The Chemical Characteristics and Metallogenic Mechanism of Beryl from Cuonadong Sn-W-Be Rare Polymetallic Deposit in Southern Tibet, China

: The Cuonadong deposit is the ﬁrst large scale Sn-W-Be rare polymetallic deposit located in southern Tibet, China, where beryl is the main beryllium-bearing mineral. In this paper, the beryl crystals in the pegmatitic Therefore, it can be concluded that beryl mineralization largely reﬂects the process of magmatic–hydrothermal mineralization. Because of a large number of mineralized areas with the similar metallogenic backgrounds to the Cuonadong deposit in the Himalayan region, it has great potential to be a new globally signiﬁcant rare metal metallogenic belt.


Introduction
Beryllium, the lightest alkaline earth metal element, is highly incompatible; thus, it is commonly present in granitic residual melts and enriched in evolved granites and granitic pegmatites [1,2]. The formation of beryllium minerals runs through the whole magmatic-hydrothermal evolution process, including beryl, chrysoberyl, phenacite and bertrandite [3][4][5][6]. Beryl is the most common beryllium mineral, which is mostly present in peraluminous metallogenic systems and can be formed at multiple stages of the magmatichydrothermal transition. According to the evolution stage of the magma-hydrothermal system, it is further divided into three types: magmatic, granitic pegmatite and hydrothermal type [1,7]. Beryl is the most important Be-hosting mineral in granitic pegmatite-type beryllium deposits [6]. During the magmatic-hydrothermal evolution of granitic pegmatite, the crystallization of beryl is predominantly controlled by temperature and pressure [6]. In addition, the stability of beryl in the melt is also affected by the activity of Si, Al and other components in the melt, such as Li, B, P, F, Rb, Cs, etc. [4,5]. Small changes in activity of Be 2+ can alter the crystallization behavior of beryl with changes in the composition and temperature of the pegmatitic melt [2,5].
In addition to being an important carrier of Be, beryl can also carry a certain amount of Li, Cs, Na, Fe, Mg, and its chemical composition and structure can be used to indicate the degree of differentiation of pegmatites and trace magmatic-hydrothermal evolution [8,9]. Studies have shown that the higher the Cs content in beryl is, the lower the Na/Cs value, Na/Li value, Mg/Fe value and Mg content are, indicating the higher degree of differentiation and evolution of granitic pegmatite [8][9][10][11].
Beryl, ideally Be 3 Al 2 Si 6 O 18 , has a ring structure consisting of layers of rings with six Si-O tetrahedra(T1), which are linked vertically and laterally by Be-O tetrahedra (T2) and Al-O octahedra(O), and the centers of the Si rings contain channels [12,13]. Various cations may be incorporated in the beryl structure, resulting in complex substitutions [6]. Therefore, the crystal structure and chemical composition of beryl can reflect the physical and chemical conditions of its formation [1,2,[9][10][11][12][13].
In this study, beryl crystals from pegmatitic and hydrothermal mineralization in the Cuonadong deposit as the research objects are investigated by energy spectrum scanning, electron probe, LA-ICP -MS, in situ microzone XRD, with a view to ascertain the physical and chemical characteristics of beryl from the two types of mineralization, and to discuss the Cuonadong deposit in the magmatic-hydrothermal evolution process of the rare metal Be metallogenic mechanism, which can provide evidence of the Himalayan metallogenic belt having further prospects in rare metals.

