You are on page 1of 44

Accepted Manuscript

Formation of hydrothermal tin deposits: Raman spectroscopic evidence for an


important role of aqueous Sn(IV) species

Christian Schmidt

PII: S0016-7037(17)30674-9
DOI: https://doi.org/10.1016/j.gca.2017.10.011
Reference: GCA 10516

To appear in: Geochimica et Cosmochimica Acta

Received Date: 15 May 2017


Accepted Date: 12 October 2017

Please cite this article as: Schmidt, C., Formation of hydrothermal tin deposits: Raman spectroscopic evidence for
an important role of aqueous Sn(IV) species, Geochimica et Cosmochimica Acta (2017), doi: https://doi.org/10.1016/
j.gca.2017.10.011

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

Formation of hydrothermal tin deposits: Raman spectroscopic evidence

for an important role of aqueous Sn(IV) species

Christian Schmidt*

GFZ German Research Center for Geosciences, Section 4.3, Telegrafenberg, 14473 Potsdam,

Germany

* Corresponding author. Email: Christian.Schmidt@gfz-potsdam.de

Tel.: +49.331.288.1406;

Fax: +49.331.288.1402

1
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

Abstract

The speciation of tin and the solubility of cassiterite in H 2O+HCl were determined at

temperatures to 600 °C using in situ Raman spectroscopy. In addition, information on the fluid-

melt partition of Sn was obtained at 700 °C and indicated a preference of the fluid only at HCl

concentrations that are much higher than in fluids exsolved from natural felsic melts. Dissolution

of cassiterite generally resulted in formation of Sn(IV) species unless reduced conditions were

generated by hydrogen permeation or carbohydrates in the starting material. The prevalent

aqueous Sn(IV) species was [SnCl4(H2O)2]0, with additional [SnCl3(H2O)3]+ and [SnCl5(H2O)]−.

The only detectable Sn(II) species was very likely [Sn(II)Cl3]−. Cassiterite solubility increased

with HCl concentration and was generally high in H2O+HCl fluids, with no strong dependencies

on temperature, pressure, or the oxidation state of tin in the fluid. The Sn(IV) concentrations at

500 and 600 °C determined from the integrated 1[Sn(IV)–Cl] band intensity are in good

agreement with literature data on the cassiterite solubility in H2O+HCl at oxygen fugacities

along the hematite-magnetite buffer. The combined results from previous experimental studies

and this study demonstrate that HCl molality is a crucial parameter for hydrothermal

mobilization and transport of tin and for cassiterite precipitation, and that pH, pressure and

temperature are less important. Current models on hydrothermal tin deposit formation need to be

augmented to include Sn(IV)–Cl complexes as significant tin-transporting species. Irrespective

of the oxidation state of tin in the fluid, cassiterite precipitates due to reaction of the

hydrothermal fluid with the wall rock (greisen or skarn formation), dilution (mixing with

meteoric water) or a decrease in the HCl activity in the aqueous liquid by boiling. A redox

reaction is only required for tin transported as Sn(II) to be converted to Sn(IV).

2
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

Keywords: tin; solubility; speciation; hydrothermal fluids; Raman spectroscopy

3
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

1. INTRODUCTION

Models of the hydrothermal formation of tin deposits generally assume that tin is

predominantly transported in the ore-forming fluids as Sn(II) chloride complexes (Heinrich

1990; Halter et al. 1998; Müller and Seward 2001). This implies that precipitation of the by far

most abundant tin ore mineral cassiterite, SnO2, requires redox reactions and H+ consumption on

a large scale via

(1)

(Heinrich 1990). However, the assumption of the prevalence of Sn(II)–Cl complexes in

hydrothermal fluids rests mainly on the interpretation of data on cassiterite solubility at

controlled oxygen fugacity (ƒO2) from quench experiments using a double capsule technique

(Kovalenko et al. 1986; Wilson and Eugster 1990; Taylor and Wall 1993). Such interpretations

have intrinsic uncertainties, e.g. because they are based on quench pH or on molality-molality

and not on activity-activity diagrams. Moreover, they assume that the measured molality of tin in

the quench fluid is the same as at high temperature and that the alloying of tin with gold or

platinum of the capsules does not affect the data. It has not yet been demonstrated by in situ

spectroscopy if tin occurs indeed predominantly as Sn(II) in a high-temperature aqueous fluid in

equilibrium with cassiterite.

There are several arguments that suggest that tin may well occur abundantly as Sn(IV)

chloride complexes in hydrothermal fluids. Firstly, in the temperature range of such fluids, it

requires fairly low ƒO2 conditions to reduce SnO2, which are lower than typical crustal values

between those defined by the assemblages hematite-magnetite (HM) and a few log units below

nickel-nickel oxide (NNO) (Fig. 1). Secondly, Sn(II) chloride solutions oxidize readily and can

be used to reduce Fe(III) in chloridic solutions

4
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

(2)

That is, Sn(IV) chloride complexes can be stable in crustal Fe(II)-bearing chloridic hydrothermal

fluids (e.g., Chou and Eugster 1977). Lastly, redox reactions are not indispensable for

hydrothermal cassiterite formation, because the cassiterite-precipitating reactions in greisens and

skarns can be written without a change in the oxidation state of tin if present as aqueous Sn(IV)

complexes. In this context, it is important to note that there is a lack of convincing evidence for

such redox reactions in numerous instances. The strata-bound Kellhuani tin deposits in Bolivia

occur in the quartzite units of the Catavi formation, and cassiterite precipitated together with

quartz and tourmaline at 410–500 °C, 50–100 MPa, later sulfides are subordinate in the ore,

whereas hematite is locally abundant (Lehmann 1985). In a careful fluid inclusion study of a

quartz-cassiterite vein with minimal wall-rock alteration (Yankee Lode, Mole Granite, eastern

Australia), Audétat et al. (1998) demonstrated a relationship between cassiterite deposition and

mixing of magmatic and meteoric fluids. Sandstone or quartzite host rocks do not provide

enough potential redox partners but cassiterite-bearing quartz veins occur in such wall rocks, e.g.

at the deposits Dajing, China (Wang et al. 2006) and Piaotang, China (Zhang et al. 2017). Other

redox reactions have been proposed if there is no significant fluid-induced alteration of the wall

rock. Coprecipitation of cassiterite and arsenopyrite has been interpreted as oxidation of Sn(II)

complexes coupled with reduction of aqueous As(III) or As(V) species

(3)

(Heinrich and Eadington 1986). In CO2-rich fluids, cassiterite precipitation may occur due to

reduction of carbon dioxide to methane (Giggenbach 1980)

(4)

5
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

However, CH4 or even CO2 are often not detectable in fluid inclusions in cassiterite (e.g.,

Campbell and Panter 1990; Schmidt and Leeder 1992; Mohamed 2013). Furthermore, there is

frequently no coprecipitation of cassiterite and arsenopyrite. Instead, cassiterite forms earlier

than the sulfides and sulfarsenides (e.g., Kelly and Turneaure 1970; Kelly and Rye 1979;

Schmidt and Leeder 1992), sometimes later than arsenopyrite (Kelly and Rye 1979; Pinto et al.

2015), or there are little sulfides and sulfarsenides or none at all in the ore (e.g., Kelly and

Turneaure 1970; Schmidt and Leeder 1992; Audétat et al. 1998; Mohamed 2013). At the

Panasqueira mine, Portugal, there are at least two generations of cassiterite in fracture fillings

that may or may not also contain arsenopyrite (Pinto et al. 2015).

In addition, it can be concluded from some experimental data that precipitation by

oxidation is not an efficient mechanism. At 500 °C, the solubility of cassiterite in H 2O+HCl at

ƒO2 controlled by the HM assemblage is only by a factor of about two lower than that along the

NNO buffer (Wilson and Eugster 1990). Similarly, Taylor and Wall (1993) observed a decrease

in the solubility of cassiterite by about 0.5 log units at 700 °C and 200 MPa when ƒO2 was

increased from the quartz-fayalite-magnetite (QFM) buffer to QFM+1.5. Kovalenko et al. (1986)

report cassiterite solubility data at ƒO2 = NNO at 500 °C that are in very good agreement with

those determined by Wilson and Eugster (1990), but the determined solubilities at higher ƒO2

(MnO2-Mn2O3 buffer) were 1.5 to 2 orders of magnitude lower.

