Introduction

Several interesting issues concerning the evolution of the subcontinental lithospheric mantle (SCLM) beneath the eastern NCC have been investigated by many workers in the last decade (Menzies et al. 1993; Menzies and Xu 1998; Xu 2001; Gao et al. 2002; Chen and Zhai 2003). The occurrence of diamonds in Paleozoic kimberlites implies the presence of a thick (150–220 km), and refractory lithospheric keel underlying the craton, while mantle xenoliths in Cenozoic basalts suggest a thin (60–120 km), and fertile oceanic-type lithosphere in the Cenozoic (Song et al. 1990; Xu 2001). Such a change of lithospheric nature implies a significant erosion of the ancient lithospheric mantle (Archaean in age; Gao et al. 2002) during the Mesozoic, and distinguishes the NCC from most other ancient cratons. The lithosphere thinning was coupled with widespread Mesozoic magmatism. What is the origin of the Mesozoic magmatism and how was it related to the lithospheric mantle evolution? What geodynamic setting change caused the sudden surge of the Mesozoic magmatism in the NCC? Understanding to these issues has been significantly hindered by lack of systematic studies on Mesozoic magmatism. In this paper, we present the results of elemental and Nd–Sr–Pb isotopic data for Mesozoic intrusive complexes from the southern Taihang orogen (Fig. 1) in eastern NCC. They will be used to trace the source of these rocks, and investigate the relationship between Mesozoic magmatism and lithospheric mantle evolution.

Fig. 1
figure 1

Geological sketch map of the southern Taihang orogen. Inset shows distribution of the Mesozoic magmatic rocks (cross) in eastern North China craton (chessboard). NSGL south–north gravity lineament.

Full size image

Geological setting and sample description

The geological setting of the NCC has been frequently described (e.g., Jahn et al. 1988; Menzies et al. 1993). The NCC is one of the oldest continental nuclei in the world, with basement of mainly Archaean to Early Proterozoic gneisses (Jahn et al. 1988). Thick sequences of middle to late Proterozoic sediments unconformably overlie the basement rocks, indicating that the craton was stabilized by the early Proterozoic at about 1.8 Ga, and much of the craton remained stable up to Triassic time. However, the eastern part of the NCC (to the east of a south–north gravity lineament; inset of Fig. 1) was intensely remobilized since early Mesozoic, with a thin lithosphere (60–120 km) and widespread Mesozoic magmatism and basin development. This is in contrast with a thick (150–220 km) lithosphere and rare magmatism in the region to the west of the lineament. The southern margin of the NCC is the Dabie ultra-high pressure (UHP) orogen that was formed during the process of a Triassic collision between the Yangtze block and NCC (Chen et al. 2002a; inset of Fig. 1).

Three suites of Mesozoic intrusions occur in the southern Taihang orogen: the Xishu monzogabbro–monzonite, the Wu’an monzonite–quartz monzonite, and the Hongshan syenite–granite (Fig. 1). Peridotite xenolths occur as enclaves in the Xishu rocks (Xu and Lin 1991). Brief petrological description for the main rock types is given below.

Monzogabbros are medium-grained, dark green rocks, made up of clinopyroxene (20%), hornblende (25%), biotite (5%), plagioclase (40, with An=59-50) and K-feldspar (8%), and accessory phases like titanomagnetite and zircon. Monzodiorites are richer in plagioclase (55%) and K-feldspar (15%), and poorer in clinopyroxene (5–10%) than monzogabbros. Monzonites mainly contain plagioclase (45%) and K-feldspar (35–40%), and small amounts of hornblende, clinopyroxene, magnetite, titanomagnetite, allanite, sphene, apatite and zircon. Syenites are pinkish-grey porphyritic rocks, containing phenocrysts of plagioclase and K-feldspar in a medium- to fine-grained groundmass. Main constituent mineral is K-feldspar (60–85%); other minerals include plagioclase (5–15%), quartz (<5%) and mafic phases such as hornblende (5–10%), biotite (5%) and clinopyroxene (5–10%), and accessory apatite, titanomagnetite, allanite, zircon and sphene.

The emplacement time for the intrusives of the Taihang orogen was dated at 129–138 Ma using the zircon U–Pb and Rb–Sr whole-rock isochron methods (Davis et al. 1998; Cai et al. 2003). This is consistent with a new zircon U–Pb age of 127±1 Ma for a diorite pluton from the Xishu suite (Peng et al. 2004). Thus, emplacement of the magmatic complexes in the Taihang orogen took place in the period of 138–125 Ma.

