Early Jurassic Rare Metal Granitic Pluton of the Central Asian Orogenic Belt in North-Central Mongolia: Tungsten Mineralization, Geochronology, Petrogenesis and Tectonic Implications

The Tukhum granitic pluton is a part of the Mesozoic composite Khentei batholith of north-central Mongolia, which belongs to the Central Asian Orogenic Belt. The shallow-seated pluton (∼900 km2) is made up of two distinct biotite granite intrusions dated at ∼191 and 183 Ma and hosts a tungsten deposit associated with the younger phase. Both intrusions are composed of ferroan A2-type granites, which are fractionated and silica-rich (>71 wt.%). Their mantle-normalized plots are relatively enriched in Cs, Rb, U, and Th and depleted in Ba, Sr, Eu, Nb, and Ti. They have εNd(t) ranging from ∼0 to +1 and Nd model ages ∼650–900 Ma. The granites were derived by partial melting of a Neoproterozoic middle/lower crustal source of felsic/intermediate composition, followed by fractional crystallization. The younger intrusion also contains leucogranites with a trace element composition indicative of a combined crystal and fluid fractionation. The source of this younger intrusion was enriched in rare metals (W, Sn). The tungsten deposit is associated with the last stages of the evolution of the granitic magma. The origin of the pluton as well as the Khentei batholith is related to a mantle plume, which provided the heat triggering a crustal melting. The plume resulted in the eastward movement of large-scale magmatism over time, from the Tarim traps (300–275 Ma) through the large Khangai magmatic center (270–240 Ma) to the Khentei batholith (230–180 Ma) in north-central Mongolia.


INTRODUCTION
Many tin-tungsten mineralization and deposits are spatially, temporally, and genetically associated with highly differentiated granitic intrusions worldwide (e.g., Sillitoe et al., 1975;Dostal and Chatterjee, 1995;Förster et al., 1999;Baker et al., 2005;Černý et al., 2005;Romer and Kroner, 2016;Korges et al., 2017). One of the regions with such an economically important association is northcentral Mongolia where mineralized granitic bodies (e.g., Yarmolyuk and Kuzmin, 2012;Syritso et al., 2018) are a part of the huge Central Asian Orogenic Belt (CAOB), which is characterized by voluminous juvenile crust unlike most Phanerozoic orogenic belts including Caledonides and Hercynides (e.g., Jahn et al., 2000). This study focuses on the granites associated with a W deposit in north-central Mongolia, where the genetic relationship of the mineralization with the granites is not clear. An investigation of the petrology, geochronology, and geochemical characteristics of the granitic pluton can constrain the petrogenesis of the granitic rocks as well as the origin of the mineralization. It can also be useful for mineral exploration, particularly if it can help to distinguish between fertile and barren granites. This is an important issue as granitic intrusions of the CAOB provide large exploration targets for W-Sn mineralization. A better understanding of the geochemical characteristics and petrogenesis of the mineralized intrusions is a significant step toward developing a strategy for mineral exploration in this region. In this paper, we present whole-rock major and trace element and Nd isotopic data as well as the U-Pb zircon ages for the granitic rocks from the Tukhum pluton (TP; northcentral Mongolia) and discuss their origin and tectonic settings. In addition, the paper contributes to the current discussion on the role of a mantle plume under this part of the CAOB (e.g., Donskaya et al., 2013;Yarmolyuk et al., 2013). The composite intrusion, situated about 100 km NW of Ulaanbaatar, hosts a tungsten deposit.

