Mineralogical and Geochemical Study on the Yaojiazhuang Ultrapotassic Complex, North China Craton: Constraints on the Magmatic Differentiation Processes and Genesis of Apatite Ores

The differentiation process of ultrapotassic magmas is enigmatic and poorly understood. The Yaojiazhuang ultrapotassic complex is concentrically zoned by late-intruded syenite in the core and early emplaced clinopyroxenite in the periphery, combining a “bi-modal” lithology. Spatially, apatite and iron oxide-apatite (IOA) ores, glimmerite and pseudoleucite occur in the upper part of clinopyroxenite. The syenite and clinopyroxenite are composed of variable amounts of clinopyroxenite, biotite, K-feldspar, magnetite, apatite with minor analcite, titanite, and primary calcite. The pseudoleucite clinopyroxenite contains mainly clinopyroxene, biotite and garnet in the matrix, and nepheline–K-feldspar intergrowth with muscovite and minor celestine in the leucite pseudomorph. Geochemically, rocks of the Yaojiazhuang complex are significantly enriched in potassium (K), light rare earth elements (LREE), and large ion lithophile elements (LILE). Crustal contamination by Archean tonalite–trondhjemite–granodiorite (TTG) gneisses basement may play an important role to convert the syenitic melts from silica-undersaturation to saturation. Fractionation crystallization is supported by the mineral crystallization sequence to explain the bimodal lithologies instead of silicate liquid immiscibility. During the magmatic evolution, decompression, fractionation of volatile-poor clinopyroxene and the enhancement by CO2 may result in the exsolution of an aqueous fluid. The late-stage interactions between existing minerals and magmatic fluids in the crystal mush could be a key process in the generation of both leucite pseudomorphs and apatite/IOA ores.


INTRODUCTION
The occurrence of ultrapotassic rocks is volumetrically rare, with the origin usually related to intraplate extension environments and collision zones (e.g., Foley et al., 1987;Miller et al., 1999;Conticelli et al., 2007Conticelli et al., , 2009Conticelli et al., , 2013. Considerable amounts of studies had been conducted on ultrapotassic rocks which are mostly focusing on their mantle source and partial melting processes (e.g., Thompson et al., 1990;Conticelli and Peccerillo, 1992;Foley and Peccerillo, 1992;Rogers et al., 1992;Zhang et al., 1995;Miller et al., 1999;Lobach-Zhuchenko et al., 2008;Gaeta et al., 2016). Metasomatized mantle sources and relatively low degree of melting process had been proposed for the generation of ultrapotassic primary melts (e.g., Dawson, 1987;Marks et al., 2008;Sun et al., 2014;Bodeving et al., 2017;Soder and Romer, 2018;Sokół et al., 2018). Nevertheless, most of the ultrapotassic rocks are not solidified from primary mantlederived liquids, thus the subsequent differentiation processes of ultrapotassic magmas in the crust including fractionation crystallization, crustal contamination, magma (un-)mixing and mingling, and exsolution/transportation of both magmatic and post-magmatic fluids could also play important roles in the formation of them (e.g., Ferlito and Lanzafame, 2010;Jeffery et al., 2013;Brenna et al., 2015;Wolff, 2017;Burchardt, 2018). However, to the best of our knowledge, these crustal processes are poorly constrained. Hence, a comprehensive illustration on the differentiation processes is crucial for our understanding about the formation of ultrapotassic rocks.
Moreover, some alkaline rocks, including ultrapotassic rocks are associated with mineral deposits, mainly apatite ± magnetite ± minor silicate minerals (phoscorite, IOA ores and monomineralic apatite ores), such as Ordovician-Silurian Misvaerdal complex in Norway (Ihlen et al., 2014), Khibiny complex in Kola Peninsula, Russia (Notholt, 1979;Kogarko, 2018) and Mushgai-Khudag Complex, South Mongolia (Nikolenko et al., 2018). Accordingly, the formation of these P-rich ores is genetically related to the differentiation of the alkaline magmas. Two major models including fractional crystallization (Hou et al., 2015) and liquid immiscibility (Kolker, 1982;Jiang et al., 2004;Mokhtari et al., 2013) had been proposed for the formation of these apatite and IOA ores. Therefore, it is of great significance to elucidate the petrogenesis of ore-bearing alkaline complex, which could potentially shed new lights on the formation of the apatite and IOA ores.
The Yaojiazhuang alkaline plutonic intrusion is a "bi-modal" ring complex with syenite in the core and clinopyroxenite in the periphery generated by ultrapotassic magmas frozen in the crust according to previous studies (Chen et al., 2013). Notably, there is also a large apatite and IOA ore deposit hosted at the upper part of clinopyroxenite and have been mined as phosphorus deposit for decades. However, the formation mechanism of bi-modal lithofacies and tons of phosphorus concentration are still in widely debates and remain poorly understood (e.g., Hou, 1990b;Mu and Yan, 1992;Zhang, 1999;Chen et al., 2013;Tang et al., 2014;Liu and Tang, 2018). Previous studies suggested that the compositionally contrasting syenite and clinopyroxenite could be the Si-rich and Fe-rich melts produced by silicate liquid immiscibility from a common ultrapotassic parental magma, supported by thermodynamic simulations (Hou, 1990b;Ma et al., 1999), and during the separation of immiscible liquids, P 2 O 5 tends to be partitioned into the Fe-rich melts and thus form the apatite and IOA ore in clinopyroxenite (Hou, 1990a). Alternatively, other authors commonly favor fractional crystallization either for Yaojiazhuang complex (Chen et al., 2013) or adjacent Fanshan complex (Cheng and Sun, 2003;Hou et al., 2015).
Additionally, the widespread hydrated minerals like biotite and the occurrence of pegmatites which has not been reported in detail indicate that during the formation of Yaojiazhuang complex, the magma ought to be enriched in fluids. Recent studies had revealed that the interactions between fluids and minerals during the last magmatic stage could significantly modify rock structures and textures, and could also act as a key mechanism in the generation of economic ore deposits, especially apatite and IOA deposits (e.g., Roedder, 1992;Huber et al., 2012;Ballhaus et al., 2015;Guo and Audétat, 2017;La Cruz et al., 2019;Knipping et al., 2019). Therefore, a combined study involving mineralogy, petrology and geochemistry studies is required to elucidate these processes. Thus in this paper, we conducted comprehensive study for the differentiation of Yaojiazhuang complex, in order to better understand the differentiation processes of ultrapotassic magmas and shed new lights on the formation of the apatite and IOA ores.

