Rare earth element abundances in the CK chondrites including the Kobe meteorite

Geochemical Journal, Vol. 36, pp. 309 to 322, 2002 Rare earth element abundances in the CK chondrites including the Kobe meteorite YUSUKE HIROTA, 1 MINAKO TAMAKI 1 and NOBORU NAKAMURA 1,2 * 1 Division of Mathematical and Material Science, Graduate School of Science and Technology, Kobe University, Nada, Kobe 657-8501, Japan 2 Depatment of Earth and Planetary Sciences, Faculty of Science, Kobe University, Nada, Kobe 657-8501, Japan (Received February 27, 2002; Accepted April 30, 2002) The Kobe (CK4) meteorite and other CK chondrites are characterized with respect to rare earth element (REE) abundances using 12 bulk samples from 10 CK chondrites including Ningqiang (CK3), Karoonda (CK4), Kobe (CK4), Y-693 (CK4), A-882113 (CK4), Maralinga (CK4), Y-82102 (CK5), Y- 82105 (CK5), Y-82191 (CK6) and A-881551 (CK6). REE, Ba, Sr, Ca, Mg, Fe, Rb and K were analyzed by high-precision isotope dilution. The CK chondrites examined exhibit systematically higher REE abundances compared to ordinary chondrites, comparable to CV and CO, flat REE patterns with minor negative anomalies up to 15% on Ce and 25% on Eu. These chondrites also exhibit an appreciable light/heavy REE gap and Yb anomaly compared to CI chondrites. CK3-4 chondrites in general are found to exhibit a larger negative Ce anomaly (mean anomaly 9.6 ± 3.8%) compared to CK5-6 chondrites (3.3 ± 1.0%), and some CK5-6 show a slightly light-ree depleted pattern with less-pronounced negative Ce anomaly. It is suggested that these CK chondrites had a larger negative Ce anomaly prior to early thermal metamorphism, which yielded the minor redistribution of light REEs. No clear correlation is found between Sm and Eu/Sm or between Eu and Sr. From these observations, it is suggested that Eu existed in the trivalent state due to higher oxidizing conditions, suppressing redistribution of REEs among minerals compared to metamorphosed ordinary chondrites during the early thermal metamorphic event on the parent body. The systematic REE anomalies observed in CK chondrites are considered to be inherited properties from the refractory precursors produced in the early solar nebula. INTRODUCTION Refractory elements such as lanthanoids (rare earth elements; REE) and actinoids are believed to be unfractionated in primitive materials such as chondritic meteorites. In fact, the REE abundances in most chondrites are nearly constant and the relative contents of REEs normalized to the solar system abundances are consistent for all REEs (Nakamura, 1974). Refractory elements including REEs appear to have been quantitatively retained in high temperature condensates without ubiquitous elemental fractionation during the condensation process in the early stage of our solar system (e.g., Grossman and Larimer, 1974). How- ever, there have been several reports of fractionated REE abundances in bulk chondritic samples. Nakamura and Masuda (1973) reported non-chondritic REE abundance patterns in Khohar (L3), Abee (EH4) and Altanta (EL6). A similar pattern in Atlanta was later confirmed by Shinotsuka et al. (1995). Condensation calculation predicts a possible fractionation of REE in the solar nebula of the early solar system (Boynton, 1975; Davis and Grossman, 1979). Indeed, large REE fractionation reflecting their gas-solid fractionation has been observed in high-temperature mineral aggregates (calcium-aluminum inclusions; CAI) in carbonaceous chondrites such as Allende (e.g., *Corresponding author (e-mail: noboru@kobe-u.ac.jp) 309

310 Y. Hirota et al. Tanaka and Masuda, 1973). It has been pointed out that the bulk abundance of REE in Allende is somewhat fractionated, reflected by a strong fractionation in CAI (Nakamura, 1974; Shinotsuka and Ebihara, 1997). Carbonaceous chondrites have traditionally been regarded as the most primitive materials accessible to us. In fact, CI chondrites have been used as the standard for the chemical composition of our solar system (Ebihara et al., 1982). Van Schmus and Wood (1967) classified carbonaceous chondrites into three groups; C1, C2 and C3, based on the petrological and chemical characteristics. With the recent increase of the number of newly recovered meteorites, new types of carbonaceous chondrites other than these groups have been identified. Under the subsequent classification scheme, these original three types are commonly named CI, CM and CO-CV, and the new types are CR, CK, CH. Compared to the original types of carbonaceous chondrites, CR, CK and CH have yet to be investigated extensively either chemically or petrologicaly. The CK chondrites are particularly important to understanding the thermal evolution of carbonaceous chondrites because this is the only group that includes thermally metamorphosed carbonaceous chondrites covering petrologic grades from type 4 to 6 (Kallemeyn et al., 1991). The Kobe meteorite is a recent CK chondrite that was recovered soon after its fall (Nakamura et al., 2000a). Based on petrological and chemical characteristics, it was classified as CK4 (Nakamura et al., 2000a, b; Tomeoka et al., 2001). Our preliminary examinations of lithophile trace elements in the Kobe meteorite indicated that one of the fragments (Kobe E) had an anomalous REE pattern indicative of nebular signatures (Nakamura et al., 2000a; Hirota et al., 2000). The primary object of present study was then to characterize the Kobe meteorite and other CK chondrites with respect to REE abundances. The Kobe meteorite is a new member of the CK class. Our study therefore focuses on determining the representative bulk REE abundances in the Kobe meteorite and the CK chondrites, identifying possible signs of nebular components that may be commonly to the CK chondrites, and clarifying the redistribution patterns of lithophile elements, particularly REEs, in highly metamorphosed CK chondrites. For this study, we selected one CK3 sample (Ningqiang), five CK4 samples (Kobe, Karoonda, Y-693, A- 882113 and