristobalite (1713-1470°C), tridymite (857-1470°C), and quartz (<857°C) (Figure 16).We found that cristobalite is the dominant silica polymorph in SIRs and host chondrules, consistent with the work of Komatsu et al. (2019) and Mart ınez and Brearley (2022), indicating formation temperatures >1470°C based on the silica phase diagram.Because there is little evidence at the chondrule/ SIR boundary that the chondrule was re-melted, we suggest that the temperature of formation of the SIR is likely at the lower end of the cristobalite stability field and that the duration of heating was short.The unique occurrence of tridymite in the SIR of EET 92042 Ch11 indicates a peak temperature of <1470°C according to the silica phase diagram (Figure 16), although there are questions about this as discussed further below.Peak temperatures around 1500°C are comparable to those proposed for igneous rims.Kring (1991) estimated IR peak temperatures between 1000 and 1500°C.Connolly and Hewins (1991) and Krot and Wasson (1995) estimated IR peak temperatures of ≥1000 and ≥1126°C, but these are minimum temperatures and do not estimate the peak temperatures reached.Matsuda et al. (2019) put upper limits on times and temperatures of heating of IRs (e.g., several days at around 1720°C or up to a year at around 1100°C) and argued that transient heating events for IRs are similar to those of chondrules.Based on the above arguments, including liquidus temperatures and silica polymorphs, we suggest that peak temperatures of SIRs were generally around or above 1500°C.
However, interpretations of the origins of silica polymorphs based on the silica phase diagram are not straightforward.Our recent experimental work (Smith, 2022;Smith & Jones, 2020) suggests that silica polymorphs are not necessarily robust indicators for peak temperature estimates.In particular, we have shown that cristobalite can crystallize at temperatures well below 1470°C in a system with SIR bulk chemistry.Ono et al. (2021) also observed primary cristobalite in experiments on a eucrite composition, during cooling at rates of 0.1-1°C h À1 from peak temperatures 1250-1300°C.In addition, the origin of tridymite in the rim on EET 92042 Ch11 is not clear.The tridymite is located right on the edge of the host chondrule and there is a sharp boundary between the chondrule and the rim (Figure 7f).Minor amounts of Ca-rich pyroxene and glass in the SIR indicate that melting occurred.The chemical composition of the rim is close to pure SiO 2 , with an estimated liquidus temperature that is likely high, around 1650°C.Clearly, this is inconsistent with crystallization of tridymite from a melt, based on the silica phase diagram.Ono et al. (2021) suggest that formation of tridymite could occur via transformation from cristobalite at very slow cooling rates (<0.1°Ch À1 ), but for reasons discussed below, we rule out such slow cooling for SIRs.
There are suggestions in the literature that the silica phase diagram, which is based on experiments by Fenner (1913), is not accurate, and that tridymite is not a stable phase in the pure silica system (Buerger, 1935;Deer et al., 1992;Lakshtanov et al., 2006;Stevens et al., 1997).For example, Stevens et al. (1997) showed that cristobalite can be formed by the heating of pure quartz but tridymite cannot.They determined that in a pure silica system, quartz is converted to cristobalite at temperatures >1300°C.For tridymite to form, a flux or mineralizer is required: for example, Na 2 WO 4 was used by Fenner (1913) and sodium carbonate and potassium carbonate were used by Stevens et al. (1997).When a flux or mineralizer is present, tridymite can form and quartz transitions to either tridymite or cristobalite between 1100 and 1600°C.To add to the complexity, Hill and Roy (1958) suggested that two forms of tridymite exist: A metastable tridymite (tridymite-M) can form a stable form of tridymite (tridymite-S) but only after converting to cristobalite first.Experiments conducted by Hill and Roy (1958) at temperatures at ~930°C for 164-279 h, and in the presence of water, formed both cristobalite and tridymite.In these experiments, tridymite was never the sole polymorph as observed in the SIR surrounding EET 92042 Ch11.The presence of tridymite in this SIR is therefore puzzling.The experimental literature would suggest that it formed from a different polymorph, in the presence of a flux or mineralizer, in the temperature range 1100-1600°C.This could potentially be attributed to an episode of aqueous alteration for the case of a CR chondrite, although temperatures are unlikely to have been this high.Also, it is unclear why a flux or mineralizer would affect this one rim and no others, and since the pervasive product of aqueous alteration of silica is an amorphous Fe-rich silicate phase, there does not seem to be any evidence that a flux reaction is responsible for formation of tridymite.Secondary heating of a different polymorph (cristobalite) into the tridymite stability field is also unlikely, as there is no petrologic evidence for secondary heating to temperatures >850°C required by the phase diagram for formation of tridymite, and it is also unlikely that only one chondrule would have been affected.Tridymite in EET 92042 Ch11 has low minor element contents, but in a comparison with literature data for tridymite in other planetary settings (Figure 17), the chemical composition is difficult to interpret in terms of its formation process.

