There has been a long debate regarding the timing of the final amalgamation of the North China Craton, which is considered to have occurred either during the Neoarchean or Paleoproterozoic era. One major point of contention is whether there existed a long-lived subduction lasting through the Neoarchean to Paleoproterozoic. The Lüliang Complex contains multiphases of magmatism and thus represents the most viable region to address this controversy. In this study, we carried geochronological and geochemical analysis on the representative granitoids. Secondary ion mass spectrometry U–Pb dating revealed four distinct granitoid groups emplaced at 2531 ± 4, 2189–2173, 2027 ± 25, and 1852 ± 41 Ma, respectively. Notably, the 2531 Ma granitic gneiss was identified for the first time in this region. Based on the geochemical characteristics, the granitoids can be divided into two types. The 2531 and 2027 Ma groups display I-type features, while the 2189–2173 and 1852 Ma groups exhibit A-type geochemical affinities. Both I-type groups exhibit enrichment in Rb, depletion in Nb, Ta, and Ti, moderate fractionated REE patterns, substantial negative Eu anomalies, low Sr/Y ratios, and positive εHf(t) (+3.51 to +5.53 and +5.59 to +7.32, respectively), indicating that they were generated from partial melting of the juvenile mafic crust. In contrast, the 2189–2173 Ma granitoids belong to A2-type and were most likely generated by the partial melting of felsic rocks in the back-arc region, while the 1852 Ma granitoids belong to A1-type and were most possibly the result of partial melting of mafic-intermediate rocks during the post-collisional stage. Based on the records of A-type granitic magmatism and the ~1950 Ma peak metamorphism throughout the Trans-North China Orogen, we propose that a long-lived subduction process (2531–1950 Ma) can mostly explain the existing geological phenomena. It is likely that the subduction between the Eastern and Western Blocks should have commenced at ~2531 Ma, followed by a long-lived subduction. The two blocks ultimately collided with each other to form the North China Craton at ~1950 Ma, which triggered post-collisional exhumation and partial melting at ~1852 Ma.
The North China Craton (NCC) is considered to be one of the oldest cratons on the planet. Despite the wide acceptance of the tectonic division of the basement of the NCC into the Eastern and Western Blocks, separated by the intervening Trans-North China Orogen (TNCO) , the tectonic evolution of the NCC remains a considerable controversy among scholars (Figure 1). Several models have been proposed to explain the tectonic evolution of the NCC, including: (1) a prolonged subduction that lasted for ~650 Ma before the Eastern and Western Blocks finally collided at ~1850 Ma [2-6]; (2) a rift event that occurred after the amalgamation of the NCC at ~2500 Ma [7, 8]; (3) two stages of subduction–rift systems that occurred at 2450–2120 Ma and 2120–1980 Ma ; (4) two oceans that formed by rifting at ~2300 Ma and that closed via subduction, successively, at ~2150 and ~1900–1860 Ma [10-13]; and (5) two microcontinents/oceanic plateaus that were accreted at ~2500 Ma, followed by a rift event and subduction polarity reversal [14-17]. Of these, the subduction polarity (eastward or westward) and whether there existed a vast ocean that lasted for ~650 Ma before the final collision remain the two major controversies.
In order to constrain the above issues, it is necessary to uncover the magmatism records from the Neoarchean to Paleoproterozoic era. Previous studies have revealed that the calc-alkaline granitoid and volcanic rocks in the Wutai greenstone belt represent the earliest evidence of arc-related magmatism in the TNCO at 2560–2520 Ma [18, 19]. Similarly aged 2560–2438 Ma granitoids have been recognized in other Terranes of the TNCO, including Xuanhua, Huai’an, Hengshan, Wutai, Fuping, Yunzhongshan, and Zhongtiao Complexes [2, 20-31]. Located in the westernmost margin of the TNCO (Figure 2), the Lüliang Complex represents the most viable area for constraining the above debates. Previous studies have indicated that the Lüliang Complex is primarily composed of Paleoproterozoic granitic plutons and meta-supracrustal rocks [24, 32-40]. In this study, we present a comprehensive analysis of the late Neoarchean to Paleoproterozoic granitoids within the Lüliang Complex, including whole-rock major and trace element compositions, secondary ion mass spectrometry (SIMS) zircon U–Pb geochronology, and in-situ zircon Hf–O isotopes. For the first time, we reported the Neoarchean magmatism in the Lüliang Complex. Combined with available petrological, metamorphic, and structural data, the results of this study led us to reconstruct an integrated accretionary process from Neoarchean to Paleoproterozoic for tectonic evolution of the TNCO and to provide new insights to further test the existing tectonic models of the NCC.
2. Geological Background and Petrography
The NCC is widely considered to be formed through the collision between the Eastern and Western Blocks along the linear structural TNCO (Figure 1 [1-5, 10, 11, 13, 21, 36, 37, 40-44]). The two blocks exhibit differences in lithology, geochemistry, structure, metamorphism, and geochronology, which have been summarized by Zhao et al.  and Kusky et al. . The nearly north‒south trending TNCO, also named the Central Orogenic Belt , is located in the central part of the NCC. There is a consensus that the TNCO represents a typical continent-to-continent collisional orogen: (1) low-grade supracrustal foreland basin deposits ; (2) structural characteristics of strike-slip ductile shear zones, large-scale thrusting and folding, transcurrent tectonics, sheath folds, and mineral lineation [4, 6, 11, 45, 46]; and (3) medium- to high-grade metamorphism, with clockwise P‒T paths involving near-isothermal decompression [44, 47-61]. From north to south, the TNCO consists of different metamorphic terranes such as Chengde, Northern Hebei, Xuanhua, Huai’an, Hengshan, Wutai, Fuping, Lüliang, Zanhuang, Zhongtiao, Dengfeng, and Taihua Complexes (Figure 1 ). The dominant rock types are Neoarchean to Paleoproterozoic tonalitic-trondhjemitic-granodioritic (TTG) gneisses, meta-supracrustal rocks, syn- to post-tectonic granitoids, and ultramafic to mafic rocks . Most of these rocks exhibit arc-related geochemical and isotopic characteristics .