Maralinga), and four CK5-6 chondrites (Y-82102, Y-82105, Y-82191, A- 771551) and analyzed small chips (typical of 30~90 mg). Based on the data obtained, we compare CK chondrites with other carbonaceous and ordinary chondrites and characterize CK meteorites with respect to elemental (mainly REE) abundances. EXPERIMENTAL Samples One of the larger fragments (Kobe E) of the Kobe meteorite (about 1.4 g) has been revealed to exhibit highly fractionated REE abundances (Nakamura et al., 2000a; Hirota et al., 2000). In this work, we analyzed another fragment (Kobe C) consisting of two chips (total weight: 0.92 g). The sample was washed twice with high purity alcohol in an ultrasonic bath, then crushed into a coarse-grained sample using an agate mortar and homogenized. This sample was labeled Kobe C- 3mix. The sample was divided into three parts; C-3mix-1 (0.65 g) for rare gas analyses (Matsumoto et al., 2001), C-3mix-2 (0.093 g) for acceleration mass spectrometry (AMS) analysis of cosmogenic nuclides (Coffee et al., 2000), and C-3mix-3 (0.19 g) for isotope dilution mass spectrometry (IDMS) and prompt gamma-ray analyses (PGA) (Oura et al., 2002). Y-693 falls into weathering category A, and its petrography has been reported by Noguchi (1993) and Nakamura et al. (1993). The weathering category of the other Antarctic CK chondrites in our study have not been reported, however, the chips of these chondrites were delivered to us as fresh sample without recognizable contamination or weathering features according to observation under a binocular microscope. The Maralinga (CK4) meteorite is a find and has been weathered to some

REE in Kobe and CK chondrites 311 Table 1. Results for CK group chondrites * 1 Analyst: H. Onoue. * 2 Weights (in mg) of samples dissolved for analysis. The values in parentheses indicate total weights of samples pulverized. * 3 Data from Kallemeyn et al. (1991).

312 Y. Hirota et al. degree, as noted by Kallemeyn et al. (1991). Following our regular procedures, all the meteorite chips were cleaned three times with high purity acetone in an ultrasonic bath before crushing into powder. Isotope dilution mass spectrometry (IDMS) The abundances of alkali elements (K and Rb), alkaline earth elements (Sr, Ba, Mg and Ca), Fe and REEs were determined by the conventional isotope dilution method similar to that reported by Nakamura (1974) and Nakamura et al. (1989). Chemical processes were performed in a clean laboratory environment. After acid decomposition using HF-HClO 4 mixture and evaporation to dryness, a few drops of HClO 4 were repeatedly applied and allowed to evaporate to dryness, and then the sample was dissolved in 6N-HCl solution and divided into samples for IDMS and Rb-Sr isotopic analyses. The IDMS fraction was further divided into three samples and spiked with Rb-K-Sr-Ba composite, REE-composite and Mg-Ca-Fe composite spikes. The spike solutions used here are the same as those of Nakamura et al. (1989). The level of impurities in most reagents used for chemical processes was much lower than in previous studies (Nakamura et al., 1989), typically less than 1/10 of previous values. Nevertheless, overall procedural blanks were largely similar to those in our previous work (Nakamura et al., 1989). The procedural blanks (in 10 12 g) were as follow: La: 1.9 3.8; Ce: 2.1 5.2; Nd: 1.0 2.1; Eu: 0.71; Dy: 0.29 0.42; Er: 0.12 0.24; Rb: 6 14; Sr: 32 54; Ba: 300 860; K: 7800. The higher blank levels of K and Ba are attributed mainly to cross contamination from different composite spike solutions. Sample sizes used for IDMS ranged from 8 to 80 mg or larger (see Table 1). Hence, the blank effects on samples were less than 0.1% for most elements except for Ba (up to 2% for the smallest sample of Ningqiang II) and were thus negligible in most cases. The mass spectrometric technique employed in this work is similar to that of Nakamura (1974) and Nakamura et al. (1989), who used a JEOL- 05RB mass spectrometer. In this work, a 5-collector Finnigan MAT 262 mass spectrometer was used for isotopic measurements. RESULTS AND DISCUSSION BCR-1 and Allende standards In order to confirm the accuracy of our IDMS procedure, the United States Geological Survey (USGS) standard (BCR-1) and a meteorite laboratory standard (Allende powder sample from our laboratory) were measured repeatedly prior to the analysis of the CK samples. Selected results are shown in Appendix (Tables A1 and A2). For the two sets of REE analyses for BCR-1, most elements are in agreement within about 0.5% (Ce within 0.8%).When compared with our previous analyses, the mean values of duplicate analyses in the present study are shifted systematically to lower values by 1.1 ± 0.5% (mean) for most REEs (2.8% for Ce). Due to the high REE concentrations in BCR-1, aliquot solutions corresponding to a few milligrams of BCR-1 were taken from a 6N-HCl stock solution containing about 50 mg of BCR-1 and weighed normally. Therefore, it is possible that the aliquot solutions were too small (<100 mg) for precisely weighing, possibly yielding such a systematic error. For Allende powder, three larger aliquot solutions (>500 mg) from a 6N-HCl stock solution were analyzed. The standard deviations of most REEs are 0.1~0.5% (1.1% for La). Based on these results, we believe that the precision of REE measurement by the present procedure is about 0.5% or better for most cases (La variable up to 1%). The standard deviations of K, Rb, Sr, Ba and Fe concentrations are 0.2~0.8%, and about 2% for Mg and Ca. This relatively large error is attributable to the weighing procedures for small aliquot solutions. Nevertheless, the Fe/Mg ratio is considered to be precise enough for later discussion because of the mixed spike solution used for IDMS. From the results of replicate analyses of BCR-1 and the Allende meteorite laboratory standards, the precision of our IDMS for the CK chondrites presented in Table 1 are considered to be better than 1% for REEs and better than 2% for most of the other trace elements.