Analytical methods

Chemical and isotopic analyses were conducted in the Institute of Geology and Geophysics (IGG, Beijing). Major elements were analyzed using XRF, with analytical uncertainty <3%. Trace elements were measured using ICP-MS; samples were digested by acid (HF + HClO4) in bombs. Analytical uncertainties are 10% for elements with abundances <10 ppm, and around 5% for those >10 ppm (Table 1).

Table 1 Chemical analyses for rocks from the south Taihang orogen, North China
Full size table

Nd–Sr–Pb isotopic analyses were also measured at the IGG (Beijing). Some of the samples were chosen for duplicate analyses of Rb–Sr and Sm–Nd isotopes in Rennes (France) to cross-check the data quality, and the results obtained from the two laboratories appear in good agreement (Table 2). Details of the analytical methods in Rennes can be found in an earlier report (e.g., Chen et al. 2002a). In Beijing, samples were dissolved using acid (HF + HClO4) in sealed Savillex beakers on a hot plate for a week. Separation of Rb, Sr and light REE was done using a cation-exchange column (packed with Bio-Rad AG50Wx8 resin). Sm and Nd were further purified using a second cation-exchange column, conditioned and eluded with dilute HCl. Mass analyses were performed using a multi-collector VG354 mass spectrometer as described by Qiao (1988). Rb, Sr, Sm and Nd concentrations were measured using the isotopic dilution method. 87Sr/86Sr ratios were normalized against 86Sr/88Sr=0.1194. 143Nd/144Nd ratios were normalized against 146Nd/144Nd=0.7219. 87Sr/86Sr ratios were adjusted to NBS-987 Sr standard=0.710250, and the 143Nd/144Nd ratios to La Jolla Nd standard=0.511860. The uncertainty in concentration analyses by isotopic dilution is ±2% for Rb, ±0.5–1% for Sr, and < ±0.5% for Sm and Nd depending upon concentration levels. The overall uncertainty for Rb/Sr is ±2% and Sm/Nd ±0.2–0.5%. Procedural blanks are: Rb=120  pg, Sr=200  pg, and Nd=50–100 pg. For Pb isotope analyses, sample powder was spiked and dissolved in concentrated HF at 800°C for 72 h. Lead was separated and purified by conventional anion-exchange technique (AG1×8, 200–400 resin) with diluted HBr. Isotopic ratios were measured using the VG-354 mass spectrometer at the IGG (Beijing). Repeated analyses of NBS981 yielded 204Pb/206Pb=0.05897±15, 207Pb/206Pb=0.91445±80, and 208Pb/206Pb=2.16170±200. The Pb isotopic data are presented in Table 3.

Table 2 Rb–Sr and Sm–Nd isotope compositions for rocks from southern Taihang, North China
Full size table
Table 3 Lead isotopic analyses for magmatic rocks from southern Taihang, North China
Full size table

Results

Major and trace elements

As shown in a classification diagram K2O + Na2O versus SiO2 (Fig. 2; Middlemost 1994), the Xishu suite is dominated by gabbroic diorite and monzonite, Wu’an suite by monzonite and quartz monzonite, and Hongshan suite by syenite–granite. The Xishu and Wu’an suites show large variations in chemical compositions (Fig. 3), but the former (with SiO2=50–61%, Mg#=73–46) are slightly less differentiated than the latter (with SiO2=54–64%, Mg#=79–40; Table 1). Both Xishu and Wu’an rocks are alkali-rich with Na2O > K2O. The Wu’an rocks show similar variation trends to the Xishu rocks in most Harker plots (Fig. 3). This could be interpreted as reflecting a genetic link between the two. Except for P2O5 that shows a convex curve, CaO, MgO, FeO and TiO2 are negatively linearly correlated with SiO2, but Al2O3 positively correlated (Fig. 3). This indicates a significant fractionation of ferromagnesian phases during magma evolution. However, XS-20 is exceptionally CaO-rich (11.54%; Fig. 3). This could reflect a plagioclase cumulate character, which is consistent with its high Sr (1,240 ppm; Fig. 4) and high modal proportion of plagioclase. Similarly, QC-6 and XS-24 deviate from the evolution trends in plots of SiO2 vs. MgO and Al2O3 by lower Al2O3 and higher MgO when compared with samples with similar SiO2 contents (Fig. 3). This is likely attributed to hornblende and Cpx cumulate in the two, as is supported by the very high Cr abundances (508 ppm for QC-6 and 502 ppm for XS-24; Table 1) and relatively high CaO (7.46 and 8.62%; Table 1) as well as our petrographical observations (main rock-forming minerals for QC-6 are hornblende 35% + Cpx 25% + plagioclase 27% + quartz 10%, and for XS-24 hornblende 30% + Cpx 38% + biotite 5% + plagioclase 20% + quartz 3%). The co-magmatic nature of the Xishu and Wuan rocks can also be seen from the plots of trace elements vs. SiO2 (Fig. 4), where they exhibit regular variations with increasing SiO2, approximately linear (U, Th, Sc, Co, and V), or curvilinear (Zr and Y), but showing scattering in Sr and Rb. All rock types of the Xishu and Wu’an suites show rather similar REE patterns (Fig. 5a, b), with highly enriched LREE and positive- to minor Eu anomalies (Eu/Eu*=1.39–0.92; Table 1). In the primitive mantle-normalized spidergrams (Fig. 5d, e), they are characterized by spikes in LILEs (e.g., Sr, Ba, Th, K) and LREE, and troughs in HFSEs (e.g., Nb and Ti).

