Fluid Inclusion and Oxygen Isotope Characteristics of Vein Quartz Associated with the Nabeba Iron Deposit, Republic of Congo: Implications for the Enrichment of Hypogene Ores

The Nabeba iron ore deposit is located at the northern part of Congo Craton, Republic of Congo. The ore deposit consists of supergene and hypogene ores, both of which are hosted in the Precambrian Nabeba banded iron formation (BIF). This study focuses on the hypogene iron ore mineralization associated with quartz veins in the Nabeba deposit, for which two hypogene ore stages have been recognized based on geologic and petrographic observations: early-stage high‐grade hematite‐rich ore (HO‐1) and late-stage magnetite‐rich ore (HO‐2). Based on microthermometric measurements and laser Raman spectroscopy of the fluid inclusions, the H2O‐NaCl ± CO2 fluids interacting with the Nabeba BIF at the HO‐1 stage evolve from high‐to‐moderate temperatures (203–405 °C) and contrasting salinities (moderate-to-low: 1–15 wt. % NaCl equiv.; high: 30–35 wt. % NaCl equiv.) to H2O‐NaCl fluids of moderate‐to‐low temperatures (150–290 °C) and salinities (1–11 wt. % NaCl equiv.) for the HO‐2 ore stage. Assuming equilibrium oxygen isotopic exchange between quartz and water, the δ18Ofluid values range from 4.7–8.1‰ for the HO‐1 stage and −2.3‰ to −1.5‰ for the HO‐2 stage. This implies the ore‐forming fluid of initially-mixed metamorphic–magmatic origin, later replenished by seawater and/or meteoric water during the formation of the HO‐2 stage. These mixtures of different fluids, coupled with their interaction with the BIF lithology followed by phase separation, are responsible for the enrichment of hypogene iron ore in the Nabeba deposit.


Introduction
Precambrian banded iron formations (BIFs) are known to host the world's largest and richest iron deposits attractive for exploration [1,2]. Iron ore deposits in BIFs commonly form by supergene weathering under tropical conditions and by interaction with low-temperature (meteoric) fluids. However, some deposits have been considered to precipitate from high-temperature hypogene hydrothermal fluids [3][4][5][6][7].
The Nabeba iron deposit, hosted in a Precambrian BIF, is located in the northwestern edge of the Congo Craton and in the Ivindo Basement Complex (IBC) of the Souanké district of the Republic of  [11] and Desthieux [12]). Abbreviations: CAR, Central African Republic; DRC, Democratic Republic of the Congo; EG, Equatorial Guinea.  [11] and Desthieux [12]). Abbreviations: CAR, Central African Republic; DRC, Democratic Republic of the Congo; EG, Equatorial Guinea.
In this paper, we present fluid inclusion and oxygen isotope data from these veins with the aim of reconstructing the nature of the hydrothermal fluids and physico-chemical processes that were involved in the iron ore enrichment.

Geological Setting
The Nabeba deposit, about 80 km from Souanké district, is a part of the Archean supracrustal rocks of IBC [11,12] (Figure 1), which includes banded-iron formations, sericitic and chloritic quartz schists, amphibolites and mafic to intermediate volcanic rocks (Figure 2), and intrusives such as syenite and alkaline microsyenite, fine-grained doleritic gabbro, porphyroid granite, and peridotite-pyroxenolite lenses [13]. In some sections of the deposit, the BIF alternates with chlorite-albite-carbonate-quartz schist and chlorite-sericite-quartz schist (greenstones) to a depth of about 80 m. The association with volcanic rocks and the dominant mineralogy of magnetite and hematite-rich bands renders the Nabeba deposit as an oxide facies Algoma-type BIF [9,10].
The Nabeba area was interpreted as a structurally-dismembered, fan-shaped set of convex North-East-striking imbricate thrusts or reverse faults, coalescing into an East-West-trending regional shear zone to the north of the deposit [8,9]. The BIFs are deformed to upright, tight to isoclinal, large-scale folds with flat plunges to the northeast and southwest. Centimeter-to decimeter-scale parasitic folds mimic the large-scale fold geometry [8] (Figure 3).

