INTRODUCTION

Chlorine, along with H2O, CO2, S, and F, is one of the most important volatile components dissolved in magmas. In magma–fluid systems, effects of Cl on phase equilibria (Filiberto and Treiman, 2009; Aranovich, 2017), the evolution of magmatic fluids (Lukanin, 2015, 2016), and metasomatic transformations (Aranovich and Safonov, 2018; Kusebauch et al., 2015; Safonov and Aranovich, 2014) are disproportionally important, given the relatively low planetary abundance of this element (Patiño Douce et al., 2011). Chloride ligands control the solubility of most ore metals in fluids of magmatic mineral deposits (Rubtsova et al., 2023 and references therein). The occurrence of Cl and its soluble salts also controls the pH of magmatic fluids (Holland, 1972; Ryabchikov, 1975). Variations in this parameter due to magmatic differentiation processes at the cooling and/or ascent of melts, as well as those induced by interaction between magmatic fluid, host rocks, and meteoric solutions, principally affect metasomatic and transport processes and the deposition of ore metals.

Chlorine solubility in silicate melts has been studied by many experimentalists (Webster et al., 2015; Dolejš and Zajacz, 2018; Chevychelov, 2019; Hsu et al., 2019; Thomas and Wood, 2023; and references therein). A large set of experimental data on systems of complex composition was used to derive empirical equations for describing Cl solubility as a function of temperature, pressure, and composition of felsic (Lukanin, 2015; Dolejš and Zajacz, 2018) and mafic (Webster et al., 2015; Thomas and Wood, 2023) melts. Witham et al. (2012) have proposed a model for Cl partitioning between mafic melts and C–O–H–S–Cl fluid. The partitioning of cations between melts and Cl-bearing fluids is much less studied, despite this problem is extremely important for modeling the compositional evolution of fluid segregated during magma degassing. No models have been proposed so far that would describe Cl solubility in melts and could be utilized in currently widely used petrological program packages for calculating phase equilibria in natural systems. This significantly hampers (or even makes impossible) reasonably accurate simulations of the evolution of magma–fluid systems.

This publication presents our newly acquired experimental data on Cl solubility in haplobasalt melts depending on their composition and chloride concentrations with known activities of the salts in the fluid phase. The paper also reports estimated Ca and Na partitioning between melt and aqueous saline fluid. Considered together with previous experimental data on the melting of model granite in the presence of aqueous (K,Na)Cl solution (Aranovich et al., 2013), these data were employed to develop a thermodynamic model for silicate melts containing K, Ca, and Na chlorides and to evaluate Ca and Na partitioning between basalt melt and fluid and K and Na partitioning between granite melt and fluid, which is principally important to reproduce the compositional evolution of fluid segregating from magma at its crystallization and to understand the nature of metasomatic processes occurring when host rocks interact with magmatic fluids.

EXPERIMENTAL EQUIPMENT AND METHODS

Experimental Equipment

The experimental runs were carried out using IHPV, with Ar as the pressure medium. The pressure was measured by a spring-tube manometer accurate to ±20 MPa. All runs were carried out at 400 MPa. The temperature was measured by two Pt–PtRh10 thermocouples, whose hot junctions were positioned at the top and bottom ends of the copper holder into which four to six platinum capsules (50 × 5 × 0.2 mm) with starting materials were placed. Temperature was measured and controlled in the course of the runs with a MINITHERM temperature responsive controller and was determined accurate to ±5°C. The runs were conducted within the temperature range of 900–1200°C. The run duration was 5 days, which is sufficient to reach equilibrium in the system (Shaposhnikov and Aranovich, 2015).