Fig. 2
figure 2

Classification diagram (Middlemost 1994) for the Xishu (filled circle), Wu’an (open square) and Hongshan (open circle) suites.

Full size image
Fig. 3
figure 3

Harker diagrams for the Xishu (filled circle), Wu’an (open square) and Hongshan (open circle) suites.

Full size image
Fig. 4
figure 4

Plots of selected trace elements vs. silica for the Xishu (filled circle), Wu’an (open square) and Hongshan (open circle) suites. Note that Hongshan syenite–granites plot in different fields from the Xishu–Wu’an monzonitic rocks.

Full size image
Fig. 5
figure 5

Chondrites-normalized REE patterns and primitive mantle-normalized spidergrams for the Xishu, Wu’an and Hongshan suites. Chondrite values are from Masuda et al. (1973), further divided by 1.2; primitive mantle values from Sun and McDonough (1989). The grey curve in (c) represents calculated pattern for the residual melt after 22% hornblende–feldspar dominated fractionation (plus minor apaptite and allanite) from a syenitic parent (HS-14) based on the Rayleigh fractionation law. REE distribution coefficients for hornblende and apatite are from Arth (1976), feldspar from Nash and Crecraft (1985), and allanite from Brooks et al. (1981). Calculated bulk distribution coefficients are: La=8.33, Ce=8.41, Nd=8.30, Sm=7.33, Eu=7.32, Gd=7.54, Dy=8.08, Er=6.86, Yb=3.56 and Lu=2.48.

Full size image

The Hongshan syenites also exhibit a significant variation in chemical compositions, with SiO2=61–67% (Table 1). Most of them have high K2O (5.5–7.6%) and Na2O (5.5–6.0%). Granites (HS-11 and HS-12) differ from syenites by higher SiO2 (>74%) and lower Al2O3 (11.5–12.6%) and alkalis (Fig. 3). A striking feature is that the Hongshan syenites–granites plot in different fields and show variation trends independent of the Xishu–Wu’an complexes. This is particularly true in plots CaO, Al2O3, TiO2 and alkalis (Fig. 3) and U, Th, Sr, Rb and Zr (Fig. 4). All syenites REE patterns are LREE-enriched, with minor to positive Eu anomalies and highly depleted HREE (Fig. 5c) and Y (Fig. 4), implying an important role of residual garnet in the source. The two granites, however, show remarkable middle REE depletion and negligible Eu anomalies, with concave-shaped REE patterns (Fig. 5c). In the spidergrams (Fig. 5f), Hongshan rocks display positive anomalies of Rb, Th, K, Sr, Zr and LREE, and negative anomalies of Ba, Ti and Nb.

