Document [original]

Geochimica et Cosmochimica Acta 376 (2024) 37–53

Available online 28 May 2024

nc/4.0/).

Arsenic-poor fluids promote strong As partitioning into pyrite

Martin Kutzschbach

, Frederik Dunkel

, Christof Kusebauch

, Ferry Schiperski

Frederik B¨

orner

, Henrik Drake

, Kevin Klimm

, Manuel Keith

Technische Universit¨

at Berlin, Applied Geochemistry, Berlin, Germany

GFZ German Research Centre for Geosciences, Potsdam, Germany

Friedrich-Alexander-Universit¨

at (FAU) Erlangen-Nürnberg, GeoZentrum Nordbayern, Erlangen, Germany

Department of Biology and Environmental Science, Linnaeus University, Kalmar, Sweden

Goethe-Universit¨

at Frankfurt, Institut für Geowissenschaften, Frankfurt am Main, Germany

ARTICLE INFO

Associate editor: Georges Calas

Keywords:

Arsenic partitioning

Pyrite

Siderite

Geothermal energy

Ore deposits

Groundwater remediation

Hydrothermal systems

ABSTRACT

Pyrite is a ubiquitous sulfide mineral found in diverse geological settings and holds great significance in the

formation of Au deposits as well as the safe utilization of groundwater due to its remarkable ability to incorporate

substantial amounts of As. However, despite its importance, there remains a dearth of fundamental data on the

partitioning of As between pyrite and fluid, which is key for accurately modeling the As distribution in these

environments.

Here, we present new insights into the partitioning behavior of As between pyrite and fluid at conditions that

mimic natural fluid systems. Pyrite was synthesized by replacement of natural siderite in hydrothermal exper-

iments at 200 ◦C and pH 5 applying a wide range of fluid As concentrations, spanning from 0.001 to 100 µg/g.

The As distribution and concentration in synthetic pyrite was analyzed by quantitative LA-ICP-MS mapping

providing a high spatial resolution and sensitivity at 2–3 µm image pixel size at a detection limit of ~1 µg/g at

the single pixel scale. Pyrite-fluid partitioning coefficients (D

(py/fluid)

) between synthetic pyrite and experimental

fluid agree with previously published data for high fluid As concentrations of 1 µg/g to 100 µg/g (D

<2000).

However, at low As concentrations in the experimental fluid (<1 µg/g), a steep increase in the D

(py/fluid)

values of

up to ~30,000 was detected, demonstrating even stronger As partitioning into pyrite. This is confirmed by the

analyses of natural pyrite that precipitated from As-poor fluids (0.3–0.4 ng/g) within a deep anoxic aquifer in SE

Sweden. The discovery holds significant implications for the mobility and scavenging of As, which in turn is

important for understanding the formation and fingerprinting of mineral deposits as well as for the secure uti-

lization of groundwater resources.

1. Introduction

Pyrite (FeS

) is the most common sulfide mineral in hydrothermal

systems of the Earth’s crust, which is due to its stability over a wide

range of pH (0–9), temperature (up to 740◦C) and redox conditions

(Nordstrom, 1982; Evangelou, 1995; Craig et al., 1998; Kusebauch et al.,

2018). In combination with its ability to incorporate various trace ele-

ments (such as As, Ag, Au, Cu, Ni, Co, Pb, Sb, Se, Te) either through solid

solution or as micro- to nano-scale inclusions, this makes pyrite a

valuable tracer for comprehending hydrothermal (ore-forming) systems

(Keith et al., 2018, Ciobanu et al., 2012; Deditius et al., 2011; Large

et al., 2009; Palenik et al., 2004; Mukherjee and Large, 2016; Reich

et al., 2013).

Arsenic has variable oxidation states (−1, +2, +3, +5), and can

substitute for both Fe and S in the crystal lattice of pyrite (Deditius et al.,

2008; Keith et al., 2018). Arsenic-rich (‘arsenian’) pyrite that commonly

forms from reduced hydrothermal fluids, as known from Carlin-type and

some epithermal deposits, mostly hosts As

−

, which substitutes for S

−

the pyrite structure (Simmons et al., 2005; Deditius et al., 2014; Keith

et al., 2018). The preferred incorporation of As

−

into pyrite from rather

reduced fluids is confirmed by high pyrite-fluid partitioning coefficients

(py/fluid)

(300–1700) for fluids with >1 µg/g As (Reich et al., 2005; Su

et al., 2008, 2009, 2012; Qian et al.,. 2013; Large et al., 2016; Kusebauch

et al., 2018). In this context, it has been shown that As enhances the

solubility of Au in the pyrite lattice, in economic Carlin-type and some

epithermal deposits (Pals et al., 2003; Palenik et al., 2004; Reich et al.,

* Corresponding author.

E-mail address: [email protected] (M. Kutzschbach).

Contents lists available at ScienceDirect

Geochimica et Cosmochimica Acta

journal homepage: www.elsevier.com/locate/gca

https://doi.org/10.1016/j.gca.2024.05.027

Received 21 February 2024; Accepted 25 May 2024

Geochimica et Cosmochimica Acta 376 (2024) 37–53

2005; Deditius et al., 2014; Kusebauch et al., 2019; Pokrovski et al.,

2019, 2021). Pyrite can host between 6 and 10 wt % of As in solid so-

lution, whereas randomly distributed arsenopyrite domains were re-

ported in pyrite with higher As contents (Reich and Becker, 2006;

Blanchard et al., 2007; Simmons et al., 2005). Accordingly, B¨

orner et al.

(2021) recently demonstrated that the general assumption of increasing

Au with increasing As in pyrite is only valid for As contents <13 wt %,

whereas decreasing Au contents were reported above this threshold.

Beyond the economic interest of using As in pyrite as a vector to-

wards Au-rich mineralization, it also poses a latent risk, as As can be

released during pyrite dissolution under more oxidizing conditions

eventually induced artificially through aquifer storage and recovery

(Jones and Pichler, 2003; Stuyfzand and Bonte, 2023) or geothermal

energy production (Burnside et al., 2016). In addition, high levels of As

in pyrite are a primary cause of alarmingly high As concentrations in

groundwater and running water associated with active or abandoned

mining and in industrial areas (Baraga˜

no et al., 2020; Oliveira et al.,

2012; Savage et al., 2000). Even low levels of As in the range of 10 µg/l

in groundwater can be considered as a potential health risk, and are

observed in many developing countries (Smith and Smith, 2004).

Hence, despite the critical importance of understanding how As is

incorporated in pyrite from fluids to constrain the sources of ground-

water As contamination and to reconstruct the ore-forming processes of

precious metal deposits, there is currently a lack of systematically

derived D

(py/fluid)

data at As levels below 1 µg/g, which are known from

natural and artificial (geothermal) pyrite-forming hydrothermal envi-

ronments (Drake et al., 2015; Schmidt et al., 2017). The availability of

such data would be particularly beneficial for modeling As mobility in

subsurface hydrothermal systems or during fluid-rock interaction to

better constrain the formation conditions of As-rich pyrite in the context

of ore formation and environmental As pollution. To address these

knowledge gaps, hydrothermal experiments were conducted under

reduced fluid conditions (−43.9 log fO

), where pyrite replaced pre-

cursor siderite by interaction with fluids of variable As concentration

(0001–100 µg/g).

2. Methods and experimental materials

2.1. Hydrothermal synthesis and starting materials

Pyrite was synthesized following Kusebauch et al. (2018) via hy-

drothermal replacement of natural siderite (FeCO

) at 200◦C and pH 5 in

S-bearing solutions via the reaction:

FeCO

(s) +2 H

S (aq) → FeS

(s) +H

O +H

(aq) +CO

(aq)(1)

Experiments were carried out in large volume (50 ml) PTFE-lined

stainless-steel autoclaves to achieve high fluid/solid ratios (~6000),

while still obtaining sufficiently sized pyrite grains for convenient in situ

chemical analysis. The large fluid reservoir was chosen to maintain

constant As concentrations in the fluid throughout the experiments. The

following starting materials were used:

(1) Siderite (FeCO

): The siderite used in this study originates from a

cryolite deposit in Ivigtut, Greenland. A euhedral crystal of

approximately 3.5 cm

was crushed and the size fraction of

250–500 µm was separated by sieving. Individual crystal frag-

ments were mostly transparent, occasionally displaying thin

(<10 µm) brown limonite coatings on cleavage planes and crystal

faces (Fig. S1). The siderite is homogeneous and depicts near end-

member composition with minor amounts of Mn (Fe

0.96(1)

0.04

(1)

), which agrees with published siderite data from Ivigtut

(Sliwinski et al., 2018). Trace element analysis by LA-ICP-MS (cf.

Section 2.4.4) revealed As concentrations below the minimum

detection limit of 0.02 µg/g in clear parts of the siderite crystal

(Table S1). The limonite coatings show detectable, but negligible

amounts of As (<1 µg/g, Fig. S1b). About 20–30 clear siderite

grains (~5 mg) without coating (Fig. S1) were handpicked for

each experiment.

(2) Thioacetamid (C

NS): Solid thioacetamide (Sigma-Aldrich) is

highly soluble in aqueous solutions and decomposes to ammo-

nium acetate and hydrogen sulfide when heated, which served as

the S donor for pyrite synthesis (Qian et al., 2013; Kusebauch

et al., 2018, 2019). Approximately 96 mg solid thioacetamide

were weighed into each PTFE beaker. This results in an H

concentration of 0.043 mol/l, assuming complete decomposition

of thioacetamide to ammonium acetate and hydrogen sulfide.

(3) pH-buffer: A sodium acetate/acetic acid buffer at pH 5 (Sigma-

Aldrich) was used for all experimental runs. Two buffer solutions

at high and low concentrations were prepared (Table 1). After

verifying the pH values with a pH probe, 25 ml of the buffer so-

lution were pipetted into each PTFE beaker. Both buffer solutions

maintained the pH during the experiments, but the higher

concentrated buffer showed a better performance, as indicated by

a pH decrease of less than 0.06 pH units compared 0.54 pH units

when using the lower buffer concentration.

(4) Arsenic solution: Stock solutions of varying As concentrations

were prepared from a NIST traceable 1000 µg/g As ICP-MS single

element standard (Sigma-Aldrich). Subsequently, 5 ml of the

stock solution were loaded into the beaker resulting in final fluid

As concentrations of 0.001, 0.01, 0.1, 1, 10, and 100 µg/g. The

filled PTFE beakers were weighed, and transferred into stainless-

steel autoclaves, which were closed firmly via a screw cap. The

assembly was placed in a muffle furnace at 200◦C for 168h. In a

series of As-free test runs using the 250–500 µm siderite size

fraction, no siderite was detected in the run products after 168h,

as determined through powder X-ray diffraction (XRD) analysis.

Based on these findings, a run duration of 168 h was selected for

the subsequent experiments. Each sealed PTFE beaker was

weighed before opening. A negligible weight loss of ~1 % was

observed for all experimental runs, which indicates that all au-

toclaves remained airtight throughout the experiment. The pH

value of the fluid was re-measured, indicating constant pH con-

ditions over the run duration (Table 1).

2.2. Fluid and solid sample extraction and cleaning procedure

The experimental fluid from the PTFE beakers was carefully pipetted

into 50 ml volumetric flasks. The beakers containing the solid run

products were rinsed in a two steps procedure by (1) a few ml of 1 %

concentrated nitric acid (Merck Suprapure®) and (2) ultrapure (UP)

H2O from an Arium pro VF ultrapure water system (Sartorius;

0.055 µS cm

−1

). Rinse solutions were transferred to the volumetric flask

that contained the experimental fluid and filled up with ultrapure water

to achieve a total volume of 50 ml. Finally, the sample solution was

filtered through a 0.45 µm syringe filter. The solid residues in the PTFE

beakers were dried in an oven at 50 ◦C for about 24h. A 50 % fraction of

the solid sample material from three experimental runs was analyzed by

XRD (cf. Section 2.4.3). Crystals from the dried solid residues were

embedded in 1-inch epoxy mounts and polished for subsequent analysis

by Electron Probe Micro Analysis (EPMA) and Laser Ablation Induc-

tively Coupled Plasma Mass Spectrometry (LA-ICP-MS). After each

experimental run, the PTFE beakers were carefully cleaned mechani-

cally and chemically (10 % HNO3 for at least 24 h) to prevent As

contamination.

2.3. Natural groundwater and pyrite from the ¨

Asp¨

o hard-rock laboratory

The Swedish Nuclear Fuel and Waste Management Co. (SKB) oper-

ates a full-scale test facility in SE Sweden (¨

Asp¨

o Hard Rock Laboratory).