Regional Geology
The Yaojiazhuang ultrapotassic complex is located at the north margin of the North China Craton (NCC), ∼160 km northwest of the Beijing city ( Figure 1A). The basement rocks of NCC are mainly mafic granulites, amphibolites and tonalitetrondhjemite-granodiorite (TTG) gneisses of ages up to 3.85 Ga (Liu et al., 1992), and are overlain unconformably by Meso-to Neo-Proterozoic sediments (Zhao et al., 2001). To the north of the NCC is the Central Asian Orogenic Belt, which strongly influenced the northern margin of NCC during Carboniferous to Permian by the progressive closure of Paleo-Asian Ocean leading to Andean-type magmatism and Himalayan-type collision, and formation of the Solonker suture (e.g., Xiao et al., 2003;Zhang et al., 2009). Following the closure was a post-collisional extension during early Mesozoic, resulting in a series of magmatic emplacement events and forming a > 1,500 km E-W trending alkaline rock belt along the northern margin of NCC (e.g., Zhang and Wang, 1997;Zhang, 1999;Zhao et al., 2008;Ren et al., 2009;Zhang et al., 2012). The major rock types can be divided mainly into three associations: (1) clinopyroxene diorite -quartz syenite -aegirine syenite -syenite -granite association, such as Shuiquangou complex; (2) nepheline syenite -aegirine syeniteurtite -ijolite association, such as Xiangshuigou and Aoyukou complexes; (3) clinopyroxenite -biotite clinopyroxenitenelsonite -syenite association, representative complexes usually host economic iron and phosphorous ore deposits such as Yaojiazhuang (this study) and Fanshan complexes (e.g., Mu and Yan, 1992;Zhang, 1999;Jiang et al., 2004;Zhang et al., 2012;Hou et al., 2015).

Geology of Yaojiazhuang Ultrapotassic Complex
The ca. 221 Ma Yaojiazhuang ultrapotassic complex (Zhang et al., 2012) is an oval-shaped intrusive body in plan with a  Ren et al., 2009). Labeled ages and analytical methods are compiled from Zhang et al. (2012) and references therein. (B) Geological map of Yaojiazhuang ultrapotassic complex modified from Hou (1990a).
Frontiers in Earth Science | www.frontiersin.org 3 August 2020 | Volume 8 | Article 357 size around 1.8 × 1.3 km ( Figure 1B). It intruded the Archean gneiss and consists of a "bi-model" lithology with syenite at the core and clinopyroxenite at the periphery. At the places where they directly contacted, mingling between syenite and clinopyroxenite is observed (Figure 2A). In some parts of the complex, K-feldspar clinopyroxenite is outcropped either between the units of syenite and clinopyroxenite or scattered adjacent to the clinopyroxenite ( Figure 2B). Besides, some clinopyroxenite host cm-sized pseudoleucite which had been previously identified as K-feldspar clinopyroxenite. At the upper part of clinopyroxenite, i.e., toward to syenite, the amount of apatite increased continuously, forming porphyritic or spheroidal apatite + magnetite ores up to around 10 cm ( Figure 2C). These apatite clusters are cemented by a mineral assemblage resemble to clinopyroxenite. Coarsening of biotite is also observed locally, thus rocks in these places become extremely fragile ( Figure 2D). Pegmatitic syenites occurred at the center of the syenite unit ( Figure 2E) and aegirine syenite is locally present. Moreover, except the main units as mentioned above, late stage intrusive activities are quite vigorous, almost Frontiers in Earth Science | www.frontiersin.org all the rocks in the Yaojiazhuang complex are cut by dykes with various compositions (Figure 2F).

Scanning Electron Microscope
Polished thin sections were firstly mapped using a Zeiss Ultra 55 FESEM equipped with an INCA MAX 20 electron dispersive spectrometry (EDS) system at the Key Laboratory of Submarine Geosciences State Oceanic Administration, Second Institute of Oceanography, Ministry of Natural Resources (SIOMNR), China. The employed accelerating voltage was 15 kV.