Sr–Nd–Pb isotopes

Most Xishu rocks have moderately high initial 87Sr/86Sr ratios (ISr) ranging from 0.7056 to 0.7071, and low ɛNd(135 Ma) values from −12.3 to −16.9. However, their data rather scatter in a plot ISr vs. ɛNd(135 Ma) (Fig. 6a). The Wu’an monzonitic rocks have ISr from 0.7059 to 0.7076 and ɛNd(135 Ma) from −13.8 to −18. The fields for the two suites cannot be completely separated, but they show some difference (Fig. 6a). By contrast, the Hongshan syenites–granites differ from Xishu–Wu’an monzonitic rocks by obviously higher ɛNd(135 Ma) values (from −8.2 to −11), and rather varied ISr (from 0.7052 to 0.7102), and roughly a flat trend can be seen for the syenite–granite suite (Fig. 6a). The very high ISr of HS-12 (0.7102) is only a face value; its high 87Rb/86Sr ratio (10.4) would produce a large uncertainty in the calculated ISr (Jahn et al. 2000).

Fig. 6
figure 6

a Plot of ɛNd(135 Ma) vs. ISr ratios. Two end members assumed for the isotopic modeling based on an AFC model (DePaolo 1981) are the EM1-type mantle (ISr=0.7055, Sr=1,100 ppm, ɛNd= −8.5, Nd=24 ppm; Chen et al. 2003; Chen and Zhai 2003) and LCC (ISr=0.710, Sr=350  ppm, ɛNd=−30, Nd=26  ppm; Jahn et al. 1988); other parameters are DNd=0.9, DSr=1.4, r=0.34. The dashed line represents a roughly lateral variation trend for the Hongshan suite. b Plot of 207Pb/204Pb vs. 206Pb/204Pb. c Plot of 208Pb/204Pb vs. 206Pb/204Pb. The Mesozoic Fangcheng basalts (Zhang et al. 2002) were plotted for reference. NHRL north hemisphere reference line. Location of EM1 and EM2 is from Zindler and Hart (1986).

Full size image

Lead isotopic ratios of the three suites are shown in Fig. 6b, c. 208Pb data are slightly more radiogenic than 206Pb, so that these rocks plot to the left of the north hemisphere reference line (NHRL). Also plotted for comparison are EM1 (enriched mantle with intermediate 87Sr/86Sr, low 143Nd/144Nd and low 206Pb/204Pb), EM2 (enriched mantle with high 87Sr/86Sr, high 206Pb/204Pb, and intermediate 143Nd/144Nd; Zindler and Hart 1986) and LCC (lower continental crust of the NCC with low Pb isotopic ratios 206Pb/204Pb=14–17.5, 207Pb/204Pb=14.6–15.4, 208Pb/204Pb=34.3–36.5; Zhu 1991). Lead isotopic data of the Mesozoic Fangcheng basalts (within the NCC; Zhang et al. 2002) were shown for reference, which were thought to reflect the Pb isotopic signature of the enriched mantle beneath NCC. All three rock suites plot between the fields of the EM1 and of LCC, suggesting a mixing process between LCC and EM1 components for their genesis. Xishu and Wu’an rocks form continuous variation trends in the Pb isotopic ratio plots, with the Xishu rocks ((206Pb/204Pb)i=16.92–17.3, 207Pb/204Pb=15.32–15.42, 208Pb/204Pb=37.16–37.63) having faintly higher Pb isotopic ratios than the Wu’an rocks (mostly 16.63–17.4, 15.28–15.44, 36.78–37.3, respectively). Hongshan syenites–granites show Pb isotopic ratios higher than Xishu–Wu’an rocks (Fig. 6b, c), plotting close to the EM1 field. But sample HS-8 has very low Pb isotope ratios (Fig. 6c), and the reason is unclear.

Discussion

Petrogenesis of the Xishu–Wu’an monzonitic rocks

The Xishu rocks are characterized by comparatively low SiO2 (mostly <58%) and high Mg# (mostly >50). Theoretically, these rocks could originate from partial melting of a basic lower crust, or of a mantle source. However, recent experimental data (Rapp and Watson 1995; Wolf and Wyllie 1994) have shown that extremely high temperatures (~1,100°C) are required to produce metaluminous low-silica (<58%) melts by dehydration melting of metabasalts in the lower crust. Regardless of the degree of partial melting, such melts are generally characterized by low Mg# (<42), which is not the case for Xishu rocks. Therefore, Xishu rocks cannot be generated by melting of basic lower crustal rocks, and a mantle source is required. This conclusion is supported by the occasional occurrence of peridotite xenoliths within the complex (Xu and Lin 1991).