REE in Kobe and CK chondrites 313 Lithophile element abundances in CK chondrites The IDMS results for CK chondrites are presented in Table 1 along with the instrumental neutron activation analysis (INAA) abundance ranges reported for larger samples (250 300 mg) of CK chondrites (Kallemeyn et al., 1991). The abundance of Mg and Fe in CK chondrites (except for Y-82191) analyzed in this work are mostly within the range of INAA values, and the atomic Mg/ (Mg + Fe) ratios of the present analyses range from 0.57 to 0.60 with mean value of 0.59 ± 0.01. This is consistent with the INAA values (0.58 0.59) Fig. 1. Abundance correlations between (a) K and Rb, (b) Sr and Rb, (c) Ca and Sr, and (d) Sr and Eu. Open circles indicate CI abundances and the solid line represents the CI abundance ratio. Solid circles indicate individual CK samples: 1 and 2, Ningqiang (CK3) I and II; 3, Kobe (CK4) C-3mix; 4, Karoonda (CK4); 5, Y-693 (CK4); 6, A-882113 (CK4); 7, Maralinga (CK4); 8 and 9, Y-82102 (CK5) I and II; 10, Y-82105 (CK5); 11, Y- 82191 (CK6); and 12, A-881551 (CK6). There is no clear correlation between any two alkali elements or between alkaline earth elements.

314 Y. Hirota et al. (Kallemeyn et al., 1991). Except for Y-82191 (CK6), most analyses of lithophile elements including REEs are within the range of INAA values. The slight deviations for Mg, Ca and K and/ or Mg/(Mg + Fe) values for Y-82102 and Ningqiang may be due to the small size (30 40 mg) of samples used for analysis in this work. We therefore suggest that despite the small sample sizes, the abundance features obtained here for major and trace elements reasonably represent the bulk CK chondrites. The homogeneous elemental distributions found in such CK samples were rather unexpected. The only exception is Y-82191 (CK6), which has much higher abundances of refractory lithophile elements including Ca, Sr, Ba and REEs (see Table 1). This may be due to the heterogeneity of the samples, although this will need to be confirmed through further analysis. The possible correlations between abundances of alkalis, alkaline earth elements and Eu are shown in Figs. 1(a) to (d). In Fig. 1(a), K ranges from 237 to 460 ppm, while Rb exhibit a much larger variation (from 0.56 to 2.0 ppm). Although many of the data points are scattered around the CI line, there is no clear positive correlation between Rb and K. Two data points (1 and 2) of the Ningqiang (CK3) meteorite fall directly on the CI line, and the Kobe (sample number 3) and Karoonda (sample number 4) meteorites plot above and bellow the line. It is also noted that both of the data points of the Maralinga meteorite (sample number 7, significantly weathered) and Y-693 (sample number 5) (weathering category A) fall close to each other, although far bellow the CI line. Therefore, there does not appear to be any systematic variation in the Rb-K characteristics with respect to Antarctic and non-antarctic meteorites with different degrees of weathering. This suggests that the observed elemental distributions are due not to the terrestrial weathering but possibly related to the thermal history of the parent meteorites. Given this observation, it will be interesting to see whether there exists any correlation between the petrologic types and the Rb-K distributions. However, no such relationships are noted in Fig. 1(a). The main host of the alkali element Na is plagioclase, which exhibits large compositional variability in the CK chondrites. Rubin (1992) reported that plagioclase has the widest compositional range: An 16 69 in ALHA82135 (CK4), An 17 81 in PCA82500 (CK4/5) and An 45 78 for type 6 CK LEW87009. Heterogeneous plagioclase composition have also been reported for the Maralinga (CK4) meteorite (An 40 75 ), Karoonda (CK4) (An 40 80 ; Noguchi, 1993), and Kobe (CK4) (An 40 70 ; Nakamura et al., 2000a; Tomeoka et al., 2001; Tachibana et al., 2002). The heterogeneous distribution of Na in plagioclase grains suggests that other alkali elements such as K and Rb may also have non-equilibrium distributions. Curtis and Schmitt (1979) reported that alkalis (Na, K and Rb) are exclusively associated with feldspar in L6 chondrites and thus a close relationship between K and Rb is expected. However, this is not the case for these samples as seen in Fig. 1. Therefore, it is conceivable that the large variation of Rb in the CK chondrites is due to the heterogeneous distributions of plagioclase grains with markedly different Rb/K ratios. If the K-Rb association had indeed been established during early thermal metamorphism, it is considered that the alkali elements have subsequently redistributed due to later thermal events such as shock heating, giving rise to local heterogeneity of the Rb/K ratio. Considering the fact that there appears to be no correlation between distributions of alkalis (Rb/ K) and petrologic types for the CK chondrites, it is interesting to note that most CK chondrites exhibit reverse zoning of plagioclase composition, likely to have been produced during the early and/ or recent thermal events such as shock heating (Tomeoka et al., 2001; Rubin, 1992). A more detailed investigation of CK chondrites will be necessary to identify the nature of these late thermal events that resulted in the observed alkali distributions. As shown in Figs. 1(b) and (c), with an exception of Y-82191, Sr in CK chondrites is much less variable than Rb and Ca. This may suggest that the host phases for Sr are distributed relatively