GEOLOGICAL SETTING
The CAOB, a large accretionary orogen bounded by the Siberian craton to the west and north and the North China and Tarim cratons to the south, stretches from the Ural Mountains to the Pacific Ocean (Figure 1). It consists of numerous orogenic belts (e.g., Sengör et al., 1993;Jahn et al., 2000), some of which are characterized by widespread granitoid magmatism ranging in age from the Late Precambrian to the Mesozoic. In northern Mongolia and southern Siberia, along the southern border of the Siberian craton, the CAOB contains voluminous Late Paleozoic to Mesozoic granitoid rocks including large batholiths (such as Khangai and Khentei) surrounded by Permian-Early Cretaceous rift zones (Figure 1). In north-central Mongolia, the CAOB features the Daurian-Khentei megadome or uplift, a northeasttrending bulge about 600 km long and 200 to 220 km wide that has been uplifting from the early Mesozoic to the Recent. The megadome contains mainly Paleozoic turbidites intruded by the Khentei batholith, the largest Late Triassic-Early Jurassic intrusion in north-central Mongolia. The batholith is composed of numerous Mesozoic plutons, which were emplaced between 230 and 180 Ma (Yarmolyuk et al., 2013;this paper). The individual plutons vary in composition from granodiorite to leucogranite with minor amounts of gabbro and diorite as well as small bodies of Li-F-rich granite occurring along the margins of the batholith. Some of the Li-F-rich granites host the W-Sn mineralization. The intrusions of the megadome show a concentrically zoned arrangement (Figure 2) where the Khentei batholith is in the center of the bulge and is surrounded by three rift zones: Western Transbaikalian, Kharkhorin (Karakorum), and North Gobi. The rift zones consist of horsts, grabens, and depressions containing Upper Permian to Upper Triassic clastic sedimentary rocks and bimodal volcanic suites (Yarmolyuk et al., 2002) and Mesozoic granitic plutons (e.g., Dostal et al., 2014Dostal et al., , 2015bAntipin et al., 2016). The age of the plutons of the rift zones range typically between 221 and 186 Ma (Yarmolyuk et al., 2002;Dostal et al., 2014).
The rift zones and adjoining marginal parts of the Khentei batholith host mineralized granites. One of such plutons in north-central Mongolia is the TP which hosts a tungsten deposit (Tsagaan Davaa). Unlike some weakly mineralized granitic bodies such as Janchivlan and Avdar (Antipin et al., 2016), which occur in the rift zones along the margins of the Khentei batholith, Early Jurassic TP is located within the batholith but along its margins (Figure 2). It is one of the shallow-seated granitic intrusions, which are discordant and are surrounded by contact-metamorphic aureoles. The TP intruded into slightly metamorphosed Cambrian to Ordovician sedimentary rocks (mainly sandstones). The contact metamorphic aureole around the TP is up to 1.5 to 2 km wide. Close to the pluton, the aureole contains quartz-biotite-feldspar hornfels. The pluton is elongated and exposed over an area of ∼900 km 2 . The internal structure of the TP suggests a dome-shaped intrusion. Our research focused on the southeastern part of the pluton (Figure 3), which hosts a tungsten deposit.
An unusual structure of the Daurian-Khentei megadome is not unique in the area. A large magmatic complex with a zonal concentric structure similar to that of the Khentei batholith occurs southwest of the study area (Figure 1). This Permian-Early Triassic body of comparable size consists of a granitoid core surrounded by rift zones (Yarmolyuk et al., 2008(Yarmolyuk et al., , 2013Kuzmin, 2011, 2012). The core of the structure, the Khangai batholith (>120,000 km 2 ), is composed mainly of granites and granodiorites emplaced between 270 and 240 Ma (Yarmolyuk et al., 2013). It lies between the Gobi-Altai and Northern Mongolian rift zones, which are similar to those around the Khentei batholith.

PETROGRAPHY
The pluton is composed mainly of fine-to coarse-grained biotite granites (Figure 4) and is composed of two intrusive phases (Figure 3). The first phase (including dated sample WOL-12) is made up of gray and yellowish gray fine-to mediumgrained porphyritic granite with feldspar phenocrysts typically 1-3 cm in size that are enclosed in a groundmass composed of K-feldspar, plagioclase, quartz, and subordinate biotite (up to ∼10%). Amphibole, Fe-Ti oxides, secondary minerals (including carbonates), and accessory minerals occur in trace amounts. Phenocrysts, which are microcline-perthite and subordinate plagioclase, account for about 25% of the volume of the rocks. This intrusive phase contains numerous xenoliths of country rocks, particularly metasedimentary rocks. The second intrusive phase (encompassing dated samples WOL-14 and WOL-15) includes dominant porphyritic coarse-grained biotite granite and subordinate leucogranite. Porphyritic biotite granite contains feldspar phenocrysts, which are also mainly microcline-perthite FIGURE 1 | Simplified geologic map of parts of the Central Asian Orogenic Belt and Siberian Craton showing the occurrence of two Late Paleozoic-Early Mesozoic batholiths and rift zones (modified after Yarmolyuk and Kuzmin, 2011) and the Tukham pluton (TP). Inset is a sketch map of the Central Asian Orogenic Belt, Siberian Craton, and surrounding geological provinces (modified after Jahn et al., 2009) showing the location of Figure 1. and subordinate plagioclase. The groundmass is made up of plagioclase, microcline, quartz, and minor biotite and accessory minerals (titanite, zircon, Fe-Ti oxides, and pyrite). The rocks also contain minor but variable amounts of secondary minerals (chlorite, sericite, and carbonate). Compared to the rocks of the first phase, these granites have notably larger grain size. The leucogranite is composed of plagioclase (albite, albite-oligoclase), K-feldspar, quartz, secondary, and accessory minerals. These rocks mostly form bands ranging in width from a few meters to ∼100 m, which pass gradationally into the biotite granite. Some leucogranites contain dominantly primary magmatic mineral assemblages while other leucogranites, namely, those that are close to the mineralization, were affected by silicification and other alteration processes such as greisenization (quartz, dark mica, and minor fluorite). We have subdivided the altered leucogranites into moderately altered and strongly altered.
Moderately altered leucogranites typically contain secondary quartz and smaller amounts of Li-bearing mica, fluorite and topaz, while the strongly altered rocks are composed mainly of a secondary mineral assemblage. It is possible that precursors of some of these strongly altered rocks were biotite granites. Porphyritic biotite granite of the second phase and leucogranite are assumed to be approximately coeval (also, e.g., Ivanova, 1976;Khasin, 1977).
leucogranites, which are altered around the mineralization. Hydrothermal alteration and mineralization appear to be controlled by faults. The main mineralized zone is >1 km long and 200-500 m wide. The individual mineralized quartz veins range in thickness from <1 cm to >1 m but on an average they are ∼0.5 m thick and can be traced for tens of meters. The tungsten-bearing veins are mostly sub-horizontal bodies, gently dipping, which are parallel with the roof of the pluton and are typically accompanied by greisen and silica alteration. Greisen is the common hydrothermal alteration type associated with mineralization and is characterized by Li-Fbearing micas, fluorite, wolframite, cassiterite, and dominant quartz. The main ore mineral in the deposit is wolframite accompanied by subordinate cassiterite and molybdenite and minor scheelite, malachite, azurite, and beryl. Some veins consist almost entirely of quartz and wolframite. The deposit contained ∼3100 t of WO 3 in ore grading 1.52% WO 3 and 740 t of WO 3 in ore grading 0.72% WO 3 . The grade ranges from 0.1 to 12.6% WO 3 . The mine went into production in 1978.