Starting Materials

The starting materials were glasses compositionally approximating the eutectics diopside (Di)–anorthite (An) (Di58An42, wt %) with varied quartz contents (0–20 wt % Qtz), Di–albite (Ab) (Di11Ab89, wt %), and Di11Ab89 + 2 wt % NaCl. The glasses of corresponding composition were prepared in open Pt capsules at 1450°C in a high-temperature tube furnace (Borisov and Aranovich, 2019) for 24 h from mixtures of Na2CO3, Al2O3, SiO2, CaCO3, Mg(OH)2, and NaCl (all chemically pure), which were carefully mixed in desired proportion in an agate mortar under acetone. The glasses thus made (25–30 mg) and desired amounts of deionized H2O and salt (NaCl or CaCl2, which had been carefully dried before loading) were loaded into Pt capsules using the weight technique. All weighings were done using a Mettler-Toledo analytic balance, which guarantees a reproducibility to ±0.02 mg. Upon their loading, the capsules were arc-welded in Ar and weighed to check possible weight losses on welding. The experimental parameters and weight proportions of the starting materials are summarized in Table 1. Four to six capsules were simultaneously run in the IHPV. The runs were quenched by rapidly dropping the capsules into cold water. After the runs, the capsules were dried and weighed to check for possible weight losses.

Table 1.   Parameters of experimental runs on Cl solubility in haplobasalt melts at 4 kbar
Full size table

Analysis of Experimental Products

The content of the capsules after runs (silicate glasses and solutions) was carefully collected into Petri dishes. The solutions were collected by decanting into measuring tubes and then diluted to an aliquot of 5 ± 0.1 mL with a required amount of deionized H2O. The solutions were analyzed for Na and K cations by flame photometry on a FPA-2-01 flame photometer in gas–air flame. Ca was analyzed by atomic absorption spectrometry on a Shimadzu 7000 spectrometer in acetylene–air flame. The glasses (quenched melts) were dried in a muffle furnace at 110°C and used to make pellets with epoxy resin. The pellets were carefully polished and then analyzed by EPMA on a JEOL Superprobe 8200 with five spectrometers at the Institute of Geology of Ore Deposits, Petrography, Mineralogy and Geochemistry, Russian Academy of Sciences. The analysis was carried out at an accelerating voltage of 15–20 kV, sample current of 10 nA, with the beam focused to a spot 5–10 μm in diameter. Natural diopside and sanidine were the standards in analysis for Si, Al, Ca, and K; Na2BeSi2O6 was the standard for Na (with a correction introduced for the possible Na losses; see Andreeva et al., 2018); and synthetic atacamite Cu2Cl(OH)3 was used for Cl.

EXPERIMENTAL RESULTS

In most of the runs, silicate melts were quenched into homogeneous glasses that contained variable amounts of fluid inclusions, which had been entrapped in the course of the runs and on quenching (Fig. 1a). The products of some of the runs at 900 and 1000°C contained minor amounts (no more than 10%) of quenched clinopyroxene crystals of composition close to diopside (Fig. 1b). The products of the experimental runs of Set 4, with Di58An42 + Qtz starting glass (Table 1), contained up to 80–90 vol % quenched clinopyroxene grains of the total volume of the run product (visual evaluations in BSE images, Fig. 1c). The quenched nature of the clinopyroxene is evident from the morphology of the feather-shaped (Fig. 1b) and dendritic (Fig. 1c) grains and from the specific composition with unequilibrated Ca/Mg > 1.

Fig. 1.
figure 1

BSE images of glasses in the products of runs (a) 2-8, (b) 5-3, and (c) 4-5. Brighter grains in Figs. 1a and 1b are quenched clinopyroxene of composition close to diopside. Images in Figs. 1b and 1c were taken at the same magnification.

Full size image

The composition of the glasses (EPMA data) and equilibrium solutions are reported in Table 2. For the runs whose products contained quench clinopyroxene, the real composition of the melt was recalculated with regard to the amount of the quench grains. Of course, this introduced an additional inaccuracy into the determination results, particularly in the runs with high contents of quench phases (Fig. 1c, these runs are marked with asterisks in Table 2).