The Wu’an monzonitic rocks are slightly more evolved than the Xishu rocks, but are also characterized by relatively low silica contents (54–64%, mostly <60%) and elevated Mg# (79–40). Consequently, they are unlikely pure crustal melts and a considerable amount of mantle-derived magma is required in their genesis. In fact, the co-magmatic nature of the Wu’an and Xishu rocks is readily suggested by the close temporal and spatial relationship of the two and the chemical trends shown in Figs. 3 and 4, as well as their similar REE patterns and spidergrams (Fig. 5). Therefore, the Wu’an rocks could originate from the Xishu rocks through coupled fractionation of ferromagnesian phases and assimilation of lower crustal components (= AFC process). This model is further verified in Fig. 7 where the two suites are closely associated as shown by the linear variation trends in ISr vs. 1/Sr (Fig. 7a), ɛNd(135 Ma) vs. 1/Nd (Fig. 7b) and Rb/Sr vs. Rb (Fig. 7c). However, another possibility cannot be precluded that the Xishu and Wu’an suites were formed corresponding to different batch of melting of the same mantle source, with the internal differentiation by AFC more developed in the latter, as suggested by Xu and Lin (1991).

Fig. 7
figure 7

Plots of a ISr vs. 1/Sr, b ɛNd(135 Ma) vs. 1/Nd, and c Rb/Sr vs. Rb for the Xishu, Wu’an and Hongshan suites, showing the Xishu and Wu’an rocks are co-magmatic, plotting on the extension from EM1 (Chen and Zhai 2003; Chen et al. 2003) to LCC (Jahn et al. 1988) and Hongshan syenites–granites derived from a separate magma chamber.

Full size image

The sharp depletion of CaO, MgO, FeO and TiO2 with increasing SiO2 (Fig. 3) indicates a significant fractionation of ferromagnesian phases like pyroxene and hornblende during magma evolution, which is also supported by the negative correlation of SiO2 with Sc, Co and V (Fig. 4). Separation of feldspar in Xishu rocks appears minor, as suggested by a positive correlation between Al2O3 and SiO2 (Fig. 3) and positive Eu anomalies (Fig. 5a). However, it is complicated for Wu’an rocks with respect to feldspar removal, as revealed by the highly varied Eu/Eu* ratios ranging from 0.92 to 1.32 (Table 1), and the kinks in plots SiO2 vs. Al2O3 (Fig. 3) and SiO2 vs. Sr (Fig. 4). The above argument suggests that feldspar removal is minor in an earlier stage, but significant in a later stage of magma evolution. Fractionation of accessory phases like zircon is important only in evolved rock types as is shown by the decreasing Zr and Y with increasing SiO2 (Fig. 4).

On the other hand, our isotopic data indicate that Xishu–Wu’an rocks were not simply produced via fractionation of mantle-derived magmas in a closed system. The least differentiated members (SiO2= 50–52%) of the Xishu suite have high Sr concentrations (570–790 ppm), highly enriched LREE patterns (Fig. 5a) and distinctive Sr-Nd isotopic signatures (ISr=0.706–0.7066, ɛNd(135 Ma)=−13 to −15). These features point to a long-term enriched SCLM source. Previous work on Mesozoic lamprophyres (Chen and Zhai 2003) and gabbros (Chen et al. 2003) revealed that the enriched SCLM source beneath the area is typically of EM1-type, with ɛNd(135 Ma)=−8.2 and ISr=0.7054–0.7058. These isotopic compositions are significantly different from those of the Xishu rocks, indicating a significant contamination of these rocks by the lower crust. As shown in Fig. 6a, the Xishu rocks plot dispersedly between the EM1 source and LCC, reflecting variable degrees of contamination by LCC. The Wu’an rocks also plot between the EM1 mantle source and LCC, with data points distributing slightly farther away from the EM1 field than Xishu rocks, and more contamination by LCC is implied. Specifically, the ɛNd(135 Ma) and ISr of the Xishu–Wu’an rocks are roughly inversely correlated (Fig. 6a), pointing to an important lower crustal incorporation. The scatter of data points is probably caused by the isotopic heterogeneity of LCC. Fan et al. (1998) suggested that the Mesozoic lower crust beneath the NCC was likely a mixture of Archaean TTG (ɛNd=−30; Jahn et al. 1988) and Mesozoic underplated “basaltic” rocks (ɛNd=−8 to −15; Zhang et al. 2002; Chen et al. 2003) from enriched mantle sources. The important input of lower crustal components in the Xishu–Wu’an rocks is evidenced by the linear evolution lines in plots ISr vs. 1/Sr (Fig. 7a) and ɛNd(135 Ma) vs. 1/Nd (Fig. 7b), which evolve from the EM1 source to LCC.