Whole-Rock Major and Trace Elements
Samples were firstly peeled off the weathered or altered surfaces before jaw-crushing and the subsequent grounding in agate mills to powders of over 200 mesh. The pseudoleucite clinopyroxenites were initially packaged in a plastic envelope and crushed by hammering to small pieces, and then those pieces of as much as pure pseudoleucite or matrix were separately collected by hands and grounded to powders of ∼200 mesh in an agate mill. Major elements were analyzed by X-ray fluorescence (XRF) on fused glass discs at the laboratory of Nanjing Hongchuang Exploration Technology Service Co., Ltd. (NHETS) in Nanjing, China. The analytical uncertainties are < 1%, estimated from repeated analyses of 10 standards (GBW07101, GBW07103, GBW07105, GBW07114, AGV-2, BHVO-2, COQ-1, DTS-2b, GSP-2, and W-2a). Trace element data was determined by inductively-coupledplasma mass-spectrometry (ICP-MS) after acid digestion of powders in high-pressure Teflon bullets performed at NHETS. The analytical precision for trace and rare earth elements is generally better than 10%, verified by repeated analyses of eight standards (GBW07103, GBW07316, BCR-2, W-2a, GSP-2, DTS-2b, BHVO-2, and AGV-2).

Electron Microprobe Analysis
Compositions of mineral phases were analyzed at SIOMNR using a JXA-8100 electron probe micro-analyzer (EPMA). Operating conditions were performed at 15 kV accelerating voltage with a 20 nA beam current. The beam size was 5 µm or 1 and up to 15 µm depending on crystal sizes. The peaks are counted by 10 s with background half of that for every element and ZAF method is employed in calibration. The standard reference samples are provided by National Technical Committee for Standardization of Microbeam Analysis (China) or produced by SPI Supplies R (America). The analytical precision was better than 5%.

PETROGRAPHY Clinopyroxenite
Medium-to coarse-grained clinopyroxenite is commonly composed of > 60 mol.% of euhedral columnar (up to 2 mm in length) clinopyroxene, as well as various amounts of anhedral biotite and K-feldspar, and minor accessory minerals mainly including apatite, magnetite and analcite (Figure 3). The oriented euhedral cumulate clinopyroxene are commonly seen as imbedded crystal hosted in biotite, titanite and K-feldspar ( Figures 3A,B) forming poikilitic texture. Besides, most of the cumulus clinopyroxene is compositionally zoned. Except occurring as oikocryst, K-feldspars also occur as fine-grained crystals coexisting with analcite in between clinopyroxene crystals ( Figure 3C). Occasionally, biotite constitutes up to 50 mol.% of the rock and becomes euhedral with size up to 3 mm thus forming biotite clinopyroxenite and even glimmerite ( Figure 3D). Euhedral apatite (0.1-0.5 mm in diameter), magnetite (0.1-1.0 mm) and occasionally ilmenite (∼0.1 mm) are presented as minor phases, and in some biotite rich area apatite can reach 15 mol.% to form apatite ores. Additionally, at some contact boundaries of minerals, magnetite is more wellformed than apatite but less than clinopyroxene ( Figure 3C). Moreover, the presence of veins of pure analcite with width of 1-2 mm are ubiquitous in the clinopyroxenite ( Figure 3E). These veins are commonly composed of (1) analcite; (2) K-feldspar with exsolved lamellae of albite, which are approximately right to the vein); and (3) biotite, occasionally hosting anhedral clinopyroxene inclusions. The clinopyroxene adjacent to these veins is euhedral and usually rimmed by a darker greenish corona of clinopyroxene.

K-Feldspar Clinopyroxenite
K-feldspar clinopyroxenite is transitional in term of mineral assemblage and crystal mode between syenite and clinopyroxenite. These rocks are dark reddish in color and relatively coarse-grained in hand specimen scale, and mainly consists of euhedral K-feldspar and clinopyroxene, with anhedral biotite. The modal proportion of the three main minerals vary from place to place. The euhedral tabular K-feldspar usually vary from 1 to 7-8 cm in length and are preferentially orientated. The occurrence of clinopyroxene and biotite are constrained in the interstices between K-feldspar crystals. They usually exhibit poikilitic texture similar to the clinopyroxenites, whereas minor titanite (1-2 mm) occurs as individual grain ( Figure 3F). Minor apatite (<1 mm) also exists as interstitial phase. In addition, some veins with 1-2 mm width are filled by analcite are commonly present in the K-feldspar clinopyroxenite as well.

Pseudoleucite Clinopyroxenite
Pseudoleucite clinopyroxenite is characterized by leucocratic pseudoleucite or cluster of pseudoleucite embedded in the melanocratic matrix ( Figure 4A), which are predominantly composed of biotite, clinopyroxene, and subordinate garnet, apatite, magnetite and K-feldspar. Most of the pseudoleucite crystals are generally euhedral to subhedral, with the typical shape of leucite with ∼1 cm in diameter ( Figure 4B). Similar to some of the pseudoleucite reported (Gittins et al., 1980;Comin-Chiaramonti et al., 2009), the pseudoleucite in Yaojiazhuang consists of dominantly fine-grained K-feldspar (∼50 mol.%, 5-40 µm), nepheline (∼30 mol.%, 5-20 µm) and muscovite (∼15 mol.%, 5-20 µm) intergrowth at the core and nearly pure fine-grained K-feldspar at the rim (Figure 4). The fringes of the intergrowth are vermicular or flame-like or feather-like and usually perpendicular to the boundaries of the pseudoleucite grains. Within some pseudoleucite, coarsening of euhedral prismatic to granular nephelines up to ∼200 µm is observed coexisting with K-feldspar, whereas many of these nephelines have been altered to kaolin and corundum ( Figure 4F). Some anhedral (sub-round to round) and zoned K-feldspars (hundreds of µm) occasionally exhibit lineation and present either across or at the edge of the pseudoleucites (Figures 4E,F). Adjacent to the K-feldspar lineation, shapeless intergrowth of nepheline with muscovite or plus analcite pools and patches usually occur. Within some pseudoleucite crystals, nephelines are totally replaced by analcites, even in the vermicular intergrowths, and in such places, the amount of muscovite also increased accordingly. Individual mineral inclusions are prevalent in the pseudoleucite, predominantly laminar biotite (5 mol.%; tens of µm) and accessory celestine (∼10 µm), titanite (up to ∼100 µm), apatite (tens of 10 µm), calcite (up to 100 µm; Figure 5A) and magnetite (∼10 µm).
Minor zoned clinopyroxene inclusions (several to > 100 of µm) only occur as clusters and are rather rare ( Figure 5B).