In these studies, the oxidation state of tin in the fluid at elevated temperature was inferred

and not measured. Direct, in situ, determination of the actual tin species in cassiterite solubility

experiments is still missing. Such information is essential because knowledge of the Sn(II) and

Sn(IV) speciation is required to identify the reactions that govern hydrothermal tin transport and

cassiterite precipitation. There is consensus in the literature that tin is transported as

6
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

chlorocomplexes. At high concentrations, Raman spectroscopy permits easy and in-situ

identification of such complexes even in high-temperature fluids confined in optical cells

(Watenphul et al. 2014), and reliable information on tin(IV) complexation at ambient conditions
119
has been provided by a study on aqueous SnCl4 solutions combining Sn NMR and Raman

spectroscopy (Taylor and Coddington 1992). At all conditions, Sn(IV) was present in

octahedrally coordinated complexes [SnCl6−n(H2O)n]n−2 (n = 0−6). Aqueous tin(II) halide

solutions are more difficult to study because they are susceptible to oxidation. Raman spectra of

Sn(II) chloride solutions with added HCl showed Sn(II)Cl3− at ambient conditions (Woodward

and Taylor 1962; Taylor 1989).

Here, Raman spectroscopy was used for in situ determination of the aqueous tin

speciation at temperatures to 600 °C in fluids produced by reacting cassiterite with H 2O+HCl in

the sample chamber of a hydrothermal diamond-anvil cell. In addition, the solubility of

cassiterite was determined from band intensities. Moreover, information on fluid-melt

partitioning of tin and on the tin speciation in the fluid was obtained at 700 °C from experiments

using Macusani glass, cassiterite and H2O+HCl solutions.

2. METHODS

The experimental setup is shown schematically in Fig. 2. A hydrothermal diamond-anvil

cell (HDAC) (Bassett et al. 1993) was used for all experiments. The cell was modified for

Raman spectroscopic analyses of fluids (Schmidt 2009; Schmidt and Chou 2012) and was

equipped with ultra-low fluorescence grade diamond anvils with a culet diameter of 0.9 mm.

Iridium gaskets were used to ensure that reaction of the H2O+HCl fluid with the gasket remained

insignificant even at 600 °C, the highest temperature of the experiments of this study. The initial

7
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

diameter of the hole in the center of the gasket that formed the sample chamber was about 400

µm, and the initial thickness was 125 µm. The temperature in the sample chamber was measured

with K-type (Ni-Cr / Ni) thermocouples with their junctions attached to the diamonds and close

to the sample. It was calibrated using repeated measurements of the temperature of the α–β

transition of quartz at atmospheric pressure (574 °C) and the ice melting temperature of distilled

water of 0.0 °C in the presence of a trapped air bubble (Schmidt and Watenphul 2010). Resistive

heating by two independent nickel-chromium coils and maintenance of an attained target

temperature within ±0.3 °C was done using TDK Lambda Z60 power supplies and Eurotherm®

2408 temperature controllers.

The sample chamber of the HDAC was loaded with a standard solution (10 mg Sn(IV) /

ml in 10% HCl, Alfa Aesar) for calibration of the Sn(IV)–Cl Raman band intensity, or, for the

experiments at elevated temperatures, with a chip of inclusion-free colorless cassiterite from the

Viloco (Araca) Mine, Bolivia, a chip of synthetic zircon, and aqueous HCl solution. For the tin

fluid-melt partitioning experiments, the sample chamber contained an additional piece of

Macusani glass. Pichavant et al. (1987) and London et al. (1988) reported the composition of this

peraluminous rhyolitic glass, which contains about 200 ppm Sn. The HCl solutions were loaded

as droplet on a platinum wire to avoid contamination from corrosion of a stainless steel syringe.

No ƒO2 buffering assemblage was added because of the high solubility of nickel, nickel oxide,

hematite and magnetite in H2O+HCl solutions, which would not only result in significant light

absorption but also in HCl consumption to a large extent. In addition, Fe(II)–Cl, Fe(III)–Cl and

Ni–Cl complexes display Raman bands in the same wavenumber region as the Sn(II)–Cl and

Sn(IV)–Cl complexes of interest (e.g. Sitze et al. 2001; Bissonette 2013) and may completely

mask the signal from dissolved tin. Isolation of the mineral buffer assemblages from the fluid by

8
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

sealing them in a platinum capsule (Wilson and Eugster 1990) is very difficult given the very

small size of the sample chamber of a diamond-anvil cell. Before the sample chamber was

sealed, an air bubble with a volume fraction of about 0.2 was generated by controlled leaking of

the liquid. Oxygen from trapped air (its mole fraction is ~3.5·10 -5) does not cause very oxidizing

conditions at high temperatures due to reaction with the diamond anvils (Chou and Anderson

2009). The actual hydrochloric acid concentration in the sample chamber after sealing was

determined from measurement of the ice melting temperature in presence of vapor. This

temperature was converted into an HCl concentration using Eqn. (1) in Schmidt et al. (2007).

During the experiments, the HDAC was flushed with an Ar + 1 % H2 or an Ar + 1 % CH4

gas mixture (Fig. 2) to prevent oxidation of metallic parts of the cell at high temperatures. This

had the additional purpose to vary the oxygen fugacity in the sample. Argon + 1 % CH4 gas was

used to keep tin in the tetravalent oxidation state. If the HDAC was flushed with Ar + 1 % H2,

Sn(IV) was gradually reduced to Sn(II) because of hydrogen permeation into the sample

chamber likely along the contact between gasket and anvil or via dislocations.

All Raman spectra were acquired using a HORIBA Jobin Yvon LabRAM HR800 Vis

Raman microprobe equipped with a Synapse® 2048 × 512 back-illuminated CCD-detector and a

NIKON MPlan 40× SLWD objective (numerical aperture 0.4), a Laser Quantum Torus 532

DPSS laser for excitation at 532.17 nm, a grating of 1800 lines/mm, a slit aperture of 1000 µm,

and a confocal pinhole diameter of 100 µm. The laser power on the sample stage was determined

before and after one or a few series of experiments and was measured with the objective

removed and had decreased slowly from 27.7 to 24.6 mW within 9 months. The Raman spectra

of the fluid were recorded at 100–600 cm-1 and 2700–4000 cm-1 with 10 accumulations of 20 s

each, and with the mechanical focus position of the laser set to 20 µm below the interface of

9
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

fluid and upper diamond culet (Schmidt 2009; Schmidt and Chou 2012). Consecutive spectra of

the fluid reacting with cassiterite were acquired. Equilibration was assumed if no change in the

Raman band intensities of two or three consecutive spectra was detectable by inspection.

Pressure was measured using the shift in the wavenumber of the 3(SiO4) Raman band of zircon

(Schmidt et al. 2013). The Raman spectra of zircon were recorded simultaneously with emission

lines of a neon lamp to calibrate the measured wavenumber of that Raman band. All Raman

spectra were processed using the PeakFit v4.11 software from SYSTAT Software Inc. The

Gaussian + Lorentzian area model was used to fit the positions of the ν3(SiO4) Raman band of

zircon and of the neon emission lines, and after subtraction of a linear baseline in the regions of

interest, as well the positions and integrated intensities of the peaks in the 1[Sn(IV)–Cl] and the

S(O–H) regions.

The cassiterite solubility was obtained from the known Sn(IV) concentration of the

standard solution and the ratio of the corrected integrated intensities Icorr of the 1[Sn(IV)–Cl]

and the S(O–H) peaks in the sampled fluid in the HDAC and in the 0.097 molal Sn(IV) in 10.28

mass% HCl calibration solution in the HDAC. Because a ratio of integrated intensities was used,

the factors related to fluid density, spectra accumulation time and laser power cancel, and thus

Icorr was calculated from

(5)

where IICS is the measured integrated intensity of the peak corrected for response function of the

spectrometer (Dubessy et al. 2012), F is the combined frequency and scattering factor (Rudolph

et al. 2008) and B the Bose-Einstein factor. The response function had been determined by the

manufacturer for each combination of grating and excitation wavelength using a calibrated white

light source, and provided as ICS intensity correction factor in the LabSpec® software. The

10
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

wavenumber dependence of the Raman scattering intensity was accounted for using the

formulation


– (6)

(Rudolph and Irmer 2007), where 0 is the wavenumber of the exciting light and j the Raman

shift. The temperature dependence of the Stokes-Raman scattering intensity was approximated

by


– (7)

where h is the Planck constant, c the speed of light, kB the Boltzmann constant and T the

absolute temperature. These corrections minimize the error associated with the substantial

change in the scattering efficiencies of the ν S(O–H) bands with temperature (Schmidt and

Seward 2017) due to the pronounced changes in the water structure (Sahle et al. 2013). No

attempt was made to incorporate the relatively small correction factor for reflection of light at the

fluid-diamond interface (Schmidt 2009).