Plots of SiO2 vs. ɛNd(135 Ma) (Fig. 8a) and SiO2 vs. ISr (Fig. 8b) are used to further evaluate the role of LCC contamination. A roughly negative correlation is observed in Fig. 8a, whereas a positive trend is seen in Fig. 8b. Again, this indicates a significant contamination of the Xishu–Wu’an rocks by LCC. Samples XS-20, XS-12 and HL-7 plot off the evolution line in Fig. 8a, which may be attributed to contamination by the coeval Hongshan suite that has significantly higher ɛNd(135 Ma) values, or, to the LCC’s isotopic heterogeneity as discussed above. The Xishu rocks show higher ɛNd(135 Ma) values than the Wu’an rocks (Fig. 8a), which, together with their lower silica contents, suggests a lower proportions of LCC components in them. Isotopic modeling was conducted based on an AFC model of DePaolo (1981; see Fig. 6a for parameters used) assuming the EM1 mantle (ɛNd(t)=−8.5, ISr=0.7055; Chen et al. 2003; Chen and Zhai 2003) and the LCC (ɛNd(t)=−30, ISr=0.710; Jahn et al. 1988) as two end-members. The result suggests 15–30% LCC contamination for the Xishu rocks, and 20–35% for the Wu’an rocks (Fig. 6a).

This model also agrees with our Pb isotopic data. As seen in Fig. 6, the data points of Xishu–Wu’an suites form linear variation between the EM1-type mantle and LCC fields; the Xishu rocks show slightly higher Pb isotopic ratios than the Wu’an rocks. Thus, the Xishu–Wu’an suites could have originated from partial melting of EM1-type mantle sources, but variably contaminated by LCC components en route to crustal levels.

Fig. 8
figure 8

Plots of a ɛNd(135 Ma) vs. SiO2, and b ISr vs. SiO2, showing negative correlation of ɛNd and positive correlation of ISr with increasing SiO2 for the Xishu–Wu’an suites and lateral trends for Hongshan suite. HS-12 is ignored in b due to its large uncertainty of calculated ISr. Field of LCC is from Jahn et al. (1988).

Full size image

Petrogenesis of Hongshan syenite–granites

The chemical variation trends (Fig. 3) of the Hongshan syenite–granites suggest they are genetically linked. In Fig. 3, kinks are observed at SiO2= ~65% in plots of SiO2 vs. CaO, MgO, Al2O3, FeO, K2O + Na2O, and, to a lesser extent, TiO2. This can be interpreted by a two-stage fractionation process. The earlier stage is characterized by fractionation of ferromagnesian phases (e.g., pyroxene and hornblende), producing the syenites, as is inferred from the negative relationship of SiO2 vs. CaO, MgO and TiO2. Feldspar removal is negligible during this stage, because Al2O3 is positively correlated with SiO2 (Fig. 3), and, particularly, all syenites show minor to positive Eu anomalies (Fig. 5c). The later stage fractionation is probably dominated by a combined removal of feldspar and hornblende, producing the granites. This is suggested by the sharp depletion of Al2O3, TiO2 and Sr with increasing SiO2 (Fig. 3, 4). The important hornblende removal is indicated by granites’ concave-shaped REE patterns with minor Eu anomalies (Fig. 5c), because hornblende removal tends to counteract the negative Eu anomaly caused by feldspar removal and result in middle REE depletion as well. Fractionation of apatite could also have partially contributed to the middle REE depletion. Granites’ LREE depletion (compared with syenites) probably reflects removal of allanite with large distribution coefficient for LREE. To prove this, we conducted a REE modeling based on the Rayleigh fractionation law. Sample HS-14 was chosen to represent the parental magma to granites. As shown in Fig. 5c, the calculated REE compositions of a residual melt after 22% fractionation of a cumulate assemblage of hornblende 51% + plagioclase 21% + K-feldspar 25% + apatite 1.7% + allanite 1.1% from the syenitic magma roughly match those for the granites (see legend of Fig. 5c for bulk distribution coefficients of the cumulate). Thus, the granites represent residual melts after a significant differentiation of syenitic magma. This is supported by the comparable Nd isotopic compositions of the two (Fig. 6a).