Syenite
The syenite is pale red in color and consists of mainly oriented euhedral to subhedral K-feldspar (>70 mol.%, 0.1-3 mm), clinopyroxene (10-15 mol.%, 0.1-2 mm). Pegmatitic syenite is observed at the center of syenite (Figure 2E), where the length of K-feldspar can be up to 10 mm. The K-feldspar crystals commonly host exsolved albite lamellae, and in those samples collected at the center of the syenite, zonation of K-feldspar is observed ( Figure 6A). Anhedral clinopyroxene grains commonly show compositional zonation, and in some samples near the clinopyroxenite, they are in round shape with a diameter about 100 µm scattering among the K-feldspars ( Figure 6B). In the syenite near the clinopyroxenite, small (<50 µm) acicular-shaped clinopyroxene crystals are observed in the hosted K-feldspar crystals ( Figure 6C). Accessory phases include anhedral biotite (0.1-0.3 mm), analcite (tens of µm), apatite (∼100 µm), titanite (tens of µm), magnetite (several to tens of µm) and calcite (∼10 µm). The calcite usually co-exists with analcite and exsolved albite ( Figure 6D). Anhedral amphibole surrounding clinopyroxene is occasionally observed in the samples from the center of the syenite ( Figure 6E). Besides, analcite veins are usually observed in the syenite ( Figure 6F). K-feldspar adjacent to the vein is euhedral and usually grows across the veins. Analcite lamellae is usually observed in these K-feldspar crystals.

Dykes
Mafic and syenitic dykes pervasively cut all the three major lithologic units in Yaojiazhuang complex. In the field, mafic dykes (commonly ∼10 cm wide) are greenish to dark gray in color. The mafic dykes are fine-grained, lithologically similar to the clinopyroxenite, consisting of mainly oriented euhedral clinopyroxene (<1 mm), biotite (<1 mm) and K-feldspar (<1 mm) of almost equal quantities (∼30 mol.%). Analcite is commonly present in the interstices to form networks (∼1 mm in width) separating the samples into different lithological parts ( Figure 7A). In these analcite-rich zones, more clinopyroxene and biotite is observed. Minor phases include apatite, magnetite, and titanite are present.
According to the spatial relationship, both early and late stage syenitic dykes (commonly around 10 cm wide) had been recognized ( Figure 7B). Nevertheless, the two stage dykes are similar in mineral assemblage and defined as clinopyroxene syenite. K-feldspar is the main phase in both of these two kinds of dykes (>80 mol.%) and hosts exsolved albite thin lamellae. Subordinary minerals are composed of biotite and clinopyroxene. Anhedral biotite mainly occurs as interstitial phases between K-feldspar grains. Small needle-like clinopyroxene is widespread in the host K-feldspar and biotite, which is interpreted as a result of quenching. Accessory minerals includes titanite, analcite and apatite. In some samples, euhedral titanite can grow up to 1-2 mm ( Figure 7C). The late stage dyke is composed of K-feldspar, clinopyroxene, biotite with considerable amount of albite. K-feldspar accounts for 70% modal and its size is generally around 1-2 mm in length except some crystals could be up to 2-3 cm ( Figure 7D). Clinopyroxene (20 mol.%) can be divided into at least two generations, one is as euhedral to subhedral zoned phenocryst up to 0.6 mm (Figure 7E), the other is as needle-shaped inclusions in the matrix commonly less than 50 µm in length. Clinopyroxene pseudomorph filled by albite, muscovite and magnetite is also observed ( Figure 7F). Subhedral to anhedral biotite (<2 mm) constitutes the rest 10% modal percent. Anhedral albite is observed co-existing with K-feldspar either as exsolved lamellae or in the interstices of clinopyroxene and biotite crystals ( Figure 7E). Euhedral titanite and apatite are also present as minor phases.

Scanning Electron Microprobe
Elements mapping (Figure 8) shows that the pseudoleucites are mainly enriched in Si, Al, Na, and K but poorer in Fe and Ca relative to the matrix, which should be caused by the appearance of different minerals. The concentrations of elements within the pseudoleucites from edges to cores are relatively homogeneous, except that the compositionally zonation of K-feldspars slightly lack in Si and K. The coarse nepheline pseudomorphs are more enriched in Ca, Na but deficient in Fe compared to other parts of the pseudoleucite.