Himalayan Metallogenic Belt
The Tethyan Himalaya Sequence (THS) is separated by the Indus Yarlung Tsangpo suture zone to the north and the South Tibet Detachment System (STDS) to the south [35][36][37], sandwiched between the Great Himalayan Crystalline Complex (GHC) and the Lhasa terrane [38] (Figure 1). As the largest highly differentiated granite belt developed in the Himalayas, the study of Cenozoic tectono-magmatic evolution and metallogenic relationship between highly differentiated leucogranites and their rare metals in this region has been a hot topic at home and abroad in recent years. Previous studies revealed that the Himalayan Metallogenic Belt is mainly composed of Pb-Zn-Ag-Au-Sb deposits [39][40][41][42][43]. Meanwhile, researchers have been paying attention to the mineralization of rare metals such as beryllium, tungsten, and tin in the Himalayas. Many rare metal minerals have been found in leucogranites and pegmatites, including lithium-rich minerals such as lithium tourmaline and spodumene [44], and other rare metal minerals such as scheelite, cassiterite, beryl, niobium rutile, columbite-tantalite, ironlepidolite, spodumene, rubidium-containing mica, etc. [44][45][46][47][48]. Beryl, as a beryllium-rich mineral, was found in the Pusila, Wengbo, Cuona, Yadong, Kangma, Gaowu, Dingjie, Xiaru, Peguco and other leucogranite bodies [47]. The latest exploration results show that Be-Rb-Nb-Ta-W-Sn and other rare metal mineralizations are developed in and around the Himalayan domes ( Figure 1). Preliminary studies show that this region has great metallogenic potential and is expected to become an important rare metal metallogenic belt in China [47][48][49][50].

Geology of Cuonadong Mining Area
The spatial distribution of the Cuonadong Sn-W-Be polymetallic deposit is strictly controlled by the structure of the Cuonadong gneiss dome, which is typically representative of rare metal mineralization in the Himalayan gneiss dome belt [14][15][16][17].

Cuonadong Gneiss Dome
The Cuonadong gneiss dome is shaped as a rhombus, which is 20 km in length, 20 km in width and approximately 400 km 2 in area. The dome can be divided into three parts: core, mantle and rim [15,19,22] (Figure 2). First, the core is located under the lower detachment fault, consisting of early Paleozoic orthogneiss and Miocene leucogranite and pegmatite. Furthermore, the mantle and core are separated by the lower ductile detachment fault. The mantle shows typical characteristics of the ductile shear zone, formed from a set of strongly deformed quartz-mica schists with mylonitized carbonate rocks metamorphosed at Paleogene. Strong contact metasomatism occurred between the marble and leucogranite or granitic pegmatite, thus forming a discontinuous skarn zone around the dome where W-Sn-Be polymetallic ore bodies are found. In addition, the rim and mantle are separated by the upper brittle detachment fault, and the rim is dominated by a set of low-grade metamorphic rocks composed of phyllite, sandy slate and carbonaceous slate. The Cuonadong dome has experienced three strong magmatism stages. The first stage occurred under the background of the dome activity of the south Tibet detachment system (STDS) in southern Tibet in the Oligocene (34~26 Ma) and was mainly distributed in the mantle and rim of the dome [50]. The lithology is mainly composed of strong deformed two-mica granites and granitic pegmatite. The second stage occurred in the lower detachment fault in the Early Miocene (24~18 Ma) [8,10]. The lithology consists of two-mica granites with weak deformation. Particularly in the vicinity of the lower detachment fault, the second-phase two-mica granite is directed with schist and marble and shows strong mineral orientations and lenticular shapes [18,20,30]. 3 The third stage occurred in the Middle and Late Miocene (16~14 Ma) with undeformed massive muscovite leucogranite and highly differentiated garnet-tourmaline-bearing muscovite leucogranite and beryl-bearing pegmatite [27,31]. The main body of the dikes is mainly distributed in a north-south direction, intrudes into various geological bodies and structural layers, and cuts the detachment fault in the form of veinlets, which is probably related to the north-south-trending rifts (NSTRs) (Figure 2).

