Paleoproterozoic crustally derived carbonate-rich magmatic rocks from the Daqinshan area, North China Craton: Geological, petrographical, geochronological and geochemical (Hf, Nd, O and C) evidence

Most carbonate-rich magmatic rocks are mantle-derived, namely carbonatite(s), with a minority of them being contaminated by crustal rocks. It is debated whether there are also carbonate-rich magmatic rocks derived solely from crustal sources. In this contribution, we report crustally derived carbonate-rich magmatic rocks, named here crustal carbonatite(s), in the Daqingshan area, Western Block of the North China Craton. The Daqingshan crustal carbonatites were previously considered to be metasedimentary marbles. However, they cut adjacent rocks and contain some enclaves that are irregular in shape and show unoriented distribution of lithologies that cannot be found in the local wall rocks. Zircons from the crustal carbonatites show clear oscillatory zoning, and contain calcite, quartz, feldspar, diopside and CO 2 inclusions. The zircons have correlated U and Th abundances and give a U/Pb age of 1951 ± 5 Ma, being the same age as metamorphic zircon cores and rims from a diopsidite enclave (1954 ± 27 Ma of core and 1944 ± 40 Ma of rim). The zircon from the crustal carbonatite has t DM (Hf) and e Hf (t) of 2353 to 2457 Ma and -3.2 to 0.7, whereas the t DM (Hf) and e Hf (t) of the core and rim zircons from diopsidite range from 2228 to 2160 Ma and 3.0 to 4.8 and 2219 to 2057 Ma and 3.2 to 7.4, respectively. The δ 18 Ο (‰) zircon-V-PDB of zircon from the crustal carbonatite ranges from -21.5 to -19.6, with O isotope equilibrium temperature being 555°C to 635 °C. The crustal carbonatite shows a large variation in chemical composition, with SiO 2 =10.2-37.3% and Total REE=48-267 ppm. The t DM (Nd) age and e Nd (t) are ∼2.5 Ga and -2.9 to -2.4. The δ18 O(‰) Rock-V-PDB and δ 13 C(‰) Rock-V - PDB vary from -19.5 to -15.2 and from -5.2 to -2.4, being distributed between primary carbonatite field and the field of dolomitic marble from the study area in δ 18 O(‰) vs. δ 13 C(‰) diagram. Combined with previous studies, we drew the conclusion that some of the carbonate-rich rocks in the study area are magmatic in origin, by anatexis of impure marble plus common contamination by calc-silicates and other materials. This is consistent with the high-P-T experiments of CaO-CO 2 -H 2 O system (Wyllie and Tuttle, 1960) and MgO-CaO-CO 2 -H 2 O system (Fanelli and others, 1986), which indicate that partial melting of limestone will happen when temperature is > 700°C and when water is also present.

geological background from the Daqinshan area, North China Craton metamorphism, consistent with the P-T estimation of the metamorphic peak stage that Tϭ750 -800°C and Pϭ0.4 -0.5 GPa (Jin and others, 1991) or Tϭ850 -900°C and Pϭ0.77-0.93 GPa (Yang and others, 2004). Besides "diopside marble" (regarded in this paper as crustal carbonatite, see below), the diopside gneiss subunit is mainly composed of feldsparϩdiopside gneiss and diopsidite, most of them belonging to calc-silicate rocks. They commonly contain diopside and titanite, with some containing garnet, hornblende, calcite, phlogopite, scapolite, wollastonite and graphite. The marble subunit is mainly composed of dolomitic marble, with interlayers of thin diopside quartzite and tremolitite. The dolomitic marble commonly contains olivine and phlogopite.
