Age and Composition of Columbite-Tantalite Group Minerals in the Spodumene Pegmatite from the Chakabeishan Deposit, Northern Tibetan Plateau and Their Implications

: The Chakabeishan (CKBS) deposit is the ﬁrst pegmatite-type Li-Be deposit discovered in the eastern North Qaidam Tectonic Belt (NQTB) of Tibetan Plateau. The correct understanding of its petrogenesis and the precise determination of its formation age are of great signiﬁcance for further regional prospecting and the discovery of new economically valuable rare-metal deposits. Therefore, a systematic study of texture, major-element composition, and U-Pb dating of columbite-tantalite group minerals (CGMs) in the spodumene pegmatite dyke from the CKBS deposit was undertaken. Three types of CGMs were identiﬁed, including concentric oscillatory ferrocolumbite (CGMs-1), homogeneous ferrocolumbite (CGMs-2), and irregular ferrotantalite (minor manganocolumbite) with abundant early ferrocolumbite replacement remnants (CGMs-3). The zoning patterns and chemical compositions in the CGMs record the complex evolutionary history of their host pegmatite from the magmatic stage (CGMs-1, disequilibrium crystallization) to the magmatic-hydrothermal transition stage (CGMs-2, equilibrium crystallization) and then to the late metasomatic stage (CGMs-3, replacement/re-equilibrium). CGMs U-Pb dating results suggest that the spodumene pegmatite dyke ( No. 15) emplaced at 230.1 ± 2.6 Ma. Subsequently, it experienced ﬂuid metasomatism at 221 ± 5.3 Ma. Based on the new age data and published geochronological data, it can be concluded that the spodumene pegmatite dykes in the CKBS deposit formed in an oceanic subduction-related setting, representing a new metallogenic event in western China. Except for the CKBS deposit, a large number of rare-metal pegmatite dykes have also been discovered in the eastern NQTB, indicating that the eastern NQTB may be an important potential rare-metal metallogenic belt that should be explored in detail and arouse painstaking attention.


Introduction
Rare metals (e.g., Li, Be, Nb, Ta, Rb and Cs) play a vital role in the modern electronic industry, medicine, new energy industry, national defense and other high-tech fields. No nation has all of the rare-metal resources it needs domestically, so international trade in rare metals is necessary [1]. If global supplies are considered to be at risk or insufficient to meet growing demand, international commerce may result in international competition for rare-metal resources. Against the background of competition for rare-metal resources, many countries have designated highly import-dependent rare metals as critical metals or strategic metals in the 21st century [1][2][3]. Granitic pegmatite (hereafter pegmatite) is one

Geological Backgrounds
The NQTB is situated on the northern margin of the Tibetan Plateau (Figure 2a), with the Qaidam Block to the south, the Qilian Block to the north, the Altyn Tagh fault to the west and the West Qinling Orogen to the east [33][34][35][36][37]. The NQTB is divided into a northern unit and a southern unit by the E-W-trending Yuka-Wulan fault (Figure 2b). The southern unit is an Early Paleozoic subduction-collision complex related to Proto-Tethys Ocean subduction. It is well known for ultra-high pressure (UHP) metamorphic rocks [37,38]. The northern unit, also known as Quanji Massif or Olongbuluke Massif, mainly consists of Proterozoic metamorphic basement and Phanerozoic sedimentary rocks [35,36,39]. More detailed information about the NQTB can be found in the work of Wang et al. [32], Chen et al. [34], Song et al. [38], Pan et al. [40], and Liu et al. [41]. In this paper, we focus on the internal zoning pattern, major-element composition, and U-Pb isotope of CGMs (AB 2 O 6 , A = Mn, Fe, B = Nb, Ta) in the spodumene-bearing pegmatite from the CKBS Li-Be deposit with the aim of revealing the evolutionary history of the Li-rich pegmatite, determining the formation age of the host pegmatite, and deciphering the relationship between the CKBS deposit and rare-metal deposits in adjacent tectonic units. The results of this study can potentially aid in the exploration of new rare metal mineralization in the eastern section of the North Qaidam Tectonic Belt (NQTB).