Two boreholes (labeled KA3105A and KA3385A, Drake et al., 2015)

were drilled horizontally at 415 and 445 m crustal depth between 1994

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

and 1995. Different borehole sections were isolated by inflatable

packers. Groundwater was constantly supplied to the packed-off bore-

hole sections by intersecting bedrock fractures. An Al-rod ran through all

packer sections and groundwater was sampled in the tunnel environ-

ment by opening valves that allowed the relatively over-pressured water

from the sections to pass through polyamide tubings attached to each

packer. The temperature was constant at 14 ◦C and the pH-value was

stable between 7.4 and 7.7. The borehole instrumentation was retrieved

after 17 years and calcite, barite, and pyrite were scraped off the

instrumentation and sampled in nitrogen-filled gas-tight bags to prevent

oxidation. The chemical composition of the groundwater and of pyrite

from one section of each of the boreholes are presented in this study. For

more information about the site and sampling procedure see Drake et al.

(2015).

2.4. Analytical methods for characterization of solid run products and

natural pyrite

2.4.1. Micro X-ray fluorescence (µ-XRF)

The composition and distribution of major elements in the initial

siderite was determined through µ-XRF mapping using a Bruker M4

Tornado at the MAGMA Lab of the TU Berlin. The instrument is equip-

ped with an X-ray focusing capillary optic and two 30 mm

silicon drift

detectors. Mappings were performed on a polished 1-inch epoxy mount

using a 20 µm spot size. The Rh-anode operated in an evacuated

chamber at 20 mbar with an acceleration voltage 50 kV and an anode

current of 600 mA. Spots were set at intervals of 50 µm and measured for

200 ms. Siderite stoichiometry was derived by integrating the signal of a

representative region of individual grains, which was processed to

quantify relevant cations by a standardless method (Flude et al., 2017;

Table S1). Concentrations of C and O were derived mathematically

assuming one CO

anion per formula unit.

2.4.2. Electron Probe Micro-Analysis (EPMA)

Electron back-scattered images, and major and minor element

compositions of the solid run products were acquired using a JXA8200

Superprobe at the GeoZentrum Nordbayern. The measurements were

performed in an evacuated chamber (<4.0 ×10

−6

mbar) with a focused

beam, an acceleration voltage of 20 kV and a beam current of 50 nA. The

following standards were used for the quantitative analysis: FeS

for Fe

and S, FeAsS for As, MnSiO

for Mn, and CaCO

for Ca. Data was ac-

quired based on the following x-ray lines, spectroscopic crystals, and

count times (in sec.) at the peak and background (lower and upper)

positions: Fe (K

, LIF, 20, 10, 10); S (K

, PETJ, 20, 10, 10); As (L

, TAP,

100, 50, 50); Mn (K

, PETH, 100, 50, 50); Ca (K

, PETJ, 20, 10, 10).

Calculated theoretical minimum detection limits are 18 µg/g (Fe), 43

µg/g (S), 115 µg/g (Mn), 23 µg/g (Ca) and 61 µg/g (As). However, the

method detection limit for As appears to be somewhat higher (~300 µg/

g; see Section 3.2).

2.4.3. Powder X-ray diffraction (XRD)

Aliquots of the run products (~1–3 mg) from all experiments were

analyzed using a Rigaku SmartLab XRD at the MAGMA Lab of the TU

Berlin. The instrument is equipped with a copper X-ray tube. For the

analysis, the run products were powdered using an agate mortar and

evenly distributed on a silicon monocrystal. Measurements were con-

ducted between 2θ =5–80◦with a step size of 0.01◦and a scan speed of

10◦per minute, while rotating the sample with 20 revolutions per

minute. Phase identification on the resulting spectra was performed

using the PDF4 database (ICDD), integrated in the Smartlab Studio II

software (Rigaku). Rietveld analysis was employed to quantify the

mineral phase contents, beginning with the most prominent phases

Table 1

Overview of pyrite synthesis experiments.

Exp.

Initial

fluid

Buffer (per 250 ml H

O) pH initial siderite

mass

w/r final As

fluid

(µg/

RSD (

Product phases (wt %)

As (µg/g) acetic acid

(ml)

Na-acetat

(mg)

initial final (mg)

V3_1 0.001 12.91 43.5 4.93 4.93 5.123 5856 0.0019 13.6 py (47) sd (21) pyh (19) mrc (13)

V3_2 0.001 12.91 43.5 4.93 4.93 5.089 5895 0.0017 19.4 py (56) pyh (24) sd (16) mrc (4)

V3_3 0.001 12.91 43.5 4.93 4.93 5.110 5871 0.0019 6.5 sd (32) py (32) mrc (20) pyh (11)

V1_5 0.01 12.91 43.5 4.96 4.98 5.160 5814 0.0106 8.7 mrc (36) py (34) pyh (16) sd (13)

V1_6 0.01 12.91 43.5 4.96 4.92 5.136 5841 0.0072 9.7 sd (54) py (35) mrc (4) pyh (8)

V1_7 0.01 12.91 43.5 4.96 4.92 4.994 6007 0.0054 17.4 sd (37) mrc (32) py (18) pyh (5) qz (8)

V2_1 0.1 12.91 43.5 4.95 4.89 5.126 5853 0.0673 7.4 sd (45) py (37) pyh (10) mrc (6) aso (tr)

V2_2 0.1 12.91 43.5 4.95 4.91 4.993 6008 0.0936 9.9 sd (85) mrc (8) py (7) mag (tr)

V2_3 0.1 12.91 43.5 4.95 4.89 5.054 5936 0.0855 2.9 sd (85) pyh (7) py (6) hem (tr)

V2_5 0.1 12.91 43.5 4.95 4.9 5.064 5924 0.0824 3.7 sd (90) mrc (5) py (4) pyh (tr)

V4_1 1 0.86 3.18 4.89 4.93 5.126 5853 0.99 1.0 py (60) hem (24) mrc (11) szo (6)

V4_2 1 0.86 3.18 4.89 4.89 5.054 5936 0.97 1.3 mrc (42) py (32) grg (15) mag (6) hem

(4)

V4_3 1 0.86 3.18 4.89 4.85 5.200 5769 0.75 1.7 mrc (66) py (29) mln (5)

V4_4 1 0.86 3.18 4.89 4.82 5.044 5948 1.07 1.0 mrc (49) py (39) mln (6) mkw (4)

V5_6 1 0.86 3.18 4.87 4.94 5.124 5855 1.238 0.7 mrc (41) py (34) mln (16) hem (7) aso,

mkw (tr)

V1_3 10 12.91 43.5 4.96 4.92 5.156 5818 9.866 9.4 sd (53) py (27) mrc (15) pyh (4)

V5_7 10 0.86 3.18 4.87 4.89 5.061 5928 7.805 0.7 mrc (45) py (42) hem (7) mln (6)

V4_5 100 0.86 3.18 4.89 4.4 5.166 5807 2.12 1.6 py (57) hem (19) mrc (13) grg (10)

V4_6 100 0.86 3.18 4.89 4.39 5.247 5718 2.31 1.8 py (71) hem (13) grg (10) sd (4) mrc (3)

V4_7 100 0.86 3.18 4.89 4.38 5.190 5780 2.15 2.9 py (47) grg (25) sd (15) mck (8) mln (6)

V4_8 100 0.86 3.18 4.89 4.35 5.124 5855 2.51 2.7 sd (47) py (35) grg (19)

To all experiments 96 mg of thioacetamide have been added (see methods section).

w/r =water to rock ratio based on the amount of siderite and total fluid volume of 30 ml assuming a density of 1 g/cm

fluid

determined by ICP-MS (see methods section). RSD =internal measurement uncertainty expressed as the relative standard deviation based on the measurement

of 3–5 replicates.

sd: siderite (FeCO

); py: pyrite (FeS

); mrc: marcasite (FeS

); pyh: pyrrhotite (Fe

(1-x)

S); grg: greigite (Fe

); mkw: mackinawite (FeS); qz: quartz (SiO

); hem: hematite

(Fe

); mag: magnetite (Fe

); mln: melanterite (Fe(H

) •H

O; aso: arsenolite (As

); szo: szomolnokite (FeSO

⋅H

O).

Phase content determined by XRD (see methods section). Sum of quantitative phase content may deviate from 100 wt % due to rounding. “tr” refers to phases with

contents ≤1 wt %.

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

(fitted properties: background, absolute x-shift, phase scale, lattice pa-

rameters, profile function, texture).

2.4.4. Micro-Raman

Micro-Raman Spectrosopy was performed using a WITec alpha 300 R

Raman imaging microscope equipped with a green laser (532nm) at the

Goethe-University of Frankfurt. The system was calibrated using the 520

−1

band of silicon wafer. Spectra were collected using a 1800 g/mm

grating and a 100×objective in the range of 130–1300 cm

−1

with a

spectral resolution of 1.3 cm

−1

and a laser spot size of ~2 µm. The laser

power was set to 1 mW in order to avoid any beam damage on the

sample. Integration time was 5 s with 10 accumulations resulting in a

total acquisition time of 50 s for each spectrum. Higher laser power of

5mW oxidised the sulfides and magnetites in the samples as indicated by

the presence of hematite in the Raman-signal. All spectra were back-

ground corrected and normalised to 1 for the most intense band in the

spectra.

2.4.5. Laser Ablation Inductively Coupled Plasma Mass Spectrometry (LA-

ICP-MS)

Quantitative trace element mapping was performed on the run

products by LA-ICP-MS during three analytical sessions (S1, S2, S3) at

the MAGMA Lab of the TU Berlin, using an Agilent 8900 ICP-MS/MS

coupled to an Analyte Excite 193 nm excimer laser (Teledyne Photon

Machines). The aerosol rapid introduction system ARIS (Teledyne

Photon Machines) was used in combination with a low-volume adapter

from Glass Expansion. Helium (99.999 %) was used as a carrier gas with

a total flow rate of 0.95 l/min (0.5 l/min cell flow, 0.45 l/min cup flow).

After plasma ignition and initialization of the He flow, m/z =42 (e.g.,

N) and m/z =31 (e.g.,

O) were recorded to monitor the

amount of air entrained in the interface tubing and ablation cell. After

counts were stable and below 30,000 cps, the Ar nebulizer gas and He

carrier gas flows were tuned to achieve m/z =248/232 ratios of <0.20

% (ThO/Th) and m/z =232/238 ratios of 100 ±1 % (Th/U), while

ablating NIST 610 in line scan mode. Small amounts (3–4 ml/min) of

high-purity (99.999 %) N

were admixed to the Ar nebulizer gas to in-

crease the sensitivity. Once stable plasma conditions were reached, an

automatic lens tune was performed on NIST 610 to maximize the

sensitivity in the mid-mass range. Laser parameters for the tuning were

50 Hz repetition rate, 5

m circular spot size, and 4 J/cm

fluence.

The single pulse response (SPR), i.e., the combined wash-in and

wash-out, was monitored by ablating NIST 610 at 1 Hz and dosage 1 in

line scan mode by recording m/z =238 (U) with a dwell time of 5 ms. By

averaging the peak shape of 300 single pulses, the SPR was determined

and was between 80 ms (S2, S3) and 110 ms for S1 (full width at 10 % of

the maximum intensity).

For mapping the elemental distribution within the target grains,

mass channels m/z =34 (S), m/z =55 (Mn), m/z =57 (Fe), m/z =72

(Ge), and m/z =75 (As) were recorded with dwell times of 20 ms for

each element in session S1, and 2.5 ms for Mn, 1 ms for Fe, and 60 ms for

As in sessions S2 and S3. Dwell times were optimized for a good signal-

to-noise ratio and to avoid carry over of the aerosol beyond the laser spot

diameter, which was between 2 and 3 µm. Applied dosages (degree of

laser shot overlap) were between 10 and 20, repetition rates between 90

and 250 Hz, and scan speeds between 18 and 37.5 µm/s. Optimization of

these parameters was based on obtaining a maximal spatial resolution

and suppressing aliasing effects, while maintaining a signal/noise ratio

of at least 20 for each mass channel. Depending on the size of the region

of interest a single mapping took between 30 and 200 min.

Grains were mapped using a raster of unidirectional scans with no

overlap between lines and a 2 s pause after each line to prevent memory

effects. At regular intervals, typically every 30 min, a set of reference

materials were ablated in 20–30 s long line scans using the same laser

and ICP-MS parameters as for the mappings. MASS-1 (USGS) was used as

the primary calibration standard and STDGL (fused glass containing

sulfides RTS-4 and CZN-3; CODES Lab) and IMER5 (Ge-Ga-S

chalcogenide glass; University of Adelaide) as secondary standards. In

each session, about 10–15 line scans of each reference material were

recorded.