Whole-Rock Analyses
The analyses of whole-rock major and trace elements of the TP samples (Table 1) were done using lithium metaboratetetraborate fusion at the Activation Laboratories Ltd. in Ancaster, ON, Canada. Major elements were analyzed by an inductively coupled plasma-optical emission spectrometer, whereas trace elements were determined by an inductively coupled plasma mass spectrometer (ICP-MS; Perkin Elmer Optima 3000). The accuracy for each element was monitored by analyzing FIGURE 3 | Geological map of the southeastern part of the Tukhum pluton and surrounding units (modified after Dejidmaa, 2003;Dorjsuren and Bujinlkham, 2004) showing the location of the Tsagaan Davaa tungsten deposit. international standards, which were run as unknown. Based on replicate analyses, the precision is generally better than 3% for most major elements and between 5 and 10% for trace elements. Major and some trace elements in the Bayan-Ulan samples (Supplementary Table S2) were determined by X-ray fluorescence at the Regional Geochemical Center at Saint Mary's University. Analytical precision as determined on replicate analyses is generally better than 5% for the major oxides and between 5% and 10% for minor and trace elements. Iron was determined as total Fe (Table 1 Supplementary Table S2).
Sm and Nd concentrations and Nd-isotope ratios of the granitic rocks ( Table 2) were determined at the Atlantic Universities Regional Facility at the Department of Earth Sciences of Memorial University of Newfoundland (St. John's, Newfoundland, Canada) using a multi-collector Finnigan MAT 262 thermalionization mass spectrometer (Pollock et al., 2015). Replicate analyses of JNdi-1 yield a mean 143 Nd/ 144 Nd = 0.512100 ± 6. The 2σ values are given in Table 2. ε Nd (t) values were calculated with respect to CHUR using a present-day 143 Nd/ 144 Nd ratio of 0.512638 and a 147 Sm/ 144 Nd ratio of 0.196593, and were subsequently age-corrected. A T DM model age (

