The Palaeozoic Nordvik salt dome, about 3×1 km in size, is located on the Nordvik Peninsula, between the Nordvik and Khatanga Gulfs (Fig. 1a). It cuts a large anticline of Triassic to Early Cretaceous age, which constitutes the major structural unit of the eastern part of the peninsula (Fig. 1b). In the core of the anticline, the salt dome is topped by a complex caprock, up to 135-m thick, made of a mélange of various brecciated blocks of Cretaceous age (dark-grey dolomite and dolomitized limestone, dark-grey thin plated limestone, grey anhydrite, greenish-grey diabase, and bluish-grey clay) in a matrix of Middle Devonian gypsum. Drillings have shown that heterogeneities continue at depth, the caprock appearing as a mega breccia formed by a series of blocks, 1–30 cm in size, separated by gypsum breccias and/or gypsum sand [2].

In more detail, three zones have been identified (Upper, Middle and Lower, respectively), which all contain fluorite, typically occurring in up to 1-m long, 1–20-cm wide calcite and gypsum veinlets, cementing meter-size limestone and dolomite blocks (Fig. 1c). Another type of occurrence corresponds to more or less spherical pockets, up to 3–4 cm in diameter. Fluorite crystals, which can reach up to 4–5 cm in size, have usually a rather irregular shape, related to available growing space. Well-formed idiomorphic cubic crystals with octahedral facets are very rare; in most cases only traces of one to three cubic facets can be seen. The colour ranges from colourless through blue to light violet and violet. Rare mineral inclusions are anhydrite, dolomite, and rutile [5]. Most crystals are only semi-transparent, due to a great number of minute fluid inclusions. Only a fraction (less than 10%) are transparent and colourless enough to be suitable for optical purposes, but these are of outstanding quality. For fluid inclusion studies, six fragments from different crystals occurring within the Middle Zone have been investigated (samples 1–6, Table 1). This zone is composed of light-grey and grey gypsum, containing minute clay impurities. Well-formed dolomite rhombohedra and large (0.5–3.0 cm) trapezohedral crystals may also occur.

Fluid inclusions have been studied in double polished plates (typical size: 25×25×0.5 mm), first optically, then by microthermometry using a THMSG-600 ‘Linkam’ freezing-heating stage equipped with Olympus-80 objective at IGEM, RAS. Precision at low temperature (cooling) is about 0.7 °C, reaching about 2 °C at high temperature (heating). Salinity in aqueous fluids was estimated from final ice melting temperature (Tm) and/or halite dissolution temperature (Ts), according to the procedure described, e.g., in [5].

Gas composition was checked beside microthermometry by Raman analysis at the ‘Centre de recherches sur la géologie des matières premières minérales et énergétiques’, UMR CNRS G2R–‘Université Nancy-1’, France, using a spectrometer Dilor RTI-30 (J. Dubessy, analyst).

Bulk inclusion analysis by gas chromatography and ion chromatography has been carried out in the Vernadsky Institute of Geochemistry and Analytical Chemistry, Russian Academy of Science, Moscow, Russia. The first technique used a chromatograph ‘Tzevet-100M’ (O.F. Mironova, analyst). The second one used a chromatograph ‘Tzvet 3006’ (N.I. Savel'eva, analyst). The detection limit of the technique is 0.001 μg μl⁻¹.

Each plate contains up to hundred inclusions, roughly identical in all six studied samples. All inclusions are isolated (primary) or short trail-bound (pseudosecondary) within the crystal host. There is no sign of fluids having been introduced from outside, but also no clear relations with the crystal growth zones. The sub-idiomorphic shape of the fluorite, growing in open cavities, indicates however that all fluids must have been present during the crystallization of fluorite, a fact that does not preclude the possibility of major post-trapping changes after initial trapping (notably for the pseudo-secondary inclusions). Three fluid types have been identified, based on phase relations at room temperature and microthermometric behaviour (Fig. 2).

Type-A corresponds to inclusions, 5 to 35 μm in size, containing at room temperature a cubic halite crystal and a spherical bubble of comparable size, each roughly 10–15% of total volume (Fig. 2a). A dark globule occurs almost systematically within halite, determined as native sulphur [5]. The gas bubble frequently shows an internal meniscus, disappearing (to liquid) upon slight heating (subcritical CO₂, cf. Table 1). Few inclusions contain minute anisotropic solids, which dissolve progressively upon moderate heating at 330–335 °C. Their composition remains unclear.