Whole-Rock Geochemistry
The whole-rock major and trace element analyses of the studied syenite, clinopyroxenite, pseudoleucite clinopyroxenite, as well as syenitic and clinopyroxenitic dykes from Yaojiazhuang alkaline complex are given in Supplementary Table 1.
The trace element and rare earth element (REE) compositions of the Yaojiazhuang intrusion are given in Supplementary  Table 1 and illustrated in chondrite-normalized and primitive mantle-normalized diagrams in Figure 10. In general, (biotite) clinopyroxenites have higher REE contents than syenites, and the matrix of pseudoleucite clinopyroxenite has the highest. K-feldspar clinopyroxenites contain moderate REE contents between (biotite) clinopyroxenites and syenites. For most REEs and trace elements, the contents in pseudoleucite are very low, which may be inherited from the extremely low partition coefficients in leucite. All the rock types are characterized by significant enrichment in highly incompatible elements and light REE. Notably on the spider diagram, (biotite) clinopyroxenites and K-feldspar clinopyroxenites show a variation of P from positive anomaly to negative anomaly, symmetrical about that of the clinopyroxenite dyke, implying the fractionation/cumulation of apatite to form apatite ore hosted in clinopyroxenites. The (biotite) clinopyroxenite and the matrix of pseudoleucite clinopyroxenite exhibit a distinct enrichment in high fieldstrength element compared to syenite and pseudoleucite. The pseudoleucite is significantly enriched in large ion lithophile elements (LILE), such as K, Rb, Ba, and Sr, commonly higher than those of syenites. In the REE patterns, all of rock types have similar slopes, whilst the concentrations of REE in clinopyroxenites and the pseudoleucite clinopyroxenite matrix are much higher than other samples. Despite two syenitic dykes, other samples do not show distinct Eu anomaly, with (Eu/Eu * ) N values generally vary around 1.
Clinopyroxenes in all lithologies of the Yaojiazhuang intrusion generally show zonings with a Mg-rich core and an Fe-rich rim ( Figure 11A). Zoned clinopyroxene has a composition of Wo 44 . 1−47 . 4 En 34 . 7−42 . 8 Fs 9.3−15 . 7 Ac 1.0−4 . 2 in the core and to Wo 35 . 3−44 . 8 En 20 . 5−29 . 5 Fs 20 . 5−28 . 8 Ac 5.2−15 . 3 in the rim (Figure 11A). Rims are usually become rich in Na, indicated by the high Ac endmember. Interestingly, the clinopyroxenes in both of pseudoleucite and the matrix of pseudoleucite clinopyroxenite are similar in composition as well. A compositional profile of a clinopyroxene crystal with oscillatory zonings in clinopyroxenite is shown in Figure 12A. Aegirine is widely recognized in the late-stage clinopyroxene syenite dykes with a composition of Wo 11 . 1 En 6.3 Fs 40 . 5 Ac 40 . 2 . A profile of a zoned clinopyroxene in this syenite dyke is shown in Figure 12C with an augite core and aegirine rim.
K-feldspars in the syenites in the center of the complex and the late-stage syenite dykes exhibit a relatively large compositional rang of Or 52 . 6−74 . 8 Ab 24 . 9−46 . 5 An 0.3−1 . 0 . In contrast, K-feldspars in other parts of the Yaojiazhuang complex  Published data of Yaojiazhuang complex are from Mu and Yan (1992) and Chen et al. (2013). The primitive mantle and C1 chondrite normalizing values are from Sun and McDonough (1989).
with composition Or 0.4−1 . 6 Ab 91 . 6−99 . 3 An 0.3−7 . 2 . The most Anrich feldspar (An = ∼7) is found in the syenites in the center of the complex, where that zonation of K-feldspar with an Or-rich core and an Ab-rich rim is commonly observed (Figure 12B).
The composition of biotite in the Yaojiazhuang complex show restricted range, with FeO tot from 18.2 to 23.6 wt.%, MgO from 10.4 to 14.2 wt.%, TiO 2 from 1.8 to 2.5 wt.%, and K 2 O from 9.4 to 9.8 wt.%. Accordingly, the biotite can be classified into the siderophylite group on the Fe/(Fe + Mg)-Al diagram ( Figure 11C).
Apatites are nearly pure fluorapatite characterized by X F [= molecular F/(F + Cl)] very close to 1.00 with high F (2.1-3.4 wt.%) relative to low Cl contents (<0.02 wt.%), which implies that they may crystallized from a high temperature, i.e., magmatic environment (Tacker and Stormer, 1989). Other oxides contents in all three lithology units exhibit no obvious differences, with

Oxides
Oxides in Yaojiazhuang complex mainly include magnetite and ilmenite, and the analyses of them are reported in Supplementary Table 3 (Figure 11E; Buddington and Lindsley, 1964), magnetite from all lithologies are defined as titanomagnetite.

Sulfates
Sulfate, i.e., celestine is observed only in the pseudoleucite, and its composition is listed in Supplementary