Geological Backgrounds
The NQTB is situated on the northern margin of the Tibetan Plateau (Figure 2a), with the Qaidam Block to the south, the Qilian Block to the north, the Altyn Tagh fault to the west and the West Qinling Orogen to the east [33][34][35][36][37]. The NQTB is divided into a northern unit and a southern unit by the E-W-trending Yuka-Wulan fault (Figure 2b). The southern unit is an Early Paleozoic subduction-collision complex related to Proto-Tethys Ocean subduction. It is well known for ultra-high pressure (UHP) metamorphic rocks [37,38]. The northern unit, also known as Quanji Massif or Olongbuluke Massif, mainly consists of Proterozoic metamorphic basement and Phanerozoic sedimentary rocks [35,36,39]. More detailed information about the NQTB can be found in the work of Wang et al. [32], Chen et al. [34], Song et al. [38], Pan et al. [40], and Liu et al. [41]. The CKBS deposit is the first pegmatite-type Li-Be deposit discovered in the eastern segment of NQTB (Figure 2b). There are ca. 800 LCT-type pegmatite dykes in the CKBS deposit, which were derived from high fractional crystallization of granitic magmas [42,43]. These pegmatite dykes occurred in the Paleoproterozoic Dakendaban Group schist Formation and the Ordovician quartz diorite (Figure 2c). The elongation of these pegmatites varies but generally shows NW-SE trending with varying dipping (23°-83°). These pegmatite dykes are ca. 10-400 m in length and ca. 0.5-5 m in width. Since its first discovery in 2018, Li-Be resources of ca. 14,200 t Li2O and ca. 7000 t BeO have been identified [31]. More detailed information on the pegmatites and ore-body in the CKBS deposit can be found in the work of Liu et al. [21], Wang et al. [32], Pan et al. [40], and Liu et al. [41]. In the past four years, the formation ages of pegmatite dykes in this deposit, with greater controversy, have been suggested, including 240.6 ± 1.5 Ma [40] and 217-215 Ma [21] by coltan U-Pb dating; 213-211 Ma by muscovite Ar-Ar dating [21,44]; 216.6 ± 0.88 Ma by lepidolite Ar-Ar dating [21]; 235.9 ± 2.3 Ma and 217 ± 1.8 Ma by zircon U-Pb dating [32].

Samples
The investigated pegmatite samples were collected from the No.15 spodumene pegmatite dyke in the CKBS deposit, which is emplaced in the Ordovician diorite and The CKBS deposit is the first pegmatite-type Li-Be deposit discovered in the eastern segment of NQTB (Figure 2b). There are ca. 800 LCT-type pegmatite dykes in the CKBS deposit, which were derived from high fractional crystallization of granitic magmas [42,43]. These pegmatite dykes occurred in the Paleoproterozoic Dakendaban Group schist Formation and the Ordovician quartz diorite (Figure 2c). The elongation of these pegmatites varies but generally shows NW-SE trending with varying dipping (23 • -83 • ). These pegmatite dykes are ca. 10-400 m in length and ca. 0.5-5 m in width. Since its first discovery in 2018, Li-Be resources of ca. 14,200 t Li 2 O and ca. 7000 t BeO have been identified [31]. More detailed information on the pegmatites and ore-body in the CKBS deposit can be found in the work of Liu et al. [21], Wang et al. [32], Pan et al. [40], and Liu et al. [41]. In the past four years, the formation ages of pegmatite dykes in this deposit, with greater controversy, have been suggested, including 240.6 ± 1.5 Ma [40] and 217-215 Ma [21] by coltan U-Pb dating; 213-211 Ma by muscovite Ar-Ar dating [21,44]; 216.6 ± 0.88 Ma by lepidolite Ar-Ar dating [21]; 235.9 ± 2.3 Ma and 217 ± 1.8 Ma by zircon U-Pb dating [32].

Samples
The investigated pegmatite samples were collected from the No.15 spodumene pegmatite dyke in the CKBS deposit, which is emplaced in the Ordovician diorite and mainly consists of spodumene, albite, feldspar, and quartz, with minor muscovite, tourmaline, and garnet (Figures 2c and 3). Although CGM grains were not directly observed within double polished thin sections and hand specimens, they were successfully separated from the investigated samples using conventional heavy liquid and magnetic methods and then hand-picked under a binocular microscope. The CGM grains were extracted, mounted in epoxy resin, and polished by standard methods. The images of back-scattered electron (BSE), transmitted light, and reflected light were obtained to ensure that the CGM grains are fresh and free of inclusions or cracks. served within double polished thin sections and hand specimens, they were successfully separated from the investigated samples using conventional heavy liquid and magnetic methods and then hand-picked under a binocular microscope. The CGM grains were extracted, mounted in epoxy resin, and polished by standard methods. The images of back-scattered electron (BSE), transmitted light, and reflected light were obtained to ensure that the CGM grains are fresh and free of inclusions or cracks. As shown in Figure 4, the lengths and widths of the investigated CGM grains are ca. 100-350 μm and ca. 50-150 μm. Based on conspicuously different zoning patterns in the BSE images, three types of CGMs were identified: concentric oscillatory CGMs-1 ( Figure  4a  As shown in Figure 4, the lengths and widths of the investigated CGM grains are ca. 100-350 µm and ca. 50-150 µm. Based on conspicuously different zoning patterns in the BSE images, three types of CGMs were identified: concentric oscillatory CGMs-1 (Figure 4a