Implications for SIR and Chondrule Formation Models
The above discussion allows us to evaluate the formation models for SIRs in the context of chondrule formation mechanisms.We concur with previous conclusions that silica-normative precursor material aggregated onto the surfaces of solidified chondrules and then subsequently melted (Krot et al., 2004;Noguchi, 1995;Rubin, 2018: Models 1 and 3, Figure 15).Lines of evidence supporting aggregation/melting models include the presence of SIRs around layered chondrules, irregular outlines of the rimmed chondrules, distinct compositional boundaries between host chondrules and their rims (e.g., in x-ray maps of elements such as Si, Mn, and Na), or between Fe,Ni mantles and rims, and the presence of silica-rich chondrules that are not layered.This model is similar to the generally accepted model for IR formation (e.g., Connolly & Hewins, 1991;Krot & Wasson, 1995;Prinz et al., 1986;Rubin, 1984Rubin, , 2018)), and it is logical that the same process was operating, since the features are clearly very closely related.Krot et al. (2004) argued that gas-melt condensation of SiO (gas) directly into a partially solidified chondrule could be necessary to explain mineralogical zonation of chondrules with silica-rich outer zones.However, we suggest that such chondrules could be explained simply by heating for a more extended time at the peak temperature, during which a silica-rich layer that had aggregated to the surface was incorporated into the chondrule melt.We observe some examples of incipient melting at the chondrule/SIR or IR/SIR boundaries, and we would expect some melting to occur when the SIR is melted.Condensation of SiO (gas) into partially molten chondrules has gained support recently, to explain mineralogical zonation of chondrules, that is, olivine-rich cores and pyroxene-rich mantles such as those in Figures 2e, 3b, and 5g (Barosch et al., 2019;Chaussidon et al., 2008;Friend et al., 2016;Jacquet et al., 2012;Libourel et al., 2006;Libourel & Portail, 2018;Tissandier et al., 2002).However, not all chondrules show this zonation, for example, Figures 2b, 5d, and 7a, and their textures are not consistent with sectioning effects of mineralogically zoned chondrules (Barosch et al., 2020).Thus, it is not a universal effect and it is not clear why some chondrules would be subject to this process while others escape this interaction with surrounding gas.It is possible that some chondrules record a progressive change in composition resulting from SiO (gas) condensation during melting, whereas others cooled and solidified before a layer of solid condensates aggregated onto the surface, to be melted during subsequent heating events.The majority of IRs and SIRs can be described best by the latter of these processes.The distinction is important because the former case (continuous condensation into a melt) only requires one heating/cooling event per chondrule, whereas a typical chondrule that has an IR overlain by an SIR requires three consecutive heating events.This puts important constraints on chondrule formation models.Additional evidence for a reheating model is that some chondrules show different types of SIR at different locations around the rim, and others have discontinuous SIRs.This could occur if small clumps of condensed solid material, with different compositions, attach themselves to a chondrule surface and melt very locally.But it is difficult to explain such an observation in a scenario in which the partially molten chondrule is exposed to SiO (gas) because we would expect this to result in a more uniform effect around the chondrule surface.
All models for SIR formation propose a continuous evolution toward more Si-rich bulk compositions (chondrule to IR to SIR).Although this is very commonly the case, there are exceptions.An example is the multilayered sequence observed in the SIR of QUE 99177 Ch7 (Figure 7a-e) which follows the progressively more Si-rich sequence of pyroxene to silica, but then this is overlain by an Si-gl SIR.Assuming that each of these layers represents separate deposition of material, the Si-gl SIR reverts to a lower bulk Si content.Fluctuating conditions need to account for layers of nearly pure SiO 2 such as those described as Si-edge (Si-e) and the tridymite rim, as well as the monomineralic silica layer in the multilayered sequence.The suggestion by Rubin (2018), that SIRs are derived from pyroxene-rich aggregates, and that silica within them is the product of fractional crystallization (Model 3, Figure 15), can account for some common SIRs, but it does not seem that it can explain the deposition of highly fractionated compositions close to pure SiO 2 .It is also not clear whether fractional condensation of silica-rich materials can extend to these more extreme compositions.