Mineralization Characteristics
Four regions with important metallogenic potential are distributed around the dome, including Xianglin, Yunazha, Rina and Qugaer ( Figure 2). Among them, the Xianglin area in the northwestern part of Cuonadong has the most abundant mineralization types and the largest inferred resources; thus, it is the key research area in this study and the main area of recent mineral exploration ( Figure 3). The main mineralization types at Xianglin include pegmatite Be, skarn W-Sn-Be and hydrothermal vein Sn-W-Be ores. Preliminary exploration of Xianglin shows that the resources of Sn, WO 3 and BeO are over 80,000 t at 0.36%, 50,000 t at 0.21% and 34,000 t at 0.08% [27,28]. 1 The pegmatite Be orebody is spatially closed to muscovite granite (16 ± 1 Ma), and a large amount of beryl-bearing pegmatite is developed on the top of the muscovite granite. The gradual and transitional contact between the two implies an evolutionary relationship. Beryl is the main beryllium-bearing mineral in the disseminated form, with crystals up to 10 cm and colors ranging from blue-green to pale green, which co-exists with quartz, plagioclase, K-feldspar, muscovite, tourmaline and fluorite ( Figure 4a). The content of beryl in pegmatite dikes is not evenly distributed, and the scale of individual veins is tens of centimeters to a few meters, which has not yet been evaluated [14,17]. In the greisen,the cassiserite-quartz vien was cut by the cas-siserite sulfide vein. 2 Skarn ore bodies are mainly W-Sn mineralization, followed by Be. Cassiterite is the dominant tin-bearing mineral, and scheelite is the main tungsten-bearing mineral ( Figure 4b). Beryllium mainly occurs in silicate minerals, such as vesuvianite, albite and garnet, through element substitutions and a small amount of Be-containing minerals such as phenacite, bertrandite [32]. However, as far as the current mineral processing technology is concerned, beryllium in these skarn minerals is difficult to fully utilize [51]. 3 Hydrothermal vein ore bodies are produced in high-angle and nearly north-south faults and detached faults, mainly exhibiting greisenization, albitization, silicification, fluoridization and pyritization. The mineralization includes cassiterite-quartz veins ( Figure 4c) and cassiterite sulfide veins (Figure 4d), which is superimposed on skarns, or occurs on the top or in the edge of granitic pegmatites, which developed later than weakly oriented mica granites, indicating that hydrothermal veins may be formed in garnet-bearing mica granites. The latest exploration results show that 11 new hydrothermal vein Sn-W-Be ore bodies have been discovered in the Xianglin area, among which one main ore body (No. Z6 orebody) has the predicted potential resources of Sn 73,800 tons at 1.75%, WO 3 19,900 tons at 0.36% and BeO 3100 tons at 0.14% [51]. Beryllium occurs mainly in beryl, and tin and tungsten occur in cassiterite and scheelite, respectively. The hydrothermal vein Sn-W-Be ore bodies are the most significant mineralization in the Cuonadong mining area [17,51].