Sampling and Analytical Methods
Sampling for this study was carried out on fifteen drill cores that intersected the Nabeba BIF of contrasting ore grades and partly associated with the quartz veins.
To determine the mineral assemblages and the mineral paragenetic sequence, forty-four thin sections were prepared. Petrographic and mineral compositional analyses were conducted on the prepared samples using the reflected and transmitted light Olympus BX 51 microscope, scanning electron microscope (SEM) (Phenom-World B.V, Eindhoven, The Netherlands), and electron probe microanalysis (EPMA) (JEOL's EPMA technologies, London, UK). A total of twelve doubly-polished sections were prepared. Petrographic descriptions were undertaken using an Olympus BX 51 microscope (Olympus Corporation, Tokyo, Japan) to define fluid inclusions types, sizes, morphologies, and assemblages. Microthermometric analyses were conducted using a Linkam THMSG600 (Linkam, London, UK) freezing-heating stage with a TMS 93 temperature controller (Linkam, London, UK), fitted on an Olympus BX 51 transmitted light microscope, at the Key Laboratory of the Department of Mineral Resources Engineering at the University of Science and Technology, Beijing. The unit operates in the temperature range of −190 • to +600 • C. The stage was calibrated with Linkam synthetic fluid inclusion standards: pure H 2 O and mixed H 2 O-CO 2 . The cyclic technique [14] was used to acquire better precision in measurements of the temperatures of phase transitions. The accuracy of the freezing measurement runs was about ±0.1 and ±1 • C for heating runs between 200 and 500 • C. The precision for microthermometric measurements below 200 • C was ±0.2 • C. The following temperatures were recorded: (a) ice melting (Tm_ice); (b) halite dissolution (Tm_halite); (c) CO 2 melting (Tm_CO 2 ); (d) partial homogenization of CO 2 -rich phase (Th_CO 2 ); (e) clathrate melting (Tm_clathrate); and, (f) total homogenization (Th_total). Calculations of fluid inclusion salinity (from Tm_ice, Tm_halite, and Tm_clathrate), density (from Th_CO 2 and Th_total), and pressure-temperature (P-T) isochore slopes were made using worksheets of Steele-MacInnis et al. [15] for aqueous inclusions (HOKIEFLINCS_H 2 O-NaCl), Steele-MacInnis [16] for aqueous-carbonic inclusions, and Hansteen and Klügel [17] for monophase CO 2 inclusions. Salinity is reported in equivalent weight percentages of NaCl (wt. % NaCl equiv.), and density as g/cm 3 .
Raman spectroscopy was used to qualitatively determine the composition of gases to complement the data acquired from fluid inclusion microthermometry. Raman spectra were obtained on a Jobin Yvon/Horiba LABRAM-HR 800 spectrometer (Horiba Jobin Yvon Ltd., Longjumeau, France) equipped with a tunable Ar + laser, with power up to 50 mW and with notch filters at 488 and 514 nm. The Raman signal was collected by a BX-51 Olympus microscope using 100× objectives. The lateral resolution was better than 2 µm, the acquisition time ranged from 10 to 120 s, and the accumulation number was 10. Collected Raman spectra were analyzed and optimized with Labspec 4.18 and Origin 8.0. Background was removed and peaks deconvoluted using the mathematical methods of Cadusch et al. [18]. Measurements were performed at the Raman Laboratory in the Beijing Research Institute of Uranium Geology.
For oxygen isotope study, seven samples of vein quartz were carefully selected from Veins A and B under a binocular microscope (~1 g for each sample) and crushed into 200 mesh. After, the crushed samples were reacted with BrF5 at 500-700 • C to liberate oxygen. The oxygen was converted to CO 2 gas on a platinum-coated carbon rod and the oxygen isotope composition of CO 2 gas was determined using a Finnigan MAT-253 mass spectrometer (Thermo Finnigan GmbH, Bremen, Germany) at the Analytical Laboratory, Beijing Research Institute of Uranium Geology. The isotopic data are reported relative to the standard mean ocean water (SMOW). The analytical precision is better than 0.2% for δ 18 O. The δ 18 O values of water in equilibrium with the hydrothermal quartz were calculated using the equation of Clayton et al. [19] using the minimum trapping temperatures of fluid inclusions in the same quartz samples.