Major-Element Analyses
Major-element concentrations were analyzed by JEOL JXA8230 at Wuhan SampleSolution Analytical Technology Co., Ltd. (Wuhan, China), using a 20 kV accelerating voltage, 20 nA beam current, and 1 μm beam diameter. Peak and background counting times for all investigated elements were 10 s and 5 s, respectively. Data were corrected by the ZAF method provided by JEOL and the following standards were used: TiO2 for Ti, CaMnSi2O6 for Ca, NaAlSi2O6 for Na, SnO2 for Sn, MnSiO3 for Mn, FeCr2O4 for Fe, Bi2Se3 for Bi, Mg3Al2Si3O12 for Al, PbS for Pb and pure metals for Ta, Nb, Pb, U and W. To reveal the chemical variations in the oscillatory zoning, EMPA mapping was performed on typical CGMs-1 grains using the JEOL JXA8230 at the same company, using a 20 kV voltage and 50 nA beam current. The dwell time is 50 ms, the image pixel lattice size is 1 × 1 μm, and the pixel size is 262 × 192.

Major-Element Analyses
Major-element concentrations were analyzed by JEOL JXA8230 at Wuhan SampleSolution Analytical Technology Co., Ltd. (Wuhan, China), using a 20 kV accelerating voltage, 20 nA beam current, and 1 µm beam diameter. Peak and background counting times for all investigated elements were 10 s and 5 s, respectively. Data were corrected by the ZAF method provided by JEOL and the following standards were used: TiO 2 for Ti, CaMnSi 2 O 6 for Ca, NaAlSi 2 O 6 for Na, SnO 2 for Sn, MnSiO 3 for Mn, FeCr 2 O 4 for Fe, Bi 2 Se 3 for Bi, Mg 3 Al 2 Si 3 O 12 for Al, PbS for Pb and pure metals for Ta, Nb, Pb, U and W. To reveal the chemical variations in the oscillatory zoning, EMPA mapping was performed on typical CGMs-1 grains using the JEOL JXA8230 at the same company, using a 20 kV voltage and 50 nA beam current. The dwell time is 50 ms, the image pixel lattice size is 1 × 1 µm, and the pixel size is 262 × 192.

U-Pb Analyses
The in-situ U-Pb dating for CGMs grains was performed at the Mineral Laser Microprobe Analysis Laboratory, China University of Geosciences, Beijing (CUGB), China, using an Agilent 7900 Quadrupole Inductively Coupled Plasma Mass Spectrometry (Agilent, Santa Clara, CA, USA) coupled with an ASI RESOlution S115LR laser ablation system (Australian Scientific Instruments, Fyshwick, ACT, Australia). During the analyses, the laser spot size was 33 µm, energy density was 5 J/cm 2 , ablation time was 40 s, repetition rate was 5 Hz, and raster scanning speed was 5 µm/s. The dwell times for each mass scan were 10 ms for 232 Th, 15 ms for 238 U and 208 Pb, 20 ms for 204 Pb and 206 Pb, and 30 ms for 207 Pb. During analysis, Coltan 139 was used to correct the ratios of 206 Pb/ 238 U, 207 Pb/ 235 U and 207 Pb/ 206 Pb (ID-TIMS age of 506 ± 2.3 Ma, [45]); NP-2 was used to monitor instrument status and experimental results. Offline data analysis and processing were undertaken with the Iolite software (Iolite Softwares Inc., Laurel, MD, USA). In addition, Isoplot v.4.15 software (Berkeley Geochronology Center, Berkeley, CA, USA) was used to plot Tera-Wasserburg diagrams and calculate weighted mean U-Pb ages [46].

Major-Element Composition
All analysis results by EMPA are listed in Table S1. The abundance ranges, average values, and standard deviations of major element compositions in the CGMs are shown in Table 1. The results of EMPA mapping for CGMs-1 grains are shown in Figure 5. The three types of CGM show marked differences in chemical composition. Among, the CGMs-1 grains are characterized by episodic Nb-and Ta-rich zones alternating to each other (   Table S1.

U-Pb Ages
LA-ICP-MS U-Pb data for the CGMs are summarized in Table 2 and illustrated in

U-Pb Ages
LA-ICP-MS U-Pb data for the CGMs are summarized in Table 2 and illustrated in

CGMs Origin and Chemical Evolution
Both primary magmatic and secondary metasomatic models have been suggested for the genesis of CGMs in the hydrous granitic magma systems. The magmatic origin of CGMs has been established by both experimental determinations and natural samples [47][48][49][50]. The researchers favoring a magmatic model suggested that the CGMs commonly