Analytical Methods
In this paper, two beryl crystals from pegmatitic and four from hydrothermal mineralization of the Xianglin area are taken as the key research objects ( Figure 5), and the location of the samples is shown in Figure 3. The samples under investigation were characterized through a combination of optical examination, combined with scanning electron microscopy, electron microprobe analyses and laser ablation inductively coupled plasma mass spectro metry.
Major-and minor-element compositions of beryls were conducted on a JEOL JXA-8230 electron microprobe at the Key Laboratory of Metallogeny and Mineral Resource Assessment, Institute of Mineral Resources, Chinese Academy of Geological Sciences. The analytical conditions were: an accelerating voltage of 15 Kv, a beam current of 20 nA, and a beam size of 5µm. The following standards were used: albite (Na), sanidine (Si,Al,K), pyrope (Mg), titanite (Ca,Ti), almandine (Fe), rhodonite (Mn), Rb-leucite (Rb) and pollucite (Cs). The analytical data were corrected using the ZAF methods.
Trace-element compositions of beryls were performed using an X-Series ICP-MS (Thermo Fisher Scientific, Braunschweig, Germany) coupled with a J-200 343 nm Yb: Fiber femto-second laser ablation system (Applied Spectra, West Sacramento, CA, USA) housed at the National Research Center for Geoanalysis, Chinese Academy of Geological Sciences. Helium gas carrying the ablated sample aerosol from the chamber was mixed with argon make-up gas and nitrogen as an additional diatomic gas to enhance sensitivity. A baffled-type smoothing device was used in front of the ICP-MS to reduce fluctuation effects induced by laser-ablation pulses and to improve the quality of the data [52]. The beryl samples were ablated for 50 s at a repetition rate of 8 Hz at 6 J/cm 2 , and ablation pits were~50 µm in diameter. The calibration was performed externally using two NIST SRM 610 and one NIST SRM 612 for every 10 samples with 29 Si, measured by electron microprobe, as the internal standards to correct for instrument drift. Data reduction was carried out with the commercial software ICPMS Data Cal 10.8 [53,54]. The detection limits of LA-ICP-MS ranged from 0.05 to 0.1 ppm for REE. Repeated analyses of the standards SRM 610 and SRM 612 indicate that both precision and accuracy were better than 10% for most analyzed elements.
The mapping of beryl was determined using a NWR 193 nm ArF Excimer laserablation system coupled to an iCAP RQ (ICPMS) at the Guangzhou Tuoyan Analytical Technology Co., Ltd., Guangzhou, China. The ICPMS was tuned using NIST 610 standard glass to yield low oxide production rates. Then, 0.7 L/min He carrier gas was fed into the cup, and the aerosol was subsequently mixed with 0.89 l/min Ar make-up gas. The laser fluence was 5 J/cm 2 , with a repetition rate of 20 or 30 Hz and a 5, 8 or 10 µm spot size. The laser scan speed was 15, 20 or 35 um/s. The raw isotope data were reduced using the "TRACE ELEMENTS" data reduction scheme (DRS). The DRS runs within the freeware IOLITE package of Paton [55]. In IOLITE, user-defined time intervals are established for the baseline correction procedure to calculate session-wide baseline-corrected values for each isotope. Blocks of two standards (one NIST 610 and one GSE-2G) and one MASS-1 sulfide standard analyses were followed by 5 to 8 unknown samples. 29 Si in beryl measured by electron microprobe was used as internal standard to correct instrument drift.
In situ micro-X-ray diffraction (XRD) experiment was conducted with a Rigaku D/max Rapis IIR micro-XRD system at the Guangzhou Institute of Geochemistry, Chinese Academy of Sciences. All measurements were carried out at the 40 kV and 250 mA (Cu Kα radiation) with a collection time of 22 min, 2θ range is 20~110 • C, and continuous scan. The scan speed was 4 • /min. The X-ray beam was about 40 µm in diameter, focused on the selected spots on the thin sections. The data point interval was 0.045 • . The software of MDI Jade 6.0 (International Centre for Diffraction Data, Denver, CO, America) was used to analyze the obtained XRD patterns for the derivation of unit-cell parameters by whole pattern fitting.

Occurrence
Beryl from pegmatite (Beryl-I) is mostly symbiotic with rock-forming minerals such as quartz, potash feldspar, albite, muscovite and fluorite, with well-developed hexagonal column shape and particle sizes of 0.1-5 cm (Figure 5a,b). BSE images of these beryl particles show a homogeneous structure and locally developed altered edges, which is metasomatized by quartz, muscovite, K-feldspa, and albite along the edges or fissures (Figure 5b,c).
Beryl from the hydrothermal veins (Beryl-II) is mainly associated with cassiterite, scheelite, quartz, muscovite, phlogopite, fluorite, albite, sphalerite, and arsenopyrite, with the sizes ranging from 0.1 to 20 mm (Figure 5d,e). It is typically characterized by the development of complex zonal structures, which are locally replaced or dissolved by later K-feldspar, albite, and scorodite (Figure 5e,f).