REE in Kobe and CK chondrites 315 homogeneously in the CK chondrites. It is also noteworthy that no clear correlation exists between Sr and Eu (Fig. 1(d)). If Eu exists in its divalent state, Eu can be expected to behave in a similar way to Sr due to the similar ionic radii. Therefore, it is considered that Eu existed primarily in the trivalent state during the thermal events, possibly establishing the petrologic characteristics of the CK chondrites. This problem will be discussed in more detail below. Detailed REE patterns The chip of the Kobe meteorite examined in this study (Kobe C-3) was expected to give more representative REE abundances for the Kobe meteorite. Furthermore, the possible existence of anomalous REE feature in CK chondrites was investigated by analyzing smaller chips (<90 mg) of various CK chondrites. As shown in Fig. 2, the Kobe meteorite (C-3mix) exhibits a flat REE pattern similar to the patterns of other CK chondrites, though with somewhat variable absolute abundance (1.3 to 2 CI). Duplicate analyses of Ningqiang (CK3) yielded consistent results, and the two chips (I and II) of Y-82102 (CK5) exhibited somewhat fractionated features in heavier REEs, indicating that the mean abundances are quite consistent with other CK. This uniformity of REE patterns in CK chondrites is surprising considering the small sample sizes used for analyses and the different petrologic types (from 3 to 6). However, this finding is consistent with the similarity in major elements between all these CK chondrites. Y-82191 (CK6) exhibits anomalously high REE abundances (about 3 times CI), which appears to be parallel the higher abundances of Ca, Sr, Ba and K and lower abundances of Mg and Fe. The Maralinga (CK4) chondrite is considered to be an anomalous CK due to the fact that it contains much more abundant chondrules and CAIs compared with other CK (Keller et al., 1992). As shown in Fig. 2, the Maralinga sample analyzed in this work has the highest Ca and Sr abundances but lowest REE abundances in a flat distribution. From these observations, we conclude that a carrier of a highly fractionated REE Fig. 2. CI-normalized lithophile element abundance patterns for bulk CK chondrites. The majority of CK chondrites have grossly flat REE patterns with abundances of 1.4 to 2 times CI chondritic, except for the Y- 82191 chondrite (CK6), which has exceptionally high REE abundances but a similar pattern. The Maralinga (CK4) and Y-82105 chondrites (CK5) exhibit a somewhat different pattern from other CK chondrites. component as found in Kobe E is quite rare amongst CK chondrites. The uniform REE abundances and negative Ce and Eu anomalies characteristic of CK chondrites is a feature that differs significantly from ordinary chondrites as well as other groups of carbonaceous chondrites (Masuda et al., 1973; Nakamura, 1974; Evenson et al., 1978; Shinotsuka et al., 1995; Shinotsuka and Ebihara, 2001). Ordinary chondrites, particularly metamorphosed chondrites, have more homogeneous Eu distributions but variable absolute common REE distribution, yielding complimentary REE patterns of positive and negative anomalies (Masuda et al., 1973). Ordinary chondrites also exhibit small but clear positive and negative Ce anomalies, as well as Yb anomalies (Masuda et

316 Y. Hirota et al. Fig. 3. Detailed CI-normalized REE patterns for CK chondrites. Data for Ningqiang (CK3) and Y-82102 is taken from the mean of two analyses. The majority of CK chondrites have systematic negative Ce and Eu anomalies and tend to exhibit irregular Yb abundances. A light/heavy REE (Sm/Gd) gap appears occasionally. The CK3-4 chondrites tend to exhibit a larger negative Ce anomaly than CK5-6. The Maralinga (CK4) and Y- 82105 (CK5) chondrites have the lowest REE abundances and exhibit only slightly positive Eu anomalies or none at all. Only Y-82105 exhibits a significantly negative Yb anomaly. al., 1973; Nakamura, 1974). REE anomalies in CK chondrites were examined by analyzing the detailed REE patterns, as shown in Figs. 3 and 4. For quantitative discussion, Ce, Eu and Yb anomalies ( Ce, Eu and Yb) are defined as the departure of the relevant data points from the lines spanning La and Nd, Sm and Gd, and Er and Lu, respectively. It is pointed out that in the 12 analyses of 10 bulk CK chondrites, most commonly observed features of REE patterns are a negative Ce anomaly typical of Ningquiang (CK3) ( Ce = 15%), a negative Eu anomaly typical of A-881551 (CK6) ( Eu = 25%), a Yb anomaly typical of Karoonda (CK4; Fig. 4. REE abundance patterns of the Kobe meteorite (CK4) and mean REE patterns for CK3-4 and CK5-6 chondrites compared to the Allende (CV) and Murchison (CM) chondrites. The CK5-6 (mean) indicates systematic light-ree depletion compared to CK3-4 (mean). The Kobe and mean CK chondrites exhibit characteristic negative Ce and Eu anomalies and a slight enrichment of Gd relative to Sm, while the Allende (CV) chondrite exhibits a negative Eu anomaly and heavy-ree depletion. Yb = +6.5%), A-882113 (CK4; Yb = +6.3%) and Y82102 (CK5; Yb = +6.6%), and a small light/heavy REE (Gd/Sm) gap typically of Kobe (CK4), Y-693 (CK4) and Y-82191 (CK6), relative to CI chondritic abundances. The Maralinga (CK4) sample has the lowest REE abundances, with a small positive Eu anomaly. Only two CK chondrites, Maralinga (CK4) and Y-82105 (CK5), exhibit a higher Sm/Gd ratio or a negative Yb anomaly. It should be pointed out that the REE anomalies observed here for CK chondrites far exceed the analytical errors mentioned in the previous section and are thus quite confident results in this work. In Fig. 4, CK3-4 (mean) and CK5-6 (mean) exhibit similar abundances of heavy REEs, although CK5-6 (mean) include a slight light-ree