The Lüliang Complex is situated at the western margin of the central part of the TNCO and is composed of Paleoproterozoic supracrustal rocks and Neoarchean to Paleoproterozoic granitoid plutons that experienced greenschist- to granulite-facies metamorphism (Figure 2 [24, 32, 33, 44, 49, 61]; this study). The meta-supracrustal rocks of the Lüliang Complex are primarily composed of graphite-bearing pelitic gneisses/schists, marbles, quartzites, phyllites, meta-basalts, sandstones, meta-conglomerates, and dolomites [10, 24, 33-35, 37, 40, 63-66]. Based on their stratigraphy from bottom to top, these rocks can be subdivided into the Jiehekou, Lüliang, and Yejishan/Heichashan Groups (Figure 2 [25, 65, 66]).
Zhao et al.  classified the granitoids that intruded into the supracrustal rocks into pre-tectonic gneisses, syn-tectonic gneissic granites, and post-tectonic granites. The pre-tectonic gneisses consist of the Yunzhongshan TTG gneisses, Gaijiazhuang gneisses, and Chijianling–Guandishan gneisses (Figure 2). The Yunzhongshan gneisses are predominantly exposed to the northeastern part of the Lüliang Complex and have undergone metamorphism to the upper amphibolite facies. The TTG, monzogranitic gneisses, and high-Mg mafic-ultramafic rocks in the Yunzhongshan area, formed at 2540–2500 Ma, are interpreted to have originated in an arc setting [24, 29, 67, 68]. Yunzhongshan gneisses are considered as comparable to the adjacent 2520–2475 Ma Hengshan and Fuping TTG gneisses [2, 22-24, 69]. The 2408–2364 Ma Gaijiazhuang gneisses are in tectonic contact with the Lüliang Group and are composed of coarse-grained porphyritic gneisses and monzonitic granite [24, 70]. The 2173–2199 Ma Chijianling–Guandishan gneisses are strongly deformed, metamorphosed, and widely distributed in the southern and western parts of the complex (Figure 2 [24, 32]). The syn-tectonic 1832 ± 11 Ma Huijiazhuang fine- to medium-grained, weakly gneissic to massive granite clearly intrudes the Chijianling–Guandishan gneisses and contains numerous xenoliths of the Chijianling–Guandishan gneisses . The Huijiazhuang granite is contemporaneous with the 1950–1800 Ma metamorphism of the TNCO [24, 39, 44]. The post-tectonic granites in the Lüliang Complex, including the 1800 ± 7 Ma Luyashan coarse-grained charnockite, 1807 ± 10 Ma Luchaogou coarse-grained prophyritic granite, and 1798–1790 Ma Tangershang/Guandishan fine-grained granite, all of which exhibit a massive or structureless texture [24, 32]. While previous studies have roughly established a geochronological framework for the Lüliang Complex, there is still a lack of systematic analysis of geochemical and isotopic data and their connection to the specific tectonic settings.
In this study, we collected twelve granitic samples from the key granitoid intrusions throughout the Lüliang Complex (Figure 2). Four age groups of granitoids were identified based on our new SIMS data (see texts below; online supplementary Table S1). The first group is 2531 Ma granitic gneiss that was collected from the Fangshan County (Figure 2). In the field, the foliated granitic gneiss is truncated by undeformed granitic vein (Figure 3(a)). Microscopic observation indicates that the 2531 Ma granitic gneiss mainly consists of plagioclase (~30%), quartz (~25%), K-feldspar (~20%), and oriented mafic minerals including biotite (~15%) and hornblende (~5%), with a typical granitic texture and foliated structure (Figure 3(b)). The second group is 2189–2173 Ma deformed granite that was collected from the southwest area of the Lüliang Complex (Figure 2). This deformed granite shows penetrative foliation and the alignment of K-feldspar + plagioclase + biotite (Figure 3(c)). The representative mineral assemblage is K-feldspar (~30%), quartz (~25%), plagioclase (~25%), and biotite (~15%), displaying a granitic texture (Figure 3(d)). The third group is 2027 Ma granitic gneiss that was collected from an outcrop near the Wangjiagou primary school of Lüliang City (figures 2 and 3(e)). The rock is medium-grained and gneissic in structure and dominantly composed of K-feldspar (~30%), quartz (~25%), plagioclase (~25%), and oriented biotite (~15%), all of which exhibit a preferred alignment to form the major foliation (Figure 3(f)). The last group is 1852 Ma granite, which was collected from an abandoned quarry located ~1 km south of the main road in Kuaili Village, Xishe County (Figure 2). The rock is medium-grained and displays variation from strong strain domain of well-foliated texture to weak strain domain of massive structure (Figure 3(g)). The dominant mineral assemblage of this group is K-feldspar (~30%), quartz (~25%), plagioclase (~25%), and biotite (~15%), displaying a typical granitic texture (Figure 3(h)).