The corrected integrated intensity of the individual 1[Sn(IV)–Cl peaks was divided by

the number of Sn–Cl bonds in the complex to approximately correct for the different scattering

efficiencies of the species and to obtain a reliable estimate of the Sn(IV) speciation. This reduces

the uncertainty in the cassiterite solubility that stems from changes in the Sn(IV) speciation with

pressure and temperature. A further advantage is that only one calibration solution needs to be

measured, and only at ambient conditions. A fully empirical calibration from measurement of

calibration solution as function of tin and HCl concentrations, pressure and temperature would in

theory be slightly more accurate, but be too time-consuming and probably not be possible at high

temperatures because of alloying of tin from the solution with noble metals (e.g., Keppler and

11
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

Wyllie 1991), here the iridium gasket. The uncertainty in the total aqueous Sn(IV) concentration

determined as described in the previous paragraphs is estimated to be about 10 to 20%.

3. RESULTS

3.1 Tin speciation

The Ar + 1 % CH4 gas mixture was used to flush the HDAC in experiments with 1.63,

2.81, 2.92, 3.21 and 4.26 molal HCl. Argon + 1 % H2 gas was used for experiments with 0.4,

0.69, and 1.54 molal HCl. Figure 3 compares a Raman spectrum of a 0.097 molal Sn(IV)

solution in 3.14 molal HCl with spectra of the fluid of the experiment with 1.63 molal HCl at

temperatures from 20 °C to 600 °C. All spectra display a prominent peak at ~325 cm-1 and a

weak to very weak feature at about 235 cm-1 the intensity of which correlates with that of the

peak at ~325 cm-1 (Fig. 3). An additional weak and broad Raman signal at about 275 cm-1 is

present in the spectra at ambient temperature at the start and the end of the experiment with 1.63

molal HCl, but may also exist as very weak feature in the spectra at elevated temperatures (Fig.

3). The peak at ~325 cm-1 has an asymmetric shape at most temperatures, and a shoulder at about

340 cm-1 is evident in the spectrum at ambient temperature after quenching.

Figure 4 shows Raman spectra of the fluid at 500 °C from cassiterite dissolution at

different HCl concentrations from 1.63 to 4.26 molal. In these experiments, the Ar + 1 % CH4

gas mixture was used to flush the HDAC. All spectra display the Raman band at ~325 cm-1. In

two experiments at 2.81 and 2.92 molal HCl, an additional peak emerged at about 275 cm-1 at the

expense of the intensity of the Raman signal at about 325 cm-1. This additional peak is fairly

broad and may consist of overlapping bands, which is better seen in Fig. 5. Figure 5 shows

Raman spectra of the fluid of the experiment with 2.81 molal HCl in the region of Sn–Cl and C–

12
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

H stretching vibrations. The intensity ratio of the peaks at ~275 cm-1 and ~325 cm-1 changed

little with time at 500 °C, and remained constant at 600 °C (Fig. 5). The evolution of the

intensity of the band at ~275 cm-1 correlates with pronounced changes in the Raman signal in the

2875–2950 cm-1 region. A broad peak at about 2760 cm-1 is observed in the spectra at 500 and

600 °C but not at ambient temperature (Fig. 5).

Figure 6 shows Raman spectra of the fluid of the experiment at 1.54 molal HCl

concentration, in which the HDAC was flushed with Ar + 1 % H2 gas mixture. At 400 °C, no

band at about 275 cm-1 is observed even after more than two hours at that temperature. This was

also the case at the start of spectra acquisition at 500 °C. Subsequently, the spectra show an

increase in the intensity in the region from 200 to 285 cm-1 coupled with a decrease in the

intensity of the Raman band at about 325 cm-1 (Fig. 6). At 500 and 600 °C, the intensity ratio of

the bands at about 275 and 325 cm-1 increases with time.

3.2 Cassiterite solubility

Table 1 lists the obtained data for the cassiterite solubility in H2O+HCl solutions at 500

and 600 °C. Only spectra from the experiments at 1.63, 3.21 and 4.26 molal HCl were used for

determination of the Sn(IV) concentration in the fluid because no Raman band at 275 cm-1 was

observed (Fig. 4). Figure 7 compares cassiterite solubilities determined in situ from Raman

spectra with the data by Wilson and Eugster (1990) at 500 and 600 °C, 150 MPa, and at ƒO2 =

HM and ƒO2 = NNO. In all six cases, there is a linear increase in the solubility with HCl

molality. Furthermore, the Sn(IV) solubility obtained from the Raman band intensities is in very

good agreement with the cassiterite solubility reported by Wilson and Eugster (1990) at 500 and

600 °C and oxygen fugacities along the hematite-magnetite buffer.

13
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

3.3 Fluid-melt partitioning of tin

Figure 8 shows Raman spectra of the aqueous fluid and the melt at 700 °C from two runs,

in which the sample chamber of the HDAC was loaded with zircon, cassiterite, Macusani glass

and H2O+HCl fluid and in which the HDAC was flushed with Ar + 1 % CH4 gas. At first glance,

the spectra of fluid and melt in the experiment at an initial HCl concentration of 4.83 molal

appear to be very similar apart from the different overall intensity, but it was verified after

acquisition of the spectrum of the melt that no bubble of aqueous fluid had formed in the

analyzed spot. Both spectra display Raman bands at ~275, 326 (fluid) or 325 (melt), 774 (fluid)

or 773 (melt), 867 (fluid) or 864 (melt), and 889 (fluid) or 898 (melt) cm-1, a broad peak at about

2760 cm-1, and an asymmetric band centered at about 3536 (fluid) or 3538 (melt) cm-1 (Fig. 8).

The integrated intensity ratio of the peaks at ~275 cm-1 and ~325 cm-1 of about 0.1 is nearly the

same in the spectra of melt and aqueous fluid. The measured integrated intensity ratio of the

peaks at ~325 cm-1 and ~3540 cm-1 is 0.16 in the spectrum of the aqueous fluid and 0.15 in the

spectrum of the melt. The two spectra differ in peaks at ~465 and ~950 cm-1 that are only

observed in the spectrum of the melt, and a broad band at ~600 cm-1 in the spectrum of the fluid

(Fig. 8).

The spectra of fluid and melt in the experiment at an initial HCl concentration of 2.9

molal show Raman bands at 329, 771 (fluid) or 770 (melt), 864 (fluid) or 863 (melt), and 903

(fluid) or 905 (melt) cm-1, a broad peak at about 2770 cm-1, and an asymmetric band centered at

about 3548 (fluid) or 3549 (melt) cm-1 (Fig. 8). The measured integrated intensity ratio of the

peaks at 329 cm-1 and ~3550 cm-1 is 0.007 in the spectrum of the aqueous fluid and 0.006 in the

spectrum of the melt. These spectra differ in peaks at ~230, ~260, ~400, ~465, and ~960 cm-1

14
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

that are observed in the spectrum of the melt but not in that of the fluid, and a peak at ~645 cm-1

in the fluid (Fig. 8).

4. DISCUSSION

4.1 Tin speciation

The peak at ~325 cm-1 originates from symmetric stretching vibrations 1[Sn(IV)–Cl]

(Taylor and Coddington 1992). The asymmetric shape of the 1[Sn(IV)–Cl] Raman band in

many spectra and the shoulder at about 340 cm-1 (Fig. 2) indicate contributions from several

species. These species may include the chloroaquostannate(IV) complexes [SnCl6]2−,

[SnCl5(H2O)]−, cis-[SnCl4(H2O)2]0, and [SnCl3(H2 O)3]+, with wavenumbers of the 1[Sn(IV)–Cl]

Raman band of 312 cm-1, 321 cm-1, 330 cm-1, and 342 cm-1 at ambient pressure-temperature

conditions (Taylor and Coddington 1992). The fractions of these species could not be obtained in

a reliable manner by peak fitting, because the 1[Sn(IV)–Cl] peaks of the chloroaquostannate(IV)

complexes are only separated by about 10 cm-1, and broaden somewhat with temperature (Fig.

3). However, based on estimates from fitting of the 1[Sn(IV)–Cl] bands acquired in the

experiment at 1.63 molal HCl assuming three Gaussian + Lorentzian peaks of equal width, most

of the integrated intensity originates from [SnCl4(H2O)2]0, with substantial fractions of

[SnCl5(H2O)]− and [SnCl3(H2O)3]+ at all temperatures (Fig. 9). It should be noted here that the

shift of the 1[Sn(IV)–Cl] band to lower wavenumbers with increasing temperature at vapor

pressure (Fig. 3) is primarily an effect of temperature and not of a change in speciation. Above

200 °C, the band shifts back to higher wavenumbers due to the substantial increase in pressure,

so that the effects of pressure and temperature on the band positions roughly cancel at 600 °C,

546 MPa.

15
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

The shoulder at about 340 cm-1 in the spectra of the experiment with 1.63 molal HCl at

ambient temperature indicates a significant [SnCl3 (H2O)3]+ concentration (Fig. 3). This shoulder

is less obvious in the spectrum of the 0.097 molal Sn(IV) solution (Fig. 3). This is probably due

to the higher HCl molality (3.14 molal), which should result in a change in the tin(IV) speciation

to complexes with a higher Cl-H2O ratio. The weak band at about 235 cm-1 can be assigned to

bending modes of the Sn(IV) complexes because its intensity correlates with that of the Sn(IV)–

Cl symmetric stretching mode. The weak Raman signal at about 275 cm-1 particularly in the

spectra at 20 °C before heating and at 22 °C after quenching but not in the 0.097 molal Sn(IV) –

Cl solution at 21 °C (Fig. 3) is probably fromSn(II)–Cl stretching. No Raman band was

detected at 400–600 cm-1, in which region a signal from Sn–OH vibrations should be located

based on the study by Taylor and Coddington (1992) who report a wavenumber of 551 cm-1 for

1[Sn(IV)–O] of [Sn(OH)6]2− in solutions with excess NaOH.