REE in Kobe and CK chondrites 317 Fig. 5. (a) Ce-La and (b) Eu-Sm correlation diagrams for the CK chondrites (solid circles) in comparison to ordinary, enstatite and carbonaceous (CO, CV) chondrites. The CK samples are: 1 and 2, Ningqiang (CK3) I and II; 3, Kobe (CK4) C-3mix; 4, Karoonda (CK4); 5, Y-693 (CK4); 6, A-882113 (CK4); 7, Maralinga (CK4); 8 and 9, Y-82102 (CK5) I and II; 10, Y-82105 (CK5); 11, Y-82191; and 12, A-881551 (CK6). Data for ordinary chondrites is taken from Masuda et al. (1973), Nakamura (1974), Okano et al. (1990) and Nishikawa et al. (1990), Nakamura et al. (1994), for enstatite chondrites from Nakamura and Masuda (1973), and for carbonaceous chondrites from Nakamura (1974) (partly unpublished). In (a), the majority of the CK chondrites plot close to or below the CI ratio, while the enstatite chondrites plot above and ordinary chondrites plot on both sides. In (b), the solid lines for H and L/LL chondrites represent correlation curves suggested by Nakamura (1974). A similar correlation is also clear for E chondrites over a wide range of compositions. However, any such correlation is much less convincing for the CK chondrites. Except for the Maralinga and Y-82105 chondrites, the CK chondrites exhibit characteristic prominent negative Ce and Eu anomalies. depletion together with a smaller Ce anomaly compared to CK3-4. The Kobe meteorite has a heavy REE pattern similar to that of Allende, while the light-ree pattern of Kobe differ significantly from that of Allende. As pointed out by Nakamura (1974) and Shinotsuka and Ebihara (1997), Allende has small but clear light/heavy REE fractionation and minor irregularities in Eu and possibly Yb. Such REE fractionation is similar to those of CAIs, typically group II CAIs. By analogy, the slightly fractionated REE patterns with negative Ce and Eu anomalies along with light/ heavy REE fractionation for Kobe and CK3-4 (mean) may indicate that these CK chondrite samples include more of the nebular components carrying highly fractionated REEs. Negative Ce anomaly in CK chondrites CK chondrites are compared with ordinary and enstatite (particularly metamorphosed) chondrites using precise IDMS data from Masuda et al. (1973), Nakamura (1974), Nishikawa et al. (1990), Okano et al. (1990) and Nakamura et al. (1994). The data is plotted in Fig. 5(a) (Ce/La-La diagram) and Fig. 5(b) (Eu/Sm-Sm diagram). The IDMS data obtained by Evenson et al. (1978) is not considered here in order to eliminate possible systematic and/or fortuitous errors between two labora-

318 Y. Hirota et al. tories, although such error is believed to be quite small (possibly smaller than 2%) (Evenson et al., 1978). Two important features can be seen in the Ce/ La-La diagram. First, La abundances in CK chondrites are systematically higher than in ordinary chondrites and comparable to CV-CO. This is also the case for other REE such as Sm (Fig. 5(b)). Second, the majority of CK chondrites have a smaller Ce/La ratio than that of the CI chondrites. Using the definition of the magnitude of the Ce anomaly Ce, all the CK chondrites analyzed in this work have negative Ce anomaly ranging from 1.8% to 14.8%. The calculated Ce values (in %) are 13.5 and 14.8 for Ningqiang (CK3), 11.0 for Kobe (CK4), 5.8 for Karoonda (CK4), 5.1 for Y-693(CK4), 3.4 for A-882113 (CK4) and 7.2 for the Maralinga (CK4) sample. The mean Ce is 9.6 ± 3.8% for CK3-4 (excluding A-882113) and 3.3 ± 1.0% for CK5-6 (excluding Y-82102 II). The highly metamorphosed type 5-6 CK chondrites have smaller Ce anomalies compared to CK3-4, and although three highly metamorphosed chondrites, Y-82102 (CK5), Y- 82191 (CK6), A-881551 (CK6), exhibit Ce/La ratios close to CI (Fig. 5(a)), these chondrites still exhibit significant negative Ce anomalies ( Ce = 1.8~ 3.7%). As recognized in Fig. 3, this is attributed to the depletion of La relative to Nd in these chondrites. It is thus considered that these metamorphosed CK had a larger negative Ce anomaly prior to the metamorphic event that caused the minor REE redistribution responsible for the slight light-ree depletion. As La and other common REEs reside primarily in Ca-phosphate and to lesser extent in Ca-pyroxene in metamorphosed ordinary chondrites (Ebihara and Honda, 1984), it is considered that the small light-ree fractionation observed here may reflect the heterogeneous distribution of these minerals in CK chondrites. There is strong evidence to suggest that all CK chondrites under consideration originally had a negative Ce anomaly of about 10%. This is in contrast with ordinary chondrites, which have both positive and negative Ce anomalies (Masuda et al., 1973; Nakamura, 1974). The systematic Ce depletion observed in bulk CK chondrites is considered to be the first case of such a REE pattern in chondrite groups, and may be important to understanding refractory precursor materials in CKgroup