Traditionally, Hongshan syenites were considered to form by fractional crystallization from the nearby Xishu–Wu’an monzonitic rocks (Huang and Xue 1990). However, this model is not supported by our new data. As shown in Fig. 6a, the Hongshan syenites have ɛNd(T) values significantly higher than those for the Xishu–Wu’an rocks. Taking into account the different evolution trends of the Hongshan rocks in Figs. 3 and 4, it can be concluded that the Xishu–Wu’an monzonitic rocks were unlikely parental to the Hongshan syenites. This is further disclosed in Figs. 7 and 8 that the Hongshan syenites do not lie on the extension defined by Xishu–Wu’an suites. Other authors (e.g., Deng et al. 1996) advocated that the syenites were derived from partial melting of lower crustal rocks at high pressures (over-thickened crust). However, recent experimental data by Litvinovsky et al. (2000) demonstrate that partial melting of quartzofeldspathic rocks in the over-thickened lower crust produces granitic melts with 72–73% SiO2, rather than syenitic liquids. Similarly, the melting experiments of Montel and Vielzeuf (1997) on a wide range of crustal materials show that syenitic melts are unlikely to be produced directly by anatexis.

We note from the plot ISr vs. ɛNd(135 Ma) (Fig. 6a) that the Hongshan syenites have Sr–Nd isotopic compositions comparable to the EM1-type mantle. Sample HS-17 almost falls in the field of the mantle source, and the remainder (except HS-12 with imprecise ISr ratio) plot on the extension toward the LCC field. Similarly, the data points of these rocks lie near the EM1 mantle source in Pb isotopic plots (Fig. 6b, c), with extension towards the LCC field. Therefore, we suggest that the parental magma (not exposed) of the Hongshan syenites-granites also originated from partial melting of the EM1-type mantle source from which the Xishu–Wu’an rocks were derived, but in a separate magmatic event and were much less contaminated by LCC than the latter. Isotopic modeling suggests that less than 10% LCC components has been incorporated in the Hongshan rocks (Fig. 6a). Actually many syenites were reported to originate from enriched mantle sources, or form via a process of hybridization between mantle-derived basaltic magmas and crustal materials, like the Bryansky syenite–granite suite from Russia (Litvinovsky et al. 2002) and the Mesozoic syenitic magmas from Namibia (Harris et al. 1999).

The evolution of the syenite–granite complex appears to occur in a system with involvement of only a small amount of LCC. This is compatible with the roughly lateral variation trends of Nd–Sr isotopic ratios with SiO2 (Fig. 8). Although the parental magma (likely an alkali-rich basalt) to the syenites is not exposed, the isotopic signature of the EM1-type source could be approximately reflected by the isotopically most “primitive” sample HS-17, with ɛNd(135 Ma)=−8.2 and ISr=0.7052. The mantle source must have long been enriched in incompatible elements prior to partial melting, as suggested by the rather low ɛNd value, and high Sr (948 ppm) and Rb (342 ppm) of the sample. In addition, the sample has relatively low (206Pb/204Pb)i ratios (<17.6), with high, positive Δ8/4 (>70) and relatively low Δ7/4 values (2.7; Table 3). These isotopic signatures agree with an EM1 source, suggesting a long-term low-μ mantle source (low U/Pb), and their positive Δ8/4 values suggest a long-term high Th/U source. Mantle enrichment may be caused by interaction of normal mantle peridotite with (1) fluids derived from a subducting slab (Maury et al. 1992) or (2) volatile-rich melts migrated from the asthenosphere (McKenzie 1989). The depletion of HFSEs (Nb, Ti) of the “primitive” sample in the spidergrams (Fig. 5f) suggests that the mantle enrichment is subduction-related, which probably happened in the early to middle Proterozoic as suggested by its high Nd model age (1.52 Ga).

Implications for lithosphere thinning

Voluminous Mesozoic igneous rocks were emplaced in East China. Their origin and the relationship with the eminent loss of lithospheric mantle during Mesozoic times has long been a disputed issue. We proposed in previous papers (Chen et al. 2002a, b, 2003; Chen and Zhai 2003) that enriched SCLM could be the main source for the Mesozoic magmas. The proportion of lower crustal material involved in their genesis was considered as less than 35% based on isotopic modeling; they thus represent significant addition of juvenile continental crust in Mesozoic times. This model is supported by our new data as well as some recent works (e.g., Qian et al. 2002). Therefore, the enriched mantle portions could be significantly consumed by melting to produce the Mesozoic magmas, triggered by upwelling of asthenosphere. This, together with thermo-mechanical and chemical erosion within the gradually upward moving lithosphere–asthenosphere interface, played an important role in thinning the lithosphere (Menzies et al. 1993; Xu 2001). The residual lithospheric mantle could be penetrated by, and mixed with, hot and dense asthenosphere, and then be removed through delamination.