3. Analytical Methods and Results
Whole-rock major and trace element geochemistry, SIMS zircon U–Pb geochronology, and Hf–O isotope analyses were performed on representative granitic samples from the Lüliang Complex. The details of the analytical techniques are described in the online supplementary Text 1.
3.1. SIMS Zircon U–Pb Geochronology
Six representative samples were analyzed using zircon U–Pb dating. Four groups of granitoids with ages of 2531, 2189–2173, 2027, and 1852 Ma, respectively, were identified.
3.1.1. 2531 Ma Granitic Gneiss
Zircon grains from Sample 16LL35-1 are euhedral, appearing light brown to light orange and transparent. The crystals are commonly 100–250 µm in length, with length/width ratios of 1.5–2.0. Cathodoluminescence (CL) images from these zircon grains exhibit weak luminescence and display narrow bands of oscillatory zones, with no obvious core–rim textures and overgrowth rims, indicating a magmatic origin (Figure 4(a)). Eleven zircon grains were selected for SIMS U–Pb analyses, two of which were abandoned due to the excessive common Pb (online supplementary Table S1). The remaining nine analyses possess high Th/U ratios of 0.28–0.62. Seven analyses fall on the Concordant line and yield an apparent 207Pb/206Pb age between 2537 and 2515 Ma and a weighted mean 207Pb/206Pb age of 2531 ± 4 Ma (MSWD = 1.9, Figure 5(a)). This age can be interpreted as the crystallization age of this sample, and it is for the first time a Neoarchean age has been reported in the Lüliang Complex.
3.1.2. 2189–2173 Ma Deformed Granite
Three representative samples including 16LL26-2, 16LL31-1, and 16LL39-2 were dated using SIMS. Zircon grains from Sample 16LL26-2 are light gray to orange and transparent and have smaller sizes of 100–150 µm in length and length/width ratios of 1–2. Most grains are subhedral and show core–rim textures (Figure 4(b)). Wide bands of oscillatory zoning are generally preserved in the core section, and the rims show nebulous to “fir-tree” zoning (Figure 4(b) ). Some zircon grains show clear wide bands of oscillatory zonation surrounded by a thin rim (Figure 4(b)). Sixteen zircon grains with oscillatory zoning were selected for SIMS U–Pb analyses and display high Th/U ratios of 0.34–0.74 (online supplementary Table S1). Nine analyses fall on the Concordant line and yield an apparent 207Pb/206Pb age of 2194–2178 Ma with a weighted mean 207Pb/206Pb age of 2189 ± 5 Ma (MSWD = 2.4; Figure 5(b)). This age of 2189 ± 5 Ma can be interpreted to represent the crystallization age of the sample.
Zircon grains from Sample 16LL31-1 are subhedral, light gray to brown, and transparent to opaque. They are commonly 100–150 µm in length with length-to-width ratio of 1.5–2 (Figure 4(b)). They display oscillatory zoning cores and dark–thin metamorphic rims, though the rims are too narrow to analyze (Figure 4(b)). Twenty analyses of typical igneous zircon grains were conducted, of which eleven analyses with high Th/U ratios of 0.38–1.78 defined a discordant line with an upper intercept age of 2169 ± 10 Ma (MSWD = 0.9; Figure 5(c)). Five analyses falling on the Concordant line yield an apparent 207Pb/206Pb age of 2186–2161 Ma with a weighted mean 207Pb/206Pb age of 2173 ± 14 Ma (MSWD = 2.5, Figure 5(c)). It is indicated that the age of 2173 ± 14 Ma represents the crystallization age of the sample.
Zircon grains from Sample 16LL39-2 are euhedral and light red to brown. While they are mostly transparent, they contain some opacity caused by radiation damage, resulting in the transformation of the zircon from crystalline to an amorphous one. The zircon grains of this sample have lengths ranging from 100 to 200 µm with a length-to-width ratios of ~2 (Figure 4(b)). The internal structure of the zircon grains is much more complicated, some of which show irregular internal structures, while some others still retain the oscillatory zonings (Figure 4(b)). Twenty-one spots on the oscillatory zoning areas were selected for U–Pb geochronological analyses. Nine available analyses exhibit high Th/U ratios of 0.20–0.74 and define a discordant line with an upper intercept age of 2178 ± 15 Ma (MSWD = 0.9; Figure 5(d)). The other four analyses on the Concordant line exhibit an apparent 207Pb/206Pb age of 2181–2163 Ma and yield a weighted mean 207Pb/206Pb age of 2175 ± 13 Ma (MSWD = 2.3; Figure 5(d)). In our interpretation, the age of 2175 ± 13 Ma can be regarded as the crystallization age of the sample.
3.1.3. 2027 Ma Granitic Gneiss
Zircon grains from Sample 16LL40-1 are light gray to brown, subhedral, and transparent to opaque. The zircon grain lengths are 150–200 µm, with a length-to-width ratio of ~2. In CL images, the opaque zircon grains show a spongy internal structure, and the translucent zircon grains still retain oscillatory zoning cores with bright and thin structureless rims (Figure 4(c)). Fourteen analyses on the oscillatory zoning areas were conducted, of which seven analyses display low common Pb, and high Th/U ratios (0.24, 0.53) were adopted and defined a discordant line with an upper intercept age of 2043 ± 21 Ma (MSWD = 0.4; Figure 5(e)). Three analyses on the Concordant line exhibit apparent 207Pb/206Pb age of 2035–2012 Ma and yield a weighted mean 207Pb/206Pb age of 2027 ± 25 Ma (MSWD = 5.3; Figure 5(e)). The age of 2027 ± 25 Ma can be interpreted as the crystallization age of this sample.