Ore Mineral Parageneses
The paragenetic sequence in the least-altered Nabeba BIF (magnetite-rich), through the itabiritetype, hypogene and supergene ores, is shown in Figure 5.
The hypogene iron ore enrichment (HO), within alternating bands of grey-to-white quartz and brown-to-dark brown iron oxide ( Figure 6), occurs below the supergene and itabirite-type mineralization. The HO is subdivided into: early high-grade hematite-rich stage (HO-1) and late magnetite-rich stage (HO-2).
High-grade hypogene hematite-rich ore of the early HO-1 stage is consistently sub-vertical and sub-parallel to BIF bedding. The ore comprises steeply South-dipping zones of hard and massive hematite and local occurrences of specularite ( Figure 7b). Hematite is commonly fine-to mediumgrained, with sheaves of fine specularite and cm-scale vugs lined with specularite. The high-grade The itabirite-type mineralization ( Figure 6) is considered to have formed primarily by metamorphic alteration of the original magnetite-rich Nabeba BIF (Figure 7a). The ore consists primarily of coarser grains of quartz and magnetite replaced by hematite. In some BIF sections, siderite-ferroan magnesite are also abundant, containing accessory pyrite, fluorapatite, xenotime, monazite, ferri-muscovite, ferroan-dolomite, arsenopyrite, sphalerite, and chalcopyrite.
The hypogene iron ore enrichment (HO), within alternating bands of grey-to-white quartz and brown-to-dark brown iron oxide ( Figure 6), occurs below the supergene and itabirite-type mineralization. The HO is subdivided into: early high-grade hematite-rich stage (HO-1) and late magnetite-rich stage (HO-2).
High-grade hypogene hematite-rich ore of the early HO-1 stage is consistently sub-vertical and sub-parallel to BIF bedding. The ore comprises steeply South-dipping zones of hard and massive hematite and local occurrences of specularite ( Figure 7b). Hematite is commonly fine-to medium-grained, with sheaves of fine specularite and cm-scale vugs lined with specularite. The high-grade hypogene massive iron ore consists mostly of granoblastic and tabular hematite (Figure 7c). Relics of magnetite and subhedral to euhedral magnetite (Figure 7d), present in this stage, represent the earliest iron oxide mineral of the BIF, enclosed in hematite (martite). The presence of hard martite-microplaty hematite is noted also in this ore stage (Figure 7e). These high-grade hypogene hematite ores are in direct contact with quartz (Figure 7c-e).
The late hypogene magnetite-rich ore of stage HO-2 (Figure 7f,g) is the product of nucleation of magnetite crystals that remained after direct removal (or leaching) of silica from primary quartz-magnetite bands by the hydrothermal fluids. In this stage, the ore is composed of massive magnetite and hematite (Figure 7f,h); and the gangue minerals consist of fine-grained siderite, ferroan magnesite, chlorite, (Figure 7g), apatite, and monazite.
The supergene ore consists of dispersed hematite-goethite ± limonite-quartz-clay, with high-grade areas dominated by hematite. Some areas also contain hematite-martite, which is the result of the replacement of the least-altered BIF by supergene hematite-goethite ( Figure 7i). Most of the goethite replaces hematite and magnetite along grain boundaries and fractures ( Figure 7i). The late hypogene magnetite-rich ore of stage HO-2 ( Figure 7f,g) is the product of nucleation of magnetite crystals that remained after direct removal (or leaching) of silica from primary quartzmagnetite bands by the hydrothermal fluids. In this stage, the ore is composed of massive magnetite and hematite (Figure 7f,h); and the gangue minerals consist of fine-grained siderite, ferroan magnesite, chlorite, (Figure 7g), apatite, and monazite.
The supergene ore consists of dispersed hematite-goethite ± limonite-quartz-clay, with highgrade areas dominated by hematite. Some areas also contain hematite-martite, which is the result of the replacement of the least-altered BIF by supergene hematite-goethite ( Figure 7i). Most of the goethite replaces hematite and magnetite along grain boundaries and fractures ( Figure 7i).

Quartz Veins Associated with Nabeba Hypogene Iron Ore
The quartz vein systems ( Figure 8) in the Nabeba deposit are constrained by different structural features that are closely associated with the formation of the hypogene ores. Two major types of quartz veins are recognized and labelled as quartz veins A and B, on the basis of crosscutting relationships that suggest the quartz vein A is older than the quartz vein B. The morphology of each vein system are defined by the lithology and deformation affecting the BIF.
Quartz vein A is considered to be related to the hypogene hematite-rich ore of stage HO-1. The vein is conformable with the S0 bedding plane of the BIF and seems to develop a divide along contact with the BIF (Figure 8a-c). The orientation of the vein is sub-parallel and concordant to the high-grade hematite-rich ore. In highly deformed sections, it is boudinaged and appears oblique to the hematite ore.  with the BIF (Figure 8a-c). The orientation of the vein is sub-parallel and concordant to the highgrade hematite-rich ore. In highly deformed sections, it is boudinaged and appears oblique to the hematite ore.
Quartz vein B is related to the hypogene magnetite-rich ore of the stage HO-2 ( Figure 8d). About 1 to 4 cm thick veinlets of quartz vein B are observed to crosscut folded bands of magnetite-rich ore and appear to be discordant with the hypogene magnetite-rich ore.