Table 2.   Glass composition (elements in wt %) and Ca mole fraction in glass (m) and fluid (fl) in the products of experiments on Cl solubility in haplobasalt melts at 4 kbar
Full size table

DISCUSSION

The Ca and Na partition between the melt and fluid is displayed in Fig. 2. The partition is obviously nonideal and weakly depends on temperature. The partition isotherm of 1200°C, which is drawn provisionally, reasonably well describes the experimental results and is very close to the curve showing exchange between fluid and plagioclase at 700°C according to (Shmulovich and Graham, 2008), which indicates that the mixing properties of the components CaAl2Si2O8 and NaAlSi3O8 in haplobasalt melt and the anorthite and albite end members in the plagioclase solid solution are similar. The significant scatter of the experimental results at different temperatures (likely due to analytical uncertainties of the phases and, particularly, the occurrence of quench crystals) did not allow us to identify any clear temperature dependence of the partitioning.

Fig. 2.
figure 2

Ca and Na partitioning between haplobasalt melt and aqueous chloride fluid: experimental data at 4 kbar. Legend: temperature is in °C, red line is the isotherm of 1200°C (provisional data); the thin continuous line shows equal Ca/(Ca + Na) values. The error brackets are shown for the runs at 1200°C. Red circles are data of experiments on Ca–Na exchange between fluid and plagioclase (Pl) according to (Shmulovich and Graham, 2008).

Full size image

An important result obtained in this study is the identification of a dependence of Cl concentration in haplobasalt melt on the Ca/(Ca + Na) ratio of this melt (Fig. 3a) and coexisting fluid (Fig. 3b): Cl solubility increases nonlinearly from 0.2–0.3 wt % in the Ca-poor melts to >2 wt % in the Na-poor ones. This result is in agreement with experimental data obtained at a lower pressure (Chevychelov, 2019; Chevychelov and Suk , 2003) and with the empirical equation for the dependence of melt chlorine capacity on its composition (Thomas and Wood, 2023). This result indicates that at the deep-sitting crystallization of basalt magma with a constant initial Cl concentration, the first fluid portions segregating from the magma should be enriched in Na, i.e., bring about the “early” sodic profile of the metasomatic processes at interaction with host rocks.

Fig. 3.
figure 3

Dependence of Cl solubility in haplobasalt melt on the Ca/(Ca + Na) ratio of (a) the melt and (b) fluid and (c) Cl solubility in haplobasalt melt depending on the Mg(Mg + Ca) ratio of the melt. Symbols are experimental data (with error bars in Fig. 3a). Red curves in Figs. 3a and 3b are correlations according to (a) quadratic and (b) cubic equations.

Full size image

Chlorine concentration in the melts negatively correlates with the Mg/(Mg + Ca) ratio (Fig. 3c). This correlation, although not clearly discernible in our runs, indicates that the early crystallization of relatively Mg-rich phases should lead to gradual Cl enrichment in the residual melt.

THERMODYNAMIC MODEL OF Cl-BEARING SILICATE MELTS

Experimental data of this study and those of (Aranovich et al., 2013; see also SupplementaryFootnote 1 1, ЕSM_1 and ESM_2), make it possible to calculate parameters of thermodynamic models for Cl-bearing basalt and granite melts, because the experiments were carried out in the presence of fluid with known mixing properties (Aranovich and Newton, 1996, 1997; Ivanov et al., 2019; Ivanov, 2023), and the equilibrium compositions of both phases (glasses, i.e., the quenched melt, and fluid) have been directly analyzed.