Zircon U-Pb Dating
Whole-rock samples, ca 1-2 kg in weight, were collected for mineral separation. Zircons were separated using conventional techniques: crushing, Wilfley concentration table, magnetic, and, finally, heavy liquid separations. Handpicked zircon grains were mounted in 1-inch epoxy-filled blocks, ground, and polished. Internal zircon structures were checked by cathodoluminescence (CL) imaging using scanning electron microscope at Charles University in Prague (Czech Republic). An Element 2 highresolution sector field mass spectrometer (Thermo Scientific, Waltham, MA, United States) coupled with a 193-nm ArF Analyte Excite (Teledyne/Cetac) excimer laser ablation system at the Institute of Geology of the Czech Academy of Sciences in Prague was used to acquire the Pb/U isotopic ratios in zircons. The laser is equipped with a HelEx II active 2-volume ablation cell. The laser was fired at a repetition rate of 5 Hz, using a spot size of 25 µm. Acquisitions for standards and unknown samples consisted of a 15-s measurement of a blank followed by U and Pb signals from zircons for another 35 s. The signal was tuned for maximum sensitivity of Pb and U, Th/U ratios close to unity, and a low oxide level, commonly below 0.2%. The total of 420 mass scans data were acquired in time-resolved-peak jumping-pulse counting/analog mode with 1 point measured per peak for masses 204 Pb + Hg, 206 Pb, 207 Pb, 208 Pb, 232 Th, 235 U, and 238 U. Due to a non-linear transition between the counting and analog acquisition modes of the ICP instrument and the fact that 238 U is usually measured in "both" mode, the raw data were preprocessed using a Python module called ExctractDat for decoding the Thermo Element ICP-MS data files (Hartman et al., 2017) and also an in-house Excel macro. As a result, the T Zr ( • C)-zircon saturation temperature estimate in • C calculated according to Boehnke et al. (2013); M = (Na + K + 2Ca)/(Al × Si).
Frontiers in Earth Science | www.frontiersin.org intensities of 238 U were left unchanged, if measured in a counting mode, and recalculated from 235 U intensities, if the 238 U was acquired in analog mode. No common Pb correction was applied to the data due to the high Hg contamination of the commercially available He carrier gas, which precludes accurate correction of the interfering 204 Hg on the very small signal of 204 Pb (common lead). The Hg impurities in the carrier He gas were reduced by using an in-house made gold-coated sand trap. Elemental fractionation and instrumental mass bias were corrected by the normalization of an internal natural zircon reference material Plešovice for samples WOL-12 and WOL-14 (337 Ma, Sláma et al., 2008) and 91500 (1065 Ma, Wiedenbeck et al., 1995) for sample WOL-15. Zircon reference materials GJ-1 (609 Ma, Jackson et al., 2004;603 Ma, Kylander-Clark et al., 2013) were periodically analyzed during the measurement for quality control, and the values for the natural standard correspond well and are less than 1.5% accurate within the published reference values. Raw data reduction and age calculations, including corrections for baseline, instrumental drift, mass bias, and downhole fractionation, were carried out using the computer program Iolite (v. 3.5; Paton et al., 2011) with the VizualAge utility (Petrus and Kamber, 2012). All presented ages are quoted at 2σ absolute. The discordance limit for the concordant population was taken as 1.5% (Supplementary Table S1) and was calculated as follows: D < 1Ga = [1 − (( 206 Pb/ 238 U)/( 207 Pb/ 235 U))] × 100.

GEOCHRONOLOGY
The age of the TP was not well established. The K-Ar wholerock dating for the granite of the TP yielded a cooling age of 190.5 ± 4.7 Ma, whereas the whole-rock Rb-Sr gave an age range of 225 to 188 Ma (Smirnov et al., 1977). In order to refine the age, we have dated the pluton by a U-Pb zircon laser ablation ICP-MS technique.
The U-Pb isotopic ratios in zircons were measured in three samples (see Supplementary Table S1 and Figure 5 for dating results). Sample WOL-12 is biotite granite from the first intrusive phase while samples WOL-14 and WOL-15 are biotite granites from the second intrusive pulse. Most of the zircon grains from WOL-12, WOL-14, and WOL-15 are slightly pale brown or clear and are predominantly prismatic grains, and their fragments or rarely needles have a length of ∼200 to 500 µm. Internal crystal interiors visible in CL imaging revealed that the most of crystals show well-developed magmatic oscillatory (or sector) zoning with only slight alteration with rare featureless unzoned cores (Figure 6). All studied samples show uniform Th/U 0.1-0.6 (average value of 0.4; see Supplementary Table S1) in both the rims and cores. This Th/U ratio is typical of a magmatic origin (Hoskin and Schaltegger, 2003 and references therein).
U-Pb zircon dating of sample WOL-12 yielded a scatter in concordant ages between c. 185 Ma and 195 Ma that constitute a single concordia age of c. 191 ± 2 Ma (2σ; 25 analyses; Figure 5). The majority of zircon crystals extracted from samples WOL-14 and WOL-15 yielded a well-defined concordia magmatic age of ∼183 ± 2 Ma (2σ; 25 analyses; Figure 5) and 183 ± 2 Ma (2σ; 21 analyses; Figure 5), respectively. Inherited ages do not significantly differ from those obtained in zircon domains with magmatic growth zoning (Figure 6). Thus, the first phase intruded at ∼191 Ma while the second phase was emplaced at ∼183 Ma.
The chemical compositions of the biotite granites were not significantly modified by secondary processes. On the other hand, there are notable differences among the leucogranites. Leucogranites (WOL-6 and WOL-10) probably retained most of their original composition while the moderately altered leucogranites (WOL-4 and WOL-5) were silicified, leading mainly to an addition of silica. The other two samples (WOL-7 and WOL-8) were strongly hydrothermally altered and accompanied by a notable change of the chemical composition. These two samples were not consequently plotted on some diagrams.
In addition to the TP rocks, 12 samples from the Bayan-Ulan granitic intrusion (see section "Relationship between petrogenesis and mineralization") were also analyzed (Supplementary Table S2). These samples do not appear to be noticeably affected by secondary processes (Dostal et al., 2015b).