CGMs Origin and Chemical Evolution
Both primary magmatic and secondary metasomatic models have been suggested for the genesis of CGMs in the hydrous granitic magma systems. The magmatic origin of CGMs has been established by both experimental determinations and natural samples [47][48][49][50]. The researchers favoring a magmatic model suggested that the CGMs commonly coexist with other rock-forming minerals, show simple textures (oscillatory zoning, homogeneous, progressive normal/reverse zoning), and display gradual or periodical variations in Ta/(Ta + Nb) and/or Mn/(Mn + Fe) from core to rim [49][50][51][52][53][54][55]. The main pieces of evidence cited for a metasomatic model are that the CGMs show patchy/complex textures with two distinctly different chemical domains (mainly expressed in the ratios Ta/Nb and Mn/Fe), and coexist with secondary minerals (e.g., wodginite, microlite), and occur in replacement bodies [19,40,48,49,54]. It is worth noting that a recent experimental study suggested that homogenous CGMs with magmatic textures could crystallize from fluidmelt interactions in a hydrous granitic system, where CGMs saturations are reached at geologically reasonable concentrations of high field strength elements (HFSE, e.g., Nb, Ta) in melts and fluid-mobile elements (FME, e.g., Mn, Fe) in hydrothermal fluids [50]. The experiment demonstrated that CGMs can form at the magmatic-hydrothermal stage and require two-end members, rather than always resulting from a single melt or fluid phase as previously reported. The question arises as to whether the CGM grains in the spodumene pegmatite from the CKBS deposit were formed through magmatic processes, metasomatic processes, or fluid-melt interactions.

Primary CGMs-1 and CGMs-2
The oscillatory zoning in the CGMs-1 ferrocolumbite is concentric and continuous, which is different from the irregular, discontinuous oscillatory zoning in metasomatic CGMs [56], indicating that CGMs-1 are of non-metasomatic origin. The textural pattern in CGMs-1 is similar to that in magmatic CGMs from granite and pegmatite worldwide, such as Shangbao granite [47], Ineigi granite [48], Penouta leucogranite [52], Separation Lake pegmatite [19], Tanco Lower pegmatite [49], Jezuitské Lesy pegmatite [57], and Koktokay No.3 pegmatite [5], implying that CGMs-1 have a syn-magmatic origin. The magmatic origin of the CGMs-1 is further supported by the coupled chemical oscillatory zoning, including periodical variations in the concentrations of Ta, Nb, Ti, Sn, W and Pb (Figures 5 and 6c). The chemical oscillatory zoning in magmatic CGMs from pegmatites is believed to result from non-equilibrium fractional crystallization due to the slow diffusion of Nb and Ta, as well as Ti, Sn, W, and Pb, relative to the rate of crystal growth in the crystallization front of the pegmatite [49,56,58,59], which is described as a process of constitutional zone refining by London et al. [4].
Unlike CGMs-1, CGMs-2 ferrocolumbite grains are homogeneous in the BSE images ( Figure 4c). The unzoned textural pattern in the CGMs-2 ferrocolumbite reflects balanced contributions of the main components [49,60,61], chiefly Nb and Ta (Figure 7a). This indicates that the CGMs-2 ferrocolumbite grains were products of equilibrium crystallization, which is significantly distinct from the non-equilibrium crystallization crystals of CGMs-1 ferrocolumbite. In the context of pegmatite, equilibrium-crystallization could take place in the condition of a silicate melt phase coexisting with an independent fluid phase [62,63], where CGMs might directly crystallize from the melt phase, or from the fluid phase, or form via melt-fluid interactions. The chemical and B-isotopic compositions in tourmaline indicated that Na-Fe-B-rich fluid exsolution occurred during the evolution of spodumene pegmatite from the CKBS deposit [43]. The Na-Fe-B-rich fluid cannot transport significant amounts of Nb and Ta to precipitate CGMs-2 ferrocolumbite directly, since melt-aqueous fluid partition coefficients for Ta and Nb are extremely low in the hydrous granitic melt system [12,[64][65][66]. The solubility of Nb and Ta could be enhanced in fluoride solutions, mainly as HF [55,67,68]. However, we have to discount a major role of such fluid on the formation of CGMs-2, since F-rich minerals (e.g., lepidolite, fluorapatite, fluorite, topaz) are either scarce or absent in the investigated pegmatite (No.15), suggesting a low-F system rather than an F-rich one [69,70]. Furthermore, if CGMs-2 had crystallized directly from an F-rich fluid phase, they should have plotted in the Mn-rich region [19,69] rather than the Fe-rich region (Figure 7a). Unlike Nb and Ta, abundant Fe (a fluid mobile element, [60]) was incorporated into the Na-Fe-B-rich fluid phase during fluid-melt immiscibility, resulting in the residual melt phase hardly crystallizing ferrocolumbite directly due to the absence of bivalent cations. Thus, the most possible mechanism for the formation of homogeneous CGMs-2 ferrocolumbite grains is fluid-melt interactions, i.e., they formed via interactions of a melt that is enriched in HFSE (e.g., Nb, Ta) with a diffused fluid enriched in FME (e.g., Mn, Fe). The elevated Mn/(Mn + Fe) trend in the CGMs-2 ferrocolumbite crystals could be ascribed to the competitive crystallization of paragenetic Fe-rich minerals (e.g., hydrothermal tourmaline, [43]).