In two of the five experiments at 500 °C with Ar + 1 % CH4 flushing, an additional peak

emerged at about 275 cm-1 at the expense of the intensity of the Raman signal from 1[Sn(IV)–

Cl] (Fig. 4, experiments at 2.81 and 2.92 molal HCl). The wavenumber of this peak is in very

good agreement with that of the strongest Raman band at 272 cm-1 of aqueous SnCl2 solutions

with added HCl reported by Chen and Grandbois (2013). They did not specify the Sn(II) species,

but the expected species is [SnCl3]− (Brugger et al. 2016). However, this species should display

peaks at about 256 and 297 cm-1 (Woodward and Taylor 1962; Taylor 1989) at ambient

conditions. On the other hand, [SnCl3]− in ionic liquids has a strong band at 266 cm-1 (Currie et

al. 2013), and an assignment to SnCl20 is not compatible with the results by Müller and Seward

(2001). In Figs. 4, 5 and 6, the observed peak at about 275 cm-1 is designated as S[Sn(II)–Cl]

because the Sn(II) species causing it is not known with certainty. Irrespective whether the

16
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

species is [SnCl3]− or not, the peak at about 275 cm-1 signifies reduction of some Sn(IV) to Sn(II)

in experiments at 2.81 and 2.92 molal HCl (Fig. 4). The redox reactions resulting in Sn(II)

formation in these two experiments were caused by carbohydrate contamination because the

formation of Sn(II) correlated with changes in the Raman signal from C–H stretching vibrations

and, at constant temperature, the Sn(II)-Sn(IV) ratio changed little with time or remained

constant (Fig. 5). The carbohydrate contamination generating reducing conditions was likely

introduced by the loaded hydrochloric acid droplet, because the effect was much smaller or not

seen at all when another HCl solution was used. The broad band at about 2760 cm-1 in the spectra

at 500 and 600 °C (Fig. 5) is very likely from H–Cl vibrations (Salant and Sandow 1931) due to

the strongly increasing formation of HCl0 ion pairs with temperature (Tagirov et al. 1997).

The increase in intensity at 200–285 cm-1 in the experiment at 1.54 molal HCl with Ar +

1 % H2 flushing (Fig. 6) is also due to formation of Sn(II) chloride complexes, probably of

[SnCl3]− (Taylor 1989; Chen and Grandbois 2013; Currie et al. 2013). The increase in the Sn(II)-

Sn(IV) ratio with time at 500 and 600 °C implies a continuous supply of the reducing agent.

Therefore, in that run, reduction to Sn(II) is not related to hydrocarbon contamination but instead

to hydrogen permeation from the Ar + 1 % H2 gas mixture into the sample chamber (Schmidt

and Chou 2012). This is supported by Raman spectra acquired in additional runs at similar

conditions. In an experiment with 0.4 molal HCl in which the HDAC was also flushed with Ar +

1 % H2, Raman bands from Sn(II)–Cl and Sn(IV)–Cl vibrations were observed in the fluid at 300

°C, vapor pressure (8.9 MPa) and at 400 °C, 92 MPa. The Sn(II)-Sn(IV) ratio increased with

time and temperature, as observed in the run at 1.54 molal initial HCl concentration. Upon

continued heating, only Sn(II)Cl2(aq) was detected in the fluid at 500 °C, 261 MPa, and 600 °C,

430 MPa. Another experiment with Ar + 1 % H2 flushing at 0.69 molal HCl showed the same

17
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

qualitative results. Dissolved Sn(IV) chloride was detected at 300 °C and vapor pressure (9.2

MPa) but no Sn(II)Cl2(aq). Both Sn(II)Cl2(aq) and Sn(IV)Cl2(aq) were present at 400 °C, 167

MPa, at which conditions the Sn(II)-Sn(IV) ratio was observed to increase with time, and only

Sn(II)Cl2(aq) was found at 500 °C, 352 MPa and 600 °C, 588 MPa.

The Sn(IV) speciation observed in high-temperature H2O+HCl fluids is very similar to

that at ambient conditions as determined by Taylor and Coddington (1992). The most abundant

Sn(IV) species at all temperatures is likely [SnCl4(H2O)2]0, with significant [SnCl5(H2O)]− and/or

[SnCl3(H2O)3]+ (Fig. 9), and an increasing fraction of [SnCl3(H2O)3]+ at low HCl concentrations.

The Raman spectra of more reduced fluids suggest that [SnCl3]− is the by far predominant Sn(II)

species at all studied conditions. Thus, the significant cassiterite dissolution reactions include

(8)

in the case of Sn(II) and

(9)

(10)

(11)

for Sn(IV). Equations (10) and (11) can be described by Eqn. (9) if [SnCl5(H2O)]− and

[SnCl3(H2O)3]+ are present in about equal quantities. Equations (9) and (11) should apply to

natural hydrothermal fluids with low HCl concentrations, at which the Sn(IV) speciation may

shift to [SnCl3(H2O)3]+ as predominant species.

Irrespective of the oxidation state of tin in the fluid, all four reactions (Eqns. (8) to (11))

imply that cassiterite precipitates if the concentration of HCl decreases, most likely as a result of

reactions with the wallrock, dilution, or partitioning of HCl into the vapor phase during boiling.

Current models on the genesis of hydrothermal tin deposits state that precipitation of cassiterite

18
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

can be caused by a decrease in chloride activity or an increase in pH. A decrease in the chloride

molalities by dilution, e.g. by mixing with meteoric water, will indeed cause oversaturation of Sn

(Duc-Tin et al., 2007). However, cassiterite precipitation by a decrease in the Cl− activity and/or

an increase in pH is inconsistent with the experimental observation that the solubility of

cassiterite in H2O+HCl±(NaCl,KCl) is determined by the HCl molality and not by the Cl−

concentration (Kovalenko et al. 1986). The HCl molality must be an essential parameter not only

because HCl and HF are required for greisenization, the characteristic alteration in hydrothermal

tin deposits, which would not take place if HCl would be substituted by NaCl, KCl, or CaCl2.

This is also because the solubility of cassiterite in H2O+HCl is about two orders of magnitude

higher than in equimolal NaCl or HF solutions (Duc-Tin et al., 2007). In other words, 0.01 molal

HCl transports as much tin as 1 molal NaCl. That NaCl concentration is a typical value of the

chloride molality in tin-ore forming fluids (Kelly and Rye 1979; Heinrich 1990; Campbell and

Panter 1990; Schmidt and Leeder 1992), but the HCl concentration can be higher than 0.01 molal

and may then govern the majority of the tin transport. Particularly at relatively low pressures

relevant for the formation of hydrothermal tin deposits, HCl concentrations present in aqueous

fluids exsolved from felsic melts are significant, to about 0.02 molal (Piccoli et al. 1999), 0.002

to 0.06 molal (Heinrich 1990), or 0.05 to 0.1 molal HCl (Kovalenko et al. 1986). Thus, HCl

dissociation will determine largely the pH of the fluid. At 50 MPa, 0.01 molal HCl, the H+ = Cl−

molality is 0.0012 corresponding to pH = 2.92 at 500 °C and 0.0039 at 400 °C corresponding to

pH = 2.41 (Tagirov et al. 1997). That is, at realistic conditions during the formation of

hydrothermal tin deposits, pH would decrease and the Cl− molality would increase during

cooling. For Sn(II)–Cl complexes, this is the opposite of what should result in cassiterite

precipitation based on Eqns. (1) and (8). However, a decrease in pH can cause SnO2 formation in

19
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

the case of the Sn(IV)–Cl complex [SnCl3(H2O)3]+, which may be predominant at low HCl

concentrations (Eqn. (11)).