Major and Trace Elements
The granitic rocks of the TP are highly siliceous and fractionated with silica contents ranging from 71 to 77 wt.% (Figure 7). The silica contents of the first intrusion vary between 72 and 75 wt.% in contrast to the second phase with 71 to 77 wt.%. On the normative quartz-alkali feldspar-plagioclase (QAP) graph (Figure 8A), they plot into the field of granite. Although there are differences between the two intrusive phases, all the rocks have molar Al 2 O 3 /(Na 2 O + K 2 O) ∼ 1-1.2 and Al 2 O 3 /(CaO + Na 2 O + K 2 O) ∼ 1, indicative of their mildly peraluminous character with the exception of the granites of the first phase, which are mildly metaluminous ( Figure 8B). The granites also have a high FeO * /(FeO * + MgO) ratio ( Figure 7A) and correspond to ferroan granites of Frost and Frost (2011). According the modified alkali-lime index [MALI = (Na 2 O + K 2 O-CaO)] of Frost and Frost (2011), the rocks are mostly alkali-calcic ( Figure 7B); the granites also have high contents of K 2 O ( Figure 7C). The variations of major and several trace elements relative to silica (Figure 7), particularly for the second phase, indicate that the granites underwent fractional crystallization. The TP granites, particularly those of the first phase, are comparable to those of numerous other intrusions of the Khentei batholith (Yarmolyuk et al., 2013).
The chondrite-normalized REE patterns of all the studied granites are enriched in light REE (LREE) and have relatively flat unfractionated heavy REE (HREE) and negative Eu anomalies. However, there are subtle differences among the rocks of the two intrusive phases. The patterns of phase 1 (Figure 9) have a relatively high (La/Yb) n ratio (∼14-18) accompanied by (La/Sm) n ∼ 4 to 6 and slightly sloping HREE with low (Gd/Yb) n ratios (∼1.8). They are closely comparable to the average of the granitic rocks of the whole Khentei batholith (Figure 9) reported by Yarmolyuk et al. (2013). The second intrusive phase has patterns with (La/Yb) n ranging from ∼7 to ∼10 and a more pronounced negative Eu anomaly than the first intrusive pulse. The leucogranites of the second pulse have variable REE patterns including a nearly flat one with (La/Yb) n ∼ 1.2 and (Gd/Yb) n ∼ 0.7. The altered leucogranites have patterns comparable to those of leucogranites. The flat HREE patterns imply that garnet was not in the source of the granites.
The primitive mantle-normalized plots of the TP granites (Figure 10) are distinctly enriched in several large ion lithophile elements, namely, Cs, Rb, Th, and U and depleted in Ba, Sr, Eu, Nb, and Ti. The altered leucogranites also have distinct positive anomalies for W and Sn. Relative to the granites of the first phase, the second-phase granites have lower Ba and Sr (Figure 7) but higher Rb, Ta, W, and Sn (Figure 10). An enrichment of Rb in the second-phase granites is reflected by the relatively low K/Rb ratio (mostly 50-150) compared to typical crustal values of ∼230  (Shaw, 1968;Taylor and McLennan, 1985). These rocks also have anomalous Ba/Rb and Rb/Sr ratios.
The temperatures of zircon saturation (T Zr ), estimated from relating the concentration of Zr to the bulk composition of the magma (Watson and Harrison, 1983;Hanchar and Watson, 2003;Boehnke et al., 2013), are variable for the various TP rocks ( Table 1). The average temperature estimates range from a relatively high value (∼805 • C) for the granites of the first phase through the granites of the second phase (∼762 • C) and leucogranites (∼755 • C) to low values for the moderately altered leucogranites (∼665 • C) and aplite (∼710 • C).

Nd Isotopes
The Nd isotopic data are given in Table 2. The initial isotopic ratios and ε Nd (t) values are age-corrected to the ages of the emplacement (phase 1 = 191 Ma; phase 2 = 183 Ma). ε Nd (t) values of the TP granitic rocks are close to chondritic values (0 to +0.8), indicating that they were derived from a reservoir with a longterm history of near-chondritic Sm-Nd values. The values are similar to those of Jahn et al. (2009) for the Early Mesozoic A-type granitoid rocks from this part of the CAOB (0 to +4) and are within the range of the granitoids of the Khentei batholith (+2 to −4; Yarmolyuk et al., 2013). The Nd model ages (T DM ) ranging from 650 to 900 Ma suggests a Neoproterozoic source.