There might be diopside marble (metasedimentary impure marble) rocks in the diopside gneiss subunit. However, most of the "diopside marble" shows magmatic features. They occur as veins or dikes ( fig. 2A) with width varying from ten centimeters to several meters and extending several hundred meters in east-west direction (subconcordant to gross lithological layering). They are homogenous ( fig. 2B) or contain enclaves (figs. 2C, 2D, and 2E). The wall rocks cut by crustal carbonatite are commonly calc-silicate rocks (figs. 2A, 2C, and 2D). Networks of veinlets emanate from main crustal carbonatite dike body into the country rocks ( fig. 2F). The enclaves are round or irregular in shape and vary in scale, being usually less than thirty centimeters in diameter. They show oriented or unoriented distribution (figs. 2D and 2E). All the features are similar to those of the crustal carbonatite in the eastern Himalayan syntaxis (Liu and others, 2006). The crustal carbonatites are mainly composed of calcite plus feldspar, quartz, clinopyroxene, orthopyroxene, phlogopite, garnet and hornblende (figs. 3A, 3B, and 3C). The different mineral assemblages of the crustal carbonatite in different locations may partly be related to assimilation of wall rock. Fine-grained orthopyroxene grew around coarse-grained clinopyroxene ( fig. 3B). The existence of phlogopite and hornblende suggest that a mixed CO 2 ϩH 2 O fluid existed during the crustal carbonatite formation or/and later geological processes. In some crustal carbonatites, clinopyroxene porphyroclasts (from enclaves) clearly show wavy extinction and calcite and other felsic minerals are fine-grained ( fig. 3C), indicating later deformation happened locally. Enclaves in the crustal carbonatite are mainly calc-silicate rocks (figs. 2D, 2E, and 3D), like the wall rocks of the crustal carbonatite. However, there are also enclaves of plagioclase two-pyroxene granulite ( fig. 3E) and other lithologies that cannot be found in the adjacent wall rocks. In many felsic enclaves, plagioclase show strong partial melting and K-feldspar extensive exsolution Rock samples were crushed to 200-mesh size for analysis. Whole-rock major element, REE and Y abundances in the whole-rock were determined by XRF (using 3080E) and ICP-MS (using Excell) at the Institute of Geological Analysis, Chinese Academy of Geological Sciences (CAGS) in Beijing. Uncertainties are estimated at ϳ3 to 5 percent and ϳ3 to 8 percent, respectively. Sm and Nd isotopic compositions were determined by isotope dilution at the Isotope Lab, Institute of Geology and Geophysics, Chinese Academy of Sciences, in Beijing. The procedures were similar to those described by Zhang and Ye (1987). The Nd model ages reported here are based on the depleted mantle model of DePaolo (1988).
Whole-rock carbon and oxygen isotopic ratios were measured at the Institute of mineral resources, CAGS. Details of the procedures were described by Hou and others (2006). A sample weighing 10 to 15 mg was reacted in vacuum with 5 ml of 100 percent orthophosphoric acid at 25°C for 2 to 3 days. Note that this gives only the isotopic composition of carbonate phase of the sample, because silicate phases cannot be dissolved by this acid. The CO 2 liberated from the samples were analyzed using a MAT 251 EM mass spectrometer. C isotope data are presented in terms of the standard '␦' notation relative to V-PDB, whereas O isotope data are presented in terms of the standard '␦' notation relative to both V-PDB and V-SMOW (␦ 18 O V-SMOW ϭ1.03086*␦ 18 O V-PDB ϩ30.86). Reference material 4416 analyzed (nϭ2) during the course of the present study yielded ␦ 13 Cϭ1.7 to 1.8 ‰ and ␦ 18 Oϭ -11.5 to -11.4 ‰ (V-PDB). [The recommended ␦ 13 C and ␦ 18 O values of the standard are 1.61 ‰ and -11.59 ‰ (V-PDB), respectively].
The zircons were dated using the SHRIMP II ion microprobe at the Beijing SHRIMP Centre, Institute of Geology, CAGS, by an analytical procedure similar to that described by Williams (1998). A 4 nA, 10 kV, negative O 2 primary ion beam was focused to ϳ30 m diameter on the zircon surface. Mass resolution during the analytical sessions was ϳ5000 (1 % definition). Each analysis took about 20 minutes, including 2 to 3 minutes initial raster ion cleaning of the surface and 5 scans through the masses of interest. Reference materials used were SL13 (Uϭ238 ppm) and TEM ( 206 Pb/ 238 U ageϭ517 Ma; Williams, 1998;Black and others, 2003). Decay constants used for age calculation are those recommended by the Subcommission on Geochronology of IUGS (Steiger and Jaeger, 1977). Data processing was carried out using the Squid and Isoplot programs (Ludwig, 2001) and ages were assessed using 207 Pb/ 206 Pb ratios corrected for common lead based on measured 204 Pb. Uncertainties for single analysis and mean age are one standard error and 95 percent confidence limits, respectively.