Raw data processing was performed using the software HDIP (Tele-

dyne Photon machines). The background was corrected using a cubic

spline function and a linear drift correction was performed using MASS-

1. To obtain quantitative mappings, each pixel was calibrated on the

MASS-1 line scans. In a second step, segments with Fe >100,000 µg/g

and S >100,000 µg/g were defined in the mapped areas, to distinguish

sulfides (pyrite, pyrrhotite) against siderite or the embedding medium

(epoxy). To correct for differences in ablation yield between calibrant

and sample, each pixel within such a segment was normalized to a sum

of Fe +S =100 wt %. Secondary reference materials were calibrated

against MASS-1 and normalized using an internal standard (Fe for

STDGL and S for IMER5).

Repeated line scans on the primary standard MASS-1 yielded an

external precision (within session reproducibility) better than 4 % 2 RSE

(RSE =relative standard error) for As. The internal precision based on a

single line scan was better than 2.5 % 2 RSE. Concentrations of As ob-

tained for IMER5 (reference value: 25.41 µg/g) were between 32.6 µg/g

– 34.1 µg/g and for STDGL (reference value: 348 ±5.6 µg/g) between

329.4 and 321.4 µg/g. The results for individual sessions overlap within

the uncertainty and demonstrate good accuracy for As between 9 %

(based on STDGL) and 35 % (based on IMER5). Deviations are attributed

to varying matrix effects between the secondary reference materials

(glasses) and MASS1 (pressed sulfide powder).

2.5. Analysis of quenched experimental and natural fluids

Arsenic concentrations in natural and synthetic fluids were measured

using an Agilent 8900 ICP-MS/MS during four analytical sessions

(S1–S4). All fluids were mixed online with a 0.5 µg/g Ge solution that

served as an internal standard to which all analyte counts were

normalized. Samples were introduced into the Ar Plasma via a Micro

Mist nebulizer and a Scott-type double-pass spray chamber. To reduce

the plasma load, samples from sessions S1 and S2 (total dissolved solids,

TDS, ~14.5 %) were diluted by a factor of 25 and samples from session

S4 (TDS ~1.5 %) by a factor of 8 through online with Ar gas dilution

using the Ultrahigh Matrix Introduction (UHMI) system of the Agilent

8900. The ¨

Asp¨

o fluid was diluted offline by a factor of 10 using ultrapure

O (Rotipuran®Ultra). To avoid carry-over effects a two-step rinsing

procedure was applied after every standard and sample using (1) 6 %

HNO

and (2) 2 % HNO

each with a rinsing time of 30 s. Rinse solutions

were prepared from diluting 65 % HNO

(Merck Suprapure®) with UP

water. To establish the external calibration, a multi-element standard

solution (Agilent environmental standard) was diluted with ultrapure

water to achieve the desired calibration levels (see Table S2). Addi-

tionally, a commercial 5 % HNO

blank (Agilent) was used as the cali-

bration blank. Accuracy was verified by measuring a second multi-

element solution (LabKings quality control standard # 26) diluted to a

concentration of 1 and 10 ng/g levels using ultrapure H

O. The ICP-MS

was operated in He-mode to avoid isobaric interferences of

. For each sample, 3–5 replicates were measured with ten mass

sweeps per replicate and a dwell time of 0.5– 2 ms for m/z =75.

Measurement uncertainties correlate with As concentrations and

vary from a few percent for high As contents up to a RSD of about 20 %

at low As concentrations (Table 1). The highest measurement uncer-

tainty for As was observed for the ¨

Asp¨

o fluid sample (32 %, RSD, n =5).

Detection limits expressed as three times the standard deviation of the

blank divided by the slope of the linear calibration regression are <1

ng/g As (S1, S2, S4), with a minimum detection limit of 0.016 ng/g As in

session S3 (Table S2). Accuracy is estimated to be better than 10 % based

on the recovery of As from the quality check solution. The machine drift

during each analytical session was <10 %, as indicated by the recovery

of the internal Ge standard from the HNO

blank solution.

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

2.6. Geochemical modeling

Geochemical modeling was performed using the PHREEQC software

(Version 3.4.0), while species plots were generated using PhreePlot

(Version 1). To ensure the accuracy of As stability data, we utilized the

Nordstrom and Archer (2003) dataset as implemented in the wateq4f.

dat database file. Thermodynamic data for all other chemical reactions

were obtained from the Lawrence Livermore National Laboratory

database (llnl.dat).

The model aimed to examine the impact of increasing As concen-

trations on As speciation. First, a solution was equilibrated with the

atmosphere in a batch setup. Subsequently, all reactants (acetic acid, Na-

acetate, thioacetamide and siderite) were added. The batch was then

sealed and heated to 200◦C, with the gas phase implemented at a volume

1.65 times that of the water phase to mimic the experimental vessel

conditions. Finally, arsenic was incrementally added in 60 steps, with

logarithmically equal steps covering the entire range of initial fluid As

concentrations employed in the hydrothermal experiments, ranging

from 0.001 mg/L to 100 mg/L. The input code of the model is accessible

through the data repository at https://doi.org/10.5281/zenodo

.11086033.

Species plots depicting arsenic speciation were generated using

PhreePlot, employing a “hunt and track” approach with a resolution of

500. This approach provided a comprehensive understanding of the

prevailing As species under the given pH and pressure conditions. The

pH was set to the experimental pH of 4.9, while the pressure was set to

15 bar in alignment with the calculated pressure values obtained from

the PHREEQC models. The x-axis and y-axis were chosen to fix the H

and O

(gas) concentrations, respectively, enabling a specific investiga-

tion of the influence of these two parameters on As speciation. The plots

showcased the dominant arsenic species, representing the species ac-

counting for the largest number of moles of the main species. For

example, if only two species were present, the species with a fraction

greater than 50 % would be depicted as the dominant species.

3. Results

3.1. Texture and mineralogy of solid run products

Analysis by XRD, EPMA and micro-Raman showed that pyrite

formed in all 21 experiments through replacement of siderite. Pyrite is

consistently found at the outermost parts of the replaced siderite grains,

while some siderite grains also display full replacement by pyrite

(Fig. 1). In experiments with lower concentrations of the acetic acid

buffer, the pyrite yield was highest with a mean phase content of 45 wt

%. In these experiments, siderite is commonly fully replaced by pyrite

(Fig. 1) and residual siderite is only detected in 3 out of 10 experiments

(Table 1). Experiments conducted at higher buffer concentration have

lower pyrite contents (mean =28 wt %). Here, pyrite typically forms

thin rims (10–100 µm) around relict siderite, which has been detected in

all experiments at low buffer concentrations (Table 1). Pyrite occurs

mainly in two configurations: as porous aggregates with occasional

framboidal and lamellar habits and as dense anhedral to euhedral

crystals (Fig. 1). The latter predominates at higher buffer concentra-

tions, while at lower buffer concentrations, porous pyrite dominates.

Euhedral pyrite up to ~40 µm in size was observed in the central part of

fully replaced siderite and close to the siderite/pyrite interface in cases

where siderite was only partially replaced (Fig. 1).

Marcasite is detected in 19 out of 21 experiments with contents

ranging from 3 to 66 wt %. It is commonly overgrown by pyrite and its

occurrence is restricted to the siderite/pyrite interface and the inner

parts of fully replaced siderite. In the majority of the experiments, pyrite

is the dominant FeS

polymorph. Other sulfides present in the run

products are pyrrhotite (Fe

1-x

S) in the high buffer experiments and

greigite (Fe

) as well as mackinawite (FeS) in the low buffer experi-

ments. While the presence of greigite and mackinawite is evident from

XRD, pyrrhotite has additionally been identified in reflected light, where

it is overgrown by pyrite (Fig. 1). Magnetite and hematite were identi-

fied close to the siderite/pyrite interface, where magnetite forms an

overgrowth on hematite and pyrite on magnetite and hematite (Fig. 1).

Other minor phases detected by XRD are the Fe-sulfates melanterite (Fe

)⋅H

O) and szomolnokite (FeSO

⋅H

O), arsenolite (As

)

and quartz. Moreover, the run products of the experiments with high

fluid As concentrations (100 µg/g), had a strikingly yellow-orange color,

which is attributed to the presence of realgar. Its absence in the XRD

data, is explained by the rapid light-induced decomposition of realgar to

amorphous pararealgar (Pratesi and Zoppi, 2015).

3.2. Arsenic contents in pyrite close to the pyrite-fluid interface

The chemical analysis of pyrite is focused to the outermost regions of

the replaced siderite, due to the complex textural and multiphase nature

of the inner parts. To ensure high spatial resolution and sensitivity, the

As concentrations in pyrite were determined by quantitative LA-ICP-MS

mapping. The following data reduction strategy was employed. In a first

step, transects with 7–25 pixels in width and perpendicular to the fluid-

pyrite interface were inspected for their Fe/S ratio (wt. %). Only those

transects were chosen, where Fe ≤S, which indicated the presence of

pyrite (S =53. 45 wt %, Fe =46.55 wt %; Fe/S =0.87). The mean sulfur

concentrations of pyrite-assigned transects varied between 50–55 wt %

(https://doi.org/10.5281/zenodo.11086033). Other Fe/S ratios with Fe

>S are either attributed to pyrrhotite (Fe/S =1.65), mackinawite (Fe/S

=1.74), greigite (Fe/S =1.31) or a mixed analysis of pyrite +Fe-oxides

and hence were discarded. Since marcasite and pyrite are indistin-

guishable by their Fe/S alone, we restricted the analyses to the first 30

µm of the replacement rim. This procedure largely avoids the analyses of

marcasite, which is restricted to the more interior parts of the re-

placements rims (Fig. 1). Moreover, it facilitates the comparison be-

tween different experiments. From these regions, the As concentration

across all pixels perpendicular to the transect is averaged at intervals of

3 µm intervals, equaling the image pixel size. As a result, 11 As con-

centration values were obtained for each transect, corresponding to the

distance from the pyrite/fluid interface. The starting point for the

quantification interval is selected to align with the moment when con-

centrations of both Fe and S initially reach a plateau (Fig. 2). This

approach was adopted to mitigate potential biases arising from data

acquisition and processing.

For the experimental pyrite, there are two main results. First, the As

concentration decreases with distance to the pyrite/fluid interface, i.e.,

towards the relict siderite core (Figs. 2, 3a, Table 2). Occasionally,

elongated As-hot spots were observed in the pyrite cores, that are

aligned approximately parallel to the outer grain boundaries (Fig. 2f),

but their As contents never exceeded those of the pyrite directly at the

pyrite/bulk fluid interface. Secondly, the As contents in the pyrite rims

systematically increase with increasing initial As contents in the

experimental fluids (Figs. 2, 3a, Table 2). For instance, in experiments

with an initial fluid As concentration of 0.001 µg/g, the mean As content

of the pyrite rims ranged from 4.9 to 20.2 µg/g. In contrast, pyrite from

fluids with 100 µg/g As exhibited As concentrations ranging from

18,239 to 60,695 µg/g (Table 2).

Moreover, the As concentration of three euhedral pyrite grains and

two pyrite overgrowth on magnetite from ¨

Asp¨

o (cf. Section 2.3) were

determined by LA-ICP-MS mapping (Fig. 4). These display a mean As

concentrations of 9.2 ±5.9 µg/g (1 SD, n =5) (Table 3). This is within

the range of the As concentration of the synthetic pyrites from the ex-

periments with the lowest initial As contents.

The comparison of As concentrations derived from randomly

selected regions in pyrite by EPMA to the bulk compositions determined

by LA-ICP-MS revealed a good correlation along a 1:1 regression line at

concentrations >500 µg/g (Fig. S2). At lower As contents, the data of the

two methods significantly deviate, which is due to the higher minimum

detection limit of As in pyrite by EPMA (300 µg/g) compared to LA-ICP-

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

Fig. 1. Typical pyrite (py) textures in hydrothermal synthesis experiments. (a) Dense and thin pyrite rim forming at the former siderite (sd)/fluid interface in an

experiment conducted at a high buffer concentration. At the pyrite/siderite interface, magnetite (mgt) is identified. (b) Close-up of the pyrite rim, additionally

showing marcasite (mrc) close to the pyrite/siderite interface. Marcasite shows cream-white to bluish color in reflected light. Hematite (hem), red internal reflections

is partly replaced by magnetite. (c) Pyrrhotite (pyh, brownish color) overgrowing magnetite and being overgrown by pyrite at the inner siderite interface (d) porous

pyrite as typically observed in experiments conducted at lower buffer concentrations. (e) Close-up of the porous pyrite rim (f) pyrite replacing both magnetite and

hematite close to the inner siderite interface. (g) Siderite grain displaying only partial replacement by pyrite. Here pyrite additionally forms lamellar crystal shapes.