Assessment on Alteration
Most syenitic rocks have LOI values less than 1 wt.%, except one syenite (YJZ08-1) and a pegmatitic syenite dyke (YJZ04-1) which are 2.5 and 1.5 wt.%, respectively. The high LOI values for the two samples are mainly attributed to the bound water H 2 O + and their moisture water contents show no significant differences to other samples (H 2 O − = 0.1-0.4), thus we suggest that the high LOI values are due to a relatively high content of hydrous minerals like analcite and biotite. Similarly, the biotite clinopyroxenite YJZ07-5, clinopyroxenite dyke (YJZ07-2) and K-feldspar clinopyroxenite (YJZ04-4) also show high contents of LOI values up to more than 3 wt.%, whereas their high LOI values are mainly constituted by high bound water up to 2.7 wt.%, consistent with the high contents of biotite and analcite on petrography. The pseudoleucite also has high LOI, high H 2 O + (3 wt.%) but relatively low H 2 O − (0.5 wt.%), owing to the widespread existing of muscovite, analcite and biotite. Hydrothermal alteration or weathering usually introduce water mobile elements, such as LILE, into rocks, thus the high contents of these elements may indicate the alteration process. However, either on the Sr vs. LOI (Figure 13A) or moisture water H 2 O − diagram (Figure 13B), no significant correlation is observed for both of clinopyroxenitic or syenitic rocks, implying their primary geochemical signatures are preserved.
There are also classic indices to evaluate the degree of chemical weathering of silicate rocks. For instance, Plagioclase Index of Alteration (PIA) established by Fedo et al. (1995) is widely employed in evaluating the alteration of plagioclase-and/or  Fedo et al., 1995) and (D) Bases to Alumina ratio (BasesAl; Coleman, 1982) diagrams to evaluate the degree of chemical alteration for Yaojiazhuang samples.
K-feldspar-rich rocks, and Bases to Alumina ratio (BasesAl; Coleman, 1982) is a power tool for mafic rocks. The higher the PIA or the lower the BasesAl, the higher the degree of alteration. There is still no obvious correlation is present for synetic rocks on the Sr vs. PIA diagram ( Figure 13C) or for clinopyroxenites and K-feldspar clinopyroxenites on the Sr vs. BasesAl diagram ( Figure 13D). The K-felspar clinopyroxenite with low BasesAl values show higher Sr contents than clinopyroxenites, which shoud be resulted from the higher proportions on Sr-bearing minerals like K-feldspar instead of alteration. Thus, their high Sr contents are able to represent their primary high LILE contents.

Crustal Contamination
Silica saturation index (SSI) was introduced by Motoki et al. (2010) to verify the role of crustal contamination by syenitic magma. The presence of pseudoleucite, nepheline and analcite indicate that the parental magma of Yaojiazhuang complex ought to be peralkaline and undersaturated in silica. The country rocks of the Yaojiazhuang complex is the Archean TTG, which are basically SiO 2 -oversaturated and meta-to peraluminous in composition (e.g., Xie et al., 2019;Chen et al., 2020). Since the Yaojiazhuang complex were emplaced into these TTG, it seems unavoidable for the parental magmas were contaminated. Hence, the elevation of SSI is expected due to assimilation of TTG. Meanwhile, such interaction between parental magmas and wall rocks could also decrease the alkali content in the magmas as low-alkali affinity of the TTG (Motoki et al., 2015). Accordingly, on the SSI vs. molar (Na + K)/Al diagram (Figure 14A), there are four out of six syenites plotted on the left-top quadrant, implying most of syenites had been largely affected by crustal contamination. This inference is further supported by the plot of SSI vs. K 2 O/(K 2 O + Na 2 O) diagram ( Figure 14B).

Composition of Parental Magma
As stated above, the clinopyroxenite, K-feldspar clinopyroxenite and syenite exhibit coherent evolution trends both mineralogically and geochemically, indicating these lithologies have a common parental magma. Such inference is also supported by the similar Sr-Nd isotopic compositions of rocks from different lithologies (Chen et al., 2013). Whereas the intrusive contacts between these lithofacies suggest that they are not derived from a single batch of magma. Therefore, it is more likely that there should be a deeper magma chamber where fractional crystallization of parental magma takes place.
Conventionally, the parental magma of Yaojiazhuang complex is thought to be ultrapotassic (Hou, 1990b;Mu and Yan, 1992;Chen et al., 2013), which is defined as K 2 O > 3 wt.%, MgO > 3 wt.%, and molar K 2 O/Na 2 O > 2 in whole-rock geochemistry according to Foley et al. (1987). While, the MgO contents for all of our syenite samples are below 3 wt.%, and those MgO enriched clinopyroxenite samples contain K 2 O less than 3 wt.%. Despite that the molar K 2 O/Na 2 O in some samples can reach more than 5, there are still some syenite and clinopyroxenite samples having smaller K 2 O/Na 2 O ratios.
However, the definition in Foley et al. (1987) is based on the whole-rock compositions of sequences of volcanic rocks which are deemed to be liquids. However, rocks from Yaojiazhuang complex are cumulates, the whole-rock composition of which are controlled by the modal proportions of cumulate minerals. Thus, the low MgO K-feldspar-dominated syenites are expected to have low MgO content, and the low K 2 O but MgO enriched clinopyroxene dominated clinopyroxenites are expected to be lower in K 2 O but higher in MgO. Nevertheless, a feeder dyke (sample YJZ07-2) which is fine-grained and both ultrapotassic in composition (MgO = 6.8 wt.%, K 2 O = 7.4 wt.%, molar K 2 O/Na 2 O = 3.5) could represent the composition of parental magma. Such inference is consistent with abundant presence of K-rich phases including biotite and K-feldspar in the complex. This dyke represents dominant phase among the basic dykes in the studied area and exhibits similar mineralogy and coherent variation trend geochemically with other dominant phases, i.e., clinopyroxenite and syenite. The moderate geochemical characteristics of the clinopyroxenite dyke between clinopyroxenites and K-feldspar clinopyroxenites indicate that clinopyroxenite and K-feldspar clinopyroxenite are derived from the cumulation of mafic minerals, and the syenite is a more evolved phase differentiated from them.