The behaviour of type-A inclusions upon freezing-heating is quite complex, described on the basis of 22 representative inclusions listed in Table 1 (all temperature symbols identical in the text and table). Upon freezing of the liquid, which occurs suddenly in the −70/−100 °C T-range, the gas bubble frequently disappears, due to the expansion of ice. Then, upon slow-rate heating, CO₂ melting (when seen) occurs around −59 °C (TmCO2). Then, the first droplet of liquid (Teut=eutectic melting) appears at −51/−55 °C, together with the formation of a rounded isotropic phase, which disappears between −11.6 and −12.8 °C. This can be taken as final ice melting (Ts), even if the refractive index of the dissolving solid remains somewhat higher than the enclosing liquid. It can thus be supposed that this ‘ice’ is a salt hydrate, most probably Ca-bearing (antarcticite?). It can also be seen that at these low temperature halite has turned to hydrohalite, which again will revert to halite between 0 and +5 °C. Upon ice melting, the gas bubble does not reappear immediately. Firstly, an irregular white mass can be seen, which disappears progressively (but rapidly) between −26 and −29 °C, together with the reappearance of the spherical bubble. Sometimes, minute bubbles are expelled from this disappearing mass, immediately merging with the large bubble: typically a gas hydrate (clathrate), to be expected in this type of inclusion (Tclath=temperature of clathrate melting).

Upon further heating (T>0°C), the first phase change to be seen is CO₂ homogenization (to liquid, +28.3 to +36 °C=ThCO2), then dissolution of the halite cube (Ts, number in brackets) between 171 and 273 °C, and finally total homogenization (to liquid=Ttoth) between 279 and 332 °C.

Type-G inclusions are gas-dominated, with only a small quantity of liquid water at both inclusion ends (Fig. 2b). Frequently this liquid phase is not visible, making only a thin film of the walls of the cavity. Its volume remains always below about 10% of the total volume. Most inclusions contain a small rounded solid, assumed to be sulphur, which typically adheres to the internal boundary of the gas meniscus (Fig. 2b). Microthermometry data are simpler than for A-type inclusions. In inclusions where a liquid phase is visible at room temperature, a solid appears at about −40 °C (freezing), which melts between −22 and −20 °C, in a somewhat indistinct and erratic way. Like for A-type inclusions, it can be assumed that this solid is a clathrate, but results were too imprecise to be reported in Table 1. Phase transitions are more clearly visible in the large gaseous bubble, with sudden freezing at about −100 °C (solid about 50% cavity volume), final (solid) melting between −58 and −62 °C, in most cases after a short, but distinct melting trajectory (TmCO2, Table 1), and homogenization, always to liquid, between −14.4 and −18.3 °C (ThCO2, Table 1).

Type-M inclusions combine all features of A- and G-types. The volume of the gas bubble ranges from 20 to 70 vol.%, and all solid phases may be present, either halite or sulphur (Fig. 2c and d). Microthermometric data are also comparable to the two other types: eutectic temperature is −55 °C (Teut, Table 1) and CO₂ homogenizes at −13.7 to −7.2 °C into liquid (ThCO2, Table 1) with CO₂ melting temperatures (TmCO2, Table 1) between −58.5 to −57.6 °C. These data indicate that type-M inclusions are nothing but mechanical mixtures between Types A and G fluids, which must have both existed together at the time of the crystal growth.

5.1 Fluid composition

5.2 Hydrothermal fluids: immiscibility, depth of unmixing

Fluid inclusion data confirm without any ambiguity the hydrothermal origin of the optical fluorite occurring within the Nordvik-Taimyr salt dome caprock. Fluorite was deposited at a temperature of about 300 °C from an immiscible fluid mixture of salty brine and dense CO₂. Other optical fluorite deposits are indeed formed at significantly lower temperatures (120 to 200 °C for Amderminskoe, 110–230 °C for Kulinkolonskoe), from low-salinity aqueous fluids (3–7 wt% eq. NaCl) [7]. High formation temperature and high salinity fluids can thus explain the clarity and crystalline perfection of the Nordvik optical fluorite. Moreover, it is suggested that the dense CO₂ fluid, which has originated by unmixing of a homogeneous CO₂–H₂O–NaCl (dominantly) fluid phase at a depth of several kilometres, has significantly contributed to the intrusive character of the salt diapir.