Fluid Inclusion Study
In this study, fluid inclusions were genetically classified to primary, pseudo-secondary, and secondary according to the criteria given by Goldstein and Reynolds [14] and Roedder [20]. Only the primary and pseudo-secondary fluid inclusions were taken into consideration. Individual fluid inclusions and those that constitute fluid inclusion assemblages (FIAs) in individual quartz crystals were selected for microthermometry and Raman spectrometry. Volumetric proportions of the liquid (L) and vapor (V) phases relative to the total volume of each studied fluid inclusion at room temperature (25 °C) were estimated based on standardized charts of Shepherd et al. [21]. Quartz vein B is related to the hypogene magnetite-rich ore of the stage HO-2 ( Figure 8d). About 1 to 4 cm thick veinlets of quartz vein B are observed to crosscut folded bands of magnetite-rich ore and appear to be discordant with the hypogene magnetite-rich ore.

Fluid Inclusion Study
In this study, fluid inclusions were genetically classified to primary, pseudo-secondary, and secondary according to the criteria given by Goldstein and Reynolds [14] and Roedder [20]. Only the primary and pseudo-secondary fluid inclusions were taken into consideration. Individual fluid inclusions and those that constitute fluid inclusion assemblages (FIAs) in individual quartz crystals were selected for microthermometry and Raman spectrometry. Volumetric proportions of the liquid (L) and vapor (V) phases relative to the total volume of each studied fluid inclusion at room temperature (25 • C) were estimated based on standardized charts of Shepherd et al. [21].

Fluid Inclusion Petrography
Five types of fluid inclusions were recognized. Quartz vein A contains all five fluid inclusion types, while quartz vein B contains Type IIa and Type IIb only.
Type I monophase CO 2 -rich inclusions are relatively rare in comparison to the other fluid inclusion types. Type I fluid inclusions are found in irregular clusters confined to individual quartz grains. These inclusions have cylindrical, irregular, or negative crystal shapes with sizes ranging from less than 5 to 10 µm. They are transparent to dark-colored and single-phase at room temperature (Figure 9a).