Studies of Cl-bearing silicate melts by the structure-sensitive methods (Dalou and Mysen, 2015; Evans et al., 2008) have shown that Cl in the melt forms clusters with major network-modifying cations. In view of this, we assumed that the standard states of chlorides in both melt and fluid are molten pure salts (NaCl, KCl, and CaCl2). Equilibrium condition of chlorides in melt and fluid are then written as

$${{a}_{i}}\left( m \right) = {{a}_{i}}\left( {fl} \right),$$
(1)

where ai is the thermodynamic activity of salt i in melt (m) or fluid (fl)

$${{a}_{i}} = {{X}_{i}}\exp \left[ {{{G}^{{{\text{ex}}}}}\left( i \right)/\left( {RT} \right)} \right].$$
(2)

In Eq. (2), Хi is the mole fraction of chloride, G ex(i) is the excess partial Gibbs free energy of salt i, T is the temperature in K, and R is the universal gas constant, 8.314 J/(K mol).

The equations for Gex(i) in Н2О–NaCl–KCl and Н2О–NaCl–CaCl2 are presented in (Aranovich and Newton, 1996, 1997) and (Ivanov et al., 2019; Ivanov, 2023), respectively. The activity values of fluid components in the experiments are reported in Supplementary 1, ESM_2.

The total Cl concentrations in the experimental glasses were no higher than 2.15 wt %, i.e., the melts can be confidently regarded as diluted solutions of the chlorides. To describe their mixing properties, we applied empirical equations that were close to Darken’s formalism (Darken, 1967; Aranovich, 1991). Because the variations in concentrations of all major components except Na and Ca in basalts and Na, K, and H2O in granites are very minor, these equations can be written as

$$G_{1}^{{{\text{ex}}}} = {{J}_{1}} + {{X}_{1}}{{W}_{{1(1)}}} + X{}_{2}W_{{2(1)}}^{{}} + {{X}_{{{{{\text{H}}}_{{\text{2}}}}{\text{O}}}}}{{W}_{{{{{\text{H}}}_{{\text{2}}}}{\text{O}}(1)}}}$$
(3a)
$$G_{2}^{{{\text{ex}}}} = {{J}_{2}} + {{X}_{1}}{{W}_{1}}_{{(2)}} + X{}_{2}{{W}_{2}}_{{(2)}} + {{X}_{{{{{\text{H}}}_{{\text{2}}}}{\text{O}}}}}{{W}_{{{{{\text{H}}}_{{\text{2}}}}{\text{O}}(2)}}}.$$
(3b)

In Eqs. (3a) and (3b), \(G_{i}^{{{\text{ex}}}}\) is the partial excess free Gibbs energy of component i, subscript indexes 1 and 2 denote salt components of melts (NaCl and CaCl2 for basalts and NaCl and KCl for granites), Xi is the mole fractions of corresponding melt components, Wi(j) is energy parameters (kJ/mol) pertaining to the excess energy of component i that characterize the total interaction of salts 1 and 2 with major components of the melts, and Ji are constants corresponding to the Henry constant. In processing the data on granite, this constant was represented according to the Gibbs–Helmholtz equation as a linear function of temperature (in K) and pressure (in kbar)

$${{J}_{i}} = {{A}_{i}} + {{B}_{i}}T + {{C}_{i}}P.$$
(4)

The experiments on Cl solubility in basalts were conducted at a single pressure P = 4 kbar and almost constant H2O content in the melt (about 5–6 wt % at 900–1200°C and 4 kbar according to the equation in Papale et al., 2006). Because of this, the last summands in Eqs. ( (3a), (3b), and (4) were omitted in fitting these data.

We assumed in the calculations that the mole fractions of salts in the melts were proportional to the bulk molar ratios of the respective cations, which leads to the following relations for the calculation of the mole fractions of salts in haplobasalt melt:

$${{X}_{1}} = {\text{ }}X{{\left( {{\text{NaCl}}} \right)}^{m}} = {\text{Na/}}{{\left( {{\text{Na}} + {\text{Ca}}} \right)}^{m}} \times {{\left[ {{\text{Cl}}} \right]}^{m}},$$
(5a)
$${{X}_{2}} = X{{\left( {{\text{CaC}}{{{\text{l}}}_{{\text{2}}}}} \right)}^{m}} = {\text{Ca/}}{{\left( {{\text{Na}} + {\text{Ca}}} \right)}^{m}} \times {{\left[ {{\text{Cl}}} \right]}^{m}}.$$
(5b)