The dotted rectangle marks the region of interest as it appears after the LA-ICP-MS mapping. (h) Close-up of the lamellar pyrite habit appearing in experiments

conducted at lower buffer concentrations. (i,j) Fully replaced siderite exhibiting the characteristic inner grain phase mixture comprising anhedral to euhedral pyrite

overgrowing magnetite and marcasite. (k) Back scattered electron image of a porous lamellar pyrite rim (l) back scattered electron image of a porous framboidal

pyrite rim m) Micro-Raman spectra collected from the run products to confirm the phase identification by reflected light. Numbers correspond to the analyses lo-

cations marked in the reflected light images in a-f. Black dotted lines are comparative spectra from the respective phases from the RRUFF database (see also

https://doi.org/10.5281/zenodo.11086033 for details). All Raman spectra are normalized to a main peak height of 1.

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

Fig. 2. LA-ICP-MS mappings of As concentrations in pyrite that forms during the replacement of siderite in hydrothermal experiments. For each envisaged initial

fluid concentration one representative example is shown. An As gradient is observed throughout the experiments, with the highest values appearing in the outermost

pyrite rims. The As concentration in pyrite increases with the As concentration of the fluids (mind the different scale for individual maps). For quantification of the As

concentration in the pyrite rims, transects (white rectangles) have been extracted, which are displayed to the righthand side of each map. Only the first 30 µm after Fe

and S have reached a pleateau are considered. The As concentrations of the bulk phase assemblage of the innermost part of the replaced grains has been derived from

averaging the pixel within the elliptic segments. For further information see Section 3.2. The image pixel size for all maps is 3 µm.

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

MS (<1 µg/g at the single pixel level). Therefore, only the As contents

derived from LA-ICP-MS analysis were considered for the data presen-

tation and interpretation.

3.3. Arsenic concentration of the core phase assemblage

The inner core regions are composed of a phase mixture of mainly

pyrite, marcasite, hematite and magnetite (Fig. 1). The intimate inter-

growth of these minerals presents challenges for conducting analyses of

As concentrations in individual phases by LA-ICP-MS. Instead, bulk As

concentrations of the core phase assemblage have been determined by

averaging As concentrations within elliptic segments assigned to the

core regions of the replaced siderite grains (Fig. 2; https://doi.org/10.5

281/zenodo.11086033). The results indicate that As concentrations at

the innermost regions are about 1–2 orders of magnitude below the

maximum As concentration measured at the pyrite rim (Fig. 3a,

Table 2). Similar to the pyrite rims, we observe a general trend of

decreasing As concentrations with decreasing As concentration in the

initial fluid.

3.4. Arsenic contencrations in fluids

The As recovery from the synthetic fluids was determined by

comparing the As contents of the initial fluid with those of the quenched

fluid following pyrite synthesis (Fig. 3b). The results showed that the

recovery of As ranged from 54 to 190 %, with an average recovery of

approximately 85 %. Recoveries near 100 % indicate that pyrite growth

and consecutive As incorporation did not change the As contentration of

the bulk fluid, leading to similar As contents between the initial and

quenched fluids. Conversely, recoveries >100 %, as observed in exper-

iments with low initial As contents (0.001 µg/g) are attributed to minor

As contamination at the 1 ng/g level likely originating from the siderite

starting material (cf. Section 2.1, Fig. S1). In runs that were conducted in

PTFE beakers previously used for fluids with 100 µg/g As, even higher

levels of contamination of ~1 µg/g were observed (Fig. 3b). We attribute

this type of As contamination to residual memory effects that remain

despite of the thorough cleaning protocol for the PTFE beakers (cf.

Section 2.2). These experiments were thus excluded from the present

study, to avoid any impact of cross-contamination effects on the pro-

cessed data. In the experimental runs V4_5 to V4_8, precipitation of

realgar resulted in a significant loss of As, with recoveries ranging from

only 2.1 % to 2.5 % (Fig. 3b).

The natural groundwater samples from two separate boreholes

within the ¨

Asp¨

o Hard-Rock-Laboratory (cf. Section 2.3) displayed low

As concentrations of 0.3 and 0.4 ng/L (Table 3).

4. Discussion

4.1. Textural evolution and passivation during replacement of siderite by

pyrite

The replacement of siderite follows an outward-to-inward progres-

sion, closely associated with pyrite formation through an interface-

coupled dissolution-reprecipitation mechanism (Fig. 5). This connec-

tion is supported by the enduring rhombic siderite habit (Fig. 1g,i),

which persists even in cases of complete siderite replacement by pyrite,

suggesting an epitaxial relationship between the two phases (Putnis,

2014). As such, the siderite replacement proceeds via a two-step process,

starting with the dissociation of H

S via:

(1) H

S → H

+HS

−

This is followed by coupled siderite dissolution and pyrite repreci-

pitation via:

(2) FeCO

+2 HS

−

→ FeS

+HCO

−

Siderite replacement is not always complete, particularly not in ex-

periments conducted at higher buffer concentrations (Fig. 1). This is

attributed to the formation of pyrite acting as a passivating agent at the

siderite/fluid interface. Similar observations were made during other

dissolution-reprecipitation reactions, such as the weathering of wollas-

tonite, feldspar and silicate glass (Nugent et al., 1998; Geisler et al.,

2010; Ruiz-Agudo et al., 2016; Kutzschbach et al., 2018). Surface altered

layers (SAL) were described in this context, that limit the fluid supply to

the inner pristine mineral/glass – fluid interface, which usually results in

a decrease of the dissolution rate. Depending on the porosity and

thickness of the SAL, further dissolution is limited by volume diffusion

through solid instead of fluid transport along connected pore space

within the SAL. Hence, in the case of incomplete siderite replacement, it

is expected that pyrite represents a passivating agent, with siderite

dissolution being increasingly controlled by the supply and removal of

2−

, H

and HCO

−

ions that can still penetrate the previously formed

pyrite rim. In experiments with high buffer concentrations, the domi-

nance of dense anhedral pyrite is observed early on (Fig. 1a), resulting in

prompt passivation of the inner siderite/pyrite interface. Pyrrhotite,

distance to pyrite/fluid interface

0 µm . . . . . . 30 µm

10000

100000

0.1

100

1000

Asfluid,initial (µg/g)

As (µg/g)

0.001 0.01 0.1 1 10 100

pyrite rim

core phase assemblage 0.001

0.01

0.1

100

0.001 0.01 0.1 1 10 100 1000

As contamination

precipitation of

realgar (AsS)

Asfluid,measured (µg/g)

Asfluid,initial (µg/g)

a) b)

Fig. 3. (a) Arsenic concentrations in pyrite close to the pyrite/bulk fluid interface. Black and white colors denote the location within the transect across the pyrite

rim, with black meaning closest to the pyrite/bulk fluid interface. Stars indicate the As concentrations of the bulk phase assemblage of the innermost part of the

replaced grains. The locations of the rim transects and core segments is illustrated in Fig. 2 and in the data repository https://doi.org/10.5281/zenodo.11086033. (b)

Plot of the As concentrations in the initial experimental fluid vs. As measured in the quenched experimental fluid. The experiments marked in green follow a 1:1 trend

(dashed line) indicating no significant loss of arsenic or arsenic contamination. Yellow experiments suffer from As loss due to precipitation of realgar. Red exper-

iments suffer from As-contamination and thus were not considered in this study. Fluid data for green and yellow experiments are compiled in Table 1.

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

Table 2

Arsenic concentrations of pyrite rims close the pyrite/bulk fluid interface and D

(py/fluid)

partitioning coefficients obtained from the hydrothermal synthesis experi-

ments. Additionally, the As concentrations (±1 SE, abs) of the core phase assemblage is presented.

Distance from pyrit-

fluid interface (µm)

As (µg/g) D

(py/fluid)

1 SE (

As (µg/g) D

(py/fluid)

1 SE (

As (µg/g) D

(py/fluid)

1 SE (

As (µg/g) D

(py/fluid)

1 SE (

V3_1

(grain 1)

(As

ini

0.001 µg/g)

(n =

13)

V3_1

(grain 2)

(As

ini

0.001 µg/g)

(n =

15)

V3_2 (As

ini

0.001 µg/g)

(n =

12)

V3_3 (As

ini

0.001 µg/g)

(n =

0 25.8 25,830 5.2 28.4 28,431 15.2 17.5 17,515 6.4 30.1 30,125 27.2

3 18.4 18,438 6.0 20.1 20,077 8.2 11.2 11,202 7.0 30.1 30,055 12.7

6 9.0 9034 6.0 15.9 15,885 13.7 5.2 5157 8.8 31.6 31,585 10.3

9 5.9 5923 5.3 11.6 11,584 19.8 2.4 2407 8.1 30.5 30,522 9.0

12 6.1 6099 10.8 9.1 9141 23.9 2.3 2264 7.6 28.9 28,858 17.3

15 4.5 4509 11.3 7.2 7195 26.9 2.3 2270 10.1 22.0 21,953 17.5

18 3.0 2998 5.2 3.8 3753 29.6 2.4 2377 13.5 14.7 14,736 23.3

21 3.1 3150 3.1 2.2 2240 30.0 3.1 3066 9.8 9.2 9218 37.4

24 3.3 3322 3.8 1.6 1562 21.8 3.5 3456 4.3 13.0 13,027 85.1

27 4.2 4190 3.7 1.3 1288 10.0 2.6 2631 7.6 7.0 6998 83.0

30 4.7 4695 6.9 1.2 1152 9.5 2.0 2018 7.2 4.9 4889 92.2

Core 3.8 ±0.2 1.1 ±0.1 0.7 ±0.1 4.9 ±0.2

V1_5 (As

ini

0.01 µg/g)

(n =

11)

V1_6 (As

ini

0.01 µg/g)

(n =

16)

V1_7 (As

ini

0.01 µg/g)

(n =

14)

V2_1 (As

ini

=0.1

µg/g)

(n =

11)

0 59.4 5942 7.2 42.4 4241 10.8 187 18,668 11.6 667 6667 5.8

3 24.0 2402 5.3 32.3 3230 7.1 193 19,342 10.2 687 6873 6.9

6 10.2 1024 4.6 26.0 2601 12.1 177 17,723 8.0 705 7055 7.3

9 6.2 621 6.8 24.7 2466 11.2 134 13,410 7.9 736 7358 6.6

12 4.8 479 11.1 20.8 2078 9.8 111 11,084 6.7 782 7821 5.3

15 4.4 443 12.1 17.7 1775 11.5 83.3 8331 6.7 757 7568 7.8

18 5.1 508 10.0 14.0 1396 12.8 42.5 4252 6.3 489 4889 10.5

21 5.8 585 10.6 8.2 815 10.0 18.6 1863 11.3 201 2014 14.3

24 6.0 596 12.8 4.1 408 10.8 9.3 931 29.1 81.0 810 16.5

27 5.9 587 12.0 2.5 252 11.0 5.4 539 32.4 48.3 483 15.4

30 6.1 612 9.7 2.0 195 6.7 3.8 379 22.1 27.5 275 19.9

Core 1.4 ±0.1 13.8 ±

4.8

0.9 ±0.1 14.6 ±

0.8

V2_2 (As

ini

=0.1

µg/g)

(n =

13)

V2_3 (As

ini

=0.1

µg/g)

(n =

15)

V2_5 (As

ini

=0.1

µg/g)

(n =

11)

V4_1 (As

ini

µg/g)

(n =

17)

0 268 2677 10.2 584 5840 6.6 316 3162 8.0 2936 2936 4.5

3 186 1859 6.5 444 4444 9.1 326 3261 10.1 991 991 3.9

6 165 1653 4.5 332 3316 8.8 337 3375 17.7 647 647 10.9

9 126 1264 8.2 305 3052 8.0 218 2182 8.4 600 600 12.6

12 68.0 680 8.3 294 2942 14.4 108 1077 13.6 469 469 9.5

15 33.1 331 10.2 173 1728 13.7 49.8 498 24.0 401 401 6.9

18 12.9 129 18.6 75.9 759 11.9 20.2 202 28.4 476 476 15.7

21 9.2 92 39.7 31.9 319 11.4 12.3 123 23.3 434 434 11.9

24 4.7 47 38.3 14.2 142 12.6 8.2 82 23.3 399 399 11.4

27 2.9 29 23.3 8.2 82 13.2 4.3 43 20.8 480 480 21.9

30 2.8 28 11.4 4.6 46 13.0 2.7 27 26.1 751 751 33.2

Core 46.5 ±

2.4

0.8 ±0.1 166 ±1 255 ±5

V4_2 (As

ini

µg/g)