In order to understand the nature of chondrule formation events further, it is helpful to place constraints on the peak temperatures and cooling rates of SIRs.This should help to establish if there is an evolution of the heating process during successive heating events.However, the conditions recorded by SIRs are currently poorly constrained.The ambient temperature is quite well defined: It can be constrained to >530°C by the absence of sulfides (Krot et al., 2002(Krot et al., , 2004)), and condensation of Na and K requires temperatures around 700°C (Wood et al., 2019).As discussed above, we estimate that peak temperatures during the heating event are >1500°C.Retention of Na and other volatile species in chondrule melts is inferred to require either flash-heating or high partial pressures of Na or K vapor (Yu & Hewins, 1998).Cooling rates of SIRs are currently poorly constrained.Their fine grain sizes compared with chondrules could be controlled by a high number of nucleation sites resulting from incomplete melting, rather than being an indicator of a difference in cooling rate.The presence of glass in SIRs does not help to constrain cooling rates, since glass can form at a range of cooling rates including slower rates around 10°C h À1 (e.g., Wick & Jones, 2012).The presence of plagioclase in some SIRs indicates that cooling was relatively slow, around 10°C h À1 , at least at lower temperatures (Wick & Jones, 2012).However, the sharp boundaries retained at the interfaces between host chondrules/IRs/SIRs suggest a limited duration at high temperatures, and relatively rapid cooling, likely hundreds of °C h À1 , at least initially from peak temperatures.In experiments conducted by Ono et al. (2021) on a eucrite composition, cristobalite was partially transformed to quartz at a cooling rate of 0.1°C h À1 .Since we did not observe quartz, we can infer that cooling rates in SIRs were not this slow.
Chondrules containing silica minerals occur in a range of chondrite groups, including other carbonaceous chondrite groups, ordinary and enstatite chondrites (e.g., Hezel et al., 2006;Kimura et al., 2005;Olsen, 1983;Olsen et al., 1981), so the presence of highly evolved bulk compositions for chondrules is not unique to CR chondrites.However, layered chondrules that include SIRs are exclusively found around type I porphyritic chondrules in CR chondrites, and indeed are common in this group.The absence of SIRs in other chondrite groups indicates that a unique process occurred in the region of the disk where type I CR chondrules formed.Such differences between chondrite groups have been attributed to localized chondrule formation regions in the protoplanetary disk (e.g.,Jones, 2012;Kita & Ushikubo, 2012).Formation ages of CR chondrules from 26 Al-26 Mg dating are >1 Myr later than chondrules from O and CO chondrites (Nagashima et al., 2018;Schrader et al., 2017;Tenner, Ushikubo, et al., 2019): This later time might give rise to chemical and physical changes that allowed formation of SIRs.
The constraints on SIR thermal histories discussed above (ambient temperature around 500-700°C, peak temperature >1500°C, initial cooling rate likely hundreds of degrees per hour) are similar to thermal histories determined for chondrule formation (e.g., Jones et al., 2018;Krot & Wasson, 1995;Rubin, 1984;Rubin & Wasson, 1987).Thus, for an accretion/melting model of SIR formation, a repeated heating process with similar characteristics must be a requirement of a viable chondrule heating model.This is already a known constraint for chondrule formation based on the presence of relict grains and IRs (e.g., Connolly & Jones, 2016).For a gas/melt condensation model (for mineralogically zoned chondrules as well as SIRs), Tissandier et al. (2002) invoked thermal activity from the protosun (the X-wind model as proposed by Shu, 1997) which is suggested to allow for the high partial pressures of SiO (gas) required, as well as reaching appropriate peak temperatures.Desch et al. (2012) calculated that a chondrule cools at ~6°C h À1 in this model.Thus, it is plausible that this model can account for a gas/melt condensation mechanism for enriching the outer regions of chondrules with silica.However, this mechanism can only be proposed for a limited number of chondrules with SIRs, and the majority of SIRs require multiple heating events.We consider it unlikely that two very different formation environments are represented in a single chondrule population in CR chondrites, particularly given the need to confine CR chondrules to a limited spatial or temporal region of the protoplanetary disk.We prefer to argue that all chondrules with SIRs formed by the same mechanism.This means the heating process must be repeatable, with chemical evolution between successive events, and with each event reaching similar peak temperatures.Since we have not, to date, observed evidence for the unmelted precursors of silica-rich rims (i.e., silica-rich accretionary rims), it appears that the process of silica enrichment in solid materials must be related to the sequence of hightemperature events in the chondrule-forming region.The question of why chemical fractionation to highly silicarich compositions occurred in the CR chondrule formation region, and not in others, is open.It may be reversed to ask why chemical fractionation did not occur in the chondrule formation region of other chondrite groups.