Chemical Composition
As Li and Be are light elements (i.e., the X-ray yield through electron excitation is poor), and it is not possible to obtain precise Li 2 O and BeO analyses on an electron microprobe. Furthermore, Be is a major element in beryl; thus, the results measured by LA-ICP-MS are inaccurate [33]. Assuming that the crystallochemical formula of beryl is X 3 Y 2 [Z 6 O 18 ], the anion method is adopted for calculation, and the theoretical number of oxygen atoms is 18 [56]. The theoretical atomic number of Be, herein, is calculated by taking the sum of the cations at the X site as 3. Meanwhile, assuming that Be at the X site is only replaced by Li, the theoretical atomic number of Be can be calculated by Be = 3-Li apfu [56]. The quantification of H 2 O was estimated as H 2 O wt.% = (0.845958 × Na 2 O in wt.%) + 0.08373 [57].
In this paper, two methods are used to obtain the Li content in beryl, one is through chemical calculation of EPMA (Li = Ca + Na + K + Cs, the number of atoms obtained by stoichiometric calculation) [58], the other is direct determination by LA-ICP-MS. By comparing the data results of these two methods, we found that Li content in beryl calculated by EPMA (0.04-0.87 wt.%) is significantly higher than that directly determined by LA-ICP-MS (1.23-2.48 wt.%). The reason is that there is complex metal replacement in beryl [59]. Li + replacement of Be 2+ results in a charge deficit, which requires monovalent alkali metal to enter the channel to balance the charge. This is the same as the replacement process of Al 2+ by Fe 2+ and Mg 2+ [34].
Beryls from the Xianglin area contain different amounts of FeO (0.17-1.31 wt.%) and MgO (0~2.06 wt.%), leading to the high content of Li obtained by chemical calculation of EPMA. It implies that the content of Li in beryls is not completely consistent with the sum of other alkali metals ( Figure 6). Thus, the content of Li calculated by using the EPMA component may lead to deviation. Moreover, the deviation of the Li content calculation value will further affect the content of BeO. Therefore, in this paper, the content of Li directly measured by LA-ICP-MS is used to calculate the content of BeO. By calculation, the BeO content of beryl in pegmatite is 11.69~12.05 wt.%, and that in the hydrothermal vein is 12.27~13.75 wt.%. The H 2 O content of the two beryls is 1.41~1.60 and 1.26~2.25 wt.%.  where T2 and T1 are the tetrahedral sites for Be and Si, respectively, O is the octahedral site, and Ch is the "channel".  A thorough analysis is presented in Sections 5.1 and 5.2.

LA-ICP-MS Mapping
LA-ICP-MS mapping is used to analyze the complex zonal structure of beryl in the hydrothermal vein of Xianglin area. Judging from the mapping, the complex bands of beryl-II are mainly caused by the difference in the contents of Be, Al, Na, Cs, and Cr ( Figure 7).

Unit Cell Parameters
The in situ micro-X-ray diffraction (XRD) analysis results (Table 3) show that the absolute values and average of the unit cell parameter a,b and the c/a ratio have a slight difference between the two types of beryls. Beryl-I has the slightly higher c value (9.2209-9.2243, avg.c = 9.2228) and c/a ratio (1.0010-1.0012, avg. 1.0011) than beryl-II (c = 9.2180-9.2206, avg.c = 9.2197; c/a =1.0004-1.0010, avg. c/a =1.0008).