(n =

15)

V4_3 (As

ini

µg/g)

(n =

19)

V4_4 (As

ini

µg/g)

(n =

14)

V5_6 (As

ini

µg/g)

(n =

19)

0 4843 4843 6.4 846 846 2.9 1054 1054 4.5 1107 1107 2.8

3 3962 3962 7.3 647 647 2.9 1157 1157 3.2 1061 1061 2.8

6 2192 2192 6.3 395 395 4.1 1055 1055 3.5 922 922 4.0

9 787 787 8.1 220 220 2.9 959 959 5.1 734 734 6.1

12 371 371 6.9 179 179 5.3 1001 1001 3.7 543 543 7.5

15 464 464 4.5 217 217 9.0 1004 1004 2.9 438 438 12.2

18 591 591 3.8 300 300 6.8 1094 1094 2.0 393 393 13.4

21 484 484 3.6 311 311 3.4 1129 1129 2.6 357 357 14.5

24 304 304 2.7 264 264 5.8 1032 1032 1.7 302 302 12.8

27 232 232 3.0 190 190 10.7 972 972 0.9 285 285 8.9

30 223 223 5.7 151 151 12.4 945 945 1.8 292 292 11.0

Core 685 ±2.1 444 ±9 360 ±8 334 ±6

V1_3 (As

ini

=10

µg/g)

(n =

20)

V5_7 (As

ini

=10

µg/g)

(n =

12)

V4_5 (As

ini

100 µg/g)

(n =

25)

V4_6 (As

ini

100 µg/g)

(n =

0 31,496 3150 18.0 9866 987 1.5 62,117 621 3.0 57,983 580 2.9

3 16,426 1643 8.7 9332 933 1.3 59,474 595 2.3 65,850 658 2.7

6 16,946 1695 5.5 8612 861 1.3 61,216 612 2.9 75,440 754 2.1

9 19,324 1932 4.0 7557 756 3.0 59,252 593 3.7 69,352 694 2.3

12 18,006 1801 6.1 7099 710 4.0 54,805 548 2.4 61,873 619 3.6

15 18,862 1886 11.6 6894 689 4.2 59,223 592 1.6 51,316 513 6.1

(continued on next page)

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

detected by XRD, is interpreted as an intermediate or precursor phase

during pyrite formation, with subsequent overgrowth by pyrite (Fig. 1c),

in accordance with previous studies (Schoonen & Barnes, 1991; Qian

et al., 2011, Yao et al., 2021). Other pyrite precursor phases, such as

mackinawite (Lennie and Vaughan, 1996; Rickard and Luther, 2007)

and greigite (Schoonen and Barnes, 1991, Hunger and Benning, 2007),

exclusively appear in experiments conducted at lower buffer concen-

trations (Table 1) and are considered as intermediate products during

hydrothermal pyrite formation. We speculate that the nature of the

precursor phase influences the crystal shape of the resulting pyrite. In

experiments with low buffer concentrations, porous pyrite dominates,

facilitating ongoing migration of bulk fluid to the siderite/pyrite inter-

face. Consequently, passivation is less effective compared to experi-

ments with higher buffer concentrations, leading to the formation of

larger cavities filled with inner fluid.

The consistent occurrence of marcasite is intriguing, as marcasite

formation is typically associated with highly acidic pH <4 (Qian et al.,

2010, 2011). However, the sodium acetate acetic acid used in the hy-

drothermal experiments buffers at pH =4.89, with the lower pH in

quenched fluids being 4.35 (Table 1). The onset of marcasite formation

at slightly higher pH <5, as observed by Murowchick and Barnes

(1986), provides a plausible explanation for marcasite formation.

Moreover, Qian et al. (2011) highlighted the importance of the fluid

saturation state with respect to pyrite/marcasite, indicating that

marcasite is favored over pyrite in S

2−

deficient solutions (SI ≪ 1000).

This situation is encountered close to the siderite/pyrite interface during

earlier stages of the replacement reaction. In the immediate vicinity, the

chemical composition of the so-called “inner fluid” is expected to

deviate from that of the surrounding bulk fluid (Fig. 5). This is due to the

passivating outer pyrite rim that limits the supply of H

S-rich bulk fluid.

Over time, the composition of the inner fluid transitions to the compo-

sition of the bulk fluid. This transformation is marked by the precipi-

tation of pyrite as the stable FeS

polymorph, which is evident from the

frequent observation of pyrite overgrowing marcasite in the core regions

of the replaced siderite (Fig. 1b, j). Accordingly, the occurrence of

Table 2 (continued)

Distance from pyrit-

fluid interface (µm)

As (µg/g) D

(py/fluid)

1 SE (

As (µg/g) D

(py/fluid)

1 SE (

As (µg/g) D

(py/fluid)

1 SE (

As (µg/g) D

(py/fluid)

1 SE (

18 23,719 2372 5.6 7299 730 5.5 60,800 608 2.9 41,851 419 6.3

21 11,856 1186 3.9 7454 745 5.9 61,094 611 1.5 47,248 472 7.2

24 3401 340 11.6 7621 762 3.9 63,363 634 1.7 43,906 439 10.5

27 1032 103 15.9 7369 737 2.5 63,672 637 3.2 43,918 439 12.7

30 701 70 11.7 6987 699 2.3 62,626 626 3.1 27,759 278 11.7

Core 1373 ±

2098 ±

48948 ±

382

63462 ±

539

V4_7 (As

ini

100 µg/g)

(n =

15)

V4_8 (As

ini

100 µg/g)

(n =

12)

0 36,146 361 3.9 35,264 353 4.5

3 41,566 416 2.9 27,931 279 7.5

6 44,536 445 2.7 17,776 178 5.0

9 47,958 480 2.2 14,534 145 5.1

12 48,769 488 1.3 14,330 143 3.5

15 46,258 463 1.1 14,689 147 5.1

18 46,033 460 0.9 14,031 140 3.6

21 46,231 462 0.7 14,483 145 3.5

24 47,169 472 1.4 15,991 160 5.7

27 45,980 460 1.6 15,013 150 2.8

30 45,001 450 0.9 16,593 166 6.2

Core 24252 ±

136

n denotes the number of pixel that were averaged perpendicular to the concentration profile (see Fig. 2).

SE denotes the standard error at the 1 s level (standard deviation divided by √n).

Fig. 4. Pyrite that formed in a borehole within the ¨

Asp¨

o hard-rock laboratory

over 17 years laboratory (see methods section). The BSE image in (a) shows a

fragment of a euheudral dense pyrite crystal that displays elevated and zoned

As concentrations as revealed by LA-ICP-MS mappings (b). Pyrite also occurs as

porous overgrowth on magnetite and here also incorporates several µg/g of As

(d). Image pixel size is 3 µm. The As concentration of the host fluid is between

0.3–0.4 ng/g. Gold particles are remnants from a gold coating required for

SIMS analyses.

Table 3

As concentrations and partition coefficients for natural pyrites and fluid from the

Asp¨

o hard-rock laboratory determined by (LA)-ICP-MS.

Pyrite crystal habit Borehole As (µg/g) D

(py/fluid)

Pyrite Fluid

Euhedral KA3105 A:3 18.5 0.0003 61,667

Euhedral KA3105 A:3 10.5 35,000

Euhedral KA3105 A:3 2.8 9333

Porous on magnetite KA3385 A:1 6.2 0.0004 15,500

porous on magnetite KA3385 A:1 8.2 20,500

mean 28,400

1 SD 20,872

(1SE) for individual pyrite are better than 10 %.

(1RSD) for fluid is 30 %.

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

hematite and magnetite at the siderite/inner fluid interface can be

explained (Fig. 1c,f,i,j). Here, the high oxygen fugacity (fO

) required

for the formation of the Fe-oxides is maintained due to the limited

interaction of inner fluid and bulk fluid. The occurrence of szomolnokite

and melanterite is attributed to oxidation after opening of the beakers,

where HS

−

oxidizes to SO

2−

, which eventually precipitates along with

dissolved Fe

as melanterite or szomolnokite. A summary of the most

important observations and conclusions regarding the textural evolution

during siderite replacement by pyrite is presented in Fig. 5.

Pyrite with lamellar habit as observed in some experiments (Fig. 1g,

h) is rather uncommon in nature. Rare natural analogue could be

“feathery” or “wispy” pyrite habits observed in the Vatukoula epi-

thermal Au-Ag-Te deposit, Fiji (Pals et al., 2003; B¨

orner et al., 2021) and

the Moonlight epithermal Au prospect, Queensland, Australia (Wind-

erbaum et al., 2012). The exact crystallization mechanism of lamellar

pyrite remains unclear, but crystal habits with high aspect ratios are

known to be associated with precursor-mediated nucleation in hydro-

thermal synthesis experiments (Kutzschbach et al., 2016).

4.2. Nature of extreme As enrichment and zoning

A distinct feature of the pyrite products is the apparent zoning in As

concentrations (Fig. 2). In general, we interpret the mechanism of As

incorporation to involve a coupled adsorption/absorption process

(Bostick and Fendorf, 2003; Han et al., 2013; Kusebauch et al., 2018).

Thermodynamic modeling results demonstrate that the activity of As

species exhibits variations depending on the fluid As concentration. At

lower As concentrations, the thioarsenite species AsS(OH)(HS)

−

domi-

nates, while As

(HS)

−

becomes increasingly significant as As con-

centrations rise, eventually becoming the dominant species at As

concentrations of 100 mg/L (Fig. 6). Similar to what has been inferred

for H

AsO

and H

AsO

−

(J¨

ornsson and Sherman, 2008; Guo et al.,

2011), it is expected that these thioarsenite species first adsorb onto the

surface of siderite/pyrite. Subsequently, they become absorbed by

newly forming pyrite through co-precipitation with iron (Fe) and sulfur

(S) derived from thioacetamide breakdown and siderite dissolution

(Bostick and Fendorf, 2003; Kusebauch et al., 2018). Changes in As

speciation during individual experiments could potentially impact the

incorporation and resulting As concentration in pyrite (Qian et al.,

2013). However, mass balance calculations indicate that even a com-

plete replacement of the siderite fraction (5 mg) would only lead to a

change of approximately 0.03 log H

S units, considering the high w/r

ratios of 6000 encountered in the hydrothermal experiments. This shift

is considered insignificant with respect to the As speciation in the fluid,

regardless of the fluid As concentration (Fig. 6). Therefore, changes in

the As speciation in the fluid during an experimental run do not influ-

ence the As zoning in the product pyrite. Nevertheless, it can be spec-

ulated that the pronounced As partitioning, particularly at lower As

concentrations, may be associated with a potentially higher affinity of

the AsS(OH)(HS)

−

species to the surfaces of siderite and/or pyrite, in

contrast to As

(HS)

−

, which becomes increasingly dominant as fluid

As concentrations decrease.

In a recent study, Xing et al. (2019) introduced a mechanism of As

incorporation into pyrite through prolonged exposure to As-bearing

fluids. The study showed that the interaction between pyrite and a

basalt-buffered fluid with an As concentration of 20 µg/g, can increase

the As contents in pyrite by an order of magnitude from 0.6 to 5.7 wt %.

The authors ascribe this to a remobilization and reprecipitation process,

which is increasingly efficient at higher fluid/rock ratios. While the

highest fluid/rock ratio in Xing et al. (2019) is 120, the fluid/rock ratios

in our experiments are even higher (~6000). Thus a similar self-

enrichment mechanism may steer the observed extreme As

Fig. 5. Schematic drawing summarizing the most important observations with respect to the textural evolution during hydrothermal replacement of siderite by

pyrite. For further information see Section 4.1.

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

partitioning from the fluid into in the rim domains of the pyrite products

in our experimental study (Fig. 2). Early forming pyrite thus is expected

to have low initial As contents, which increases over time by continuous

interaction with the surrounding large fluid reservoir. Consequently,

early-formed pyrite experiences a larger time-integrated fluid/solid

ratio, than pyrite that formed later. In analogy to the findings by Xing

et al. (2019), this would lead to higher As concentrations in early pyrite

compared to late pyrite, corresponding to the As zoning between the rim

and core domains in pyrite from our experiments (Fig. 2).