chondrites. Negative Eu anomaly and metamorphic REE redistributions in CK Nakamura (1974) suggested that there is a negative correlation between Eu/Sm and Sm in metamorphosed ordinary (H and L) chondrites. The present authors reexamined this correlation for ordinary (including LL) chondrites using additional data collected after Nakamura (1974) and investigated whether a similar correlation exists for metamorphosed CK and enstatite chondrites. Referring to the Eu/Sm ratios plotted relative to CI ratio against CI-normalized Sm in Fig. 5(b). The majority of ordinary chondrites follow the respective correlation lines. Furthermore, the negative correlation is apparent for enstatite chondrites over a much wider range, although data for enstatite chondrites is still limited. The data points for CK chondrites are strongly scattered, and no clear correlation as seen for ordinary and enstatite chondrites can be readily observed. Only two CK chondrites, Maralinga (CK4) and Y-82105 (CK5), appear to follow a trend similar to that of the L and LL groups. However, this may be accidental because Maralinga Eu has no correlation with Sr in Fig. 1(d). It is noteworthy that most CK chondrites plot below the CI-chondritic Eu/ Sm ratios, with the exception of Maralinga (CK4). Such a systematic Eu depletion may be the first case for bulk chondrites. Ordinary and enstatite chondrites, as well as other carbonaceous chondrite groups, fall both above and below the CI Eu/Sm ratios and thus have complementary Eu- Sm distributions (data of other carbonaceous chondrites is not shown in Fig. 5(b)). The negative Eu/Sm-Sm correlation for ordinary and enstatite chondrites may be interpreted as reflecting the homogeneous distributions of divalent Eu and more variable distributions of trivalent Sm, which may have been produced dur-

REE in Kobe and CK chondrites 319 ing the early thermal metamorphism that established petrologic characteristics. The main hosts of Eu +2 and Sm +3 are plagioclase and phosphates, respectively (Curtis and Schmitt, 1979; Ebihara and Honda, 1984). It is suggested that plagioclase grains developed from the chondrule mesostasis had captured divalent Eu more efficiently in the more reducing conditions prevalent during the early thermal metamorphism and become more homogeneously distributed in bulk chondrites. Caphosphate carrying Sm and other common REEs on the other had are considered to become more heterogeneously distributed in bulk chondrites in regions a few grams in size. The negative correlation between Eu/Sm and Sm is thus considered as a measure of metamorphic redistributions of divalent Eu and trivalent REE as represented by Sm. CK chondrites carry a greater component of oxidizing mineral assemblages including significant amounts (several %) of magnetite but scarce metals were formed in more oxidizing conditions compared to ordinary as well as enstatite chondrites. It is thus considered that CK plagioclase captures much less Eu compared to ordinary and enstatite chondrites. As noted above, the Eu/Sm-Sm correlation is much less convincing for CK chondrite. It is therefore suggested that the majority of Eu in CK chondrites exists in the trivalent state, resulting in the poor correlation between Eu, Sm and Sr. Another factor that may have controlled the redistribution of Eu-Sm is the shorter duration of high temperature metamorphism. As typical for the Kobe meteorite, CK chondrites exhibit features indicative of much higher metamorphic temperatures compared to ordinary chondrites (Tachibana et al., 2002; Noguchi, 1993). The homogeneity of the Mg/Fe in olivine and pyroxenes (Tachibana et al., 2002), and the very low noble gas abundances (Matsumoto et al., 2001) suggest that the Kobe meteorite is equivalent to petrologic type 6 of ordinary chondrites (estimated equilibration temperature by Tachibana et al. (2002) is about 800 C), however grain coarsening of matrix are limited, resulting in classification of the Kobe meteorite as CK4 (Nakamura et al., 2000a). In order to understand redistribution of REEs in metamorphosed chondrites during early thermal metamorphism and recent post shock heating, diffusivities of trivalent REE are considered using the diffusion coefficients of the cations in apatite and plagioclase, which are possible host phases of trivalent REEs and divalent Eu, respectively. The measured diffusion coefficient of Sm in natural fluoroapatite (~10 21 m 2 /s at 800 C; Cherniak, 2000) is several orders of magnitude larger than that of NaSi-CaAl interdiffusion in plagioclase (~10 29 m 2 /s at 800 C; Grove et al., 1984). Therefore, it seems likely that REEs in apatite diffuse much faster than through NaSi- CaAl interdiffusion in plagioclase. In view of the wide variation of An content (An 30 80 ) in CK chondrites, it is possible that diffusive redistributions of REEs in CK chondrites has been limited to light-rees, which have larger diffusivities than heavy REEs (Cherniak, 2000). This can also be understood from the slightly light- REE-depleted patterns in highly metamorphosed CK chondrites. As the diffusivity of divalent Eu has yet to be reported, we assume that it is comparable to Sr but higher than NaSi-CaAl interdiffusion. If this is the case, Eu may have been in equilibrium among minerals during the period of 10 6 years of thermal metamorphism, as demonstrated by Curtis and Schmitt (1979) for the L-6 chondrites. However, a considerable fraction of the CK Eu may have existed in the trivalent state, in which case there is no correlation with Sr or clear trend in the Eu/Sm-Sm diagram (Fig. 5(b)). From the above discussion, it is clear that the CK chondrites have the highest REE abundances among chondrite groups and unfractionated relative to CI. However, based on this newly obtained data, CK chondrites in general exhibit a negative Ce anomaly (typical of CK3-4), a negative Eu anomaly (typical Eu = 10%), a variable Yb anomaly, and a small light/heavy REE gap. It is concluded that this specific REE fractionation pattern originated in the early solar nebula and remains relatively unmodified by the secondary thermal events. It is worth pointing out that these