4.2 Cassiterite solubility

Data for the cassiterite solubility in H2O+HCl solutions are only reported at 500 and 600

°C (Table 1). At lower temperatures, equilibrium was not attained within a few hours, but the

cassiterite solubility in H2O+HCl must still be high (see spectra in Fig. 6 at 400 °C in

comparison to that at 500 °C). Much of the tin dissolved at high temperature remains in the fluid

after quenching, as already noted by Duc-Tin et al. (2007). In the experiment at 1.63 molal HCl,

the concentration of dissolved tin had decreased by 52% 65 hours after cooling from 600 to 22

°C (Fig. 3). This suggests that decreasing temperature is not an efficient cassiterite precipitation

mechanism, which is in agreement with previous experimental studies. The data by Wilson and

Eugster (1990) indicate no significant change in the solubility of cassiterite in H 2O+HCl between

400 and 700 °C at an oxygen fugacity along the nickel-nickel oxide buffer and even an increase

with cooling from 600 to 500 °C with the oxygen fugacity controlled by the hematite-magnetite

assemblage. Likewise, Kovalenko et al (1986) and Taylor and Wall (1993) did not observe a

significant effect of temperature on SnO2 solubility at 500 to 750 °C. Moreover, the very good

agreement of the data by Wilson and Eugster (1990) and the data obtained in the present study at

much higher pressures indicates that the effect of pressure on SnO2 solubility in H2O+HCl is

insignificant.

The HCl concentrations used by Wilson and Eugster (1990) and in the present study

studies are much larger than in natural hydrothermal fluids, but the linear increase of the

solubility with HCl molality (Fig. 7) suggests that there are no drastic changes in the tin

20
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

speciation over a large range in HCl molality. The Sn(IV) solubility obtained from the Raman

band intensities is in very good agreement with the cassiterite solubility reported by Wilson and

Eugster (1990) at 500 and 600 °C and oxygen fugacities along the hematite-magnetite buffer

(Fig. 7). This confirms the validity of the assumption by Wilson and Eugster (1990) that the

oxidation state of tin is +4 at ƒO2 = HM but not their conclusion that ƒO2 ≤ NNO are required for

significant tin concentrations in solution. This is because the Sn(IV) molalities determined in situ

in this study are very high. In fact, they are close to the theoretical maximum solubility as Sn(IV)

chlorocomplexes, e.g. of 1 mol Sn / kg H2O at 5 molal HCl as [SnCl5(H2O)]−. The slopes of the

linear fits shown in Fig. 7 are thus compatible with [SnCl4(H2O)2]0, SnCl5(H2O)]− and [SnCl3]− as

possible predominant Sn(IV) species. At 500 °C, an aqueous fluid with a HCl concentration of

0.02 molal (Piccoli et al. 1999; Heinrich 1990) can dissolve 0.0081 molal tin at ƒO2 = NNO

(Wilson and Eugster 1990), probably as [SnCl3]− or as mixture of Sn(II)–Cl and Sn(IV)–Cl

complexes (this study). At 500 °C and ƒO2 = HM, this fluid can carry 0.0042 molal tin (Wilson

and Eugster 1990) as Sn(IV)–Cl complexes, most likely [SnCl4(H2O)2]0 or [SnCl3(H2O)3]+ (this

study). These tin concentrations are higher than those or similar to those found in fluid inclusions

formed before cassiterite precipitation (Audétat et al. 1998). Consequently, both Sn(II)–Cl and

Sn(IV)–Cl complexes can play a significant role in the hydrothermal transport of tin.

4.3 Fluid-melt partitioning of tin

The weak peaks at ~775 cm-1, ~865 cm-1, ~890 cm-1, and ~950 cm-1 in the spectra of the

melts shown in Fig. 8 stem from Si–O vibrations of silica Q0, Q1 and Q2 species, where the Qn

number denotes the number of oxygen atoms per SiO4 tetrahedron that are shared with

neighboring tetrahedra (Spiekermann et al. 2012; Steele-MacInnis and Schmidt 2014). That is,

21
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

the melt shows less silica polymerization than expected for a SiO 2-rich composition, which is

probably due to the high water content. The broad band at ~600 cm-1 in both spectra of the fluid

shown in Fig. 8 can be assigned to H6Si2O70(aq) (Zotov and Keppler 2000). Peaks at ~230, ~260,

~400, and ~465 cm-1 that occur only in the spectra of the melt are probably from crystals (mostly

quartz), which precipitated at about 600 °C when the sample was heated.

Both experiments show that Sn(IV)–Cl species were still stable at 700 °C in both fluid

and melt. In the experiment at an initial HCl concentration of 4.83 molal, there was no increase

in the integrated intensity ratio of the peaks at ~275 cm-1 and ~325 cm-1 with time, which points

to carbohydrate contamination as cause for reduction of a portion of Sn(IV) to Sn(II) (likely from

insufficient cleaning of the polished glass chip). At both initial HCl concentrations, the

integrated intensity ratios of the peaks at ~325 cm-1 and ~3540 cm-1 are nearly the same in the

spectra of melt and aqueous fluid. This correlation suggests that the concentration of Sn(IV)–Cl

complexes in the melt is coupled to the high content of dissolved water. It should be noted here

that, in the absence of chlorine and in particular of HCl, the speciation of tin is very different in

peraluminous haplogranitic glasses quenched from water-saturated melts, with Sn(II) > Sn(IV) at

~QFM to ~QFM + 2.4, Sn(II) forming highly ionic moieties and Sn(IV) being in octahedral

coordination with oxygen (Farges et al., 2006).

The ratio of the integrated ν1[Sn(IV)–Cl] band intensities in fluid and coexisting melt at

700 °C is 6.62 at an initial HCl concentration of 4.83 molal and 5.16 at an initial HCl

concentration of 2.9 molal. Within the error of the method, these numbers can be taken as fluid-

melt distribution coefficients. This is because the correction for the different reflection

coefficients of aqueous fluid and melt for the incident and the exiting light for the fluid-diamond

interface (Schmidt 2009) is small as the indices of refraction of fluid and melt should be fairly

22
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

similar (Steele-MacInnis and Schmidt 2014). Figure 10 compares the fluid–melt distribution

coefficients of Sn obtained in this study based on in-situ spectroscopy with literature data from

quench experiments (Nekrasov et al. 1980; Keppler and Wyllie 1991; Hu et al. 2008). There is

substantial scatter in these data, which may in part be related to the different techniques,

pressures and compositions of the glasses used in the experiments. Note that London et al.

(1988) did not analyse tin in their fluid-melt partitioning experiments, so that the data for

Macusani glass could only be compared to data for haplogranitic compositions. However, all

results show consistently that tin partitions into the fluid only at high HCl concentrations, but

increasingly into the melt with decreasing HCl molality (Fig. 10).

5. CONCLUSIONS

The pronounced partitioning of tin into the melt at the relatively low HCl concentrations

of aqueous fluids exsolved from peraluminous felsic melts of 0.002 to 0.1 molal (Kovalenko et

al. 1986; Heinrich 1990; Piccoli et al. 1999) indicates that tin in hydrothermal deposits does not

primarily stem from exsolved magmatic fluid. Instead, these HCl-bearing magmatic aqueous

fluids rather act as agent to mobilize tin from the exocontact of the granite intrusion or from

already crystallized granite. This implies that pre-enrichment of Sn in these rocks is a controlling

parameter for the formation of a hydrothermal tin deposit, which is in full agreement with

geological observations (Mingram 1996; Romer and Kroner 2015). Particularly “tin granites”

show such enrichment and host predominantly Sn(IV) substituting for Ti(IV) or Fe(III) in oxides

(e.g. magnetite, ilmenite, rutile, cassiterite) or silicates (biotite, titanite, amphibole) (Taylor,

1979; Lehmann, 1990). The oxidation state of tin mobilized from these minerals by magmatic

fluids is still uncertain, because the Sn(II)-Sn(IV) coexistence in high-temperature aqueous fluids

23
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

as function of oxygen fugacity is insufficiently constrained from experiments. Based on this

study, it is somewhere in the range between ƒO2 = HM and a few log units below ƒO2 = NNO

(Fig. 1).

The results of this Raman spectroscopic study and the comparison to solubility data from

the literature (Wilson and Eugster 1990; Duc-Tin et al., 2007) indicate that Sn(IV) chloride

complexes are stable in hydrothermal fluids over a large range of temperatures up to magmatic

conditions, pressure, and oxygen fugacities. This is not to say that Sn(II) complexes are

insignificant but that current models on hydrothermal tin deposit formation need to be expanded

to include the implications of [SnCl4(H2O)2]0 and [SnCl3(H2O)3]+ as possible prevalent tin-

transporting species. Any such model must consider fluid-wall rock reactions, which are a

characteristic feature of most hydrothermal tin deposits, most prominent in greisens and skarns.

In the case of greisenization, a type of phyllic alteration, where HCl and/or HF-bearing

hydrothermal fluids react with feldspar to form muscovite and quartz and/or topaz and/or

fluorite, a cassiterite-forming reaction involving [SnCl4(H2O)2]0 may be

(12)

or in a skarn, e.g.