In-situ Lu-Hf isotopic composition of zircon was measured with a Geolas-193 laserablation microprobe, attached to a Neptune multi-collector ICPMS, at the State Key Laboratory of Lithospheric Evolution, Institute of Geology and Geophysics, Chinese Academy of Sciences, in Beijing. A 193 nm UVArF excimer laser ablation system was used for laser ablation analysis. The instrumental conditions and analytical procedure were described by Wu and others (2006). Ablation times were about 26s for 200 cycles of each measurement, with a 8Hz repetition rate, a laser power of 100mJ/pulse and a spot size of ϳ60 m. Zircon standard 91500 was used as the reference ( 176 Hf/ 177 Hfϭ0.282306; Woodhead and others, 2004). Standard 91500 analyzed (nϭ7) before and after the unknowns yielded 176 Lu/ 177 Hf (corr) values of 0.000277 to 0.000296 (with an average of 0.000283) and 176 Hf/ 177 Hf (c) values of 0.282259 to 0.282304 (with an average of 0.282276). All the Lu-Hf isotope results are reported with 95 percent confidence limits. The calculation of Hf model ages was based on a depleted-mantle source with a present-day 176 Hf/ 177 Hfϭ0.28325, using the 176 Lu decay constant (1.865ϫ10 Ϫ11 year Ϫ1 ) of Scherer and others (2001). The calculation of ε Hf (t) values was based on zircon SHRIMP U-Pb ages and the chondritic values of Blichert-Toft and Albarede (1997).
In-situ oxygen isotopic composition of zircon was measured with the Cameca IMS1270 at CRPG-CNRS in Nancy, France using a Cs ϩ primary ion beam accelerated at 10 kV with an intensity of 10 nA. The analytical procedure was similar to that described   (table 1). Samples with high CaO and MgO contents and low SiO 2 , Al 2 O 3 , TFeO, NaO 2 and K 2 O contents are mainly composed of calcite plus a few silicate minerals, such as plagioclase, K-feldspar, quartz and diopside. Although some minerals, such as leucite and nepheline, are given in CIPW model mineral calculations (table 1), they do not actually occur in these rocks. Sample NM0612Ϫ1 has the highest MgO content (14.78 %) and contains some olivine. The mineralogy and chemical composition is similar to that of dolomitic marble in the north portion of the area. They have varying Total REE contents (Total REEϭ48 -267 ppm), and show weakly negative to negligible Eu-anomalies (Eu/Eu* ϭ0.54 -1.06) and moderately fractionated REE patterns ((La/Yb) n ϭ12.9 -21.1) (table 1; fig. 4A). The crustal carbonatite has high Sr and Ba contents (407-2071 ppm and 213-1482 ppm, respectively) and low in Cr (9.1-44.7 ppm). In the MORB-normalized spider diagram, large ion lithosphere elements (LILE) show relative enrichment, but the high field strength elements (HFSE), especially Nb and Ti, are strongly depleted ( fig. 4B).
Their O and C isotopic compositions are similar to those of the crustal carbonatite and the dolomitic marble, respectively (fig. 5).