Secondary CGMs-3
The textures, such as corrosion, crosscutting, embayment, and overgrowth (Figure 4d-i), exclusively indicate the partial decomposition of CGMs-3a by later fluids/melts and the generation of a more Ta-rich CGMs-3b along with the margins and micro-cracks of crystals [19,49,58]. Since the CGMs-3a domains are comparable to CGMs-1 and CGMs-2 in texture patterns and chemical compositions (Figures 4 and 7), it is thus reasonable to infer that CGMs-3a are pristine domains of CGMs-1 and CGMs-2 remaining after alterations of CGMs-3b. The texture patterns in CGMs-3 are comparable to those in CGMs from hydrous granitic melts worldwide, such as Shihuiyao granite [61], Abu Dabbab granite [71], Big Whopper pegmatite [19], Kamativi pegmatite [69], Presqueira pegmatite [70], Mu-fushan pegmatite [72], and Nanyangshan pegmatite [73], which was suggested to be of later fluid-induced or melt-induced replacement origin. If melt-induced model (Ta-rich melt replaced) were applied to CGMs-3, highly evolved injected melt could react with early crystallized rock-forming minerals and form replacement phenomena (e.g., muscovite replacement, [49]). However, replacement phenomena are not observed in the investigated spodumene pegmatite where all main rock-forming minerals boundaries are sharp (Figure 3). Thus, the most possible metasomatic medium for CGMs-3 was a special fluid phase, which might be regional metamorphic [74] or exsolved from the more evolved melt in the very last stages of fractionation [5,19,56,70]. There is no evidence for the pegmatite formation being connected to regional metamorphism either at the outcrop or thin section scale, suggesting the CGMs-3 recorded the variable conditions within the pegmatite-forming melt itself. This is further supported by the extremely wide range of composition in CGMs-3 expressed as ferrocolumbite, ferrotantalite and manganocolumbite (Figure 6a), since metamorphic fluid could homogenize the composition of CGMs whereas magmatic hydrothermal alteration results in diversification of the mineralogy of Nb and Ta [74]. Accordingly, it can be concluded that the special fluid for CGMs-3 formation was magmatic fluid that exsolved from the pegmatite-forming melt at the end of magmatic crystallization. Unfortunately, the available data in the study and reliable data in the associated papers are insufficient to constrain the physicochemical properties of the special fluid, warranting further work on this topic, especially fluid inclusions. As several case studies have shown [47,73], we suggest that the Ta-rich special fluids may be similar to hydrosilicate fluids that were rich in both H 2 O and SiO 2 and had the capacity to dissolve Ta that accumulated in the terminal melts at temperatures ranging from 600 • C to relatively lower temperatures. It is worth noting that the special fluid phase is not the same fluid phase related to CGMs-2 formation in terms of textural relationships between CGMs-2 and CGMs-3, i.e., two fluid exsolution events occurred. By studying fluid inclusions, Alfonso and Melgarejo [75] have confirmed that fluid pulse exsolution could occur during the evolution of rare-metal pegmatite.
According to the study of the internal zoning patterns and element compositions, the formation process of CGMs and the evolution of the spodumene pegmatite are specifically reconstructed as follows (Figure 9): after the pegmatitic melt was emplaced into the shallow crust, Nb was initially saturated and primary CGMs-1 crystals with oscillatory zonation were crystallized directly from the melt in disequilibrium conditions. Subsequently, during the magmatic-hydrothermal transition stage, the homogeneous CGMs-2 crystals were generated via interactions of the evolved melt (enriched in Nb and Ta) with the diffused fluid (enriched in Fe and Mn). Finally, a special magmatic Ta-rich fluid (hydrosilicate?) exsolved from the imminently exhausted melt and led to the formation of secondary CGMs-3 grains by substituting the early formed CGMs-1 and CGMs-2.