3.1.4. 1852 Ma Granite
Zircon grains of Sample 16LL42-1 are light yellow to light brown, euhedral, transparent, and partially opaque due to radiation damage. The zircon grains have lengths of 150–200 µm, with a length/width ratio of ~2. The CL images of the zircon grains are characterized by magmatic-origin oscillatory zoned cores with dark and structureless rims (Figure 4(d)). Twenty-two zircon grains were selected for U–Pb zircon analysis, of which ten analyses were discarded due to high common Pb contents (online supplementary Table S1). The remaining twelve analyses have high Th/U ratios of 0.09–1.62 and define a discordant line intersecting the Concordant line at 1852 ± 41 Ma (MSWD = 4.4; Figure 5(f)). One Concordant analysis yielded an apparent 207Pb/206Pb age of 1826 ± 8 Ma, which is consistent with the upper intercept age. The age of 1852 ± 41 Ma is regarded as the crystallization age. Zircon U–Pb age analyses are listed in online supplementary Table S1.
3.2. Whole-Rock Major and Trace Element Compositions
A total of twelve granitic samples were analyzed for whole-rock major and trace element geochemistry. They exhibit low loss on ignition (LOI) values of 0.34–1.07 wt%, suggesting a limited post-magmatic or metamorphic alteration (online supplementary Table S2). All the samples show relatively high SiO2 values ranging from 70.27 to 76.91 wt%, variable Fe2O3T (1.37, 3.43 wt%) values, high K2O (3.22, 6.37 wt%) and Na2O (2.69, 4.29 wt%) contents, and high alkalis (K2O+Na2O = 7.19–9.60 wt%) but low CaO (0.50, 2.05 wt%) and MgO (0.22, 0.82 wt%, excluding one value of 2.40 wt%) values, and low Mg# values (0.11, 0.38, excluding one value of 0.73; online supplementary Table S2). The studied granitoids yield Al2O3 contents of 11.80–14.46 wt% and A/CNK (Al2O3/[CaO+Na2O+K2O] mol%) values of 1.04–1.10, defining a weakly peraluminous character (Figure 6(a)). The K2O/Na2O ratios range from 0.75 to 2.10. These granitoids also possess Rittman Index of 1.69–3.11 and plot in the high-K calc-alkaline field on the SiO2 versus K2O diagram (Figure 6(b)). On the TAS diagram, all samples fall within the granite region (Figure 6(c)).
On the primitive mantle-normalized spider diagram, all the studied samples are selective enrichment in large ion lithophile element (LILE) of Rb and depletion in high field strength elements (HFSE) of Nb, Ta, and Ti, marked negative Sr and Ba and positive Th, U, and Pb anomalies (Figure 7(a) and 7(c)). The total rare earth element (REE) contents of the 2531 and 2027 Ma age groups ranged from 156 to 283 ppm, which are lower than those of the 2189–2173 and 1852 Ma age groups that have total REE contents of 239–666 ppm (online supplementary Table S2). All the samples show fractionated right-declining chondrite-normalized REE patterns (Figure 7(b) and 7(d)). The three older groups (2531, 2189–2173, and 2027 Ma) are characterized by enrichments in light REEs with (La/Yb)N ratios of 7.01–70.14, relatively flat heavy rare earth elements (HREEs) with (Gd/Yb)N ratios of 0.86–7.28, and strong negative Eu anomalies (Eu/Eu* = 0.15–0.61; Figure 7(b) and 7(d)). By contrast, the 1852 Ma age group shows significantly fractionated REE patterns, with (La/Yb)N ratios of 145.43–165.24 and (Gd/Yb)N ratios of 6.67–7.49, as well as moderate negative Eu anomalies (Eu/Eu* = 0.36–0.43, Figure 7(d)). Whole-rock major and trace element compositions are listed in online supplementary Table S2.
3.3. Zircon Hf–O Isotope Compositions
The samples that were previously analyzed for U–Pb geochronology were further conducted for Hf isotope analysis (online supplementary Table S3). The crystallization ages were used to calculate the Hf two-stage model age (TDM). The 2531 Ma age group yields homogeneous 176Hf/177Hf ratios of 0.281294–0.281357, corresponding to positive εHf(t) values of +3.51 to +5.53 and TDM ages of 2656–2755 Ma (Figure 8). The 2189–2173 Ma age group shows a wide range of 176Hf/177Hf ratios between 0.281187 and 0.281554, mostly positive εHf(t) values of −0.31 to +3.76, except an outlier with an εHf(t) of −8.37. This age group exhibits TDM ages of 2450–2653 Ma (Figure 8). The 2027 Ma granitic gneiss exhibits concentrated positive εHf(t) values of +5.59 to +7.32, which are close to the depleted mantle line (Figure 8). This sample shows a narrow range of initial 176Hf/177Hf values of 0.281703–0.281762 and TDM ages of 2154–2240 Ma (except for one outlier with an εHf[t] value of +4.63). The εHf(t) values of the 1852 Ma granite are below the chondrite evolution line and ranging from −4.98 to −1.53 (Figure 8). The scattered initial 176Hf/177Hf ratios of this group vary from 0.281482 to 0.281594, corresponding to the TDM ages of 2451–2622 Ma and affinities to the 2189–2173 Ma deformed granite.