320 Y. Hirota et al. REE features coincide with the highly fractionated REE abundances of the fragment E of the Kobe meteorite. It is thus possible that the CK chondrites carries common refractory component similar to that of Kobe E. As noted by Hirota et al. (2000), the anomalous REE pattern of Kobe E is quite similar to those of spinel-hibonite inclusions in the Murchison (CM2) meteorite reported by Ireland (1988) and Sahjiupal et al. (2000), although the Kobe meteorite itself contains only rare CAIs. Acknowledgments We thank Mr. R. Hirata for providing the Kobe meteorite for this study. The Ningqiang specimen was provided courtesy of Prof. D. Wang. The Karoonda specimen (powder) was provided by Dr. A. Pring care of Dr. K. Yoneda. The Antarctic CK chondrites including Y-693 (CK4), A-882113 (CK4), Y-82102 (CK5), Y-82105 (CK5), Y-82191(CK6) and A-881551 (CK6) were obtained from the National Institute of Polar Research, Japan. We are indebted to Prof. H. Nagasawa for constructive criticism and Prof. M. Ebihara for continuous encouragement, constructive comments and assistance in manuscript preparation, Dr. K. Yamashita for critical reading of the manuscript and laboratory support, and Mr. H. Onoue for analyses of the Ningqiang meteorite. This work was supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Science and Technology of Japan. REFERENCES Boynton, W. V. (1975) Fractionation in the solar nebula: condensation of yttrium and the rare earth elements. Geochim. Cosmochim. Acta 39, 569 584. Caffee, M. W., Nishiizumi, K., Matsumoto, Y., Matsuda, J., Komura, K. and Nakamura, N. (2000) Noble gases and cosmogenic radionuclides in the Kobe CK meteorite. Meteoritics & Planetary Science 35, Supplement, A37 A38. Cherniak, D. J. (2000) Rare earth element diffusion in apatite. Geochim. Cosmochim. Acta 64, 3871 3885. Curtis, D. B. and Schmitt, R. A. (1979) The petrogenesis of L6 chondrites: Insight from the chemistry of minerals. Geochim. Cosmochim. Acta 43, 1091 1103. Davis, A. M. and Grossman, L. (1979) Condensation and fractionation of rare earths in the solar nebula. Geochim. Cosmochim. Acta 43, 1611 1632. Ebihara, M. and Honda, M. (1984) Distribution of rare earth elements and uranium in various components of ordinary chondrites. Meteoritics 19, 69 77. Ebihara, M., Wolf, R. and Anders, E. (1982) Are C1 chondrites chemically fractionated? A trace element study. Geochim. Cosmochim. Acta 46, 1849 1861. Evenson, N. M., Hamilton, P. J. and O nions, R. K. (1978) Rare-earth abundances in chondritic meteorites. Geochim. Cosmochim. Acta 42, 1199 1212. Grossman, L. and Larimer, L. W. (1974) Early chemical history of the solar system. Rev. Geophys. Space Phys. 12, 71 101. Grove, T. L., Baker, M. B. and Kinzler, R. J. (1984) Coupled CaAl-NaSi diffusion in plagioclase feldspar: experiments and applications to cooling rate speedometry. Geochim. Cosmochim. Acta 48, 2113 2121. Hirota, Y., Oura, Y., Nakamura, N., Ebihara, M., Misawa, K. and Yoneda, S. (2000) Kobe, the second observed fall of CK chondrite carrying normal and anomalous REEs; bulk chemical compositions. Meteoritics & Planetary Science 35, Supplement, A74 A75. Ireland, T. R. (1988) Correlated morphological, and isotopic characteristics of hibonites from the Murchison carbonaceous chondrites. Geochim. Cosmochim. Acta 52, 2827 2839. Kallemeyn, G. W., Rubin, A. E. and Wasson, J. T. (1991) The compositional classification of chondrites: V. the Karoonda (CK) group of carbonaceous chondrites. Geochim. Cosmochim. Acta 55, 881 892. Keller, L. P., Clark, J. C., Lewis, C. F. and Moore, C. B. (1992) Maralinga, a metamorphosed carbonaceous chondrites found in Australia. Meteoritics 27, 87 91. Masuda, A., Nakamura, N. and Tanaka, T. (1973) Fine structures of mutually normalized rare-earth patternsof chondrites. Geochim. Cosmochim. Acta 37, 239 248. Matsumoto, Y., Matsumoto, T., Matsuda, J. and Nakamura, N. (2001) A preliminary report on noble gases in the Kobe (CK) meteorite: a carbonaceous chondrites fall in Kobe City, Japan. Antarct. Meteorite Res. 14, 61 70. Nakamura, N. (1974) Determination of REE, Ba, Fe, Mg, Na, and K in carbonaceous and ordinary chondrites. Geochim. Cosmochim. Acta 38, 757 775. Nakamura, N. and Masuda, A. (1973) Chondrites with peculiar rare-earth patterns. Earth Planet. Sci. Lett. 19, 429 437. Nakamura, N., Yamamoto, K., Noda, S., Nishikawa, Y., Komi, H., Nagamoto, H., Nakayama, T. and Misawa, K. (1989) Determination of picogram quantities of rare-earth elements in meteoritic materials by directloading thermal ionization mass spectrometry. Anal. Chem. 61, 755 762.