Fractional Crystallization
The bimodal clinopyroxenitic and syenitic series had been speculated to be formed via silicate liquid immiscibility (Hou, 1990b). The magma immiscibility mechanism has also been proposed to explain the petrogenesis of similar bimodal complex like in Hamilton Montana (Lelek, 1979). However, evidence from mineralogy is not consistent with the immiscibility model. For example, equilibrated immiscible Fe-and Si-liquids ought to crystallize same mineral assemblage but with different proportions (e.g., Veksler et al., 2007). According to this criterion, this is not the case for the clinopyroxenite and syenite in Yaojiazhuang as albite is absent in clinopyroxenite and garnet is exclusively seen in the matrix of pseudoleucite clinopyroxenite. Major phase in fractionation crystallization or cumulation can be confirmed by geochemical data plotting. For instance, the Zr/Sm vs. Sm/Yb (Figure 14C), Dy/Yb vs. SiO 2 diagram ( Figure 14D) and Sr vs. Ba diagram (Figure 14E) all suggest that a clinopyroxene dominant fractional crystallization/cumulation. Besides, the coupled decrease in Ba and Sr also suggest that the fractionation of K-feldspar plays a dominant role in the differentiation of syenitic rocks, which had also been illustrated by the positively correlation between K 2 O/(K 2 O + Na 2 O) and SSI ( Figure 14B). Combining the rock and mineral textures as well as geochemical characteristics, the crystallization sequence is inferred as in Figure 15, clinopyroxene/apatite/oxides → biotite/titanite → leucite/K-feldspar/analcite/aegirine → albite. K-feldspars crystallized as primary phases from the melt including the euhedral crystals in syenite and those interstitial phases in between clinopyroxenes in clinopyroxenite. In addition, some K-feldspar could also be formed by break down of leucite in pseudoleucites. Analcite can be precipitate either as primary mineral in the intergrowths with K-feldspar or secondary phases replacing leucite and/or nepheline.
(2) reaction of primary leucite with a Na-rich liquid.
(3) introduction of Na + from hydrous glass or aqueous vapor by ion exchange. Here, we will briefly discuss these theories as the possible mechanism to produce the pseudoleucites in Yaojiazhuang complex.
Conventionally, problem arises to the subsolidus breakdown of leucite because although experiments shows that leucite could accommodate about 40 wt.% NaAlSi 2 O 6 in the solid solution (Taylor and MacKenzie, 1975), natural leucite has never been observed to contain excess amount of Na to form the intergrowth with sodium-bearing phases on decomposition (Viladkar, 2010). The abundant sodium-rich phase like nepheline and analcite in the mineralogy of the pseudoleucite suggest they are generated from a pre-existing Na-rich phase. Therefore, the pseudoleucite in Yaojiazhuang samples cannot be explained by subsolidus break down of potassic leucite alone, but processes introducing sodium into leucite to form a metastable sodic leucite prior to breakdown seem to be necessary (Taylor and MacKenzie, 1975).
Reaction between leucite and Na-rich melt/liquid was proposed as mechanism to produce a nepheline-K-feldspar pseudomorph and former leucite disappears (Bowen and Ellestad, 1937;Edgar, 1984). An outstanding feature of this theory is that, the leucite crystal morphology can hardly be well preserved during the reaction with melt/liquid (Taylor and MacKenzie, 1975). We observed in our samples, the pseudoleucite generally exhibit the pentagonal icositetrahedron morphology of leucite in macro scale, whereas many of their rims are curved and modified in micro scale, suggesting that such a reaction process cannot be totally ignored (Figure 4). Indeed, the pseudoleucite is considered as a late-stage phase after mafic minerals due to these vermicular nephelinefeldspar or analcite-feldspar intergrowths are also present as anhedral interstices in the matrix or clinopyroxenites. The Na content in zoned clinopyroxenes are also observed progressively increase from core to rim, supporting to an increasingly sodic crystal mush in the late stage. Moreover, various minor mineral phases (e.g., biotite, clinopyroxene, apatite, titanite) occur within the pseudoleucite and show no obvious compositional differences to those in the matrix, suggesting they are appeared to be primary magmatic inclusions. These phases introduce additional elements into the pseudoleucites, which advises investigations only based in traditional petrogeny's residua system (SiO 2 -NaAlSiO 4 -KAlSiO 4 , quartz-nepheline-kalsilite) is not enough (Edgar, 1984).
Alkali ion exchange reaction is a reaction of pre-existing potassic leucite and sodic glass and/or aqueous vapor by cation substitution under subsolidus conditions, resulting soda leucite and later exsolution without modification of previous crystal morphology (Taylor and MacKenzie, 1975). However, this mechanism seems not to be the case for Yaojiazhuang samples, because our pseudoleucites generally have K-feldspar rims but nepheline-feldspar cores (Figure 4E), which means the bulk composition in the cores are much richer in Na than the rims. Nevertheless, the presence of a magmatic sodium rich fluid could be possible. Amphibole hygrometer suggests water content of melt during late-stage evolution exceeds 3 wt.%, therefore leucites are possible to start nucleation at the late stage of crystallization when the cooling rate is expected to be speed up at hydrous undercooling condition, and form the crystal skeletons (rims) resemble those rapid-grow hollow textures (Faure et al., 2003). While, following the formation of anhydrous rims, further crystallization of cores is tend to be impeded due to the increase of water activity and the residue melt compositionally equals to nepheline + K-feldspar. With a rapid degassing, the final liquid crystallizes a vermicular and fingerprinted intergrowth of cotectic nepheline and K-feldspar (Gittins et al., 1980). The leucite skeletons may latterly be transformed to K-feldspar and nepheline or react to form analcite in the presence of exsolved fluids (Henderson and Gibb, 1977;Wilkinson and Hensel, 1994). The exsolved fluid by degassing may also result in the coarsening of some nephelines ( Figure 4F) by self-alteration (Ballhaus et al., 2015), and introduce other hydrophilic elements such as Ca into nephelines to form cancrinite before subsequent weathered to kaolin. Complementary evidences for the activity of magmatic sodium-rich fluids during the late-stage magmatic evolution are the transformation of initial leucite to analcite in the matrix (Figure 5F), the widespread albite lamellae in K-feldspars and the pegmatitic syenites at the top of the intrusion ( Figure 2E). In summary, the presence of pseudoleucite and hydrous primary minerals provided solid evidences for the activities of fluids. Fluids saturation and exsolution may have been achieved during decompression of replenished magmas from deep and during crystallization of anhydrous minerals like clinopyroxene.