zircon petrography and shrimp u/pb dating Zircon crystals from a crustal carbonatite sample (NM0409) are translucent, purplish in color, round, stubby and columnar in shape and commonly large, with Note: Sm and Nd contents in ppm, age in Ma. Zircon crystals of a diopsidite sample, NM0407-5, which is an enclave in the crustal carbonatite at the same location as sample NM0409, are translucent, yellow in color, stubby and columnar in shape and commonly Ͼ 200 m. Core-rim textures are common. A few cores show weak zoning ( fig. 10A), many cores are inhomogeneous with a dark hair-like texture (not mineral inclusions). This texture is uncommon in terrestrial zircons, and is normally regarded as a shock feature related to an impactrelated event (Corfu and others, 2003). There are calcite, diopside, feldspar, titanite and quartz inclusions in the cores as determined by Raman. These are the main minerals of the rock, suggesting that the cores are newly-formed, but not of inherited origin. Furthermore, the existence of CO 2 inclusions ( fig. 7C) indicates that the metamorphism resulting in formation of the core took place in a CO 2 -rich environment. This is consistent with the diopsidite (NM0407-5) being a calc-silicate rock. The rims are darker than the cores in CL image and do not show the hair-like texture. They embay the cores, implying that fluid played an important role for rim formation and that at least some of the rims formed by recrystallization of the cores. In the rims  Rollinson (1993), with addition of other data (Taylor and others, 1967;Wang, 1994;Keller and Hoefs, 1995;Ray and others, 1999). Note that ␦ 18 O is plotted relative to both the SMOW and PDB scales. The isotopic composition of a number of different carbon reservoirs is plotted along the right-hand side of the diagram. Hydrothermal calcites from the mid-ocean ridges show mixing between mantle-derived carbon (M) and seawater carbon (S); M-V hydrothermal means hydrothermal calcites from Mississippi Valley-type deposits. O isotope of zircon from crustal carbonatite sample NM0409 is also plotted along the upper side of the diagram.
phism. Only a few of the rims are wide enough to measure using a ϳ30 m primary beam. Their analyses give U and Th contents and Th/U ratios ranging from 56 to 263, 1 to 34 ppm and 0.02 to 0.13, respectively (table 4). The U and Th contents and Th/U ratios of the rims are obviously lower than those of the zircon from the crustal carbonatite sample ( fig. 8), although both of them show similar features in CL images. This may suggest that the rims did not form directly from crustal carbonatitic magma but that CO 2 -rich fluid with high U and low Th played an important role in their formation. Five rim analyses (6.1, 7.1, 8.1, 9.2 and 12.1) yield close to concordia ages, with 207 Pb/ 206 Pb ages ranging from 1912 to 1983 Ma and a 207 Pb/ 206 Pb weighted mean age of 1944 Ϯ 40 Ma (MSWDϭ7.9) ( fig. 11). The ages of the core and rim of the zircon from diopsidite sample NM0407-5 and the zircon from crustal carbonatite sample NM0409 are the same within error although their U and Th contents and Th/U ratios are different to each other. The rim of the zircon from diopsidite sample NM0407-5 is interpreted as being formed by CO 2 -rich fluid activity relating to crustal carbonatite magmatism in terms of all the features mentioned above.  calculated as 555°C in terms of equilibrium equation and coefficient given by Valley (2003). If using the equation by Zheng (1993aZheng ( , 1993b, the equilibrium temperature calculated is 635°C. Both the calculated temperature values are lower than that proposed for the pervasive anatexis of the khondalite unit, probably due to later alteration of the rock. Twenty-five spots were analyzed on 13 zircon grains from crustal carbonatite sample NM0409 for Lu-Hf isotopes. Among them, twenty-three analyses show small variations in Lu and Hf isotopes, with t DM (Hf) and d Hf (t) ranging from 2353 to 2457 Ma and -3.2 to 0.7, with average t DM (Hf) and d Hf (t) values being 2392 Ma and -1.5, respectively (table 6, fig. 12). This indicates that the source rock of the crustal carbonatite had a long crustal residence time, which is consistent with conclusion reached by the Nd isotope data. Analyses 5 and 21 on the same zircon grain are different from others in Lu-Hf isotope composition; they have younger t DM (Hf) of 2190 to 2246 Ma and higher d Hf (t) values of 2.3 to 3.8.