(13)

and for [SnCl3(H2O)3]+

(14)

or


(15)

24
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

For Sn(II)–Cl species, redox reactions are required to precipitate cassiterite in a greisen or skarn,

and H2 is released, e.g.

(16)

and in skarns

(17)

The released hydrogen will react with CO2, if present

(18)

and the resulting methane may be trapped in fluid inclusions. In summary, important causes of

cassiterite precipitation include fluid-wall rock reactions, dilution (first and foremost by mixing

with meteoric water), and boiling due to partitioning of HCl into the vapor phase.

The present study confirms that the HCl molality is a crucial parameter for the

hydrothermal transport of tin and the precipitation of cassiterite. Redox conditions, pH, pressure

and temperature are less important or insignificant for the solubility of cassiterite. The very good

agreement of the solubility data at ƒO2 = HM by Wilson and Eugster (1990) and from this study

at oxidized conditions (no hydrogen permeation and no carbohydrate contamination) verifies that

all or most dissolved tin is in the tetravalent oxidation state at ƒO 2 of HM and that oxidation of

tin is not an efficient cassiterite precipitation mechanism.

ACKNOWLEDGMENTS

This study was initiated by the observations made during two visits of the Panasqueira Mine. The

author thanks Manuel Pacheco and Felipe Pinto (Beralt Tin & Wolfram, Barroca Grande,

Portugal) for making these visits possible and for their guidance through the mine, Pilar

25
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

Lecumberri-Sanchez (University of Arizona) for arranging the first visit and for discussions,

Rongqing Zhang (Guangzhou Institute of Geochemistry, Chinese Academy of Sciences) for

information about field observations in tin deposits, Gleb Pokrovski (Géosciences

Environnement Toulouse) for a sample of Macusani glass, and Max Wilke (University of

Potsdam) for a program to calculate oxygen fugacities.

REFERENCES

Audétat A., Günther D. and Heinrich C. A. (1998) Formation of a magmatic-hydrothermal ore


deposit: Insights with LA-ICP-MS analysis of fluid inclusions. Science 279, 2091–2094.
Bassett W. A., Shen A. H., Bucknum M. and Chou I-M. (1993) A new diamond-anvil cell for
hydrothermal studies to 2.5 GPa and from -190 to 1200 °C. Rev. Sci. Instrum. 64, 2340–
2345.
Bissonette K. (2013) Raman investigation of nickel chloride complexation under hydrothermal
conditions. Ph. D. thesis, Univ. of Guelph.
Brugger J., Liu W., Etschmann B., Mei Y., Sherman D. M. and Testemale D. (2016) A review of
the coordination chemistry of hydrothermal systems: do coordination changes make ore
deposits ? Chem. Geol. 447, 219–253.
Campbell A. R. and Panter K. S. (1990) Comparison of fluid inclusions in coexisting
(cogenetic?) wolframite, cassiterite, and quartz from St. Michael’s Mount and Cligga
Head, Cornwall, England. Geochim. Cosmochim. Acta 54, 673–681.
Chen X. and Grandbois M. (2013) In situ Raman spectroscopic observation of sequential
hydrolysis of stannous chloride to abhurite, hydroromarchite, and romarchite. J. Raman
Spectrosc. 44, 501–506.
Chou I-M. and Eugster H. P. (1977) Solubility of magnetite in supercritical chloride solutions.
Am. J. Sci. 277, 1296–1314.
Chou I.-M. and Anderson A. J. (2009) Diamond dissolution and the production of methane and
other carbon-bearing species in hydrothermal diamond-anvil cells. Geochim. Cosmochim.
Acta 73, 6360–6366.
Currie M., Estager J., Licence P., Men S., Nockemann P., Seddon K. R., Swadzba-Kwasny M.
and Terrade C. (2013) Chlorostannate(II) ionic liquids: Speciation, Lewis acidity, and
oxidative stability. Inorg. Chem. 52, 1710–1721.
Dubessy J., Caumon M.-C., Rull F. and Sharma S. (2012) Instrumentation in Raman
spectroscopy: elementary theory and practice. In: Raman Spectroscopy applied to Earth
Sciences and Cultural Heritage (eds. J. Dubessy, M.-C. Caumon, F. Rull), EMU Notes in
Mineralogy, Vol. 12, pp. 83–172.
Duc-Tin Q., Audétat A. and Keppler, H. (2007) Solubility of tin in (Cl, F)-bearing aqueous fluids
at 700 °C, 140 MPa: A LA-ICP-MS study on synthetic fluid inclusions. Geochim.
Cosmochim. Acta 71, 3323–3335.

26
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

Farges F., Linnen, R. L. and Brown G. E. Jr. (2006) Redox and speciation of tin in hydrous
silicate glasses: A comparison with Nb, Ta, Mo and W. Canadian Mineralogist 44, 795–
810.
Giggenbach W. F. (1980) Geothermal gas equilibria. Geochim. Cosmochim. Acta 44, 2021–
2032.
Halter W. E., Williams-Jones A. E. and Kontak D. J. (1998b) Modelling fluid-rock interaction
during greisenization at the East Kemptville tin deposit: Implications for mineralisation.
Chem. Geol. 150, 1–17.
Heinrich C. A. (1990) The chemistry of hydrothermal tin(-tungsten) ore deposition. Econ. Geol.
85, 457–481.
Heinrich C. A. and Eadington, P. J. (1986) Thermodynamic predictions of the hydrothermal
chemistry of arsenic, and their significance for the paragenetic sequence of some
cassiterite-arsenopyrite-base metal sulfide deposits. Econ. Geol. 81, 511–529.
Hu X., Bi X., Hu R., Shang, L. and Fan W. (2008) Experimental study on tin partition between
granitic silicate melt and coexisting aqueous fluid. Geochem. J. 42, 141–150.
Kelly W. C. and Turneaure F. S. (1970) Mineralogy, paragenesis and geothermometry of the tin
and tungsten deposits of the Eastern Andes, Bolivia. Econ. Geol. 65, 609–680.
Kelly W C. and Rye R. O. (1979) Geologic, fluid inclusion, and stable isotope studies of the tin-
tungsten deposits of Panasqueira, Portugal. Econ. Geol. 74, 1721–1822.
Keppler H. and Wyllie P. J. (1991) Partitioning of Cu, Sn, Mo, W, U, and Th between melt and
aqueous fluid in the systems haplogranite-H2O-HCl and haplogranite-H2O-HF. Contrib.
Mineral. Petrol. 109, 139–150.
Kovalenko N. I., Ryzhenko B. N., Barsukov V. L., Klintsova A. P., Velyukhanova T. K.,
Volynets M. P. and Kitayeva L. P. (1986) The solubility of cassiterite in HCl and HCl +
NaCl (KCl) solutions at 500 °C and 1000 atm under fixed redox conditions. Geochem.
Int. 23, 1–16.
Lehmann B. (1985) Formation of the strata-bound Kellhuani tin deposits, Bolivia. Mineral.
Deposita 20, 169–176.
Lehmann B. (1990) Metallogeny of tin. Springer, Berlin Heidelberg.
London D., Hervig, R.L. and Morgan, VI, G.B. (1988) Melt-vapor solubilities and elemental
partitioning in peraluminous granite-pegmatite systems: experimental results with
Macusani glass at 200 MPa. Contrib. Mineral. Petrol. 99, 360–373.
Mingram, B. (1996) Geochemische Signaturen der Metasedimente des erzgebirgischen
Krustenstapels. PhD Thesis, Univ. Gießen, 1995 (Scientific Technical Report STR 9604,
GFZ Potsdam)
Mohamed M. A.-M. (2013) Evolution of mineralizing fluids of cassiterite–wolframite and
fluorite deposits from Mueilha tin mine area, Eastern Desert of Egypt, evidence from
fluid inclusion. Arab. J. Geosci. 6, 775–782.
Müller B. and Seward T. M. (2001). Spectrophotometric determination of the stability of tin(II)
chloride complexes in aqueous solution up to 300°C. Geochim. Cosmochim. Acta 65,
4187–4199.
Nekrasov I. Y., Epel'baum M. B. and Sobolev V. P. (1980) Partitioning of tin between melt and
chloridic fluid in the granite-SnO(SnO2)-fluid system (in Russian). Dokl. Acad. Nauk
SSSR 252, 977–981.
Ohmoto H. and Kerrick D. (1977) Devolatilization equilibria in graphitic systems. Am. J Sci.
277, 1013–1044.