Microthermometry and Raman Spectrometry
The microthermometric data from the analysis of the fluid inclusions is summarized in Table 1. The total homogenization temperatures decreased from quartz vein A (HO-1 stage) through quartz vein B (HO-2 stage).
Quartz vein A host all the fluid inclusion types: I, IIa, IIb, IIIa, and IIIb. The range in total homogenization temperatures of the fluid inclusions in quartz vein A is 203-405 °C. The temperatures show a unimodal distribution, whereas their corresponding salinity displays a bimodal distribution of high salinity values (30-35 wt. % NaCl equiv.) and high densities (1.01-1.10 g/cm 3 ) for Type IIIa, and low-to-moderate salinity values for the Type IIa, IIb, and IIIb fluid inclusions (Figure 10a). The Types IIa and IIb fluid inclusions have salinity values between 1-15 wt.% NaCl equiv., with lower calculated densities that range between 0.59 to 0.98 g/cm 3 (Figure 10a) (Table 1). Type IIIb has salinity values between 2.8-11.2 wt. % NaCl equiv., calculated on the basis of corresponding clathrate temperatures at 3.5-8.6 °C (i.e., T_clathrate < 10 °C). The almost pure CO2 content of the Type IIIb fluid inclusions is implied from the temperatures of CO2-solid melting data of −58.2 to −56.7 °C Tm_CO2 and from the clathrate melting below 10 °C. However, one fluid inclusion sample did register Tm_CO2 of −59.9 °C, which suggests the likely presence of other carbonic phases such as CH4 or N2. The homogenization temperature of CO2 to the liquid phase of the Type IIIb fluid inclusions ranges between 14.2 and 29.9 °C with the calculated densities ranging from 0.86-1.05 g/cm 3 ( Table 1). The total concentration of CO2, X(CO2), of these aqueous-carbonic inclusions range between 7 to 15 mol % (Table S1). The unimodal distribution in temperature values but wide variations in salinity values can be attributed to fluid mixing of multiple fluids and/or phase separation.
Laser Raman analysis confirmed only CO2-rich carbonic phases in the Types I and IIIb fluid inclusions of quartz vein A (Figure 11a,b). The nature of occurrence of the Type I and IIIb fluid Type IIa fluid inclusions are two-phase (L + V) aqueous inclusions at room temperature (Figure 9b,c). The vapor phase occupies <50 vol% of the total volume. This type does not contain daughter minerals and homogenize to the liquid on heating. Type IIa fluid inclusions are irregular-to negative-crystal-shaped, they occur isolated or as random clusters in quartz veins A and B or as intragranular growth planar arrays (in quartz vein A). Thus, the Type IIa fluid inclusions are considered as primary or pseudo-secondary. Their sizes range from 5.5 to 24 µm.
Type IIb fluid inclusions are also two phase (L + V) (Figure 9d) aqueous fluid inclusions, but the vapor phase occupies >50 vol% of the total volume at room temperature. They also occur in clusters or in trails within individual quartz grains. Hence, they are likely to be primary or pseudo-secondary. The Type IIb fluid inclusions do not contain daughter minerals and homogenize to the vapor on heating. They are regular to irregular in shape and vary in their size range from 7 to 24 µm.
Type IIIa fluid inclusions are liquid-rich, three-phase (L + V + S) fluid inclusions (Figure 9e). At room temperature, they are composed of aqueous liquid, vapor and halite. They occur as clusters in quartz vein A or along the small intragranular trails in quartz vein B within individual quartz grains. As such, they are interpreted as primary or pseudo-secondary. Sizes range from 7.6 to 22.5 µm.
Type IIIb fluid inclusions are aqueous-carbonic and composed of three phases (L + L + V) (Figure 9f): aqueous liquid, CO 2 liquid, and vapor phases. These inclusions have irregular to negative crystal shapes and occur as random three-dimensional clusters and in intragranular trails. They are also interpreted as pseudo-secondary inclusions. Their sizes range from 7 to 24.6 µm.
In quartz vein A, the coexistence of Type IIIa with Type IIa and Type IIb aqueous inclusions in the same individual vein quartz crystals suggests heterogeneous trapping of the fluid inclusions, and that phase separation via the immiscibility and/or boiling phenomena occurred in the hydrothermal fluids from which quartz vein A (HO-1 stage) precipitated. Moreover, the coexistence of aqueous-carbonic Type IIIb and the carbonic Type I fluid inclusions with the other types of fluid inclusions (e.g., Type IIa and IIb) in the A veins also suggest heterogeneous trapping of the fluid inclusions and immiscibility characteristics of the aqueous-carbonic hydrothermal fluid (e.g., Figure 9c,d) in the hydrothermal system. For quartz vein B, the coexistence of only Types IIa and IIb fluid inclusions suggests that phase separation in the hydrothermal fluid continued locally during the fluid evolution at the late HO-2 stage.