The expression for the mole fractions of salts in model granite melt assumes the form

$${{X}_{1}} = X{{\left( {{\text{NaCl}}} \right)}^{m}} = {\text{ Na/}}{{\left( {{\text{Na}} + {\text{K}}} \right)}^{m}} \times {{\left[ {{\text{Cl}}} \right]}^{m}},$$
(5c)
$${{X}_{2}} = X{{\left( {{\text{KCl}}} \right)}^{m}} = {\text{K/}}{{\left( {{\text{Na}} + {\text{K}}} \right)}^{m}} \times {{\left[ {{\text{Cl}}} \right]}^{m}}.$$
(5d)

In Eqs. ( (5a)–(5d), symbols of elements denote their mole fractions in melts normalized to eight oxygen ions.

Experimental data in Table 2 and Supplementary 1 were used in evaluating parameters involved in Eq. (3) with regard to (4) and (5a)–(5d). The calculations were done by means of linear regression, because Eqs. (3a) and (3b) are linear with respect to the parameters to be refined. The results are presented in Table 3, and Supplementary 2 lists all statistical characteristics of the regression. The parameters of the basalts have large uncertainties and should be considered as tentative (the multiple regression coefficients are 0.82 and 0.68 for NaCl and CaCl2), which is related to the aforementioned uncertainties in the determined compositions of the experimental melts. Also, the equation used to calculate the activities of salts in Н2О–NaCl–CaCl2 fluid (Ivanov, 2023) may probably need to be refined (Makhluf et al., 2023). The parameters calculated for model granite are much more reliable (the multiple regression coefficients are 0.94 and 0.91, respectively, for NaCl and KCl, see Supplementary 2). Selected calculation results on granite obtained with parameters from Table 3 are presented in Figs. 4 and 5.

Table 3. Thermodynamic parameters (kJ/mol) involved in Eq. (3)
Full size table
Fig. 4.
figure 4

Cl concentration in model granite melts at 900°C and 10 kbar depending on (a) the H2O content and (b) K/(K + Na) ratio. Blue rhombs are experimental points (Aranovich et al., 2013; see Supplementary 1, ESM_1, ESM_2), and red curves are calculations.

Full size image
Fig. 5.
figure 5

Isobaric–isothermal (800°C, 5 kbar) section of the phase diagram of the pseudoternary system albite Ab–Н2О (w)–NaCl (hlt). Fields of different color show the stability of various phase associations. Symbols: F(salt) is NaCl-bearing fluid, and Melt is NaCl-bearing melt.

Full size image

For haplogranite, Cl solubility in melt seems to show a complex dependence on the melt composition (Figs. 4a, 4b): it slightly increases with increasing H2O content at a high pressure (Fig. 4a), which is opposite to the dependence identified in shallow-depth melts (at 1–3 kbar; e.g., Webster et al., 2015). The most likely reason for this is drastic changes in the activity–composition relations in the saline aqueous fluid (Aranovich and Newton, 1996, 1997). This trend also indicates that salts dissolved in melt partly occur in the form of clusters related to molecular H2O.

The decrease in Cl solubility with an increase in the K/(K + Na) ratio of the melt (Fig. 4b) is in good agreement with the results obtained in low-pressure (1–3 kbar) experiments (Chevychelov, 2019). An important implication of this result is that, when Cl-bearing granite magmas reach saturation, fluids relatively enriched in K2O are the first to separate from these magmas, and these fluids can induce potassic metasomatism when interacting with host rocks.