REE in Kobe and CK chondrites 321 Nakamura, N., Morikawa, N., Hutchison, R., Clayton, R. N., Mayeda, T. K., Nagao, K., Misawa, K., Okano, O., Yamamoto, K., Yanai, K. and Matsumoto, Y. (1994) Trace element and isotopic characteristics of inclusions in the Yamato ordinary chondrites Y- 75097, Y-793241 and Y-794046. Proc. NIPR Symp. Antarct. Meteorites 7, 125 143. Nakamura, N., Ebihara, M., Hirota, Y., Oura, Y., Yoneda, K., Kojima, H., Tomeoka, K., Kojima, T., Komura, K., Clayton, R. N., Mayeda, T. K. and Wang, D. (2000a) The Kobe meteorite: Preliminary results of bulk chemical compositions, petrography, cosmicray induced radionuclides, oxygen isotopes and classification (abstract). Lunar Planet. Sci. 31 #1234, Lunar and Planetary Institute, Houston (CD-ROM). Nakamura, N., Kojima, H., Haramura, H., Tomeoka, K., Clayton, R. N., Mayeda, T. K. and Wang, D. (2000b) The Kobe meteorite: classification and consortium studies (abstract). Antarctic Meteorites XXV, 99 101. Nakamura, T., Tomeoka, K. and Takeda, H. (1993) Mineralogy and petrology of the CK chondrites Yamato-82104, Yamato-693 and a Carlisle Lake-type chondrites Yamato-82002. Proc. NIPR Symp. Antarct. Meteorites 6, 171 185. Nishikawa, Y., Nakamura, N. and Misawa, K. (1990) Investigation of the weathering effect on Rb-Sr systematics and trace element abundances in Antarctic and non-antarctic meteorites. Mass Spectrometry 38, 115 123. Noguchi, T. (1993) Petrology, and Mineralogy of CK chondrites: implications for the metamorphicsm of the CK chondrites parent body. Proc. NIPR Symp. Antarct. Meteorites 6, 204 233. Okano, O., Nakamura, N. and Nagao, K. (1990) Thermal history or the shock-melted Antarctic LLchondrites from the Ymamato-79 collection. Geochim. Cosmochim. Acta 54, 3509 3523. Oura, Y., Ebihara, M., Yoneda, S. and Nakamura, N. (2002) Chemical composition of the Kobe meteorite; Neutron-induced prompt gamma ray analysis study. Geochem. J. 36 (Special Volume of Kobe Meteorite), this issue, 295 307. Rubin, A. E. (1992) A shock-metamorphic model for silicate darkening and compositionally variable plagioclase in CK and ordinary chondrites. Geochim. Cosmochim. Acta 56, 1705 1714. Sahjiupal, S., Goswami, J. N. and Davis, A. M. (2000) K, Mg, Ti and Ca isotopic compositions and refractory trace element abundances in hibonites from CM and CV meteorites: Implications for early solar system processes. Geochim. Cosmochim. Acta, 64, 1989 2005. Shinotsuka, K. and Ebihara, M. (1997) Precise determination of rare earth elements, thorium and uranium in chondritic meteorites by inductively coupled plasma mass spectrometry a comparative study with radiochemical neutron activation analysis. Anal. Chim. Acta 338, 237 246. Shinotsuka, K. and Ebihara, M. (2001) Detailed abundances of rare earth elements among carbonaceous chondrites (abstract). Lunar Planet. Sci. 32 #1008, Lunar and Planetary Institute, Houston (CD-ROM). Shinotsuka, K., Hidaka, H. and Ebihara, M. (1995) Detailed abundances of rare earth elements, thorium and uranium in chondritic meteorites: An ICP-MS study. Meteoritics 30, 694 699. Tachibana, Y., Kitamura, M., Hirajima, T. and Nakamura, N. (2002) Equilibration temperature of the Kobe meteorite. Geochem. J. 36 (Special Volume of Kobe Meteorite), this issue, 323 332. Tanaka, T. and Masuda, A. (1973) Rare-earth elements in matrix, inclusions, and chondrules of the Allende meteorite. Icarus 19, 523 530. Tomeoka, K., Ohnishi, I. and Nakamura, N. (2001) Silicate darkening in the Kobe CK chondrites: evidence for shock metamorphicsm at high temperature. Meteoritics & Planetary Science 36, 1535 1545. Van Schumus, W. R. and Wood, J. A. (1967) A chemical-petrologic classification for the chondritic meteorites. Geochim. Cosmochim. Acta 31, 747 765. APPENDIX (see pp. 322)

322 Y. Hirota et al. Table A1. Results of duplicate analyses for BCR-1 Elements Run 1 Run 2 Mean (A) Literature (B)* Deviation(%) of A from B La (ppm) 24.87 24.85 24.86 25.07 0.8 % Ce (ppm) 52.54 52.95 52.75 54.30 2.8 % Nd (ppm) 28.58 28.75 28.67 29.17 1.7 % Sm (ppm) 6.493 6.510 6.501 6.630 1.9 % Eu (ppm) 1.974 1.971 1.973 1.981 0.4 % Gd (ppm) 6.739 6.772 6.755 6.796 0.6 % Dy (ppm) 6.441 6.448 6.444 6.520 1.1 % Er (ppm) 3.690 3.685 3.688 3.730 1.1 % Yb (ppm) 3.356 3.370 3.363 3.383 0.6 % Lu (ppm) 0.4939 0.4920 0.4930 0.5010 1.6 % 1.1 ± 0.5% (mean)** *Nakamura et al. (1989). **Ce datum is excluded. Table A2. Results of replicate analyses for Allende Elements Run 1 Run 2 Run 3 Mean Mg (%) 14.47 13.68 14.20 14.12 ± 0.33 Ca (%) 1.705 1.622 1.684 1.670 ± 0.035 Fe (%) 22.83 22.74 22.88 22.82 ± 0.06 K (ppm) 260.7 261.7 263.6 262.0 ± 1.2 Rb (ppm) 0.991 1.005 1.010 1.002 ± 0.008 Sr (ppm) 12.28 12.44 12.25 12.32 ± 0.08 Ba (ppm) 3.995 3.979 3.989 3.988 ± 0.007 La (ppm) 0.4520 0.4421 0.4497 0.4479 ± 0.0042 Ce (ppm) 1.107 1.094 1.098 1.100 ± 0.005 Nd (ppm) 0.857 0.855 0.858 0.857 ± 0.0012 Sm (ppm) 0.2752 0.2732 0.2760 0.2748 ± 0.0012 Eu (ppm) 0.1053 0.1051 0.1052 0.1052 ± 0.0001 Gd (ppm) 0.3605 0.3593 0.3609 0.3602 ± 0.0007 Dy (ppm) 0.4384 0.4372 0.4389 0.4382 ± 0.0007 Er (ppm) 0.2808 0.2797 0.2808 0.2804 ± 0.0005 Yb (ppm) 0.3015 0.3001 0.2994 0.3003 ± 0.0009 Lu (ppm) 0.04160 0.04143 0.04167 0.04157 ± 0.00010