Microthermometry and Raman Spectrometry
The microthermometric data from the analysis of the fluid inclusions is summarized in Table 1. The total homogenization temperatures decreased from quartz vein A (HO-1 stage) through quartz vein B (HO-2 stage).
Quartz vein A host all the fluid inclusion types: I, IIa, IIb, IIIa, and IIIb. The range in total homogenization temperatures of the fluid inclusions in quartz vein A is 203-405 • C. The temperatures show a unimodal distribution, whereas their corresponding salinity displays a bimodal distribution of high salinity values (30-35 wt. % NaCl equiv.) and high densities (1.01-1.10 g/cm 3 ) for Type IIIa, and low-to-moderate salinity values for the Type IIa, IIb, and IIIb fluid inclusions (Figure 10a). The Types IIa and IIb fluid inclusions have salinity values between 1-15 wt.% NaCl equiv., with lower calculated densities that range between 0.59 to 0.98 g/cm 3 (Figure 10a) ( Table 1). Type IIIb has salinity values between 2.8-11.2 wt. % NaCl equiv., calculated on the basis of corresponding clathrate temperatures at 3.5-8.6 • C (i.e., T_clathrate < 10 • C). The almost pure CO 2 content of the Type IIIb fluid inclusions is implied from the temperatures of CO 2 -solid melting data of −58.2 to −56.7 • C Tm_CO 2 and from the clathrate melting below 10 • C. However, one fluid inclusion sample did register Tm_CO 2 of −59.9 • C, which suggests the likely presence of other carbonic phases such as CH 4 or N 2 . The homogenization temperature of CO 2 to the liquid phase of the Type IIIb fluid inclusions ranges between 14.2 and 29.9 • C with the calculated densities ranging from 0.86-1.05 g/cm 3 ( Table 1). The total concentration of CO 2 , X(CO 2 ), of these aqueous-carbonic inclusions range between 7 to 15 mol % (Table S1). The unimodal distribution in temperature values but wide variations in salinity values can be attributed to fluid mixing of multiple fluids and/or phase separation.
Laser Raman analysis confirmed only CO 2 -rich carbonic phases in the Types I and IIIb fluid inclusions of quartz vein A (Figure 11a,b). The nature of occurrence of the Type I and IIIb fluid inclusions in only quartz vein A, and not in B (HO-2 stage), may suggest phase separation of the Type IIIb fluid to the Type I monophase carbonic inclusions and the Type II aqueous fluid inclusions.
For quartz vein B of the HO-2 stage, only Types IIa and IIb fluid inclusions have been found. The aqueous inclusions are predominantly of a low salinity (0.7-10.6 wt. % NaCl equiv.; Figure 10b) and theirs densities range from 0.76 to 0.97 g/cm 3 (Table 1). This range in salinity is similar to that of quartz vein A. Total homogenization temperatures of 150 to 290 • C are lowest in comparison to those of quartz vein A (Figure 10b). Both distribution patterns in the homogenization temperatures and salinities of fluid inclusions in the Quartz vein B grains are generally unimodal.
Since heterogeneous trapping is implied from the textural distribution of these coexisting primary fluid inclusions in quartz veins from A through B, the homogenization temperatures of these fluid inclusions would serve as minimum trapping temperatures of the fluid inclusions [22]. Therefore, in the absence of other independent temperature estimates, these minimum temperatures would represent the true trapping temperatures of the fluid inclusions during the various stages of quartz development and hypogene iron mineralization in the Nabeba deposit (see discussion for details). fluid inclusions would serve as minimum trapping temperatures of the fluid inclusions [22]. Therefore, in the absence of other independent temperature estimates, these minimum temperatures would represent the true trapping temperatures of the fluid inclusions during the various stages of quartz development and hypogene iron mineralization in the Nabeba deposit (see discussion for details).
In Figure 12, the plot of the minimum homogenization temperature and the 18 Ofluid values is illustrated. It shows a fluid mixing trend from high to low temperature regimes for the hydrothermal fluid components that represent the two ore stages of the Nabeba hypogene iron mineralization.
In Figure 12, the plot of the minimum homogenization temperature and the δ 18 O fluid values is illustrated. It shows a fluid mixing trend from high to low temperature regimes for the hydrothermal fluid components that represent the two ore stages of the Nabeba hypogene iron mineralization. The δ 18 Ofluid values for water in equilibrium with quartz, calculated using the equation of Clayton et al. [19] are 4.7-8.1‰, and from −2.3‰ to −1.5‰, respectively. In Figure 12, the plot of the minimum homogenization temperature and the 18 Ofluid values is illustrated. It shows a fluid mixing trend from high to low temperature regimes for the hydrothermal fluid components that represent the two ore stages of the Nabeba hypogene iron mineralization. Ttmin: Minimum trapping temperature (°C).

Figure 12.
Plot of homogenization temperature versus calculated oxygen isotopic values of the oreforming fluid in equilibrium with vein quartz A and B in the Nabeba deposit. The compositions of primary magmatic water [23,24], metamorphic fluids [25], the Mesozoic meteoric water in the northern Xilamulun district, northeastern China [26], and seawater [27] are also shown for comparison. Figure 12. Plot of homogenization temperature versus calculated oxygen isotopic values of the ore-forming fluid in equilibrium with vein quartz A and B in the Nabeba deposit. The compositions of primary magmatic water [23,24], metamorphic fluids [25], the Mesozoic meteoric water in the northern Xilamulun district, northeastern China [26], and seawater [27] are also shown for comparison.