Geodynamic setting

What is the cause of the sudden surge of magmatism in the Mesozoic? A popular model holds that the lithospheric destruction was related to the loss of physical integrity of the craton, caused by the Triassic collision between the NCC and Yangtze block (Menzies et al. 1993; Xu 2001; Gao et al. 2002). The Mesozoic magmatism was thus considered by many (e.g., Zhang et al. 2002; Mao et al. 2003) as post-collisional magmatism developed in an intra-continental extensional regime. We alternatively suggest that the Mesozoic magmatism was probably related to subduction of the paleo-Pacific slab based on the following two pieces of evidence: (1) The earliest Mesozoic intrusion in the Jiaodong Peninsula (easternmost margin of the NCC; inset of Fig. 1) was dated at ~170 Ma (zircon U–Pb; Wang et al. 1998) although many younger intrusions with ages ranging from 157 to 120 Ma are also present in the area. However, the magmatism in the Taihang orogen happened in a short time span from 138 to 125 Ma (Davis et al. 1998; Cai et al. 2003; Peng et al. 2004), with the earliest intrusion emplaced at ~138 Ma. A younging trend of magmatism from ~170  Ma in coastal area to ~138  Ma in inland NCC is thus noted, suggesting an inlandward movement of continental arc magmatism. This is consistent with the context of the northwestwards subduction of the paleo-Pacific slab beneath East Asia, which started in late Triassic (Arakawa and Shinmura 1995). And (2) The Mesozoic magmatic belts in the NCC are NE-trending (inset of Fig. 1), approximately parallel to the subduction zone (Maruyama 1997). Moreover, these rocks show strong arc magma signatures (Wang et al. 1998; Chen et al. 2002b). This model is particularly supported by the extensive accretion of arc complexes in the eastern margin of the East Asian continent during Jurassic times (Maruyama 1997). Taking into account previous work that the Mesozoic magmatism in NE China (Wu et al. 2003) and SE China (Zhou and Li 2000; inset of Fig. 1) was linked to the subduction of the paleo-Pacific slab, we propose that entire Eastern China is part of the East Asian continental arc, a model also advocated by Sengor and Natal’in (1996).

The genesis of the Mesozoic magmatism in southern Taihang can be described briefly as below. Subduction of the paleo-Pacific slab beneath the East Asian continent transformed the eastern part of NCC into an active continental margin. As a consequence, a back-arc extensional setting was developed in Taihang area (Fig. 1). This, in turn, induced the upwelling of asthenosphere. Meanwhile, fluids released from subducting slab penetrated and interacted with the overlying enriched portions of SCLM, lowering the solidus significantly. This, in conjunction with the high heat flow from the asthenosphere, triggered intense melting of the enriched SCLM, producing voluminous basaltic magmas. These mantle-derived magmas underplated in the lower crust, followed by differentiation and some contamination of lower crustal components, generating the magmatic complexes in the southern Taihang orogen.

Conclusions

  1. 1.

    The Xishu and Wu’an suites show many similar geochemical features such as LREE-enriched REE patterns with minor to positive Eu anomalies, and highly enriched isotopic compositions, pointing to their co-magmatic origin. They originated from partial melting of an EM1-type mantle source, followed by a coupled fractionation of ferromagnesian phases and significant contamination by LCC. The Wu’an rocks are chemically more evolved and isotopically slightly more enriched than the Xishu rocks, reflecting more involvement of crustal components.

  2. 2.

    The Hongshan syenites–granites show higher ɛNd(135 Ma) and Pb isotopic ratios than the Xishu–Wu’an rocks though their ISr ratios are similar. They were formed from a separate parental magma that also originated from the EM1-type mantle source, but with minor incorporation of lower crust components during magma evolution. The granites are residual melts from the syenitic magma via a feldspar–hornblende-dominated fractionation.

  3. 3.

    A back-arc extensional regime was developed in the eastern part of NCC responding to the subduction of the paleo-Pacific slab beneath East China. Fusion of the enriched SCLM portions was induced by upwelling of asthenosphere and infiltration of slab-derived fluids. This, along with mechanical and chemical erosion within lithosphere–asthenosphere boundary contributed to the Mesozoic lithosphere thinning.