Five samples were subjected to in-situ SIMS oxygen isotope analysis (16LL26-2, 16LL31-1, 16LL39-2, 16LL40-1, and 16LL42-1). Analyses with U–Pb discordances >10% or high common Pb were excluded. The 2189–2173 Ma deformed granite has δ18O values ranging from 3.22‰ to 6.31‰ (Figure 9), with an average of 5.21‰ ± 0.23‰ (n = 24; 2σ). The 2027 Ma granitic gneiss shows δ18O values of 4.09‰–5.15‰ and an average of 4.76‰ ± 0.18‰ (n = 4, 2σ; Figure 9). The 1852 Ma granite has only two analyses near the Concordant line, displaying δ18O values of 5.16‰–5.40‰ (Figure 9), with an average of 5.28‰ ± 0.24‰ (n = 2, 2σ). The zircon Hf–O isotope compositions are listed in online supplementary Tables S3 and S4.
4.1. Classification of the 2531–1852 Ma Granitoids
The igneous rocks that have undergone strong alteration typically exhibit certain characteristics, such as high LOI values or significant Ce anomalies (|Ce/Ce*−1|>0.1 ). However, the studied samples show commonly low LOI values of 0.34–1.07 wt% and Ce/Ce* ratios of 0.90–1.35 (online supplementary Table S2), indicating weak alteration . Chappell and White  proposed the classification for the protoliths of granitic rocks and divided them into two main groups: S-type (sedimentary) and I-type (igneous). Of the I-type group, Loiselle and Wones  introduced the A-type subgroup, which was further studied by Collins et al. . The granitoids in this study can be determined as I-type granites based on the following lines of evidence: (1) absence of Al-rich minerals, such as muscovite or magmatic garnet; (2) low A/CNK values (1.01, 1.10), indicating a weakly peraluminous characteristic (Figure 6(a)); (3) lower P2O5 (0.02, 0.11 wt%) and Al2O3 (11.80, 14.46 wt%) contents than typical S-type granite (Al2O3>14 wt%; online supplementary Table S2 [76, 77]); and (4) δ18O values (<6.31‰) of magmatic zircon grains (Figure 9), indicating the involvement of limited sediments in the magma sources.
The primary components of granite are mainly influenced by the geochemical characteristics of its source region. Consequently, major elements are not effective in discriminating A-type granite from other types, particularly those with higher SiO2 contents (>72% ). A high Ga/Al ratio has been utilized as a practical marker to distinguish A-type granite from other types . All the studied granitoids display high 10,000*Ga/Al (>2.6) values in the A-type granite field (Figure 10a–10b). The Ga/Al ratio is most discernible for those strongly alkaline samples but less discernible for subalkaline samples . This phenomenon can be attributed to the fact that the studied granitoids are felsic nonperalkaline, which may represent the highly fractionated I- or S-type granites [78, 79]. In addition, it is shown that the Ga/Al ratios of both I- and S-type granites tend to increase during the process of fractional crystallization, leading to comparable Ga/Al ratios as those observed in A-type granite . Thus, the Zr+Nb+Ce+Y contents are more effective for differentiating highly fractionated S- or I-type granites from A-type granite . As shown in Figure 10(c)–10(d), the 2189–2173 Ma deformed granite and 1852 Ma granite plot within the A-type field due to their high Zr+Nb+Ce+Y values, while the 2531 and 2027 Ma granitic gneisses overlap the area of fractionated felsic granites and unfractionated I- and S-type granites (OGT). This observation is consistent with the lower whole-rock Zr/Hf and Nb/Ta ratios of the 2531 and 2027 Ma granitic gneisses (Figure 10(e)). Moreover, fractional crystallization is a cooling process, as it progresses, the Ga/Al ratio of I- or S-type granites gradually increases, while the Zr content decreases and the Nb content increases; however, such correlations are not evident in A-type granite . This trend is consistent with the negative correlation between Ga/Al and Zr and the positive correlation between Ga/Al and Nb observed in the 2531 and 2027 Ma granitic gneisses (Figure 10(a)–10(b)). Furthermore, the 2189–2173 Ma deformed granite and 1852 Ma granite exhibit higher calculated zircon saturation temperatures (845℃, 931℃) than those of the 2531 and 2027 Ma granitic gneisses (801℃, 826 ℃; online supplementary Table S2). Additionally, the total REE contents of the 2189–2173 Ma deformed granite and 1852 Ma granite are significantly higher than those of the 2531 and 2027 Ma granitic gneisses (Figure 7, online supplementary Table S2). Therefore, it can be concluded that the 2189–2173 Ma deformed granite and 1852 Ma granite possess the characteristics of A-type granite, while the 2531 and 2027 Ma granitic gneisses display I-type features.