Petrogenesis
According to the chemical composition, the TP granitic rocks correspond to the A-type granites ( Figure 11A). Specifically, they resemble the A2 group (sensu Eby, 1992), which typically represents magmas derived from lower to middle continental crust, or underplated crust ( Figure 11B). These characteristics are consistent with their ferroan and alkali-calcic features (Frost and Frost, 2011). Collectively in the TP granites, TiO 2 , FeO * , Al 2 O 3 , MgO, CaO, P 2 O 5 , Ba, and Sr decrease with increasing SiO 2 (Figure 7), indicating crystallization of plagioclase, ferromagnesian minerals (biotite), and accessories (Fe-Ti oxides and apatite). Negative anomalies of Sr, Ba, Eu, Nb, and Ti on the primitive mantle normalized plots also support such a process (Figure 10). Covariations of Ba versus Sr (Figure 12) indicate that the fractionation of feldspars played a major role during the magma evolution. However, the TP granitic rocks represent two magma pulses derived from distinct sources at different times. While the rocks of the first intrusion closely resemble the average composition of the Khentei batholith, the granites of the second phase have different geochemical characteristics (including T Zr ) and underwent more extensive fractional crystallization.
The origin of A2-type granitic rocks has been typically attributed to either fractional crystallization or crustal melting (Eby, 1992;Bonin, 2007). These processes can also be invoked for the generation of the TP granitic rocks, which intruded during the last stages of the emplacement of the Khentei batholith and of the plutons of the rift zones surrounding the megadome. More specifically, the petrogenetic processes are (1) extensive fractional crystallization of a mantle-derived mafic or intermediate magma with or without crustal contamination and (2) partial melting of middle or lower crustal source rocks.
Compositional similarities of the granites of the first intrusion to the average of the Khentei batholith and to an estimate of the total crust of East China (Figures 9, 10), peraluminous/metaluminous characteristics of the granitic rocks, and the lack of intermediate rock types (Daly gap) negate an FIGURE 9 | Chondrite-normalized rare-earth element diagrams for (A) phase 1 TP granite (WOL-12) and the average of granitic rocks of the Khentei batholith (Yarmolyuk et al., 2013); (B) phase 2 TP granites (WOL-14 typical biotite granite; WOL-4 and 5-altered leucogranites); (C) TP strongly altered granites; and (D) granites of the Bayan-Ulan pluton. Normalizing values are after Sun and McDonough (1989).
origin by fractional crystallization of mantle-derived mafic or intermediate magma but are consistent with a process where rising mafic magma triggers partial melting of the crustal material FIGURE 10 | Primitive-mantle normalized incompatible element abundances for the granitic rocks of the Tukham and Bayan-Ulan plutons. (A) Phase 1 TP granite (WOL-12) and an estimate of total crust of Eastern China (Gao et al., 1998); (B) phase 2 TP granites (WOL-14 typical biotite granite; WOL-4 and WOL-5 altered leucogranites); (C) TP strongly altered granites; and (D) granites of the Bayan-Ulan pluton. Elements are arranged in the order of decreasing incompatibility from left to right. Normalizing values are after Sun and McDonough (1989).  Whalen et al. (1987) showing the A-type characteristics of the granitic rocks of the Tukham pluton. I and S field is for I-and S-type granites. (B) Y-Nb-Zr/4 diagram of Eby (1992) for the granitic rocks of the Tukham pluton showing the fields of the A1 and A2 granitoids. The dashed green curves show the fields (BU) for the Bayan-Ulan pluton.
from which the granitic magma inherit their geochemical characteristics. This is a common process during lithospheric extension and is a consequence of magmatic underplating and crustal melting (e.g., Huppert and Sparks, 1988). Thus, the TP granites are probably the results of partial melting of the middle/lower crust of granodioritic-tonalitic composition (cf. Frost and Frost, 2011). The Nd model ages ( Table 2) suggest that the sources were of Neoproterozoic age. The differences in the T Zr between the two intrusive phases reflect, in part, different melting conditions. The low T Zr (<800 • C) of felsic magma of the second phase is considered to reflect water-fluxed melting as opposed to the relatively high temperature (>800 • C) during anhydrous melting of magma of the first phase (Miller et al., 2003;Collins et al., 2016). The low T Zr for the mineralized leucogranites (∼665 • C) is probably due to the activities of fluids.