Crystal Chemical Features
According to the crystallochemical formula of beryl, various cations are incorporated in the beryl structure, resulting in complex substitutions. As divalent cations, such as Fe 2+ and Mg 2+ , substitute trivalent Al 3+ in the Y site (octahedral), leading to a charge deficit, the long-radius monovalent cations and divalent cation Ca 2+ enter the channel to balance the charge [59][60][61][62][63][64]. Since the Ca 2+ and K + contents of the two types of beryls from the Xianglin area are negligible (Table 1), the occupying position of Ca 2+ and K + is not discussed in this paper. The atoms per formula unit of Be 2+ in the X site of Beryl-I and Beryl-II is 2.70 and 2.89 apfu, respectively, showing obvious unsaturation. Li + is the main substitute cation in the X site, and the atoms per formula unit are 0.30 and 0.11 apfu, respectively, indicating that the replacement degree of the X site ion of Beryl-I is higher than that of Beryl-II. The atoms per formula unit of Al 3+ in the Y site of Beryl-I is 1.94 apfu, and the isomorphic substitution is 0.03 apfu, manifesting that the substitution degree of the Y site of Beryl-I is low, whereas the atoms per formula unit of Al 3+ is 1.76 apfu, and the substitution of the Y site is 0.27 apfu, indicating that the substitution degree of the Y site of Beryl-II is relatively higher than that of Beryl-I.
Ideally, the sum of divalent cations in the Y site (R 2+ ) and monovalent alkali cations in channel (A + ) of beryl is close to 1:1 [65]. It can be illustrated with Figure 8 that all the points of Beryl-I lie at a high position above the 1:1 slope line, showing an obvious substitution in the channel. The points of Beryl-II are mostly located near the slope of R2+ = A+, and some points lie at a high position above the slope, indicating a certain amount of A+ entering the X and Y sites.
Linear fitting is performed on the main cation contents of the two types of beryl at the X site, Y site and channel (Figure 9). In Beryl-I, the negative correlation between Li and Na, Li and Cs, Na and Cs, and Fe and Al indicates that the mechanism of element substitution in Beryl-I includes (Na,Cs)Li -1 Be -1 channel-tetrahedral substitution, (Na,Cs)Fe 2+ -1 Al -1 channel-octahedral substitution and NaCs -1 the mutual substitution of alkali metal ions in the "channel". By comparison, in Beryl-II, Na and Li are weakly negatively correlated, Al is negatively correlated with Mg + Fe 2+ , Mg, Fe 2+ , and Mg is also negatively correlated with Fe 2+ , which can be inferred that the element substitution mechanism of Beryl-II is composed of NaLi -1 Be -1 channel-tetrahedral substitution and Na(Fe 2+ ,Mg) -1 Al -1 channel-octahedral substitution.  Linear fitting is performed on the main cation contents of the two types of beryl a the X site, Y site and channel (Figure 9). In Beryl-I, the negative correlation between Li an Na, Li and Cs, Na and Cs, and Fe and Al indicates that the mechanism of element substi tution in Beryl-I includes (Na,Cs)Li□-1Be-1 channel-tetrahedral substitution, (Na,Cs)Fe 2+ □ 1Al-1 channel-octahedral substitution and NaCs-1 the mutual substitution of alkali meta ions in the "channel". By comparison, in Beryl-II, Na and Li are weakly negatively corre lated, Al is negatively correlated with Mg + Fe 2+ , Mg, Fe 2+ , and Mg is also negatively cor related with Fe 2+ , which can be inferred that the element substitution mechanism of Beryl II is composed of NaLi□-1Be-1 channel-tetrahedral substitution and Na(Fe 2+ ,Mg)□-1Al channel-octahedral substitution. Fe is one of the most important chromogenic ions of beryl. I addition to replacing Al 3+ in the Y (octahedral) site, Fe appears in the channel, making i appear light green, blue-green, green-blue and light blue [56]. Emerald is chromatized b Fe, V and Cr [59], and a certain amount of Fe, V and Cr is detected in Beryl-I and Beryl-I Thereby, there is likely to be a potential for further emerald prospecting at the Xiangli area. vs. Al 3+ plot. Fe is one of the most important chromogenic ions of beryl. In addition to replacing Al 3+ in the Y (octahedral) site, Fe appears in the channel, making it appear light green, blue-green, green-blue and light blue [56]. Emerald is chromatized by Fe, V and Cr [59], and a certain amount of Fe, V and Cr is detected in Beryl-I and Beryl-II. Thereby, there is likely to be a potential for further emerald prospecting at the Xianglin area.