SUMMARY AND CONCLUSIONS
Silica-rich igneous rims (SIRs) are a unique feature of CR chondrites.We concur with previous observations that SIRs are common as outer layers on MgO-rich POP chondrules.We studied SIRs in the least-altered CR chondrites, which means that interfaces between SIRs and their associated host chondrules, metal mantles, or igneous rims (IRs) are observed clearly.The majority of SIRs we studied have sharp boundaries at these interfaces, although we do observe evidence for incipient melting in some instances.Also in agreement with previous observations, we find that Mn, Cr, and Na contents of bulk SIRs, and their constituent phases, are generally higher than in the host chondrules.
We describe different textural types of SIRs.The most abundant type, which we name "common SIRs" consists of silica, low-Ca pyroxene, Ca-rich pyroxene, AE glass/mesostasis AE plagioclase AE Fe,Ni metal.We also describe SIRs that do not contain a silica mineral phase, but they have relatively high bulk SiO 2 contents and contain a very SiO 2 -rich glass (Si-gl SIRs).We include these as SIRs within the definition of the term, silica-rich igneous rims.Another distinct and common texture is where a narrow layer of almost pure silica is observed at the outer edge of a chondrule: We name this feature Si-edge (Si-e).In one example, a narrow layer of silica is itself overlain by an Si-gl SIR.For all these textural types of SIRs, and for silica present in the host chondrules, silica is cristobalite.However, we observed one SIR which consists almost entirely of tridymite.It is important to recognize these different types of SIRs in order to understand the evolution of the local chemical environment.
Igneous textures, and sharp interfaces between SIRs and the layers they overlie, argue for formation of SIRs as a result of accretion of SiO 2 -rich solid material to the outer edge of an existing chondrule, followed by a short duration heating event that melted the SIR, and cooling that was rapid enough to preserve the sharp interfaces.Peak temperatures of heating are inferred to be 1500°C or higher, based on the presence of cristobalite and computed liquidus temperatures.We considered an alternative model in which SiO (gas) condenses onto the outer surface of a partially molten chondrule.While this model is a plausible mechanism to form a limited number of SIRs, we do not consider that it is likely for the majority of SIRs.The discontinuous and variable nature of SIRs around individual chondrules is difficult to explain if the chondrule was partially molten and surrounded by a gas, which would presumably be uniform around the chondrule and around all local objects.
While IRs are common in other chondrite groups, the high abundance of SIRs in CR chondrites indicates a very specific environment for chondrule formation, in which condensing solids evolved to very silica-rich compositions during the time when a repeatable heating mechanism was active.This unique setting may or may not be related to the 1 Ma younger ages of CR chondrules compared with other chondrite groups.Marrocchi et al. (2022) showed that there are two populations of chondrules in CR chondrites and argued that larger ("CR-like") chondrules evolved from smaller ("CO-like") chondrules by addition of CI-like dust.While the chondrules we studied are mostly in the larger group of Marrocchi's work (>800 lm diameter), we did not systematically search for SIRs on small chondrules, so we cannot be certain that SIRs only occur on the larger "CR-like" chondrules.Although the model is similar, chemical fractionation for formation of SIRs must progress to much higher Si/Mg ratios than those of CI chondrite.It is clear that chondrule heating models for the CR chondrites must account for the observations of repeated heating during quite extreme chemical evolution of the local nebular gas.