Figure 5 shows an isobaric–isothermal (800°C, 5 kbar) section of the phase diagram of the system albite Ab–Н2О (w)–NaCl (hlt), which was calculated using the Perple_X program complex (Connolly, 2005, 2017, ver. 7.1.0) with the thermodynamic database (Holland and Powell, 2011), mixing parameters for Ab melt–H2O from (White et al., 2014), and a model for H2O–NaCl fluid according to (Aranovich and Newton, 1997). According to the model [Eqs. (3a) and (3b)] and parameters from Table 3, the expression for NaCl activity in this system assumes the form (T is temperature in K, and P is pressure in bar)

$$\begin{gathered} RT\ln a{{\left( {{\text{NaCl}}} \right)}^{m}} = 14{\kern 1pt} 072 + 13.851T - 0.83P \\ + \,\,75{\kern 1pt} 341X{{\left( {{\text{NaCl}}} \right)}^{m}}-3561X{{\left( {{{{\text{H}}}_{{\text{2}}}}{\text{O}}} \right)}^{m}} \\ + \,\,RT\ln X{{\left( {{\text{NaCl}}} \right)}^{m}}. \\ \end{gathered} $$
(6)

The expressions for the partial excess mixing Gibbs energy for Ab and H2O were appended with terms pertaining to interaction with NaCl according to a regular solution model.

The diagram in Fig. 5 illustrates the fairly complex relations even in this very simple fluid–mineral system with an aqueous saline solution. It is important to mention the very narrow stability field of albite melt: the maximum NaCl concentration in the fluid at which melting is possible (at given P–T parameters) is X(NaCl) = 0.15, Х(H2O) = 0.85, and hence, no Cl can be incorporated into melt in the absence of H2O [X(H2O) = 0] (compare with the diagram in Webster et al., 2015, Fig. 1). NaCl concentration in the melt decreases with increasing H2O content (Fig. 6a) only in the region where fluid occurs in equilibrium with melt (±albite), which is obviously caused by the decrease in NaCl concentration in the coexisting fluid. Therewith the maximum (and constant) salt concentration in the melt (0.76 wt % NaCl) is reached within the invariant (at constant P and T) field F(salt)–Melt–Ab (red rhomb in Fig. 6a).

Fig. 6.
figure 6

Relation between water and salt concentrations in fluid-saturated melts in the system Ab–NaCl–H2O at (a) 5 kbar, 800°C and (b) 1200°C. Symbols are calculated data, curves are interpolations. The rhomb in Fig. 6a is the composition in the invariant albite–melt–fluid field.

Full size image

To elucidate in more detail how H2O content affects NaCl solubility in melt, we have calculated concentrations of these components in albite melt at 5 kbar and 1200°C, a temperature at which silicate melt is stable at any salt concentration in the fluid, up to anhydrous NaCl melt (Tenner et al., 2007). Figure 6b shows the calculation results for an arbitrarily selected bulk salt concentration, X(NaCl)= 0.5, and variable H2O content, Х(H2O) = 0 to 0.4. As seen in this figure, the dependence of salt concentration in albite melt on the water content at these calculation conditions is principally different from that in Fig. 6a: at a low H2O content (no higher than 1 wt %), the NaCl concentration is slightly lower than in equilibrium with anhydrous salt melt, and this concentration then slowly increases. Therewith the H2O/NaCl ratio rapidly increases (and tends to infinity for pure H2O fluid).

EVOLUTION OF MAGMATIC FLUID SEGREGATING FROM GRANITE

Figure 5 shows that the field of fluid-undersaturated melts [without F(salt] is very small and is bounded by the maximum salt concentration X(NaCl) = 0.04 (which corresponds to approximately 0.2 wt % in recalculation to Cl). Even at this fairly low Cl concentration (this value is close to that most commonly found in melt inclusions, e.g., Kovalenko et al., 2000), the first fluid portions segregating from melt are high-concentration brines with X(NaCl)= 0.15, which corresponds to approximately 36 wt % NaCl. Although the fraction of brine in the overall budget of segregated fluid is relatively small, its ability to extract many ore metals can be very significant (e.g., Tattitch and Blundy, 2017; Kouzmanov and Pokrovski, 2012). As the melt fraction decreases, the fluid segregated from it should become progressively more diluted (Fig. 6a), up to almost pure H2O during final crystallization stages.