Implications for the Formation of the Apatite and IOA Ores
The widespread interstitial biotite in clinopyroxenite and presence of glimmerite imply that the late-stage melt could be significantly hydrous. Dissolved aqueous fluid could be oversaturated at some point and exsolved from the crystallizing magma during decompression (e.g., Moore et al., 1998;Dixon et al., 1995;Newman and Lowenstern, 2002;Ballhaus et al., 2015) or during isobaric crystallization of anhydrous minerals (Candela, 1997), such as clinopyroxene and leucite in this study. Moreover, the presence of magmatic calcite also indicates that the parental magma could also be CO 2rich (Mu et al., 1999), which suppresses the H 2 O solubility in melt, resulting in exsolution of H 2 O at a relatively lower concentration (Caricchi et al., 2018;Edmonds and Woods, 2018). Hence, in the case of Yaojiazhuang, the scenario involving exsolution of aqueous fluids during the magmatic stage is expected. Besides, these aqueous fluids should also be abundant in F, due to F-rich biotite is widespread throughout the complex.
Experimental studies indicate that phosphate saturation in silicate melts is mainly controlled by SiO 2 and CaO, and addition of H 2 O and/or F in silicate magmas leads to saturation of apatite (Tollari et al., 2006(Tollari et al., , 2008. The Yaojiazhuang complex is characterized by relatively high CaO and low SiO 2 content. Hence, it is thus expected that the magmatic liquids during the differentiation of the complex may reach high concentrations in both Fe and P. Once the concentration of H 2 O and/or F was elevated by fractionation or other processes, saturation of apatite will be reached. Magnetite will also crystallize induced by decrease in P concentration in the melt (Tollari et al., 2006), rather than elevation of oxygen fugacity during the late stage of differentiation. In other words, apatite and magnetite preferentially co-crystallized from the magmas. Besides, the exsolution of aqueous fluids is considered as a potential mechanism in the formation of porphyritic or spheroidal structure rocks hosting economic ore deposits (Huber et al., 2012;Ballhaus et al., 2015;Guo and Audétat, 2017). As described above, the IOA ores occurred as rocks with spheroidal structure in the field. Therefore, we propose that the fluids exsolution could also be a viable mechanism to concentrate not only magnetite but also apatite by wetting (Ballhaus et al., 2015) and formed the apatite and IOA ore in Yaojiazhuang complex. Notably, Knipping et al. (2015a;2015b) have already advocated that flotation, coalescence and transport of magnetite-bubble pairs within a magma chamber is the key process for the formation of IOA ores. However, their model has only focused on enrichment of iron oxide instead of iron oxide plus apatite, i.e., IOA ores. Nevertheless, we admit that indeed, the role of exsolved fluids could be more important than we thought. Hence, if this is the case, the occurrence of apatite and IOA ores in Yaojiazhuang provides direct petrologic evidence for such a model involving fluids exsolution and concentration of iron oxide and apatite.

CONCLUSION
(1) The bimodal lithologies are more likely to be formed via crystallization fractionation of an ultrapotassic parental magma instead of silicate liquid immiscibility. Fractionations of both clinopyroxene and K-feldspar, and repeated magma replenishment in the crustal chamber played an important role during the differentiation of primitive ultrapotassic magmas. Crustal contamination may also had been involved during the solidification of the complex. Except that, during the emplacement of Yaojiazhuang magma, decompression, fractionation of anhydrous clinopyroxene and the enhancement by CO 2 may result in the exsolution of an aqueous fluid.
(2) The genesis of pseudoleucite is related to original leucite crystallization from an evolved, silica-undersaturated magma followed by the interaction with sodium-rich fluids during the late-magmatic stage. Transportation of aqueous fluid within crystal mush is believed to be a key factor in generating the leucite pseudomorph with ambiguous crystal boundary. (3) The differentiation under fluid-rich condition in the crust could be responsible for the petrogenesis of Yaojiazhuang ultrapotassic complex, thus the concentration of apatite and IOA ores is presumed to be closely related to the fluid exsolution process as well.

DATA AVAILABILITY STATEMENT
The original contributions presented in this study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
TH and RP designed the research, conducted the field trip, and all the analysis works together. Following data treatment were undertaken by RP and TH. All authors actively participated in the subsequent discussions and interpretation of the data as well as in the preparation of the manuscript.