FIGURE 1 .
FIGURE 1. Representative Raman spectra obtained in this study.Cristobalite is from QUE 99177 Ch7, tridymite is from EET 92042 Ch11, and glass is from EET 92062 Ch7.Spectra are offset vertically, for clarity.(Color figure can be viewed at wileyonlinelibrary.com)

FIGURE 2 .
FIGURE 2. BSE images (a, c, d, f) and combined EDS elemental x-ray maps (b, e) for chondrules with common SIRs in CR chondrites.X-ray maps combine Si (red), Mg (green), and Ca (blue): Olivine is green, low-Ca pyroxene is brown, silica is red, and Fe,Ni metal is black.Ca-rich pyroxene, plagioclase, and chondrule mesostasis are blue, pink, and violet: Colors may depend on the device on which the image is observed.(a-c) EET 92062 Ch8.(d-f) EET 92042 Ch14.Yellow boxes show the areas in the panels indicated.Cpx, Ca-rich pyroxene; Cris, cristobalite; Fe-Alt, Fe-rich alteration; Gl, glass; Lpx, low-Ca pyroxene; Mes, mesostasis; Met, Fe,Ni metal; Mx, matrix; Pig, pigeonite.(Color figure can be viewed at wileyonlinelibrary.com)

FIGURE 3 .
FIGURE 3. BSE images (a, c, e, f, h) and combined EDS elemental x-ray maps (b, d, g) for chondrule QUE99177 Ch5 which has a well-defined metal mantle (MM), an igneous rim (IR), and variable silica textures in different regions of the SIR: (e, f) common SIR textures, (h) Si-e SIR texture.X-ray maps combine Si (red), Mg (green), and Ca (blue): Olivine is green, low-Ca pyroxene is brown, silica is red, and Fe,Ni metal is black.Ca-rich pyroxene, plagioclase, and chondrule mesostasis are blue, pink, and violet: Colors may depend on the device on which the image is observed.Yellow boxes show the areas in the panels indicated.Cpx, Ca-rich pyroxene; Cris, cristobalite; Fe-Alt, Fe-rich alteration; Lpx, low-Ca pyroxene; Met, Fe,Ni metal; Mx, matrix; Ol, olivine; Pl, plagioclase.(Color figure can be viewed at wileyonlinelibrary.com)

FIGURE 4 .
FIGURE 4. BSE images (a, c, d) and combined EDS elemental x-ray map (b) for plagioclase-rich chondrule, EET 92062 Ch2 which has a common SIR: x-ray map combines Si (red), Mg (green), and Ca (blue): Olivine is green, low-Ca pyroxene is brown, silica is red, and Fe,Ni metal is black.Ca-rich pyroxene, plagioclase, and chondrule mesostasis are blue, pink, and violet: Colors may depend on the device on which the image is observed.Yellow boxes show the areas in the panels indicated.Cpx, Ca-rich pyroxene; Cris, cristobalite; Fe-Alt, Fe-rich alteration; Lpx, low-Ca pyroxene; Met, Fe,Ni metal; MM, metal mantle; Mx, matrix; Pl, plagioclase.(Color figure can be viewed at wileyonlinelibrary.com)

FIGURE 7 .
FIGURE 7. BSE images (b-d, f-g), combined EDS elemental x-ray map (a), and WDS oxide SiO 2 map (e).(a-e) Composite rim on QUE 99177 Ch7, including a common SIR texture (b), a silica-glass Si-gl texture (c), and a multilayered (ML) region with monomineralic layers of low-Ca pyroxene and cristobalite, followed by an Si-gl layer (d, e).(f, g) Tridymite rim on EET 92042 Ch11.Orange-dashed line shows the chondrule/SIR boundary.A smooth alteration zone (Alt) occurs between the tridymite and matrix.EDS x-ray map in (a) combines Si (red), Mg (green), and Ca (blue): olivine is green, low-Ca pyroxene is brown, silica is red, and Fe,Ni metal is black.Ca-rich pyroxene, plagioclase, and chondrule mesostasis are blue, pink, and violet: Colors may depend on the device on which the image is observed.Yellow boxes show the areas in the panels indicated.Alt Met, altered metal; Cpx, Ca-rich pyroxene; Cris, cristobalite; Fe-Alt, Fe-rich alteration; Gl, glass; Lpx, low-Ca pyroxene; Met, Fe,Ni metal; Mx, matrix; Pl, plagioclase; Trid, tridymite.(Color figure can be viewed at wileyonlinelibrary.com)