Relationship Between Petrogenesis and Mineralization
The petrogenesis of peraluminous fluorine-rich leucogranites such as those of TP is still under dispute, although these rocks can be associated with Sn-W-U-Ta mineralization (Linnen, 1998;Cerný et al., 2005;Linnen and Cuney, 2005). These rocks are enriched in incompatible trace elements such as Rb, Cs, Th, U, W, and Sn and depleted in Ti, Ba, Sr, Nb, and Eu. They also have unusual mineralogical characteristics, such as the presence of topaz and fluorite. These rocks commonly represent the late phase of late-orogenic or anorogenic granite plutons. Two main genetic hypotheses are magmatic and metasomatic, i.e., whether their geochemical characteristics are due to magmatic processes or to post-magmatic fluid-rock interaction. Furthermore, if the characteristics are due to magmatic activities, then the chemical composition could reflect either a distinct source or unusual evolution, involving, e.g., fluorine-rich fluids.
The smooth variation trends for some major and trace elements of the granites and leucogranites of the second intrusion (Figure 7) and their similarities to those from non-mineralized plutons indicate that most of the chemical compositions reflect primary magmatic evolution, including extensive fractional crystallization. On the other hand, the enrichment of some elements such as Rb (accompanied by low K/Rb), a kinked shape REE pattern, and the occurrence of topaz and fluorite among others probably reflect fluid fractionation/interaction involving fluorine. This suggests that trace elements in some rocks were modified by late or post-magmatic fluid-rock interaction (Webster et al., 2004;Salvi and Williams-Jones, 2005;Thomas et al., 2005;Dostal et al., 2015a). The petrographic observations (such as the presence of various pseudomorphs and replacement textures; Ivanova, 1976) also indicate that mineral assemblages of the leucogranites are the results of both magmatic processes as well as late-or post-magmatic fluid-melts interactions. The interaction of the leucogranites with fluids led to the enrichment and redistribution of several incompatible trace elements but has not modified their major and some trace element composition. These highly evolved melts commonly moved/escaped into the apical part of the pluton along the fracture zones and are now represented by the leucogranites.
The experimental studies (e.g., Syritso et al., 2018) indicate that the W preferentially partitions into the fluid phase, which commonly escapes from a magma chamber to the host rocks forming wolframite-bearing quartz veins or greisen that are typically structurally controlled. Their frequent association with fractionated granites indicates that they are closely related, although the W mineralization is post-magmatic and related to the hydrothermal solutions. They were probably derived from the same highly fractionated magma chamber. The W mineralization is also associated with greisen. Greisen represents granites, which were altered by fluids released from the magma during the late stages of its evolution. The geochronological study of Syritso et al. (2018) on several similar tungsten deposits from eastern Transbaikalia concluded that tungsten mineralization took place almost synchronously with the crystallization of associated granites. It also appears that in this part of CAOB including TP, the W mineralization is related mainly to Mesozoic plutons, where it was emplaced along the faults in apical parts of intrusions (Syritso et al., 2018).

Comparison With Granites of the Rift Zones
To evaluate the differences between the granitic rocks of the TP, a representative intrusion of the Khentei batholith, and those of the rift zones, the TP rocks are compared with those of the Bayan-Ulan pluton (221 Ma; Dostal et al., 2015b) from the North Gobi rift zone (Figure 2). For comparison, which might also reveal a contrast between the ore-bearing and barren granites, we have augmented the Bayan-Ulan whole-rock analyses of Dostal et al. (2015b) with the new data (Supplementary Table S2). The Bayan-Ulan pluton (BU), located about 200 km east of Ulaanbaatar, outcrops over an area of ∼1000 km 2 . The intrusion (Dostal et al., 2015b) is composed mainly of coarse-grained leucocratic granite containing mesoperthite with interstitial quartz and minor to trace amounts of Fe-rich alkali amphiboles and, rarely, Fe-rich biotite. The granites have relatively uniform chemical compositions; they are highly fractionated alkaline A2-type granites with silica ranging from 75 to 78.5 wt.% (Figure 7). Compared to the TP, the BU granites are lower in MgO, CaO, and MnO but have a higher FeO * /(FeO * + MgO) ratio (>0.95). They are predominantly peraluminous (Figure 8). Chondrite-normalized REE plots display a slight LREE enrichment with (La/Yb) n ∼ 2-7 and a distinct negative Eu anomaly (Figure 9). The primitive mantlenormalized patterns (Figure 10) are enriched in Th, U, and Rb and relatively more depleted in Ba, Sr, Eu, and Ti. Most BU granites have ε Nd (t) values of +1.4 to +1.7 and T DM model ages ∼650-1050 Ma ( Table 2). Whereas the BU model ages overlap those of TP rocks, their ε Nd (t) are higher. The BU rocks were derived from a different source and underwent a different fractionation path. Dostal et al. (2015b) inferred that they were formed by the partial melting of underplated Neoproterozoic mildly alkaline basalts in the lower crust, followed by fractional crystallization. In addition, the fluids and fluid fractionation played a significantly smaller role during the evolution of the BU intrusion compared to the TP. The main Fe-Mg mineral in the BU rocks is alkali amphibole, while in the TP rocks, it is biotite. The BU rocks are not associated with a W or Sn mineralization. A comparison of the three intrusions (two TP pulses and BU) implies that the granites hosting rare metal mineralization required a distinct crustal source enriched in these elements, a high degree of fractionation, and an involvement of fluorine-rich fluids during their evolution.