Extensive empirical data on the composition of melt and fluid inclusions in minerals highlight the complex evolutionary histories of mineral deposits related to granitoid magmatism (see recently published reviews in Blundy et al., 2021; Goldfarb and Pitcairn, 2022). As an example, Fig. 7a shows estimates of H2O and silica concentrations in melt inclusions hosted in quartz from granites of the Verkhneurmiiskii massif of the Badzhal volcano-plutonic zone, which hosts tin–tungsten deposits (Bortnikov et al., 2019). The diagram shows two clearly distinct trends: one with a positive (A) and the other with a negative (B) correlation between these parameters. Trend A is readily explained by the successive capture of inclusions during the crystallization differentiation in a deep magma chamber (Fig. 7b): the early quartz, which crystallized at 780–760°C, entrapped melt portions that contained about 7 wt % Н2О. As the melt content decreased to 5%, its H2O concentration increased to 11 wt %, the maximum possible value at these P–T parameters. Trend B in Fig. 7a can be explained by the continuing growth/recrystallization of the quartz when the magma ascended from a deep chamber (at approximately 16–18 km) to shallower depths. This ascent was associated with the segregation of CO2, brines, and aqueous fluid, which progressively depleted in salts, from the magma. This scenario is also well consistent with the variations in Cl concentration in the melt inclusions (Fig. 8): Cl enriched the melt during its deep-seated crystallization and reached a saturation concentration (0.2 wt %), after which it was segregated from the melt in the form of brine and/or more and more diluted solution.

Fig. 7.
figure 7

Correlation trends for H2O and silica concentrations in (a) quartz-hosted melt inclusions in granites from the Verkhneurmiiskii massif (Bortnikov et al., 2019) and (b) model calculations of the isobaric (Р = 5 kbar) crystallization of granite melt. The quartz fraction in Fig. 7b is shown relative to the whole quartz amount that crystallized from the melt. Calculation in Fig. 7b pertain to trend A in Fig. 7a.

Full size image
Fig. 8.
figure 8

Correlation trends for Cl and silica concentrations (wt %) in melt inclusions. Lozenges pertain to the composition of the earlier inclusions shown in Fig. 7a and, supposedly, corresponding to the respective evolution in a deep chamber, squares are inclusions captured during decompression.

Full size image

CONCLUSIONS

New experimental data have been obtained on Cl solubility in haplobasalt melts. It has been demonstrated that an increase in NaCl concentration in the fluid results in a decrease in Cl solubility in the melt. The data obtained on Ca and Na partitioning between melt and fluid make it possible to model the evolution of the Ca/Na ratio in the course of crystallization of basaltic melts. Cl solubility in haplogranite melt decreases with increasing K/Na ratio in the melt and fluid. We have also found out that Cl solubility in model granite under a high pressure (10 kbar) increases with increasing H2O content up to a maximum at an H2O mole fraction of about 0.6. Inasmuch as the experiments were conducted in the presence of aqueous chloride fluids whose thermodynamic properties are known, we could calculate the thermodynamic parameters of solute species (NaCl, KCl, and CaCl2) in silicate melts. Calculations for the simplest magma–fluid system Ab–H2O–NaCl highlight the complex character of the phase relations and, hence, also the evolution of H2O and NaCl concentrations in aluminosilicate melt. We used published data on the composition of melt inclusions in quartz from granites of the Verkhneurmiiskii massif of the Badzhal volcano-plutonic zone (Bortnikov et al., 2019) to demonstrate the complex evolution of the magmatic–fluid system.