b
Includes pyroxene from both the ML and the Si-gl SIRs.c Average of three point analyses and seven 10 9 10 pixel extracted areas.d Average of one point analysis and one 10 9 10 pixel area.

FIGURE 8 .
FIGURE 8. Olivine compositions from chondrules that have SIRs.(a) Individual analyses and (b) average olivine composition of each chondrule.Data from Krot et al. (2004) are average olivine compositions from three chondrules in different chondrites that have SIRs.(Color figure can be viewed at wileyonlinelibrary.com)

FIGURE 9 .
FIGURE 9. Pyroxene compositions from SIRs (filled circles) and host chondrules (open circles).Low-Ca pyroxene data from Krot et al. (2004) are average compositions from four chondrules and four SIRs, and Ca-rich pyroxene data are the average compositions from six chondrules and four SIRs.(Color figure can be viewed at wileyonlinelibrary.com)

FIGURE 10 .
FIGURE 10. (a-d) Compositions of low-Ca pyroxenes in SIRs (filled circles) and host chondrules (open circles).For QUE 99177, PR = pyroxene rim.Data from Krot et al. (2004) are average compositions of low-Ca pyroxene from four chondrules and four SIRs.(Color figure can be viewed at wileyonlinelibrary.com)

FIGURE 11 .
FIGURE 11. (a-d) Compositions of Ca-rich pyroxene in SIRs (filled circles) and host chondrules (open circles).Data from Krot et al. (2004) are average compositions of Ca-rich pyroxene from six chondrules and four SIRs.(Color figure can be viewed at wileyonlinelibrary.com)

FIGURE 13 .
FIGURE 13. (a-c) Compositions of silica in SIRs (filled circles) and host chondrules (open circles).All silica grains are cristobalite, apart from in EET 92042 Ch11, where silica is tridymite.Data from Krot et al. (2004) are representative analyses of silica grains from two SIRs.(Color figure can be viewed at wileyonlinelibrary.com)

FIGURE 14 .
FIGURE 14. (a-e) Glass and mesostasis compositions in SIRs (filled circles) and the host chondrule of MET 00426 Ch3 (open orange circles).Data from Krot et al. (2004) are average analyses of mesostasis from three chondrules and three SIRs.(Color figure can be viewed at wileyonlinelibrary.com)

TABLE 2 .
Properties of SIRs and their host chondrules.

TABLE 3 .
Average compositions of olivine in chondrules surrounded by SIRs, determined by EPMA.EET 92042 Ch14 is a single EPMA point analysis and the rest are analyses extracted from quantitative WDS maps.No. of analysis areas refers to the number of 10 9 10 pixel boxes in WDS maps that were combined to give the reported analysis.<n indicates the detection limit for each oxide.P 2 O 5 and NiO were also measured but values were below detection. a

TABLE 4 .
Average compositions of low-Ca pyroxene in SIRs and chondrules, determined by EPMA.

TABLE 5 .
Average compositions of Ca-rich pyroxene in SIRs and chondrules, determined by EPMA.

TABLE 6 .
Average compositions of plagioclase in SIRs and chondrules, determined by EPMA.
Note: <n indicates the detection limit for each oxide.NiO was also measured, but values were below detection except for NiO in QUE 99177 Ch7 SIR which was 0.34 wt%. a No. of analysis areas refers to the number of 10 9 10 pixel boxes in WDS maps that were combined to give the reported analysis.

TABLE 7 .
Average compositions of glass and mesostasis, determined by EPMA.

TABLE 8 .
Average compositions of silica, determined by EPMA.