O isotope compositions were not analyzed on zircons from diopsidite sample NM0407-5, but twenty spots were analyzed on 15 zircon grains for Lu-Hf isotopes. Fourteen core analyses show small variation in t DM (Hf) and d Hf (t) ranging from 2228 to 2160 Ma and 3.0 to 4.8, with average t DM (Hf) and d Hf (t) values being 2183 Ma and 4.1, respectively (table 6). They are different from zircon from crustal carbonatite sample (NM0409) in giving younger t DM (Hf) ages and higher in d Hf (t) values (table 6), suggesting that the metamorphic core zircon formed in a more depleted environment than the magmatic zircon of crustal carbonatite sample NM0409. Six rim analyses give t DM (Hf) age and d Hf (t) value of 2219 to 2057 Ma and of 3.2 to 7.4 (average being 2149 Ma and 5.0) similar to the core zircon (table 6, fig. 12). This suggests that the rim formed in a similar Lu-Hf isotope environment to the cores, although crustal carbonatite magmatism seems to play an important role in rim formation.

evidence for magmatic origin of the crustal carbonatite
In many locations the crustal carbonatite cuts wall rocks (calc-silicate and quartzofeldspathic gneisses) on different scales, with some being in complex vein networks ( fig. 2). This feature cannot be interpreted as folding or bondinage of marble layers. Some enclaves in the crustal carbonatite are irregular in shape and show unoriented distribution. Such enclave-hostrock relationship cannot be interpreted as a result of strong deformation. Furthermore, some enclaves, such as granulite and felsic gneiss, in the crustal carbonatite cannot be found in local wall rock, indicating that these exotic enclaves were transported from elsewhere by the crustal carbonatitic magma. The Daqingshan crustal carbonatite is similar in field occurrence to the carbonatite dike on Telyachi island, Kola Peninsula. In earlier studies, that dike, together with its xenoliths, was considered to be a conglomerate belonging to a Devonian volcanogenicsedimentary sequence. Later investigation established it as intrusive in origin (Claesson and others, 2000).
The petrographic characteristics of the zircon from crustal carbonatite sample NM0409 give another important evidence for magmatic origin. 1) The zircon contains calcite, quartz, feldspar and diopside, which are the main mineral assemblage of the rock. 2) The zircon contains abundant fluid inclusions (figs. 7A and 7B), some of them have been identified as CO 2 , suggesting that the environment from which the zircon is crystallized is rich in fluid.
3) The zircons have t DM (Hf) ages from 2353 to 2457 Ma (average ϭ2392 Ma), which are similar to t DM (Nd) age (2486 -2462 Ma) of the crustal carbonatite whole rock. 4) Some zircons show clear oscillatory zoning (figs. 6B and 6D), a feature of typical magmatic zircon. 5) Both Th and U of the zircon show a good linear relationship, with Th/U ratio ranging commonly from 0.2 to 0.4 (table 4). Features 1 to 3 indicate that the zircons are not detrital in origin but formed during metamorphic and/or magmatic processes. These three features may suggest that the zircon formation is related to CO 2 metasomatism. To our knowledge, however, no metamorphic zircons with so clear oscillatory zoning (feature 4) have been reported. In fact, oscillatory zoning is due to physical/chemical disequilibrium of crystals growing in a fluid medium-usually a magma. Therefore, zircons with this kind of oscillatory zoning must be crystallized from a magmatic liquid. This interpretation is in line with the good linear relationship between Th and U of the zircon (feature 5). Some metamorphic zircons from high-grade metamorphic rocks show zoning structures, but they are commonly sector zoning, not oscillatory zoning, and formation of zircon with zoning structure may also be related to anatexis resulted from high-grade metamorphism. Therefore, we contend that the zircons from sample NM0409 are mainly products of crustal carbonatitic magmatism although it cannot be excluded that some zircons formed by CO 2 metasomatism within a metamorphic fluid regime, as indicated by some analyses showing departure from U-Th correlation relationship in figure 8.
In terms of internal structures and U-Pb ages of zircons, the rim of zircon from diopsidite sample should form later than its metamorphic core but at the same time as zircon from crustal carbonatite sample. However, the rim and core of zircon from diopsidite sample are similar to each other but different from zircon from crustal carbonatite sample, in showing a more depleted Hf isotope composition, therefore suggesting formation in different Lu-Hf isotope environments. This supports the view that the rim of zircon from the diopsidite enclave formed by CO 2 -rich fluid activity relating to crustal carbonatite magmatism.