4.2. Petrogenesis of the 2531–1852 Ma Granitoids
4.2.1. 2531 Ma and 2027 Ma I-type Granitic Gneisses
As mentioned above, the 2531 and 2027 Ma granitic gneisses are I-type granites. And the I-type granite is thought to form through three possible mechanisms: (1) direct generation via fractional crystallization of mantle-derived basaltic magma ; (2) mixing of mantle-derived basaltic magma with crustal materials ; and (3) partial melting of mafic crustal rocks . The granite that originated from extreme differentiation of mantle-derived magmas is usually found in association with a large amount of coeval mafic rocks and a series of igneous rocks with successive compositions from basaltic to granitic [83-85]. But such phenomena are absent in this region. Coeval (~2531 and ~2027 Ma) mafic magmatism is rare within the Lüliang Complex, except for an amphibolite with an upper intercept age of 2051 ± 68 Ma . Thus, the possibility that the 2531 and 2027 Ma granitic gneisses were generated from fractional crystallization of mafic magma can be precluded. In addition, no petrologic or field evidence, for instance, component changes of minerals or micro-mafic enclaves within these granitoids support the magma mixing origin for the 2531 and 2027 Ma I-type granitic gneisses. Hf isotopic features are further opposed to the magma mixing origin because the 2531 and 2027 Ma I-type granitic gneisses exhibit homogeneous and positive εHf(t) values varying from +3.51 to +5.53 and +5.59 to +7.32 (Figure 8, online supplementary Table S3), respectively. Thus, partial melting of Neoarchean juvenile mafic crust might be the most likely mechanism for the 2531 and 2027 Ma I-type granitic gneisses of this study. The studied 2531 and 2027 Ma I-type granitic gneisses exhibit flat HREEs patterns, strong negative Eu anomalies (Figure 7(b)), lower Sr/Y ratios, and high Y values (Figure 10(f)). These features suggest a residue with abundant plagioclase with little or no garnet, which is indicative of melting at a shallow crustal level (<0.8 GPa or <30 km ). The 2027 I-type granitic gneiss displays low δ18O values (4.79‰, 5.15‰; except for one value of 4.09‰ of the discordant analysis 40-1-11) within the range of mantle-like zircon grains (5.3‰ ± 0.3‰; Figure 9; online supplementary Table S4 ). O isotopes of the 2027 Ma I-type granitic gneiss suggest that the source region lacked significant recycled sediments or supracrustal rocks. In conclusion, the 2531 and 2027 Ma I-type granitic gneisses originated from the partial melting of juvenile mafic rocks at a relatively shallow crustal depth.
4.2.2. 2189–2173 Ma and 1852 Ma A-type Granites
Several petrogenetic models have been proposed for the origin of A-type granite, including (1) extreme differentiation of mantle-derived basaltic magma [88-91]; (2) partial melting of mafic–intermediate rocks in the middle–lower crust ; and (3) partial melting of felsic, infracrustal igneous rocks with H2O contents similar to those of I-type granites in the shallow crust [78, 93, 94].
Mantle-derived A-type granite shares similarities with classic peralkaline A-type granite, which are identified by the presence of alkali minerals, such as amphibole, annite-rich biotite, and pyroxene, and often exhibit a narrow range of εHf(t) values that overlap or are slightly lower than those of the depleted mantle [75, 88, 91]. The 1852 Ma A-type granite displays extended negative εHf(t) values of −4.98 to −1.53 (online supplementary Table S3). The 2189–2173 Ma A-type deformed granite shows a wide distribution of εHf(t) values, mostly positive ranging from +0.47 to +3.76 (except for two outliers of −0.31 and −8.37; Figure 8; online supplementary Table S3). However, these values remain significantly lower than those of the depleted mantle line. In addition, both the 2189–2173 Ma A-type deformed granite and 1852 Ma A-type granite yield A/CNK values of 1.02–1.10, A/NK values >1.13 and a lack of alkali mafic minerals, suggesting weakly peraluminous characteristics (Figure 6(a), online supplementary Table S2). The above features are inconsistent with mantle-derived A-type granite, ruling out the first mechanism.
Zircon grains from the 2189–2173 Ma A-type deformed granite display lower δ18O values (3.22‰, 6.32‰) than mantle-like zircon grains (5.3‰ ± 0.3 ‰; Figure 9; online supplementary Table S4 ). This 18O-depletion signature in zircon grains results from two possible mechanisms: (1) meteoric–hydrothermal alteration at high temperature [95, 96]; and (2) zircon crystallization from low-δ18O magmas [97-100]. Given the slow rate of zircon growth and low oxygen diffusion rate [95, 101-103], it is unlikely for zircon to undergo isotopic re-equilibration during subsolidus (<600℃) meteoric–hydrothermal alteration within a geologically reasonable time frame [104-106]. The studied zircon grains with Concordant U–Pb ages yield low U contents (average of 304 ppm) and high Th/U ratios (0.29, 1.62, average of 0.66; online supplementary Table S1) and generally preserve typical oscillatory zonings in CL images (Figure 4), suggesting a magmatic origin and the absence of significant radiation damage or late alteration . Therefore, the low-δ18O signature of the 2189–2173 Ma A-type deformed granite is most likely inherited from their parental magma, which remelting of preexisting low-δ18O materials (e.g., isotopic exchange between source rock and meteoric water/seawater at high temperature, namely, high-temperature water–rock reaction) rather than post-magmatic meteoric–hydrothermal alteration. The formation of A-type granite requires a high heat flow and extensional environment that increased fracture permeability. This allows surface water to reach the crustal depths of 8–10 km for deep hydrothermal circulation. This creates conditions favorable for generating low-δ18O magmas . These observations consequently imply that the source of the 2189–2173 Ma A-type deformed granite may have similar H2O content as I-type granite. Furthermore, the 2189–2173 Ma A-type deformed granite exhibits moderate REE differentiation, substantial depletion of Eu (Figure 7(d)), decreased Sr/Y ratios, and elevated Y concentrations (Figure 10(f)), implying that they were formed through partial melting at a relatively shallow crust level (<0.8 GPa or <30 km). This is attributed to the absence of garnet but the presence of plagioclase in the residual source . The major and trace element geochemistry, as well as the Hf–O isotopic signatures of the 2189–2173 Ma A-type deformed granite suggest a strong resemblance to A-type granite formed by partial melting of felsic igneous rocks with normal H2O contents in the shallow crust.