Classification
As the ions are replaced, the unit cell parameters of the minerals also change. Based on the difference in the ratio of c/a, beryl is divided into three types: 1 n-beryl (normal beryl, c/a = 0.997~0.998), ideal pure beryl without replacement; 2 o-beryl (octahedral beryl, c/a = 0.991~0.996), of which Al 3+ in the center is partially replaced; 3 t-beryl (tetrahedral beryl, c/a = 0.999~1.003), of which a large number of metal ions entering the structure channel to balance the electricity price [4]. The c/a ratios of Beryl-I and Beryl-II from the Xianglin area range from 1.0004 to 1.0012 (Table 3), which shows that the beryls from the Xianglin area are t-beryl. Significantly, the c/a ratios of Beryl-I and -II show a slight decreasing trend, which is probably caused by the accommodation of alkali elements and substitution of Be 2+ and Li + in the beryl structure along the c-axis [66]. It may reflect the hydrothermal process in the late stage of pegmatite evolution [9,10].
Judging from the plot of Mg/Fe vs. Cs/Na (Figure 11), there are obvious evolutionary differences between the two beryls. The content of Cs 2 O of beryl-I in pegmatite ranges from 0.31 to 1.77 wt% (avg. = 1.01 wt%), and the ratio of Cs/Na(apfu) = 0.10-0.44 (avg. = 0.28), with almost no MgO detected. It is indicated that beryl-I was formed under highly differentiated evolution conditions [4]. By contrast, the content of MgO in beryl-II is higher, ranging from 0.46 to 2.25 wt.%, avg. = 1.48 wt.%, and the ratio of Mg/Fe(apfu) ranging from 2.73 to 17.31, avg. = 6.33, with the low content of Cs 2 O inferring that Beryl-II probably experienced hydrothermal alkali-metasomatism [10].