Previous Geochronology Review
Zircon U-Pb data revealed that spodumene pegmatite dykes in the CKBS deposit were emplaced at ca. 217 Ma [32]. However, these zircons experienced obvious metamictization and metasomatism, indicating that their U-Pb ages cannot represent the crystallization time of the host pegmatite [76]. The uncertainty of zircon U-Pb age in pegmatite-system has been demonstrated in many case studies, e.g., [21,76]. CGMs grains are regarded as an excellent U-Pb dating alternative for pegmatite-system due to their high U and low common Pb contents [24,45]. Pan et al. [40] were the first to report CGMs U-Pb age for spodumene pegmatite dykes in the CKBS deposit, where eleven homogenous dark CGMs grains yielded a weighted mean 206 Pb/ 238 U age of 240.6 ± 1.5 Ma. The textural patterns in the CGMs grains documented by Pan et al. [40] are comparable with those in CGMs-2, indicating that they formed at the magmatic-hydrothermal stage and their U-Pb ages could represent consolidation ages of the host pegmatite. Recently, Liu et al. [21] reported obviously younger CGMs U-Pb ages (217-215 Ma) for spodumene pegmatite dykes in the CKBS deposit. After careful comparisons, it can be found that these CGMs grains of 217-215 Ma commonly show bright BSE images (Ta-rich domains), which are similar to the CGMs-3b domains in this study and the CGM grain of 211 Ma documented by Pan et al. [40], indicating they are of metasomatic origin and their U-Pb ages represent the time of late metasomatic events. This is further supported by the fact that the U-Pb ages of CGMs with bright BSE images (217-215 Ma) are comparable with muscovite Ar-Ar ages (217-212 Ma) and lepidolite Ar-Ar ages (ca. 217 Ma) [21,44], since the Ar-Ar isotopic systems in the muscovite/lepidolite are susceptible to being reset by late metasomatic events related to CGMs-3 formation due to the low closure temperature (350-400 • C, Burrt and Phillips [77]).

Previous Geochronology Review
Zircon U-Pb data revealed that spodumene pegmatite dykes in the CKBS deposit were emplaced at ca. 217 Ma [32]. However, these zircons experienced obvious metamictization and metasomatism, indicating that their U-Pb ages cannot represent the crystallization time of the host pegmatite [76]. The uncertainty of zircon U-Pb age in pegmatite-system has been demonstrated in many case studies, e.g., [21,76]. CGMs grains are regarded as an excellent U-Pb dating alternative for pegmatite-system due to their high U and low common Pb contents [24,45]. Pan et al. [40] were the first to report CGMs U-Pb age for spodumene pegmatite dykes in the CKBS deposit, where eleven homogenous dark CGMs grains yielded a weighted mean 206 Pb/ 238 U age of 240.6 ± 1.5 Ma. The textural patterns in the CGMs grains documented by Pan et al. [40] are comparable with those in CGMs-2, indicating that they formed at the magmatic-hydrothermal stage and their U-Pb ages could represent consolidation ages of the host pegmatite. Recently, Liu et al. [21] reported obviously younger CGMs U-Pb ages (217-215 Ma) for spodumene pegmatite dykes in the CKBS deposit. After careful comparisons, it can be found that these CGMs grains of 217-215 Ma commonly show bright BSE images (Ta-rich domains), which are similar to the CGMs-3b domains in this study and the CGM grain of 211 Ma documented by Pan et al. [40], indicating they are of metasomatic origin and their U-Pb ages represent

Geochronogical Framework
In terms of the origin and evolution model for the CGMs crystals as discussed above, it would be accepted that the U-Pb ages of CGMs-1 and CGMs-2 represent the crystallization time of the host pegmatite whereas the secondary CGMs-3 U-Pb age is representative of late metasomatic time. One data point obtained for CGMs-1 has no apparent statistical significance, and its 206 Pb/ 238 U age of 252 ± 6.4 Ma hardly represents the crystallized age of host spodumene pegmatite. Because the cooling rate of pegmatite-forming melts after emplacement is rapid [4,78], it is reasonable to ignore the CGMs-1 U-Pb data and use the CGMs-2 intercept age as the crystallization time of the host pegmatite. Accordingly, we reconstruct the geochronological framework for the spodumene pegmatite dyke (No.15) in the CKBS deposit: a Li-B-rich pegmatite-forming melt phase was emplaced into the eastern NQTB and started to crystallize CGMs-1 and CGMs-2 at 230 ± 2.6 Ma. Subsequently, CGMs-1 and CGMs-2 grains were replaced to form CGMs-3 by a special fluid phase at 221 ± 5.3 Ma. Please note that the U-Pb ages of the primary and secondary CGMs are consistent within errors (minimum gap is ca. 1 Ma).
Comparing the new U-Pb data (Table 2, Figure 8) with published geochronological data [21,32,40,44], it can be concluded that the crystallization ages of the spodumene pegmatite dykes in the CKBS deposit are different, with a discrepancy of ca. 6-15 Ma within errors. As the consolidation of an individual pegmatite dyke should be finished in a short time [4,78], the age discrepancy of ca. 6-15 Ma reflects that (i) the Li-rich pegmatites in the CKBS deposit are not a cogenetic pegmatite group (defined byČerný [79]) although they are spatially close together, i.e., they might be derived from different sources, as documented in the pegmatites of the Fregeneda area [80] and the Kamativi area [69]; (ii) the pulse emplacements rather than a single emplacement of Li-rich pegmatite-forming melts in the CKBS deposit, i.e., multiple batches of Li-rich pegmatitic melts were emplaced into the shallow crust during a period of ca. 6-15 Ma. The latter scenario is difficult to reconcile with geological observations that a granitic magma chamber must be large enough to survive over ca. 6 Ma (1200 km 2 for ca. 10 Ma, [81]). So far, no contemporaneous S-type granite with such a large volume has been found around the CKBS deposit. The scenario of the pegmatites not being a cogenetic pegmatite group is more likely, which is supported by the numerous independent 255-235 Ma granitoids exposed around the CKBS deposit [21,42].