Compared with the 2189–2173 Ma A-type deformed granite, the 1852 Ma A-type granite exhibits highly fractionated REE patterns, medium negative Eu anomalies (Figure 7(d)), and higher Sr/Y ratios with lower Y contents (Figure 10(f)), indicating the presence of residual garnet and plagioclase in the source. The presence of garnet further suggests that the partial melting occurred under pressures of 1.0–1.4 GPa or at a depth of 33–50 km in a slightly thickened crust . The negative εHf(t) values of −4.98 to −1.53 indicate an ancient crust source for the 1852 Ma A-type granite. Thus, the 1852 Ma A-type granite was most likely derived from an ancient mafic–intermediate crust at the middle–lower crust.
4.3. Tectonic Process of the TNCO Before 1950 Ma
Multistage magmatic events were recorded in the Lüliang Complex, such as ~2410–2360 Ma Gaijiazhuang granitic gneiss [10, 11, 24, 109], ~2210–2100 Ma Chijianling–Guandishan granitoid gneisses [24, 32, 33, 35-38, 64, 110-112], ~2070–2020 Ma granitoid [32, 33, 38, 66, 111], and ~1870–1850 Ma Huijiazhuang gneissic granite. Most of these magmatic events are considered to have been generated in subduction-related environments [33, 111, 113]. As mentioned above, thus far, there has been no documentation of ~2500 Ma magmatism in the Lüliang Complex, with the sole exception of the Yunzhongshan TTG gneisses. The 2531 Ma I-type granitic gneiss from this study was first found and complement the mosaic geochronology in the Lüliang Complex. In addition, the 2531 Ma I-type granitic gneiss displays geochemical affinities to those originating from an arc setting. This is evidenced by their selected enrichment in Rb of LILE, depletion in Nb, Ta, and Ti of HFSEs, and fractionated REE patterns with negative Eu anomalies. Additionally, multiple terranes in the TNCO have recorded magmatic events during the Neoarchean period. For instance, the Datong-Huai'an Complex contains the TTGs emplaced at 2538–2497 Ma ; the Hengshan Complex contains the diorites, TTGs and volcanics with the age range of 2538–2483 Ma ; the Wutai Complex has granitoids with emplacement ages of 2560–2519 Ma ; the Fuping Complex has the TTG rocks emplaced at 2513 ± 13 Ma . The Yunzhongshan area contains granitoid gneisses and metamorphosed volcano-sedimentary sequences formed at 2535–2486 Ma ; and the Zhongtiao Complex contains the Zhaizi and Xiyao TTGs formed at 2560–2536 Ma [30, 31]. All these granitoids are interpreted to have originated in tectonic settings associated with subduction-related environment, such as island arc, continental arc, or back-arc basin. Therefore, we suggest that a subduction-related regime is a more favorable interpretation for the geochemical and isotopic features of the 2531 Ma I-type granitic gneiss. The newly reported Neoarchean magmatism of the Lüliang Complex plays a pivotal role in bridging the northern and southern parts of the TNCO, establishing a new evidence chain of Neoarchean subduction system.
In recent years, more and more studies have reported the existence of Paleoproterozoic A-type granites in the TNCO. Some of the A-type granites are interpreted to have formed in a rift setting, and detailed information on these A-type granites in the TNCO is summarized in Table 1. The generation of A-type granitic magma requires extensional decompression resulting from high heat flow, which can occur in different tectonic settings. These include within-plate rifting, post-orogenic delamination, and subduction [79, 114]. Therefore, a scheme involving a long-lived subduction–accretion system provides a clear explanation for the multistage A-type granitic magmatism throughout the tectonic evolution of the TNCO. The process of continental arc-related subduction involves a prolonged lithospheric extension, which is characterized by intermittent episodes of transient contraction . During the extension phase, the retreat of the subducting slab triggers the lithospheric stretching, which leads to the decompression and melting of the asthenosphere. The upwelling asthenosphere beneath the back-arc region provides an external heat source for the formation of A-type granitic magma. The cycles of rapid alternation between contraction and lithospheric extension, also called tectonic switching, may document multiple episodes of A-type granitic magmatism and enhance efficient continental growth [115, 116]. In addition, Figure 8 presents a compilation of Hf isotopic and U–Pb chronological data for granitoids and equivalent felsic igneous rocks that were emplaced along the TNCO from the Neoarchean to Paleoproterozoic. The εHf(t) values that are close to or intersect with the depleted mantle line indicate that they were mostly derived from the partial melting of the juvenile crust . Hf isotopic data further suggest the existence of juvenile crust or continental growth during the period of 2600–2000 Ma (Figure 11).
Eby  divided A-type granites into the A1 and A2 groups based on their Y/Nb ratios. The A1 group is typically associated with continental rifts or intraplate magmatism, whereas the A2 group is believed to represent magma generated during continent–continent collision or island-arc processes . The A-type granites identified in the TNCO predominantly exhibit A2-type features prior to 1950 Ma, whereas the A1-type signature became more prominent after 1950 Ma (Figure 12; Table 1). King et al.  proposed a further categorization of A-type granites into two subgroups (aluminous and peralkaline) based on the presence or absence of alkali mafic minerals. The aluminous group is derived from the partial melting of felsic crustal source rocks and has a similar H2O content as I-type granite, while the peralkaline group is formed through the fractionation of mafic magma . The studied 2189–2173 Ma A2-type deformed granite, as well as others in the TNCO that were emplaced prior to 1950 Ma, exhibit metaluminous to peraluminous characteristics and fall into the aluminous A-type subgroup. This implies that the source of A-type granites was not affected by any substantial mantle-derived material, and the heat source mainly came from the asthenosphere. The low δ18O signature observed in the 2189–2173 Ma A-type deformed granite suggests that they were derived from preexisting materials with low δ18O values (Figure 9). This also implies that the TNCO during this period was not an entirely “anhydrous” environment.