Tectonic Implications
After the collision of the Siberian and North China cratons, which led to the closure of the Mongol-Okhotsk basin, north-central Mongolia and an adjoining part of southern Siberia (eastern Transbaikalia) witnessed, during the Late Permian to Early Jurassic, an emplacement of large, concentrically zoned magmatic structures. One of these magmatic structures is the Khentei (Daurian-Khentei) megadome (Figure 2).
The emplacement of the A2-type granitic plutons of the Khentei batholith and of the surrounding rift zones is related to rifting and partial melting. The origin of the granitic rocks requires a heat source to produce an elevated temperature. Such a heat source is usually attributed to one of the following processes: (1) delamination of the lower lithosphere, (2) thinning of lithosphere during rifting, and (3) rising of a mantle plume. The application of these tectonic models to the region has been under discussion (e.g., Yarmolyuk and Kuzmin, 2011;Donskaya et al., 2013;Li et al., 2013). We favor the mantle plume as the heat source for magmatic activities in the area as it explains well the spatial and temporal variations in Eastern Mongolia and Transbaikalia, particularly the eastward movement of large-scale magmatism over time. Yarmolyuk and Kuzmin (2011) and Dostal et al. (2014) among others noted the eastward migration of the magmatic centers through time from the Tarim traps (300-275 Ma; >250,000 km 2 ) through the Khangai magmatic center at 270 to 240 Ma to the Khentei (230-183 Ma) in central Mongolia (Figure 13). The Tarim traps are Permian continental flood basalts, which form a main part of the Tarim Large Igneous Province located in the Tarin Basin. The migration of the magmatic activities is consistent with the movement of Mongolia over a stationary mantle plume (Mongolian plume of Kuzmin et al., 2010;Yarmolyuk and Kuzmin, 2011) and is supported by paleomagnetic data (Kuzmin et al., 2010).
The plume model can also account for the zonal arrangement of the Khentei magmatic center ranging from granites and granodiorites in the core of the batholith through Li-F-rich granites along the margin of the batholith (either within or just outside of the batholith) and finally to alkaline and peralkaline lavas and alkaline granitoid intrusions in the outer margin of the rifts (Yarmolyuk and Kuzmin, 2011). A larger degree of melting that took place at a shallower depth above the core of the magmatic center generated granites, while a deeper and smaller degree of melting took place at the margins and produced alkaline and peralkaline rocks.

CONCLUSION
The TP, one of the constituent bodies of the Mesozoic Khentei batholith, is made up of two distinct intrusions dated at ∼191 and 183 Ma, suggesting that the magmatic activity of the batholith lasted longer than the range of 230 to 195 Ma proposed by Yarmolyuk and Kuzmin (2011). The TP rocks are silica-rich A2type granites, which are mainly alkali-calcic. On the primitive mantle normalized plots, they are enriched in Rb, Cs, U, and Th and depleted in Ba, Sr, Eu, Nb, and Ti. The TP rocks have ε Nd (t) ∼ 0 to +1 and Nd model ages ∼650-900 Ma. The isotopic values are within the range of other Mesozoic granites of the Khentei batholith. The granites were derived by the partial melting of middle/lower crustal Neoproterozoic rocks, followed by fractional crystallization. However, the presence of F-rich minerals in the rocks of the second intrusion as well as W-Sn mineralization suggests that the source of the parent magma of this intrusion was enriched in several rare metals including W and Sn. The younger second intrusion also contains leucogranites with trace element compositions indicative of combined crystal and fluid fractionation during the late stages of the evolution. The bulk of the tungsten mineralization of the TP is hosted in quartz veins and resulted from an escape of fluids from a magma chamber. The mineralization is associated with the evolution of the granitic magma. An increase of the pressure of the fluids in the magma chamber and their escape caused a brecciation of the rocks and triggered an emplacement of quartz veins hosting the W mineralization.
The partial melting required a heat source. The Late Paleozoic to Early Mesozoic granitic province in north-central Mongolia, which includes two batholiths, Khangai and Khentei, and the eastward migration of the magmatic center through time, is consistent with a mantle plume as a heat source for magmatism.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
JD developed the idea, wrote the bulk of the manuscript, processed data, and collected samples. MS wrote a part of the manuscript, processed data, and created figures. OG contributed to the ideas and concept and wrote a part of the manuscript. RC collected and prepared samples and created figures. All authors contributed to the article and approved the submitted version.

FUNDING
This study was supported by NSERC Canada Discovery grant to JD and by the institutional support RVO67985831 to MS.