evidence for crustal source of the crustal carbonatite
If the Daqingshan carbonate-rich rocks are of magmatic origin, there are at least two different opinions on their origin. 1) Carbonate-rich magma came from mantle source but was contaminated by crustal rocks; 2) The impure marble of the khondalite unit underwent strong metamorphism and anatexis, culminating in formation of carbonate-rich magma. The carbonate-rich magmatic rock is high in Sr and Ba contents, which is characteristic of mantle carbonatite (Pandit and others, 2002). However, some wall rock calc-silicate rocks are also high in these two elements (table  1). CaO and MgO decrease and Al 2 O 3 , K 2 OϩNa 2 O, REE, Zr, Th, Sr, Ba, Rb and other elements increase with SiO 2 (fig. 13). The extension of some element abundances (for example, MgO, CaO, K 2 OϩNa 2 O and Zr) for carbonate-rich magmatic samples towards a compositional field defined by the calc-silicate wall rocks can be used as evidence for mixtures of carbonatite and the wall rock (Vuorinen and Skelton, 2004), but it can also be interpreted by compositional variation of the original impure marble or assimilation of crustal carbonatitic magma by wall rock. Carbon isotope study indicate that mantle-derived carbon ␦ 13 C(‰) V-PDB distribution is essentially bimodal with peaks at -5 and -25 (Deines, 2002). In the ␦ 18 O(‰) vs. ␦ 13 C(‰) diagram, all the  11. Concordia diagram of SHRIMP U-Pb data for zircon from diopsidite (NM0407-5) unfilled and gray color represent core and rim analyses, respectively. The 207 Pb/ 206 Pb weighted mean age is given for the core.
carbonate-rich magmatic samples are distributed between primary carbonatite field and dolomitic marble field ( fig. 5), being consistent with the opinion that the magma is from mantle source with assimilation of marble (Demény and others, 1998). However, there is graphite of organic origin in the khondalite unit in the Daqingshan area. Therefore, the low ␦ 13 C(‰) Rock-V-PDB value of the carbonate-rich magmatic rock can also be interpreted as influence of the graphite. Wang (1994) analyzed 15 graphite samples and 6 calcite samples separated from graphite ores from the area, with ␦ 13 C(‰) Rock-V-PDB ranging from -6.4 to -29.0 and 1.1 to -13.4, respectively ( fig. 5). The large variation of the C isotope compositions of both the graphite and calcite suggests isotope exchange between different mineral phases. On the other hand, the low ␦ 18 O(‰) Rock-V-PDB value of the carbonate-rich magmatic rock compared with the  . 5), or to influence of meteoric water (Deines, 1989). In any case, the large compositional variation of the crustal carbonatite is related to influence of crustal material, being consistent with Hf isotope composition of the crustal carbonatite zircon and wholerock Nd isotope composition of the crustal carbonatite.
However, the following phenomena are more suitable for the interpretation that the carbonate-rich magmatic rock in the Daqingshan area was produced by anatexis of impure marble. 1) The crustal carbonatites show closely temporal and spatial relationships with the calc-silicate rock which is definitely of metasedimentary origin, although the former have also been found locally in quartzo-feldspathic gneisses. 2) The magmatic zircon of the crustal carbonatite and metamorphic core and rim zircons of the diopsidite enclave give the same age (ϳ1950 Ma) within error, indicating that they are the products of one and the same tectonothermal event. 3) From experimental petrology (Wyllie and Tuttle, 1960;Fanelli and others, 1986), impure marble could start producing carbonate melts at the metamorphic temperatures of Ͼ 750°C. Thus melt production of this type should be expected. 4) Completely exotic enclaves are found in the crustal carbonatite, indicating transport of such materials from elsewhere. 5) As mentioned before, mantle-derived carbonatites usually show closely spatial and temporal relationships with other mantle-derived rocks (Bizimis, ms, 2001;Bell and Rukhlov, 2004;Downes and others, 2005). However, no contemporaneous mantlederived rocks, such as kimberlite, lamprophyres and pyroxenite, have yet been found in the Daqingshan area. 6) Mantle-derived carbonatites are characterized by very high abundace of LREE, Nb, Ta, U and Th and depletion of Hf, Zr and Ti ( fig. 14) (Bizimis, ms, 2001). In contrary to this, the crustal carbonatites are low in LREE, Nb, Ta, U and Th, with the lowest Total REE content being 48 ppm (table 1; fig. 14). Their Hf and Zr do not show depletion features. On the other hand, the crustal carbonatite shows Nb depletion, a widespread feature of continental rocks (Pearce, 1983).