27
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

Piccoli P. M., Candela P. A. and Williams T. J. (1999) Estimation of aqueous HCl and Cl
concentrations in felsic systems. Lithos 46, 591–604.
Pichavant M., Valencia Herrera J., Boulmier S., Briqueu L., Joron J. L., Juteau M., Marin L.,
Michard A., Sheppard S. M. F., Treuil M. and Vernet M. (1987) The Macusani glasses,
SE Peru: evidence of chemical fractionation in peraluminous magmas. In: Magmatic
Processes: Physicochemical Principles (ed. B. O. Mysen), The Geochemical Society
Special Publication no. 1, pp. 359–373.
Pinto F., Vieira R. and Noronha F. (2015) Different cassiterite generations at the Panasqueira
deposit (Portugal): Implications for the metal zonation model. 13th SGA Biennial
Meeting, Nancy, France. (abstr.).
Robie R.A. and Hemingway B.S. (1995) Thermodynamic properties of minerals and related
substances at 298.15 K and 1 bar (105 Pascals) pressure and at higher temperatures. U.S.
Geological Survey Bulletin 2131, Washington D.C., 461 pp.
Romer R. L. and Kroner U. (2015) Sediment and weathering control on the distribution of
Paleozoic magmatic tin–tungsten mineralization. Miner. Deposita 50, 327–338.
Rudolph W. W. and Irmer G. (2007) Raman and infrared spectroscopic investigations on
aqueous alkali metal phosphate solutions and density functional theory calculations of
phosphate-water clusters. Appl. Spectrosc. 61, 1312–1327.
Rudolph W. W., Irmer G. and Königsberger E. (2008) Speciation studies in aqueous HCO3––
CO32– solutions. A combined Raman spectroscopic and thermodynamic study. Dalton
Trans., 900–908.
Sahle C. J., Sternemann C., Schmidt C., Lehtola J., Jahn S., Simonelli L., Huotari S., Hakala M.,
Pylkkänen T., Nyrow A., Mende K., Tolan M., Hämäläinen K. and Wilke, M. (2013)
Microscopic structure of water at elevated temperatures and pressures. Proc. Nat. Acad.
Sci. 110, 6301–6306.
Salant E. O. and Sandow A. (1931) Modified scattering by hydrogen halides. Phys. Rev. 37,
373–378.
Schmidt C. (2009) Raman spectroscopic study of a H2O + Na2SO4 solution at 21-600 °C and 0.1
MPa to 1.1 GPa: Relative differential ν 1-SO42– Raman scattering cross sections and
evidence of the liquid-liquid transition. Geochim. Cosmochim. Acta 73, 425–437.
Schmidt C. and Seward T. M. (2017) Raman spectroscopic quantification of sulfur species in
aqueous fluids: Ratios of relative molar scattering factors of Raman bands of H 2S, HS–,
SO2, HSO4–, SO42–, S2O32–, S3–, and H2O at ambient conditions and information on
changes with pressure and temperature. Chemical Geology 467, 64–75.
Schmidt C. and Chou I.-M. (2012) The Hydrothermal Diamond Anvil Cell (HDAC) for Raman
spectroscopic studies of geological fluids at high pressures and temperatures. In: Raman
Spectroscopy applied to Earth Sciences and Cultural Heritage (eds. J. Dubessy, M.-C.
Caumon, F. Rull), EMU Notes in Mineralogy, Vol. 12, pp. 247–276.
Schmidt C. and Watenphul A. (2010). Ammonium in aqueous fluids to 600 °C, 1.3 GPa: A
spectroscopic study on the effects on fluid properties, silica solubility, and K-feldspar to
muscovite reactions. Geochim. Cosmochim. Acta 74, 6852–6866.
Schmidt C. and Leeder O. (1992) Temperature and pressure conditions during the formation of
pegmatites and pneumatolytic tin-tungsten mineralizations in Mongolia. Neu. Jb.
Mineral. Abh. 165, 29–52.

28
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

Schmidt C., Steele-MacInnis M., Watenphul A. and Wilke M. (2013) Calibration of zircon as a
Raman spectroscopic pressure sensor to high temperatures and application to water-
silicate melt systems. Am. Mineral. 98, 643–650.
Schmidt C., Rickers K., Bilderback D. H. and Huang, R. (2007) In situ synchrotron-radiation
XRF study of REE phosphate dissolution in aqueous fluids to 800°C. Lithos 95, 87–102.
Sitze M. S., Schreiter E. R., Patterson E. V. and Freeman, R. G. (2001) Ionic liquids based on
FeCl3 and FeCl2. Raman scattering and ab initio calculations. Inorg. Chem. 40, 2298–
2304.
Spiekermann G., Steele-MacInnis M., Schmidt C. and Jahn S. (2012) Vibrational mode
frequencies of silica species in SiO2-H2O liquids and glasses from ab initio molecular
dynamics. J. Chem. Phys. 136, 154501–154501-13.
Steele-MacInnis M. and Schmidt C. (2014) Silicate speciation in H2O-Na2O-SiO2 fluids from 3
to 40 mol% SiO2, to 600 °C and 2 GPa. Geochim. Cosmochim. Acta 136, 126–141.
Tagirov B. R., Zotov A. V. and Akinfiev N. N. (1997) Experimental study of dissociation of HCl
from 350 to 500 °C and from 500 to 2500 bars: Thermodynamic properties of HCl0(aq).
Geochim. Cosmochim. Acta 61, 4267–4280.
Taylor M. J. (1989) Raman spectrum of the [SnI3]− ion and other halogenostannate(II)
complexes including mixed halides. J. Raman Spectrosc. 20, 663–666.
Taylor M. J. and Coddington J. M. (1992) The constitution of aqueous tin(IV) chloride and
bromide solutions and solvent extracts studies by 119Sn NMR and vibrational
spectroscopy. Polyhedron 11, 1531–1544.
Taylor J. R. and Wall V. J. (1993) Cassiterite solubility, tin speciation, and transport in a
magmatic aqueous phase. Econ. Geol. 88, 437–460.
Taylor R. G. (1979) Geology of tin deposits. Elsevier, Amsterdam.
Wang Y.W., Wang J. B., Wang L. J. and Chen Y. Z. (2006) Tin mineralization in the Dajing tin–
polymetallic deposit, Inner Mongolia, China. J. Asian Earth Sci. 28, 320–331.
Watenphul A., Schmidt C. and Jahn S. (2014) Cr(III) solubility in aqueous fluids at high
pressures and temperatures. Geochim. Cosmochim. Acta 126C, 212–227.
Wilson, G. A. and Eugster H. P. (1990) Cassiterite solubility and tin speciation in supercritical
chloride solutions. Geochem. Soc. Special Publ. No. 2, 179–195.
Woodward L. A. and Taylor M. J (1962) Raman effect and solvent extraction. Part III. Spectra of
the trichlorostannite and tribromostannite ions. J. Chem. Soc., 407–410.
Zhang R., Lu J., Lehmann B., Li C., Li G., Zhang L., Guo J., Sun W. (2017) Combined zircon
and cassiterite U–Pb dating of the Piaotang granite-related tungsten–tin deposit, southern
Jiangxi tungsten district, China. Ore Geol. Rev. 82, 268–284.
Zotov, N. and Keppler H. (2002) Silica speciation in aqueous fluids at high pressures and high
temperatures. Chem. Geol. 184, 71–82.

29
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

Figure captions

Figure 1. Calculated oxygen fugacities at 200 MPa as a function of temperature for several

buffer assemblages based on data tabulated by Robie and Hemingway (1995). The

curve for the SnO2-SnO equilibrium is located several log units below the Ni–NiO

buffer (Lehmann 1990). The dashed lines are logƒO2 versus temperature curves

from Ohmoto and Kerrick (1977) and give the stability limit of graphite in

equilibrium with C-O-H fluid (long dashes) and the X(CH4)=0.1 composition

contour for a C-O-H fluid in equilibrium with graphite (short dashes).

Figure 2. Sketch of the experimental setup (not to scale). λ0 = excitation wavelength.

Figure 3. Measured Raman spectra of the fluid of the cassiterite + 1.63 molal HCl

experiment along a heating path from 20 °C to 600 °C and at 22 °C 65 hours after

cooling from 600 °C. The HDAC was flushed with Ar + 1 % CH4 gas. A

spectrum of a 0.097 molal Sn(IV) solution in 10.28 mass% HCl at 21 °C, 0.1 MPa

is shown for comparison. The intensity is corrected for the response function of

the spectrometer. For clarity, the spectra were shifted along the intensity axis. Pv =

vapor pressure. The shadowed bands denote the positions of the 1[Sn(IV)–Cl]

peaks of four chloroaquostannate(IV) complexes at ambient conditions (Taylor

and Coddington 1992).

Figure 4. Measured Raman spectra of the fluid at 500 °C from experiments in which

cassiterite was reacted with in H2O+HCl solutions with the initial HCl

30
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

concentration varying from 1.63 to 4.26 molal. The HDAC was flushed with Ar +

1 % CH4 gas. The spectra were shifted along the intensity axis. The intensity is

corrected for the response function of the spectrometer. The shadowed bands

denote positions of 1[Sn(IV)–Cl] and S[Sn(II)–Cl] peaks at ambient P-T

conditions.