Tectonic Environment
Extensive Early-Middle Triassic arc igneous plutons are exposed near the studied area in the eastern NQTB, including the Wulan gabbro (245-242 Ma), the Tianjunnanshan granites (246 Ma), the Qinghaihunanshan granite (238 Ma), the Xugeigou granite (254 Ma), the Qiluoshan granodiorite (251 Ma), the Chahannuo hornblende diorite (249 Ma), the Chahannuo granite (248 Ma), the Chahanuo granodiorite (243 Ma), the Chahanhe granite (240 Ma), the Shailekeguolei granodiorite (249 Ma), and the Narong granitic porphyry (245 Ma) [21,27,40]. The formation ages and Nd-Hf isotopic compositions of these arc igneous plutons in the eastern NQTB are comparable to those of arc igneous rocks in the Buerhanbuda area (250-238 Ma) in the East Kunlun and the Tongren-Zeku area (242-234 Ma) in the West Qinling, indicating that the eastern NQTB, East Kunlun, and West Qinling were a complete active continental margin system during the Early-Middle Triassic, which was related to the northward subduction of the Paleo-Tethys oceanic plate [21,26]. Post-collision A-type granite plutons in the eastern NQTB (220-215 Ma) and East Kunlun (228-218 Ma) indicate that the Paleo-Tethys Ocean was closed in the Late Triassic [21]. Therefore, it would be accepted that the Middle Triassic pegmatites and adjacent Early-Middle Triassic granitoids in and around the CKBS deposit were formed in an oceanic subduction-related setting. The conclusion is consistent with Boroujerd pegmatites [82].

Regional Mineralization Comparison
Regionally, there are three important rare-metal metallogenic belts related to Tethys evolution around the CKBS deposit in the eastern NQTB (Figure 10a), including the worldclass MYKB [21,24], the Guanpo-Danfeng mineralization belt (GDMB) [73,83,84] and the Tugman-Tashisayi mineralization belt (TTMB) [76,85]. The CKBS Li-Be deposit in the eastern NQTB and the pegmatite-type rare-metal deposits in the MYKB, GDTB and TGMB are products of extreme fractionation of peraluminous granitic intrusions [21,25,42,83,85,86]. Therefore, whether the CKBS deposit can be connected with one of the spatially adjacent MYKB, GDTB, and TTMB remains an interesting question, which is of great significance for regional prospecting. The following three reasons support the conclusion that the CKBS deposit is not connectible to the three spatially adjacent rare-metal mineralization belts. First, the formation ages of rare-metal deposits in the MYKB, GDMB, and TTMB are 224-144 Ma, 420-394 Ma, and 468-350 Ma, respectively, which are different from the crystallization age of the CKBS deposit (ca. 240-230 Ma, Table 3, Figure 10b-d). However, it is interesting to note that the late metasomatic ages of pegmatite dykes in the CKBS deposit overlap with the crystallization ages of some deposits in the MYKB (Figure 10d). Second, regionally comprehensive studies showed that the geological environments were late/post-orogenic regimes when rare-metal deposits formed in the MYKB, GDMB, and TTMB [25,73,83,85], whereas the dynamic environment for the formation of spodumene pegmatite in the CKBS deposit was an active continental margin, indicating that they were controlled by different geological events. Third, Hf isotopic compositions in zircons are believed to be unaffected during metamictization, which thus could trace the source of host granitic rock [86]. The ε Hf (t) values in zircons from the CKBS deposit varied from −15.5 to −11.63 [32] while those from the rare-metal deposits in the MYKB and GDMB ranged from −13.6 to −7.23 [86,87] and from −7.8 to −7.0 [83], respectively, indicating that the sources of these pegmatite-granite are different.
GDMB, and TTMB are 224-144 Ma, 420-394 Ma, and 468-350 Ma, respectively, which are different from the crystallization age of the CKBS deposit (ca. 240-230 Ma, Table 3, Figure  10b-d). However, it is interesting to note that the late metasomatic ages of pegmatite dykes in the CKBS deposit overlap with the crystallization ages of some deposits in the MYKB (Figure 10d). Second, regionally comprehensive studies showed that the geological environments were late/post-orogenic regimes when rare-metal deposits formed in the MYKB, GDMB, and TTMB [25,73,83,85], whereas the dynamic environment for the formation of spodumene pegmatite in the CKBS deposit was an active continental margin, indicating that they were controlled by different geological events. Third, Hf isotopic compositions in zircons are believed to be unaffected during metamictization, which thus could trace the source of host granitic rock [86]. The εHf(t) values in zircons from the CKBS deposit varied from −15.5 to −11.63 [32] while those from the rare-metal deposits in the MYKB and GDMB ranged from −13.6 to −7.23 [86,87] and from −7.8 to −7.0 [83], respectively, indicating that the sources of these pegmatite-granite are different.  Table 3. This diagram does not include the muscovite Ar-Ar ages of the Aketasi, Dangba and Tugmanbei deposits due to their far difference from the results of other geochronological methods. Meanwhile, due to the large error values (±8, MSWD = 6.7, Li et al. [23]), the cassiterite U-Pb age for the 505 Li is also ruled out. The ages of Triassic granitoids (grey area) in the eastern NQTB are from Wu et al. [91] and Sun et al. [42]. Table 3. The formation ages revealed by various dating methods for pegmatite-type rare-metal deposits in the TTMB, MYKB, GDMB and the eastern NQTB.