The middle section of the TNCO comprises the Lüliang, Wutai, Fuping, and Zanhuang Complexes that align along a W-E-trending (Figure 1). The A-type granites from these four complexes, which are of the same age, exhibit average zircon saturation temperatures of 879°C (this study; online supplementary Table S2), 884°C , 890°C , and 893°C , respectively. The gradual increase in crystallization temperature from west to east suggests that the Lüliang Complex is closer to the fore-arc region compared with other complexes. Moreover, the detrital zircon ages of metasedimentary rocks from the Lüliang Complex exhibit an age population of 2790–2600 Ma , while the xenolithic zircon ages from the Wutai Complex show an age population of 2763–2660 Ma , and the gneissic rocks from the Hengshan Complex represent old continental crustal components with an age range of 2712–2701 Ma . These ages are in line with the 2700–2600 Ma crustal growth event in the Eastern Block as reported in previous studies [121, 122]. Therefore, we can infer that the Lüliang Complex was situated along the western margin of the Eastern Block, indicating the existence of eastward subduction of an ancient ocean between the Eastern and Western Blocks (Figure 13). In summary, we propose that the 2531–2027 Ma granitoids from the Lüliang Complex were formed in a magmatic arc environment, and the eastward subduction had already started at 2531 Ma.
4.4. Final Amalgamation of the NCC
It is widely believed that the collision between the Eastern and Western Blocks occurred at ~1950 Ma, which is supported by the numerous metamorphic zircon ages. For instance, granulite-facies metamorphism occurred at 1950–1920 Ma in the Lüliang Complex [44, 47, 49, 61] and is interpreted as the peak-stage metamorphism due to continental collision. Metamorphic ages of 1960–1950 and ~1920 Ma are also reported in the Hengshan Complex [50, 52, 55]. The former is interpreted as representing prograde or peak stages of metamorphism, while the latter indicates the cooling stages. In the Wutai area, prepeak or peak-stage metamorphism has been constrained at ~1950 Ma [50, 51]. The Huai'an Complex has a metamorphic history ranging from the HP granulite-facies stage of 1960–1900 Ma . Decompression, external heat supply, and addition of hydrous fluid are the most significant factors influencing the formation of granite . The available data from 1890–1800 Ma post-peak metamorphism and the clockwise near-isothermal decompression P‒T path in the TNCO (e.g., Huai’an, Hengshan, Wutai, Fuping, Lüliang, and Taihua Complexes [44, 47-58, 60, 61]) support the idea that post-collisional extension and the associated asthenospheric upwelling were the mostly possible heat source that resulted in the formation of 1852 Ma A-type granite.
Based on the available data, we propose a tectonic scenario involving a long-lived subduction process for the tectonic evolution of the TNCO from late Neoarchean to Paleoproterozoic (Figure 13). At ~2531 Ma, an eastward subduction began in a vast ocean between the Eastern and Western Blocks and led to the formation of a continental arc along the western margin of the Eastern Block (Figure 13(a)). Following this, the subducting slab retreated, resulting in regional extension and the formation of back-arc basins at 2189–2173 Ma (Figure 13(b)). This long-lived subduction process involved several cycles of tectonic switching from slab subduction to roll-back, which effectively triggered multiple episodes of subduction-related magmatism and contributed to the generation of continental crust (Figure 13(b)). Finally, the collision between the Eastern and Western Blocks along the TNCO resulted in the peak-stage metamorphism at ~1950 Ma (Figure 13(c)). The collision persisted until the collapse of mountain roots, which triggered asthenospheric upwelling and post-collisional magmatism at ~1852 Ma (Figure 13(d)).
Four groups of granitoids have been recognized in the Lüliang Complex, including 2531 Ma I-type granitic gneiss, 2189–2173 Ma A2-type deformed granite, 2027 Ma I-type granitic gneiss, and 1852 Ma A1-type granite. The newly reported 2531 Ma I-type granitic gneiss represents the oldest known granitoid in the Lüliang Complex. This granitoid, along with the 2027 Ma I-type granitic gneiss, is thought to have originated from the partial melting of juvenile mafic crust in a magmatic arc environment. The 2189–2173 Ma A2-type deformed granite was formed by partial melting of felsic igneous rocks in the shallow crust with normal H2O content, which most likely developed at the back-arc basin. The 1852 Ma A1-type granite was formed by partial melting of mafic–intermediate rocks in middle–lower crust at a post-collisional setting. The evolution of the Lüliang Complex is characterized by the eastward subduction of a vast ocean between the Eastern and Western Blocks, which started during the 2531 Ma and lasted until 1852 Ma. This prolonged period of subduction recorded multiple episodes of subduction-related magmatic events. The collision between the two Blocks occurred at ~1950 Ma to form the TNCO, resulting in the final amalgamation of the NCC.
We are grateful to journal editors Prof. Tamer S. Abu-Alum and Prof. Songjian Ao, as well as two anonymous reviewers for their constructive comments that helped us improve the manuscript. We would like to thank Qing Yang, Wanfeng Zhang, and Boqin Xiong for their assistance during the SIMS U–Pb geochronology and O isotopic analyses. This study is financially supported by the Natural Science Foundation of China (42025204, 41890381, 42202225, and 41972197).
Conflicts of Interest
The authors declare that there is no conflict of interest regarding this paper.
The data that support the findings of this study are available from the published papers cited in the references list and included as supplementary material within the article.