In conclusion, the carbonate-rich magmatic rock in the study area formed by anatexis of impure marble associated with the calc-silicate rocks at peak conditions during Paleoproterozoic metamorphism. Exsolution and partial melting of feldspar and carbonate minerals indicate their high-grade metamorphism. The khondalite unit of the Daqingshan area underwent high-grade facies metamorphism, with Tϭ750 -800°C and Pϭ0.4 -0.5 GPa (Jin and others, 1991) or Tϭ850 -900°C and Pϭ0.77-0.93 GPa (Yang and others, 2004) at the metamorphic peak stage, resulting in extensive anatexis of the khondalite unit and local formation of granites derived from the meta-sedimentary rocks of the khondalite unit. According to high-P-T experiments of CaO-CO 2 -H 2 O system (Wyllie and Tuttle, 1960) and MgO-CaO-CO 2 -H 2 O system (Fanelli and others, 1986), partial melting of limestone will happen when temperature is Ͼ700°C with the presence of water. This has been observed in limestones around a granitic intrusion (Lentz, 1999).
It is possible that some of the Daqingshan impure marble underwent partial melting, forming crustal carbonatitic magma with a high H 2 OϩCO 2 volatile content. Some water was introduced into the carbonate system (impure marble and melting products) by dehydration of silicate rocks. CO 2 produced during metamorphic and melting processes resulted in CO 2 metasomatism and some CO 2 volatile escaped from the carbonate system, as suggested by the formation of the rim of zircon from diopsidite sample. These factors decreased the melting point of the impure marble. We emphasized the important role of CO 2 metasomatism in the processes of highgrade metamorphism, anatexis of impure marble, and the evolution of crustal carbonatite. We also considered that crustal carbonatitic magmatism and CO 2 metasomatism should be closely related to each other, showing different aspects of one and the same "remobilized crustal carbonatitic melt" activity. Some silicate minerals, such as diopside and olivine, in the crustal carbonatite are not crystallized from magma but refractory residual phases. Assimilation of wall rock changed the mineral and chemical composition of the crustal carbonatite. In all the crustal carbonatite samples analyzed in this study, sample NM0409 seems to be the best one to represent crustal carbonatitic magma. No inherited zircons have been found in crustal carbonatite sample NM0409. Dissolved zirconium was crystallized from anatectic carbonatitic magma to form magmatic zircon. In the crustal carbonatite, zirconium mainly occurs in zircon and no other minerals contain much zirconium. This may be the reason why zircon is relatively high in content although the rock is low in zirconium (55 ppm).  (Bizimis, ms, 2001). Normalization values from Sun and McDonough (1989).

age of the khondalite unit
The crustal carbonatite is considered to form by anatexis of impure marble of the khondalite unit. Its whole rock t DM (Nd) age and zircon t DM (Hf) age are 2.49 to 2.46 Ga and 2.35 to 2.46 Ga. Although no whole rock Nd isotope composition analyses have been done, the core and rim of the zircon from diopsidite enclave (NM0407-5) give t DM (Hf) ages of 2.23 to 2.16 Ga and 2.22 to 2.06 Ga. Combined with SHRIMP zircon U-Pb dating of the crustal carbonatite and diopsidite, their formation time can be constrained between 2.15 and 1.95 Ga. Therefore, the diopside gneiss subunit should have formed in the Paleoproterozoic, and not in the Archean as previously proposed. Our argument is consistent with the dating of detrital zircon from meta-sediments of the khondalite unit others, 2006a, 2006b;others, 2006a, 2006b). Combined with the metamorphic age of 1850 Ma identified in the khondalite rocks of the Helanshan area, western portion of the Khondalite Belt (Dong and others, 2007), it is clear that the tectonothermal event occurred between 1950 and 1850 Ma. from the Daqinshan area, North China Craton