Minerallized Belt Dating Method Age (Ma) References
Except for ca. 800 pegmatite dykes in the CKBS deposit, significantly abundant pegmatite dykes have also been found in the eastern segment of NQTB (Figure 11), e.g., ca. 360 in the Shaliuquan area, ca. 100 in the Chahannuo area and ca. 700 in the Shinaihai area [28,33]. Recently, several granite-pegmatite-related rare metal deposits/mineralization points have also been discovered in these pegmatite-exposure areas, including the Shaliuquan Nb-Ta deposit, Asiheta Be deposit, Shinaihai Nb-Ta-Rb deposit, Honglingbei Li-Be mineralization point, Yematannan Li mineralization point, and so on ( Figure 11). In addition to the pegmatite in the CKBS deposit, pegmatite dykes in other deposits/mineralized points in the eastern NQTB are very poorly understood, which makes it difficult to quantitatively decipher the rare-metal metallogenic regularity at the moment. However, the following three commonalities can be concluded for the pegmatite dykes in the eastern NQTB based on the results of preliminary field investigations [28][29][30][31][32][33]40]: (1) mostly intruded into schist, gneiss, and marble of Paleoproterozoic Dakendaban Group; (2) distributions were controlled by secondary fracture and/or ductile shear zone; (3) closed related to adjacent granite, some even occurred in granite. Therefore, we suggest that the eastern NQTB may be a new potential rare-metal mineralization belt, which should be explored in detail and arouse painstaking attention.

Conclusions
(1) Three-stage crystallizations of CGM crystals are identified in terms of variations in internal zoning patterns and major element compositions. First, oscillatory CGMs-1 ferrocolumbite grains crystallize at the magmatic stage, which are products of non-equilibrium fractional crystallization and are characterized by episodic variations in Ta/(Ta+Nb), TiO2, WO3 and PbO. Second, homogeneous CGMs-2 ferrocolumbite grains are formed at the magmatic-hydrothermal transition stage, which result from equilibrium crystallization via fluid-melt interactions and are characterized by the relatively stable Ta/(Ta+Nb) and moderately elevated Mn/(Mn+Fe). Third, irregular/complex CGMs-3 crystals commonly consist of Nb-rich domains (CGMs-3a) and Ta-rich domains

Conclusions
(1) Three-stage crystallizations of CGM crystals are identified in terms of variations in internal zoning patterns and major element compositions. First, oscillatory CGMs-1 ferrocolumbite grains crystallize at the magmatic stage, which are products of nonequilibrium fractional crystallization and are characterized by episodic variations in Ta/ (Ta + Nb), TiO 2 , WO 3 and PbO. Second, homogeneous CGMs-2 ferrocolumbite grains are formed at the magmatic-hydrothermal transition stage, which result from equilibrium crystallization via fluid-melt interactions and are characterized by the relatively stable Ta/(Ta + Nb) and moderately elevated Mn/(Mn + Fe). Third, irregular/complex CGMs-3 crystals commonly consist of Nb-rich domains (CGMs-3a) and Ta-rich domains (CGMs-3b) and are formed by the metasomatism of early-formed CGMs-1 and CGMs-2 with special fluids. This metasomatism causes considerably elevated Ta/(Ta + Nb) in CGMs-3b.
(3) The CKBS deposit is not likely to be connected to the adjacent rare-metal mineralization belts (MYKB, GDTB and TTMB). The occurrence of abundant pegmatite dykes and obvious mineralization information indicate that the eastern NQTB may be a new potential rare-metal mineralization belt in western China, which should be paid more attention to.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/min13020201/s1, Table S1: Chemical compositions of CGMs in the spodumene pegmatite from the CKBS Li-Be deposit analyzed by EMPA.

Data Availability Statement:
The authors declare that all analytical data supporting the findings of this study are available within the paper and its supplementary information files or cited peerreview references.