Trapping Mechanism and Implications for Temperature, Pressure, and depth Conditions of Hypogene Iron Ore Mineralization
Straightforward estimates of the trapping pressure can only be carried out on T-P-density isochore diagrams when the actual trapping temperature is known, or if phase separation due to boiling or immiscibility occurred in the fluid system at the time of capture [28]. The microthermometric data and coexistence of Types I and IIIb with the other types of fluid inclusions in quartz veins A suggest heterogeneous trapping of the fluid inclusions and fluid immiscibility characteristics (e.g., Figure 8c,d) in the hydrothermal system. The coexistence of Type IIa and IIIa fluid inclusions that yielded similar homogenization temperatures but varying salinity in quartz vein A (HO-1 stage) are direct evidences of fluid immiscibility and/or boiling during vein quartz precipitation in these stages. Quartz vein B of the HO-2 stage does not have Type IIIa fluid inclusions. However, local phase separation of the fluid during fluid mixing at this stage can also be assumed since the Types IIa and IIb fluid inclusions coexist as gas-poor and gas-rich inclusions, respectively. Given that the fluid inclusions are heterogeneously-trapped, only the minimum homogenization temperatures correspond to true trapping temperatures [22]. Therefore, true trapping temperatures for vein quartz in quartz vein A and quartz vein B based on the aqueous fluid inclusions are 203 and 150 • C, respectively. Consequently, using the diagram from Driesner and Heinrich [29], trapping pressures were estimated at~20 bars for hypogene hematite-rich ore associated with quartz vein B (HO-1 stage); and less than 10 bars for hypogene magnetite-rich ore associated with quartz vein B (HO-2 stage) ( Figure 13). Hydrostatic pressure conditions suggest depths of 0.2 and 0.1 km, respectively.
The estimation of the minimum P-T conditions for entrapment of Type IIIb aqueous carbonic inclusions (H 2 O-CO 2 -NaCl system) in quartz A veins at the HO-1 stage requires a plot of P-T isochores that were constructed using the minimum and maximum fluid inclusion densities. Virtually all of the Type IIIb fluid inclusions followed the P-T path, which is represented by the isochore traced by the minimum density. The hypothesis that aqueous carbonic Type IIIb fluids were heterogeneously trapped and contained a minimum of~7 mol% CO 2 (Figure 14a) was essential to constrain the P-T range. The intersection between the minimum density isochore and the isopleth for 7 mol% CO 2 , which separates the liquid immiscibility region from the H 2 O-CO 2 system, provided the minimum temperature (T t min ) and the minimum pressure (P t min ) of trapping in the H 2 O-CO 2 -NaCl system for the Type IIIb fluid inclusions of stage HO-1 at T t min = 215 • C and P t min =~800 bars (Figure 14b

Nature and Origin of the Hypogene Ore-Forming Fluids
The evolution of ore-forming fluids in the Nabeba deposit can be illustrated by plotting the microthermometric data for all types of fluid inclusions as a function of the homogenization temperature and salinity ( Figure 15). The distribution of the microthermometric data of fluid inclusions in quartz samples from the HO-1 stage and the HO-2 stage shows decreasing homogenization temperatures and salinities over time ( Figure 15).
For stage HO-1, a large variation in fluid inclusion salinity values can be interpreted as reflecting the entrapment of high temperature, higher salinity Type IIIa fluid inclusions; low to high temperature, low to moderate salinity fluid Type IIa and IIIb fluid inclusions; and then, moderate temperature, low salinity Type IIb fluid inclusions (Figures 10 and 15). These observed variations are best explained by the progressive trapping of multiple hydrothermal fluid pulses whose temperature and salinity values varied as a function of time and proximity to the central fluid source [25] and/or due to phase separation. In fact, most of the fluid inclusions in this stage correspond to Types IIa and IIIa, and display a positive correlation between homogenization temperatures and salinities. These characteristics and relationships indicate that the fluid at this stage contained a mixture of magmatic (high salinity aqueous) and metamorphic (low to moderate salinity aqueous-carbonic) components. Based on the quartz-water isotopic equilibrium fractionation factor at 203-215 • C, the hydrothermal vein quartz from the HO-1 stage was calculated to have δ 18 O fluid values ranging from 4.7% -8.1% (Table 2), which is consistent with a mixed metamorphic and magmatic origin for the fluids that formed the high-grade hypogene hematite-rich ore ( Figure 12).
Quartz vein B of the HO-2 stage is characterized by the low to moderate salinity and temperature of fluid inclusions of Type IIa and Type IIb. These two types are characterized by lower homogenization temperatures and lower salinities compared to stage HO-1 fluid inclusions, which may imply cooling and dilution of the hypogene fluids ( Figure 15). This hypothesis is corroborated by the lower δ 18 O fluid values (−2.3% to −1.5% ; Table 2) for quartz vein B of the HO-2 stage, in comparison to quartz vein A of the HO-1 stage. Such low to moderate salinity fluids with low δ 18 O fluid values and low temperatures can result from the mixing of magmatic and/or metamorphic fluids by low salinity, low temperature seawater or meteoric water [30]. The lower trapping temperatures of the fluid inclusions in the quartz of this HO-2 stage, whose timing coincides with the leaching of silica and the net removal of Fe-rich carbonate from BIF, suggests that the retrograde solubility of siliceous and Fe-rich carbonate minerals could have played an important role in the formation of magnetite-rich ore at the Nabeba deposit.
Based on the estimated shallow depths of the hypogene ore formation and the idea that metamorphic fluids are likely to maintain their CO 2 -rich component at shallower depths compared to magmatic fluids [31], we suggest that at greater depths distal from the site of deposition of the hypogene ores, a high-salinity aqueous magmatic-hydrothermal fluid mixed with a metamorphic-hydrothermal fluid containing the Type IIIb aqueous-carbonic fluid inclusions. Then, as this mixed fluid ascended to the site of ore deposition and interacted with the Nabeba BIF lithology, it underwent phase separation, via fluid immiscibility and/or boiling of the magmatic fluid component, to heterogeneously trap Type IIa, IIb, and IIIa, whereas the metamorphic fluid component underwent fluid immiscibility to heterogeneously trap Type I and Type IIa fluid inclusions. The deduction for the behavior of the metamorphic fluid component was based on the irregular to negative crystal shapes of the Type I and Type IIa fluid inclusions and their occurrence as intragranular planar arrays in quartz vein A.
Furthermore, this fluid evolved to a later HO-2 stage as low to moderate salinity and temperature fluids; a process that was influenced by the contribution from low salinity seawater fluid (−2.3% to −1.5% ) and/or meteoric water (Figure 12), to the mixed metamorphic and magmatic hydrothermal fluids, to form the hypogene magnetite-rich ore at stage HO-2. Given the similarity in fluid salinities of the HO-1 and HO-2 stages, it is believed that the cool seawater fluid mixed with the stage HO-1 fluid prior to the formation of the HO-2 stage fluid, thus implying multiple fluid mixing and interactions with the BIF units during the Nabeba hypogene iron ore formation. It also suggests that the Nabeba BIF was likely submarine during the formation of the hypogene ores. Subsequently, it was only after uplift of the BIF to the subaerial position that the supergene ores were later formed by supergene processes. it was only after uplift of the BIF to the subaerial position that the supergene ores were later formed by supergene processes. Figure 15. Homogenization temperature versus salinity plot of fluid inclusion data associated with hypogene iron ore enrichment in the Nabeba deposit.