Figure 5. Raman spectra of the fluid in the regions of Sn–Cl and C–H stretching vibrations

recorded in the experiment with 2.81 molal HCl. The intensity is corrected for the

response function of the spectrometer. The HDAC was flushed with Ar + 1 %

CH4 gas. t0 = time at which the target temperature was attained. The spectrum at

600 °C, t0 + 34 min was slightly shifted along the intensity axis because it is

almost identical with that at 600 °C, t 0 + 7 min. The shadowed bands denote

positions of 1[Sn(IV)–Cl] and S[Sn(II)–Cl] peaks at ambient P-T conditions.

Figure 6. Raman spectra of the fluid at the indicated P-T conditions from the experiment

with 1.54 m HCl. t0 = time at which the target temperature was attained. The

HDAC was flushed with Ar + 1 % H2 gas. The intensity is corrected for the

response function of the spectrometer and the spectra are shifted along the

intensity axis. The shadowed bands denote positions of 1[Sn(IV)–Cl] and

S[Sn(II)–Cl] peaks at ambient P-T conditions.

31
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

Figure 7. Cassiterite solubility at 500 and 600 °C as function of HCl concentration. Squares

denote data from this study determined by Raman spectroscopy. Data from

Wilson and Eugster (1990) are shown as circles and triangles, with linear fits.

Figure 8. Measured Raman spectra of the aqueous fluid and the melt at 700 °C from

experiments in which cassiterite reacted with melt produced from Macusani glass

and with aqueous fluid at two initial HCl concentrations of 2.9 and 4.83 molal.

The HDAC was flushed with Ar + 1 % CH4 gas.

Figure 9. Estimated Sn(IV) speciation in the experiment at 1.63 molal initial HCl

concentration. The HDAC was flushed with Ar + 1 % CH4 gas.

Figure 10. Fluid–melt distribution coefficients Df/m of Sn as function of the HCl molality.

Data from this study based on in-situ spectroscopy (700 °C, 800 to 1050 MPa,

Macusani glass) are shown as squares and compared with literature data from

quench experiments by Keppler and Wyllie (1991) at 750 °C, 200 MPa, NNO for

haplogranite (circles) and by Nekrasov et al. (1980) at 850 °C, 100 MPa, NNO for

quartz+albite melt (diamonds) and rapid quench experiments by Hu et al. (2008)

at 850 °C, 100 MPa, NNO for haplogranite (triangles). The shaded bar denotes

the range of HCl concentrations in aqueous fluids exsolved from peraluminous

felsic melts (Kovalenko et al. 1986; Heinrich 1990; Piccoli et al. 1999).

32
Raman spectroscopy on tin in H2O+HCl fluids at high P and T

Table 1. Cassiterite solubility determined by Raman spectroscopy at 500 and 600 °C and at

oxidising conditions, i.e. Sn(II) was not detectable or only present in small concentrations. CHCl

= HCl concentration in the fluid after loading as obtained from cryometry (Schmidt et al. 2007);

T = temperature; P = pressure determined from the shift in the wavenumber of the 3(SiO4)

Raman band of zircon (Schmidt et al. 2013); CSn = tin concentration in the fluid determined from

the integrated intensities of the 1[Sn(IV)–Cl] and S[(O–H) Raman bands in consecutive spectra

at the same P, T, and CHCl.

CHCl, molal T, °C P, MPa molal Sn


1.63 500 370 0.3015
1.63 500 370 0.2975
3.21 500 194 0.673
3.21 500 194 0.658
4.26 500 932 0.663
4.26 500 932 0.687
4.26 500 932 0.671
1.63 600 546 0.2249
1.63 600 546 0.2196
1.63 600 546 0.2191
3.21 600 822 0.473
3.21 600 822 0.448
4.26 600 1330 0.547
4.26 600 1330 0.57

33
-10
te
m ati ite
he gnet hite
-20 ma grap

log ƒO2
-30 O2
Sn n
Ni–NiO S O2
XCH4=0.1 Sn nS
-40 S

P = 200 MPa
-50 Fig. 1
300 400 500 600 700 800
(online)
Temperature (°C)
to spectrometer
objective lens

Raman-scattered incident
photons laser beam
from sample (0 = 532 nm)

heater
wires
inlet for
gas
H2O + HCl fluid thermo- mixture:
couples Ar + 1% H2
iridium gasket or
Ar + 1% CH4

zircon
(pressure diamond
anvils
sensor) cassiterite (ultra-low Fig. 2
fluorescence)
(online)
7
[SnCl4(H2O)2]0
[SnCl5(H2O)]−

[SnCl6]2− [SnCl3(H2O)3]+
6
20 °C,
0.1 MPa (PV)
100 °C,
5
0.1 MPa (PV)
200 °C,
iintensity IICS (103 counts)

1.7 MPa (PV)


4

300 °C,
87 MPa
3

400 °C,
227 MPa
2 500 °C,
370 MPa

600 °C,
65 h after 546 MPa
1 quench from
600 °C
22 °C,
0.1 MPa
0.097 molal Sn(IV)
in 3.14 molal HCl 21 °C, 0.1 MPa
0
200 300 400 Fig. 3
Raman shift (cm-1) (online)
8
[SnCl5(H2O)]− [SnCl4(H2O)2]0
7
[SnCl6]2− [SnCl3(H2O)3]+
500 °C

intensity IICS (103 counts)


6
4.26 molal HCl
932 MPa
5 S[Sn(II)–Cl] 3.21 molal HCl
194 MPa
4 1.63 molal HCl
370 MPa
3

2
2.92 molal HCl
1 2.81 molal HCl 567 MPa
463 MPa
0 Fig. 4
200 300 400
(online)
Raman shift (cm-1)
4 1
[SnCl5(H2O)]− [SnCl4(H2O)2]0
2.81 molal HCl
[SnCl6]2− 400 °C, 300 MPa
[SnCl3(H2O)3]+ 24.3 °C, 0.1 MPa, before heating
intensity IICS (103 counts)

3
S[Sn(II)–Cl] 600 °C, 960 MPa, t0 + 7 min
600 °C, 960 MPa, t0 + 34 min
500 °C, 463 MPa, t0 + 43 min
2 0.5 500 °C, 463 MPa, t0 + 95 min
400 °C,
600 °C, t0 + 7 min 300 MPa (C−H)
600 °C, t0 + 34 min
500 °C, t0 + 43 min
1 500 °C, t0 + 95 min

24.3 °C, 0.1 MPa,


before heating
(H−Cl)
0 0
200 300 400 2700 2800 2900 3000 Fig. 5
-1
Raman shift (cm ) Raman shift (cm-1) (online)
3
1.54 molal HCl
[SnCl5(H2O)]− [SnCl (H O) ]0
4 2 2
[SnCl6]2−
S[Sn(II)–Cl] [SnCl3(H2O)3]+

19.6 °C, 0.1 MPa,


intensity IICS (103 counts)

2 before heating

400 °C,
200 MPa,
t0 + 12 min
500 °C, 405 MPa,
t0 + 129 min
400 °C,
200 MPa,
500 °C, 405 MPa, t0 + 159 min
1 t0 + 12 min

600 °C, 608 MPa,


t0 + 12 min
600 °C, 608 MPa,
t0 + 129 min

0
200 300 400
Raman shift (cm-1) Fig. 6
(online)
this study, from ν1[Sn(IV)–Cl]
Raman band intensity
500 °C 600 °C

cassiterite solubility (molal Sn)


2 Wilson and Eugster (1990),
150 MPa, HM buffer
500 °C 600 °C
Wilson and Eugster (1990),
150 MPa, NNO buffer
500 °C 600 °C

1
194 MPa 932 MPa

370
MPa 1330 MPa
822 MPa
546 MPa
0
0 1 2 3 4 5 Fig. 7
HCl concentration (molal) (online)
4
S[Sn(IV)–Cl]
700 °C

intensity IICS (103 counts)


3 fluid, ~1050 MPa, 4.83 molal HCl
melt, ~1050 MPa, 4.83 molal HCl
fluid, 800 MPa, 2.9 molal HCl
melt, 800 MPa, 2.9 molal HCl

0
Fig. 8
200 500 1000 3000 3500
-1
(online)
Raman shift (cm )
80
1.63 molal HCl
70
60
% Sn(IV)–Cl species

50 [SnCl4(H2O)2]0
[SnCl3(H2O)3]+
40
[SnCl5(H2O)]−
30
20
10
0
0 100 200 300 400 500 600 Fig. 9
Temperature (°C) (online)
2

0
log Df/m

-1

-2 Nekrasov et al. (1980)


Keppler and Wyllie (1991)
-3 Hu et al. (2008)
this study
-4
0 1 2 3 4 5 Fig. 10
HCl concentration (molal) (online)

You might also like