Metallogenic Mechanism
The solution and saturation of beryllium in granitic magma are mainly controlled by five factors [4][5][6][7]: temperature, pressure, melt composition, fractionation, and the volatile components such as B, P, and F.
The geological characteristics, metallogenic chronology, stable isotopes and fluid inclusions of the Cuonadong Be-Sn-W polymetallic deposit show that the ore-forming materials and fluids of the deposit are derived from Cuonadong leucogranite [14][15][16][17][18][19][20][21][22][23][24]. Cuonadong leucogranite, a typical representative of the Himalayan leucogranite belt, is a complex pluton formed by the multi-stage evolution of homologous magma, including the early strongly deformed two-mica granite, the middle weakly deformed two-mica leucogranite and the late undeformed muscovite leucogranite, of which the chemical composition gradually evolved to be of higher silicon, more alkalis and more volatile [19][20][21][22][23][24]. Beryllium, an incompatible element, would preferentially enter the melt phase during the differentiation and evolution of magma. With the continuous process of this interaction, volatile compounds and ore-forming fluids can often be released from the late magmatism [68]. Highly differentiated leucogranites generally have a high alumina saturation index (ASI), of which beryllium content is positively proportional to ASI value [5,69]. In addition, the beryllium content has a weak to moderate negative correlation with the potassium content, and the reduction of potassium is due to the homogeneous reaction of volatile anions such as F, P, and B with the Na-containing magmatic melt [5]. That is, the more potassium phases (e.g., K-feldspar, mica) that gradually dominate the crystallization mode, the more sodium remains in the melt [70]. Flux components (F, B, P) in pegmatite melts can form some stable fluoride, borate and phosphate migration with rare metals [4,71]; thereby, the transitional fluids with the coexistence of magmatic fluid and residual magma, rich in Be, Sn, W, B, F, H 2 O, migrate to the top of the granitic or surrounding rocks [72] (Figure 12). Due to the intrusion of pegmatite magma into the surrounding rock, there is a significant temperature difference between the two, resulting in undercooling of the melt-liquid phase and a decrease in the solubility of rare metal minerals [69]. Meanwhile, some minerals rich in flux components (such as fluorite, tourmaline, and apatite) crystallize [73], leading to a great decrease in contents of flux components such as F, B, and P in the ore-forming system and Beryl-I crystallization. It is also the reason why Beryl-I is mainly produced in the contact between albitite pegmatite and the surrounding rock. With the continuous differentiation and crystallization of the leucogranitic magma, a large number of ore-forming materials such as Be, W and Sn are brought out. When the ore-bearing fluid migrates to the marble layer with pressure releasing, extensive water-rock reactions occur between the ore-bearing fluid and marble layer [72], and then the skarn be-W-Sn ore body is formed (Figure 12). Beryllium mainly occurs in skarn minerals such as vesuvianite, scapolite and garnet modified by late hydrothermal solution, in the form of isomorphisms [32].
Influenced by the east-west extensional structure, some nearly north-south normal faults and secondary faults were developed in the middle Miocene   [27]. Oreforming fluids migrated along these faults, and with the decrease in temperature and the mixing of some hydrothermal water, the hydrothermal vein-type Be-Sn-W ore (chemical) bodies were formed [72] (Figure 12). It can be seen from the above discussion (Section 5.2) that beryl in hydrothermal vein-type ore bodies is related to hydrothermal activity. The microthermometry results of fluid inclusions show that the temperature of ore-forming fluid of hydrothermal vein orebodies ranges from 197 to 343 • C [28]. Under such mineralization conditions, beryllium mostly exists in the form of fluoride, chloride, and complexes containing fluorine or chlorine, while beryllium tends to migrate as beryllium carbonate complexes if the activity of fluorine in the solution is low and the activity of carbonic acid is high [74,75]. There are many wide fluorite veins in the Cuonadong mining area [17,28], indicating that the ore-forming fluid is extremely rich in fluorine. Therefore, it can be inferred that beryllium is most likely transported as beryllium fluoride (BeF) or fluoride complex ions (e.g., [BeF 4 ] 2− ) [74], in this area. The result of beryl-II's mapping shows that the development of these complex zonings is mainly caused by the differences among the contents of elements such as Na, Cs, Be, Al, Cr, and As (Figure 7). It probably reflects the unbalanced crystallization environment and the heterogeneity of ore-forming fluids, indicating that extensive fluid immiscibility and metasomatism has occurred in the metallogenic system [72], and it has undergone the superimposed transformation of later ore-forming fluids, such as strong alkaline metasomatism in the late period of beryl mineralization [10,76]. Field geological survey studies have shown that a variety of alterations, such as greisenization and albitization, often develop in the hydrothermal vein orebodies [77]. During the process, liquid H 2 O is bound to be generated, which cannot only reduce the temperature, but can also react with beryllium complexes in the ore-forming fluid and combine with Si and Al, thereby forming hydrothermal beryl. The probable reaction formula is as follows [77]: Beryl occurrences in the pegmatites and hydrothermal vein ore bodies in the Cuonadong mining area indicate that beryllium mineralization is characterized by magmatichydrothermal transition. Some granitic rare metal deposits in South China have similar characteristics; that is, the pegmatic crusts usually develop on the top of the granitic rock mass, while hydrothermal mineralization occurs at the surroundings of the rock mass, and the hydrothermal mineralization pattern is closely related to the surrounding rocks [10,78,79]. A large number of leucogranites similar to the Cuonadong leucogranite are developed in the Himalayas, and the Tethys Himalayan sedimentary sequence is mainly a set of clastic rocks and carbonate rocks originating from passive Indian continental margin [32]. In conclusion, there is further great potential and prospects for the pegmatite, skarn and hydrothermal-vein rare metal deposits related to leucogranite in the Himalayas.

Conclusions
The results of this study allow for the following conclusions for beryls from the Cuonadong deposit.
(1) In terms of composition, beryls in the Cuonadong deposit are alkaline beryls, among which beryl-I is Li-Cs beryl, and beryl-II consists of Na and Na-Li beryl. Structurally, they are t-beryls. This indicates that beryl has experienced magmatism to hydrothermal alkali-metasomatism in the late stage of pegmatitic magmatism during formation. (2) The mechanism of element substitution in Beryl-I includes (Na,Cs)Li -1 Be -1 channeltetrahedral substitution, (Na,Cs)Fe 2+ -1 Al -1 channel-octahedral substitution and NaCs -1 the mutual substitution of alkali metal ions in the "channel", whereas that in Beryl-II consists of NaLi -1 Be -1 channel-tetrahedral substitution and Na(Fe 2+ ,Mg) -1 Al -1 channel-octahedral substitution.