In nature, the formation of porous (i.e., framboidal, lamellar) pyrite

is commonly related to fast nucleation rates induced by strong physi-

cochemical gradients (e.g., temperature), whereas euhedral pyrite

typically precipitates at more stable fluid conditions. Moreover, fram-

boidal pyrite is often associated with microbial processes (e.g. Widodo

et al., 2010, Beck et al., 2011). Natural euhedral pyrite is often found to

be depleted in various trace elements compared to coexisting porous

pyrite (Deditius et al., 2011; Meng et al., 2020; Grant et al., 2018;

Rom´

an et al., 2019; Falkenberg et al., 2021, B¨

orner et al., 2021). This is

in line with the observations in our study (Fig. 2b) and the results ob-

tained by Prokofiev et al. (2022), who observed that coarsening during

pyrite recrystallization in synthesis experiments resulted in decreasing

As contents in pyrite. Hence, we conclude that the As enrichment at the

rim compared to the core of the synthetic pyrites, was enhanced by fast

nucleation rates at disequilibrium conditions during early and rapid

formation of porous pyrite as opposed to euhedral and dense pyrite.

Alternatively, the As depleted core domains could be explained by the

pyrite rim that acted as a filter adsorbing most of the As from the

experimental fluid, which limited the amount of As available at the

siderite-pyrite interface (Fig. 5). We note that the aforementioned pro-

cesses are not mutually exclusive and might act together to produce the

observed extreme As enrichment and As gradients in the product pyrites

of our experiments. The elongated As-hot spots that are occasionally

found in the pyrite cores (Fig. 2f) are interpreted to result from early

pyrite formation along open cleavage planes in the siderite starting

material (Fig. S1).

4.3. As partitioning in low versus high As fluids

The precipitation of pyrite or any other phase did not significantly

impact the As concentration of the bulk fluid. This conclusion is sup-

ported by the high overall recovery of As from the quenched fluid,

facilitated by employing a high fluid-to-solid ratio (~6000).

Fig. 6. (a) Dominant solid phases as a function of oxygen fugacity and H

S activity under As free conditions demonstrating that pyrite is the stable phase in the

hydrothermal experimental system. Due to the high w/r ratio of 6000, the change in H

S activity due to pyrite formation is insignificant. (b) Dominant aqueous

arsenic species at the experimental oxygen fugacity (log fO

=-43.9) and temperature (T =200 ◦C) as a function of arsenic concentration. The range of fluid As

concentrations resembles the range investigated in this study. At lower As concentration the thioarsenite species AsS(OH)(HS)

−

dominates, while As

(HS)

−

be-

comes progressively more significant as As concentrations increase, ultimately becoming the dominant species at As concentrations of 100 mg/l. Predominance

regions of all stable aqueous species as a function of oxygen fugacity and H

S activity for lowest and highest As concentrations are presented in c) and d). Pre-

dominance regions for all As concentrations are accessible via the supplementary material (Fig. S3). Details of the PHREEQC model are presented in the methods

section and the input code can be accessed through the data repository https://doi.org/10.5281/zenodo.11086033.

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

Consequently, we calculated pyrite-fluid partitioning coefficients (D

(py/

fluid)

) using the As concentration of the initial experimental fluid (As

fluid,

initial

) and the As concentrations measured in pyrite close to the pyrite/

fluid interface (Fig. 7, Table 2). The innermost part of the replaced

siderite grains is not considered for calculating partitioning coefficients

for any of the present minerals. This due to (1) The multiphase and

intertwined nature of the innermost phase assemblage hinders the ac-

curate determination of As concentrations in individual phases with

certainty and (2) the chemistry of the inner fluid likely differs from the

bulk experimental fluid. This is evidenced by the occurrence of marca-

site, which we ascribed to a lower SI compared to the bulk fluid (see

Section 4.1). Similar to the anticipated S

2−

deficiency in the inner so-

lution, it is expected that the As concentrations are lower compared to

those in the bulk solution. However, the exact As concentration of the

inner fluid remains unknown.

Currently, data for D

(py/fluid)

values in reduced hydrothermal systems

(e.g. log fO

= − 43.9, Fig. 6) are limited to fluids with relatively high As

contents of 1–100 µg/g (Fig. 8), to which we refer as the high-As range in

the following. Our study confirms previous findings within the high-As

range (>1 µg/g As

fluid, initial

), where most D

(py/fluid)

values is <2000

(Figs. 7 and 8, Table 2; Reich et al., 2005; Su et al., 2008, 2009, 2012;

Qian et al., 2013; Large et al., 2016; Kusebauch et al., 2018). This is true

also for experiments with an initial As concentration of 100 µg/g. In

these experiments, realgar was identified as an additional byproduct (cf.

Section 3.1). We conclude due to the good agreement of our and pre-

viously published D

(py/fluid)

values (Fig. 8), that realgar forms as a

quench phase after the experimental runs during cooling (cf. Section

2.1) and does not affect the As budget and partitioning during the

experimental runs. In addition, an increase in the D

(py/fluid)

values is

observed in the high-As range with decreasing initial As fluid contents,

particularly with respect to the outermost parts of the pyrite rims

(Fig. 7). This trend agrees with previous results from hydrothermal ex-

periments conducted at pH 4.7 and 5.8 (Fig. 8; Kusebauch et al., 2018).

In fluids with initial As contents <1 µg/g, the D

(py/fluid)

values further

increase until reaching a maximum value of ~30000 in experiments

with the lowest As concentrations of 0.001 µg/g As in the initial

experimental fluids. These values are at least an order of magnitude

higher than the D

(py/fluid)

values of the high-As fluids (≥1 µg/g) from the

same pyrite domain. It is crucial to acknowledge that the computed D

(py/

fluid)

values for the inner parts of the pyrite rims should be interpreted as

minimum values. This is because the pyrite/siderite interfacial fluid

likely had lower As contents compared to the surrounding bulk fluid,

which we relate to As-scavenging by the early forming pyrite rims (cf.

Section 4.2).

A similar trend of increasing D

(py/fluid)

values is displayed by the data

presented by Prokofiev et al. (2022), which however exhibit an overall

shift towards lower D

(py/fluid)

values at a given fluid As concentration

(Fig. 8). This is attributed to the different experimental (cf. Section 2.1)

set up of Prokofiev et al. (2022), where FeS (i.e., pyrrhotite) or FeOHHS

were replaced by pyrite at elevated fluid temperatures of 350◦C. This is

important, as As preferentially partitions into the fluid relative to pyrite

at higher temperatures (Xing et al., 2019), which explains the overall

lower D

(py/fluid)

compared to our experimental runs.

Experiments with fluid As concentrations ranging from 1 to 100 µg/g

were conducted at lower buffer concentrations, resulting in the pre-

dominant formation of porous pyrite. Conversely, experiments with

fluid As concentrations below 1 µg/g were performed at higher buffer

concentrations, leading to the formation of dense pyrite. These differing

pyrite morphologies could introduce additional effects on the deter-

mined empirical D

(py/fluid)

values. Kusebauch et al. (2019) observed that

in experiments where both porous and dense euhedral pyrite formed,

the As concentration in euhedral pyrite was approximately 1.2–2.5

times higher than in porous pyrite (see their Fig. 6). This disparity was

attributed to the non-equilibrium conditions of rapid replacement re-

actions that form porous pyrite, as opposed to the slower growth of

dense euhedral pyrite under partially equilibrated conditions. Interest-

ingly, this general observation aligns with the As concentrations found

in the ¨

Asp¨

o pyrite, where, on average, euhedral pyrite contains about 1.5

times more As compared to porous pyrite (Table 3). However, our D

(py/

Asfluid,initial (µg/g)

DAs

(py/fluid)

0.001 0.01 0.1 1 10 100

1000

10000

100000

100

Distance to pyrite/fluid interface

0 µm . . . . . . . . . . 30 µm

Fig. 7. As partitioning coefficients D

calculated from the As concentration in

pyrite divided by the As concentration in the intital fluid fluid. Only LA-ICP-MS

data are shown. Black and white colors denote the location within the transect

across the pyrite rim, with black meaning closest to the pyrite/bulk fluid

interface. Note (1) the increasing partitioning of As into pyrite with decreasing

As concentrations in the fluid and (2) the decreasing D

with increasing

distancing to the pyrite/bulk fluid interface. Data are compiled in Table 2.

Fig. 8. Comparison of D

(py/fluid)

values obtained in this study to other experi-

mental studies and natural systems. Data of Kusebauch et al., 2019, Qian et al.,

2013 and Prokofiev are from replacement experiments using siderite, magnetite

and FeS/FeOHHS as precursor phases. Results from Bostick and Fendorf (2003)

are from sorption experiments. Data for natural Carlin type gold deposits

(CTDG) are from two localities in China (Su et al., 2008; Su et al., 2009, Su

et al., 2012) and Nevada (Large et al., 2016; Reich et al., 2005). Data for

geothermal systems are from Libbey and Williams-Jones (2016) and Har-

dard´

ottir et al. (2009) who studied fluids and pyrites from the Reykjanes

geothermal system in Iceland. Data for the seafloor massif sulfide are from an

actively forming hydrothermal mound on the TAG segment of Mid-Atlantic

ridge Grant et al. (2018). Here the mean value and the 5th and 95th percen-

tile are shown. The epithermal system is represented by epithermal veins at the

Rosia-Poieni porphyry Cu-Au deposit in Romania (Kouzmanov et al., 2010,

Deditius et al., 2014). Tabulated data is available from the data repository

https://doi.org/10.5281/zenodo.11086033.

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

fluid)

values determined for porous pyrite at fluid As concentrations be-

tween 1 and 100 µg/g are in good agreement with previously published

data (Fig. 8), in particular with the results of Kusebauch et al., 2019,

whose partitioning coefficients where exclusively based on dense

euhedral pyrite. Even if we consider a 2.5-fold increase in the D

(py/fluid)

for fluid As concentrations between 1 and 100 µg/g, the trend of

increasing D

(py/fluid)

with decreasing fluid As concentrations remains

evident. The D

(py/fluid)

values for fluid As concentrations of 0.001 µg/g

exceed those for fluid As concentrations of 100 µg/g by an order of

magnitude, up to a factor of approximately 50 (Table 2).

4.4. Comparison with natural systems and implications

4.4.1. Groundwater

The formation of stable arsenopyrite-like surface layers (FeAsS) due

to As adsorption on pyrite and siderite is described as an effective

mechanism for mitigating As contaminations under reducing conditions

in groundwater (Bostick and Fendorf, 2003; Bostick et al., 2004;

J¨

ornsson and Sherman, 2008). This is confirmed by experiments per-

formed by Bulut et al. (2014), who achieved a reduction of As concen-

tration in the fluid by up to 99 % due to As adsorption on fine-grained

pyrite. The initial As concentration in the fluids investigated by the

aforementioned studies was between ~4 µg/g (Bostick et al., 2004) and

10 µg/g (Bulut et al., 2014), which also lies in the As range of our

experimental fluids. Our extremely high D

(py/fluid)

values (up to

~30,000, Table 2) for low-As fluids (<1 µg/g) demonstrate that As

adsorption on pyrite is also effective in less contaminated fluids that

display As concentrations in the ng/g to µg/g range. This is underlined

by the ¨

Asp¨

o pyrite that shows bulk As concentrations between 2.8 and

18.5 µg/g, despite of its formation from fluids with very low As contents

of only 0.0003–0.0004 µg/g (Table 3). Interestingly, the formation of the

Asp¨

o pyrite is related to microbial activity that provides H

S from SO

2−

reduction (Drake et al., 2015, Yu et al., 2019, Pidchenko et al., 2023)

and bio-enhanced As trapping by pyrite has already been successfully

tested as a technically feasible strategy for groundwater remediation at

an industrial site in Florida (Saunder et al., 2018). Saunder et al. (2018)

showed that the As content of a fluid can be reduced from 0.3–0.5 µg/g

to 0.05 µg/g within a few weeks by precipitation of pyrite.

To assess the effect of D

(py/fluid)

on the As sequestration potential, we

calculated the As decrease in a 1 ng/g As fluid as a function of the fluid

mass, based on the formation of 100 mg of pyrite (Fig. 9; Table S4). The

results show that for 1 l of water (fluid/solid ratio =10,000) and at a

(py/fluid)

value of 30,000, the As concentration in the fluid would

experience a decrease of ~75 % to 0.25 ng/g. By contrast, previously

published D

(py/fluid)

values that are in the range of 100–1000 (Fig. 8)

would result in decrease of As concentration in the fluid of only 1–10 %.

This calculation demonstrates the potential of siderite replacement in

effectively scavenging As from groundwater into the pyrite structure, in

addition to adsorption on the pyrite surface. Such information plays a

critical role in designing (cost) efficient remediation strategies.