Comparison with Other BIF-Hosted Iron Ore Deposits.
A comparison of the results from our study with other major iron ore deposits around the World reveals similarities with those of the Weld Range hypogene iron ore deposit in Australia [25], the Krivoy Rog iron ores in Ukraine [32], and the Serra Norte deposit in the province of Carajás, Brazil [7]. The general models for these deposits implied variations in the BIF-water interaction by earlystage high temperature magmatic ± metamorphic fluids, followed by mixing with low temperature Figure 15. Homogenization temperature versus salinity plot of fluid inclusion data associated with hypogene iron ore enrichment in the Nabeba deposit.

Comparison with Other BIF-Hosted Iron Ore Deposits.
A comparison of the results from our study with other major iron ore deposits around the World reveals similarities with those of the Weld Range hypogene iron ore deposit in Australia [25], the Krivoy Rog iron ores in Ukraine [32], and the Serra Norte deposit in the province of Carajás, Brazil [7]. The general models for these deposits implied variations in the BIF-water interaction by early-stage high temperature magmatic ± metamorphic fluids, followed by mixing with low temperature basinal brines, seawater, or meteoric waters. However, these results differ from those of Mount Tom Price [3] and Paraburdoo [33] in the Hamersley Province, where studies on hypogene orebodies demonstrated the participation of hot (T < 400 • C) and highly saline Ca-rich basinal brines, and cooler, low-salinity meteoric hydrothermal fluids during formation of high-grade magnetite-and hematite-rich ores.

Conclusions
Fluid inclusion and oxygen isotope studies of the Nabeba iron ore deposit have contributed to the understanding of the nature and evolution of the fluids and processes responsible for the hypogene iron ore enrichment. The initial fluids that interacted with the Nabeba BIF was of mixed metamorphic and magmatic hydrothermal origin, giving rise to hypogene hematite-rich ore of the HO-1 stage. Subsequently, influx of low temperature seawater and/or meteoric water resulted in the replenishment of the magmatic-metamorphic fluids, giving rise to the magnetite-rich ore mineralization at the HO-2 stage. In addition, water-rock interaction with the BIF units and phase separation may have also contributed to the hypogene iron ore enrichment. The high-grade hypogene hematite-rich ores formed at the paleodepth of 0.2 to~2 km, while the magnetite-rich ores formed at a shallower paleodepth of about 0.1 km.