4.4.2. Hydrothermal systems

The comparison of our experimental D

(py/fluid)

values with those

derived from natural hydrothermal systems by coupled pyrite-fluid

analysis yielded comparable, but also strongly diverging results

(Fig. 8). The D

(py/fluid)

values from ¨

Asp¨

o (this study) and from the

actively forming TAG seafloor massive sulfide deposit at the Mid-

Atlantic Ridge (Grant et al., 2018) are representative for natural (hy-

drothermal) systems with low-As fluids (<1 µg/g) and are in perfect

agreement with our experimentally derived results (Fig. 8). Similarly,

(py/fluid)

values for natural Carlin-type systems (Su et al., 2008, 2009,

2012; Large et al., 2016, Reich et al., 2005) are representative for high-

As fluids (>1 µg/g) and also agree with our D

(py/fluid)

values and those

that were previously published. Importantly, the pyrite formation tem-

peratures of ¨

Asp¨

o (~14 ◦C; Drake et al., 2015), TAG (300–350 ◦C; Grant

et al., 2018), Carlin-type systems (180–220 ◦C, Su et al., 2012) and our

experimental approach (200 ◦C) cover a wide range applicable to

various natural systems. The comparable D

(py/fluid)

values of our

experimental study and these natural systems therefore suggest that

fluid temperature is only a subordinate effect on the As fluid-pyrite

partitioning. Xing et al. (2019), argue that As preferentially partitions

into the fluid at higher temperatures. However, their results strictly

apply to equilibrium conditions. In our study, As incorporation is more

likely to be kinetically controlled. Based on our data, particularly in

these non-equilibrium scenarios, which encompass the majority of the

lower temperature natural systems, it becomes evident that the primary

factor governing As partitioning into pyrite is the concentration of As in

the fluid. This finding agrees with the data presented by Prokofiev et al.

(2022), where increasing D

(py/fluid)

values are observed with decreasing

fluid As content (Fig. 8, Section 4.3). However, the discrepancy between

the D

(py/fluid)

values from the high-temperature experiments (T =350◦C)

conducted by Prokofiev et al. (2022) and our own at T =200◦C suggests

that temperature and fluid concentration effects may as well be super-

imposed (see Fig. 8).

While the calculated mean D

(py/fluid)

for the active Reykjanes

geothermal system on Iceland align with our experimental data within

the specified As concentration of the fluid (0.11–0.15 µg/g; Hardard´

ottir

et al., 2009), it is important to note that the Reykjanes data exhibits a

considerable variability, that surpasses the range observed in our study

(Fig. 8). The Reykjanes geothermal system is actively boiling (Har-

dard´

ottir et al., 2009), which typically results in the preferential parti-

tioning of As into the vapour phase (Pokrovski et al., 2002, 2013).

Importantly, Libbey and Williams-Jones (2016) showed that the trace

element composition of pyrite from the Reykjanes geothermal system

varies in response to fluid boiling, as known from many other

geothermal and epithermal systems, and which typically results in

strong compositional variations in pyrite chemistry (Rom´

an et al., 2019;

Nestmeyer et al., 2021; B¨

orner et al., 2021; Grosche et al., 2023). The

values of D

(py/fluid)

derived from our experiments under non-boiling

fluid conditions should thus be applied with caution to boiling hydro-

thermal systems.

In addition, the competitive incorporation of As between co-genetic

hydrothermal minerals can also strongly control the final As budget of

pyrite, and as a result the related D

(py/fluid)

values. While pyrite is the

most important host for As in Carlin-type ores (Deditius et al., 2014) and

seafloor massive sulfides at mid-ocean ridges, such as TAG (Keith et al.,

Fig. 9. Calculation of the decrease of the As concentration in a fluid with an

initial As concentration of 1 ng/g after formation of 100 mg of pyrite as a

function of the total fluid mass assuming a Rayleigh fractionation process. The

(py/fluid)

of 30,000 represents the maximum partitioning coefficient observed

in this study for fluid with a low As concentration of 1 ng/g. Detailed results of

the modeling are tabulated in Table S4 and the data repository available at

https://doi.org/10.5281/zenodo.11086033.

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

2016; Grant et al., 2018), hydrothermal systems related to other

geological settings (e.g., epithermal, porphyry) commonly host various

As-bearing minerals (e.g., enargite, luzonite, tennantite-tetrahedrite,

loellingite) that may compete with pyrite regarding the As incorpora-

tion (Einaudi et al., 2003; Simmons et al., 2005; Hedenquist et al., 2017;

Falkenberg et al., 2022). The Rosia Poieni porphyry Cu deposit in

Romania, which features an epithermal overprint, is such an example.

Here, enargite may suppress the As incorporation into pyrite leading to

the comparatively low D

(py/fluid)

values (Fig. 8; Kouzmanov et al., 2010;

Deditius et al., 2014).

The high D

(py/fluid)

values reported in this study imply that pyrite

originating from As-poor fluids likely is highly As-enriched, which

challenges whether elevated As contents in natural pyrite are generally

linked to an As-rich fluid phase. For instance, the D

(py/fluid)

values of the

high-As fluid range (>1 µg/g) would yield a fluid As content of

0.024–0.130 µg/g for a pyrite with 40 µg/g As. By contrast, our new

(py/fluid)

values for the low-As range (<1 µg/g) demonstrate that fluid

As contents could have been in a much lower range of only 0.001–0.005

µg/g.

5. Summary and conclusions

Our study demonstrates that the fluid-pyrite As partitioning is highly

sensitive to the initial fluid As content, which is particularly true for the

low-As range (<1 µg/g), as indicated by strongly increasing D

(py/fluid)

values (up to 30,000) with decreasing fluid As concentrations. The

resulting strong As enrichment in pyrite can be attributed to a complex

interplay of nucleation rate, exposure time of pyrite to the experimental

fluid, the fluid flow regime towards the fluid-siderite interface and

sorption phenomena at the siderite/pyrite surfaces.

Importantly, our D

(py/fluid)

values surpass previously published data

on pyrite-fluid partitioning, which has significant implications for nat-

ural systems. For instance, our results suggest that the replacement of

siderite by pyrite through a coupled dissolution reprecipitation reaction

may serve as an efficient process for groundwater remediation, even in

less severely contaminated aquifers. Hence, a better understanding of

the dynamics of fluid-pyrite interaction may help to develop more

effective strategies for managing and remediating As contamination in

groundwater. The applicability of our new D

(py/fluid)

values also extends

to natural hydrothermal systems operating at higher temperatures (up to

200 ◦C). Here, our findings reveales that the D

(py/fluid)

values are pre-

dominantly controlled by the initial As concentration of the fluid and

superimpose potential temperature effects. Hence, our new D

(py/fluid)

values provide a valuable tool for a more reliable reconstruction of the

fluid As concentrations based on pyrite chemistry, which enables a more

precise prediction of the behavior and fate of As in low-As (ng/g-µg/g)

fluid environments.

Finally, our study highlights the effectiveness of quantitative LA-ICP-

MS mapping in providing precise and accurate elemental distributions in

geomaterials even at the micrometer scale. This analytical approach

provides valuable insights into the mechanisms that govern fluid-rock

interaction, contributing to a better understanding of fluid-related

geochemical processes in general.

Data availability

Data are available through Zenodo at https://doi.org/10.5281/zeno

do.11086033.

Declaration of Generative AI and AI-assisted technologies in the

writing process

During the preparation of this work the author(s) used ChatGPT

(OpenAI) in order to improve spelling and readability. After using this

tool, the author(s) reviewed and edited the content as needed and take

full responsibility for the content of the publication.

CRediT authorship contribution statement

Martin Kutzschbach: Writing – review & editing, Writing – original

draft, Visualization, Supervision, Project administration, Methodology,

Investigation, Data curation, Conceptualization. Frederik Dunkel:

Writing – review & editing, Investigation. Christof Kusebauch: Writing

– review & editing, Methodology, Investigation, Conceptualization.

Ferry Schiperski: . Frederik B¨

orner: Writing – review & editing,

Investigation. Henrik Drake: Writing – review & editing, Resources,

Investigation. Kevin Klimm: Writing – review & editing, Investigation.

Manuel Keith: Writing – review & editing, Writing – original draft,

Project administration, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial

interests or personal relationships that could have appeared to influence

the work reported in this paper.

Acknowledgments

We acknowledge Paul B¨

ottcher and Cordelia Lange for assistance

during the hydrothermal experiments, XRD analyses and sample prep-

aration, as well as Nico Müller for preparing the polished epoxy mounts.

This study was financially supported by a DFG grant (KE 2395/1-1) to

Manuel Keith and grants of the Swedish Research Council (contracts

2021-04365), Formas (contract 2020-01577), J. Gust. Richert founda-

tion (contract 2023-00850), the Crafoord Foundation (contract

20210524) awarded to Henrik Drake. We thank Johannes Giebel for

providing the Ivigtut siderite from the mineral collection of the TU

Berlin and Benjamin Wade for providing the IMER5 reference material.

Also we acknowledge the Swedish Nuclear Fuel and Waste for providing

access to the ¨

Asp¨

o fluid samples. Finally, we thank Prof. Thomas Neu-

mann for the opportunity to work at the MAGMA Lab of the Technische

Universit¨

at Berlin.

Appendix A. Supplementary material

The supplementary material contains major and trace element

composition of starting siderite from Ivigtut (Table S1), the figures of

merit for As-fluid analyses by LA-ICP-MS/MS (Table S2), the results of

the EPMA analyses of pyrite from all hydrothermal experiments

(Table S3), and the results of the geochemical modeling of the As con-

centration in fluids (Table S4). Additionally, it contains reflected light

images and trace element mappings of starting siderite (Fig. S1), the

correlation of As concentrations by EPMA and LA-ICP-MS (Fig. S2), and

predominance regions of As species for all investigated fluid As con-

centrations generated using PhreePlot (Version 1), Fig. S3. Supple-

mentary material to this article can be found online at https://doi.

org/10.1016/j.gca.2024.05.027.

References

Baraga˜

no, D., Boente, C., Rodríguez-Vald´

es, E., Fern´

andez-Bra˜

na, A., Jim´

enez, A.,

Gallego, J.L.R., Gonz´

alez-Fern´

andez, B., 2020. Arsenic release from pyrite ash waste

over an active hydrogeological system and its effects on water quality. Environ. Sci.

Pollut. Res. 27, 10672–10684.

Beck, M., Riedel, T., Graue, J., K¨

oster, J., Kowalski, N., Wu, C., Wegener, G.,

Lipsewers, Y., Freund, H., B¨

ottcher, M., Brumsack, H.-J., Cypionka, H., Rullk¨

otter, J.,

Engelen, B., 2011. Imprint of past and present environmental conditions on

microbiology and biogeochemistry of coastal Quaternary sediments. Biogeosciences

8, 55–68.

Blanchard, M., Alfredsson, M., Brodholt, J., Wright, K., Catlow, C.R.A., 2007. Arsenic

incorporation into FeS

pyrite and its influence on dissolution: a DFT study.

Geochim. Cosmochim. Acta 71, 624–630.

B¨

orner, F., Keith, M., Smith, D.J., Barry, T.L., Neumann, T., Klemd, R., 2021.

Fingerprinting fluid evolution by trace elements in epithermal pyrite, Vatukola Au-

Te deposit, Fiji. Ore Geol. Rev. 137, 104314.

Bostick, B.C., Fendorf, S., 2003. Arsenite sorption on troilite (FeS) and pyrite (FeS

Geochm. Cosmochim. Acta 67, 909–921.

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

Bostick, B.C., Chen, C., Fendorf, S., 2004. Arsenite retention mechanisms with estuarine

sediments of Pescadero, CA. Environ. Sci. Technol. 38, 3299–3304.

Bulut, G., Yenial, Ü., Emiroglu, E., Sirkeci, A.A., 2014. Arsenic removal from aqueous

solution using pyrite. J. Clean. Prod. 84, 526–532.

Burnside, N.M., Banks, D.B., Boyce, A.J., 2016. Sustainability of thermal energy

production at the flooded mine workings of the former Caphouse Colliery, Yorkshire,

United Kingdom. Int. J. Coal Geol. 164, 85–91.

Ciobanu, C.L., Cook, N.J., Utsunomiya, S., Kogagwa, M., Green, L., Gilbert, S., Wade, B.,

2012. Gold-telluride nanoparticles revealed in arsenic-free pyrite. Am. Mineral. 97,

1515–1518.

Craig, J.R., Vokes, F.M., Solberg, T.N., 1998. Pyrite: physical and chemical textures.

Miner. Deposita 34, 82–101.

Deditius, A.P., Reich, M., Kesler, S.E., Utsonomiya, S., Chryssoulis, S.L., Walshe, J.,

Ewing, R.C., 2014. The coupled geochemistry of Au and As pyrite from hydrothermal

ore deposits. Geochim. Cosmochim. Acta 140, 644–670.

Deditius, A.A., Utsunomiya, S., Renock, D., Ewing, R.C., Ramana, C.V., Becker, U.,

Kesler, S.E., 2008. A proposed new type of arsenian pyrite: composition,

nanostructure and geological significance. Geochim. Cosmochim. Acta 72,

2919–2933.

Deditius, A.P., Utsunomyia, S., Reich, M., Kesler, S.E., Ewing, R.C., Hough, R., Walshe, J.,

2011. Trace metal nanoparticles in pyrite. Ore Geol. Rev. 42, 32–46.

Drake, H., Tullborg, E.-L., Whitehouse, M., Sandberg, B., Blomfeldt, Astrom, M.E., 2015.

Extreme fractionation and micro-scale variation of isotopes during bacterial sulphate

reduction in deep groundwater systems. Geochim. Cosmochim. Acta 161, 1–18.

Einaudi, M.T., Hedenquist, J.W., Inan, E.E., 2003. Sulfidation state of fluids in active and

extinct hydrothermal systems: transitions from porphyry to epithermal

environments. Spec. Publ. – Soc. Econ. Geol. 10, 285–314.

Evangelou, V.P., 1995. Pyrite Oxidation and Its Control. CRC Press, Boca Raton, p. 292.

Falkenberg, J.J., Keith, M., Haase, K.M., Bach, W., Klemd, R., Strauss, H., Yeo, I.A.,

Rubin, K.H., Storch, B., Anderson, M.O., 2021. Effects of fluid boiling on Au and

volatile element enrichment in submarine arc-related hydrothermal systems.

Geochm. Cosmochim. Acta 307, 105–132.

Falkenberg, J.J., Keith, M., Haase, K.M., Sporer, C., Bach, W., Klemd, R., Strauss, H.,

Storch, B., Peters, C., Rubin, K.H., Anderson, M.O., 2022. Spatial variations in

magmatic volatile influx and fluid boiling in the submarine hydrothermal systems of

Niuatahi Caldera, Tonga Rear-Arc. Geochem. Geophys. Geosyst. 23,

e2021GC010259.

Flude, S., Haschke, M., Storey, M., 2017. Application of benchtop micro-XRF to

geological material. Mineral. Mag. 81, 923–948.

Geisler, T., Janssen, A., Schreiter, D., Stephan, T., Berndt, J., Putnis, A., 2010. Aqueous

corrosion of borosilicate glass under acidic conditions: a new corrosion mechanism.

J. Non Cryst. Solids 356, 1458–1465.

Grant, H.L.J., Hannington, M.D., Petersen, S., Frische, M., Fuchs, S.H., 2018. Constraints

on the behavior of trace elements in the actively-forming TAG deposit, Mid-Atlantic

Ridge, based on LA-ICP-MS analyses of pyrite. Chem. Geol. 498, 45–71.

Grosche, A., Klemd, R., Denkel, K., Keith, M., Haase, K.M., Voudouris, P.C., Alfieris, D.,

Wiedenbeck, M., 2023. Sources, transport, and deposition of metal(loid)s recorded

by sulfide and rock geochemistry: constraints from a vertical profile through the

epithermal Profitis Ilias Au prospect, Milos Island, Greece. Miner. Deposita 58,

1101–1122.

Guo, H., Li, Y., Zhao, K., Ren, Y., Wei, C., 2011. Removal of arsenite from water by

synthetic siderite: behaviours and mechanisms. J. Hazard. Mater. 186, 1847–1854.

Han, D.S., Song, J.K., Batchelor, B., Abdel-Wahab, A., 2013. Removal of arsenite (As(III))

and arsenate (As(V)) by synthetic pyrite (FeS

): synthesis, effect of contact time, and

sorption/desorption envelopes. J. Colloid Interface Sci. 392, 311–318.

Hardard´

ottir, V., Brown, K.L., Fridriksson, Th., Hedenquist, J.W., Hannington, N.D.,

Thorhallsson, S., 2009. Metals in deep liquid of the Reykjanes geothermal system,

southwest Iceland: implications for the compositionof seafloor black smoker fluids.

Geology 37, 1103–1106.

Hedenquist, J.W., Arribas, R.A., Aoki, M., 2017. Zonation of sulfate and sulfide minerals

and isotopic composition in the far southeast porphyry and leptano epithermal Cu-

Au deposits, Phllipines. Resour. Geol. 67, 174–196.

Hunger, S., Benning, L.G., 2007. (2007) Greigite: a true intermediate on the polysulfide

pathwayto pyrite. Ceochem. Trans. 8, 1.

Jones, G.W., Pichler, T., 2003. Relationship between pyrite stability and arsenic mobility

during aquifer storage and recovery in Southwest Florida. Sci. Total Environ. 41,

723–730.

J¨

ornsson, J., Sherman, D.M., 2008. Sorption of As (III) and As (V) to siderite, green rust

(fougerite) and magnetite: implications for arsenic release in anoxic groundwaters.

Chem. Geol. 255, 173–181.

Keith, M., H¨

ackel, F., Haase, K.M., Schwarz-Schampera, U., Klemd, R., 2016. Trace

element systematics of pyrite from hydrothermal vents. Ore Geol. Rev. 72, 728–745.

Keith, M., Smith, D.J., Jenkin, G.R.T., Holwell, D.A., Dye, M.D., 2018. A review of Te and

Se systematics in hydrothermal pyrite from precious metal deposits: insights into

ore-forming processes. Ore Geol. Rev. 96, 269–282.

Kouzmanov, K., Pettke, T., Heinrich, C.A., 2010. Direct analysis of ore-precipitating

fluids: combined IR microscopy and LA-ICP-MS study of fluid inclusions in opaque

ore minerals. Econ. Geol. 105, 351–373.

Kusebauch, C., Oelze, M., Gleeson, S., 2018. Partitioning of arsenic between

hydrothermal fluids and pyrite during experimental siderite replacement. Chem.

Geol. 500, 136–147.

Kusebauch, C., Gleeson, S.A., Oelze, M., 2019. Coupled partitioning of Au and As into

pyrite controls formation of giant Au deposits. Sci. Adv. 5, eaav5891.

Kutzschbach, M., Wunder, B., Rhede, D., Koch-Müller, M., Ertl, A., Giester, G.,

Heinrich, W., Franz, G., 2016. Tetrahedral boron in natural and synthetic HP/UHP

tourmaline: Evidence from Raman spectroscopy, EMPA, and single-crystal XRD. Am.

Mineral. 101, 93–104.

Kutzschbach, M., Guttman, P., Marquardt, K., Werner, S., Henzler, K.D., Wilke, M., 2018.

A transmission X-ray microscopy and NEXAFS approach for studying corroded

silicate glasses at the nanometer scale. Phys. Chem. Glasses-Eur. J. Glass Sci.

Technol. Part B 59, 11–26.

Large, S.J.E., Bakker, E.Y.N., Weis, P., W¨

alle, M., Ressel, M., Heinrich, C.A., 2016. Trace

elements in fluid inclusions of sediment-hosted gold deposits indicate a magmatic-

hydrothermal origin of the Carlin ore trend. Geology 44, 1015–1018.

Large, R.R., Danyushevsky, L., Hollit, C., Maslennikov, V., Meffre, S., Gilbert, S.,

Foster, J., 2009. Gold and trace element zonation in pyrite using a laser imaging

technique: implications for the timing of gold in orogenic and Carlin-type sediment-

hosted deposits. Econ. Geol. 104, 635–668.

Lennie, A.R., Vaughan, D.H., 1996. Spectroscopic studies of iron sulfide formation and

phase relation at low temperatures. Mineral Spectroscopy: A Tribute to Roger G.

Burns. MW Schaefer, pp. 117–156.

Libbey, R.B., Williams-Jones, A.E., 2016. Relating sulfide mineral zonation and trace

element chemistry to subsurface processes in the Reykjanes geothermal system,

Iceland. J. Volcanol. Geoth. Res. 310, 225–241.

Meng, X., Li, X., Chu, F., Zhu, J., Lei, J., Li, Z., Wang, H., Chen, L., Zhu, Z., 2020. Trace

element and sulfur isotope compositions for pyrite across the mineralization zones of

a sulfide chimney from the East Pacific Rise (1–2◦S). Ore Geol. Rev. 116, 103209.

Mukherjee, I., Large, R.R., 2016. Pyrite trace element chemistry of the Valkerri

Formation, Roper Group, McArthur Basin: evidence for atmospheric oxygenation

during the Boring Billion. Precambr. Res. 281, 13–26.

Murowchick, J.B., Barnes, H.L., 1986. Marcasite precipitation from hydrothermal

solutions. Geochim. Cosmochim. Acta 50, 2615–2629.

Nestmeyer, M., Keith, M., Haase, K., Klemd, R., Voudouris, P., Schwarz-Schampera, U.,

Strauss, H., Kati, M., Magganas, A., 2021. Trace element signatures in pyrite and

marcasite from Shallow Marine Island Arc-related hydrothermal vents, calypso

vents, New Zealand, and Paleochori Bay, Greece. Front. Earth Sci. 9, 641654.

Nordstrom, D.K., Archer, D.G., 2003. Arsenic thermodynamic data and environmental

geochemistry. In: Welch, A.H., Stollenwerk, K.G. (Eds.), Arsenic in Ground Water.

Springer, Boston, MA.

Nordstrom, D.K., 1982. Aqueous pyrite oxidation and the consequent formation of

secondary iron minerals. In: Kittrick, J.A., Fanning, D.S., Hosser, L.R. (Eds.), Acid

Sulfate Weathering, Vol. 17. Soil Science Society of America Special Publication, pp.

37–56.

Nugent, M.A., Brantley, S.L., Pantano, C.G., Maurice, P.A., 1998. The influence of natural

mineral coatings on feldspat weathering. Nature 295, 588–591.

Oliveira, M.L., Ward, C.R., Izquierdo, M., Sampaio, C.H., de Brum, I.A., Kautzmann, R.

M., Silva, L.F., 2012. Chemical composition and minerals in pyrite ash of an

abandoned suphuric acid production plant. Sci. Total Environ. 430, 34–47.

Palenik, C.S., Utsunomiya, S., Reich, M., Kesler, S.E., Wang, L., Ewing, R.C., 2004.

“Invisible” gold revealed: direct imaging of gold nanoparticles in a Carlin-type

deposit. Am. Mineral. 89, 1359–1366.

Pals, D.W., Spry, P.G., Chryssoulis, S., 2003. Invisible gold and tellurium in arsenic-rich

pyrite from the Emperor gold deposit, Fiji: implications for gold distribution and

deposition. Econ. Geol. 98, 479–493.

Pidchenko, I., Christensen, J.N., Kutzschbach, M., Ignatyev, K., Puigdomenesch, I.,

Tullborg, E.-L., Roberts, N.M.W., Rasbury, E.T., Northrup, P., Tappero, R.,

Kvashnina, K.O., Sch¨

afer, T., Suzuki, Y., Drake, H., 2023. Deep anoxic aquifers could

act as sinks for uranium through microbial-assisted mineral trapping. Commun.

Earth Environ. 4, 128.

Pokrovski, G.S., Zakirov, I.V., Roux, J., Testemale, D., Hazemann, J.-L., Bychlov, A.Y.U.,

Golikova, G.V., 2002. Experiemental study of arsenic speciation in vapor phase to

500 C: implications for As transport and fractionation in low-density crustal fluids

andvolcanic gases. Geochim. Cosmochim. Acta 66, 3453–3480.

Pokrovski, G.S., Borisova, A.Y., Bychkov, A.Y., 2013. Speciation and transport of metals

and metalloids in geolical vapors. Rev. Mineral. Geochem. 76, 165–218.

Pokrovski, G.S., Kokh, M.A., Proux, O., Hazemann, J.-L., Bazarkina, E.F., Testemale, D.,

Escoda, C., Boiron, M.-C., Blanchard, M., Aigouy, T., Gouy, A., Parseval, P.,

Thibaut, M., 2019. The nature and partitioning of invisible gold in the pyrite-fluid

system. Ore Geol. Rev. 109, 545–563.

Pokrovski, G.S., Escoda, C., Blanchard, M., Testemale, D., Hazemann, J.-L., Gouy, S.,

Kokh, M.A., Boiron, M.-C., de Parseval, F., Aigouy, T., Menjot, L., de Parseval, P.,

Proux, O., Rovezzi, M., B´

eziat, D., Salvi, S., Kouzmanov, K., Bartsch, T., P¨

ottgen, R.,

Doert, T., 2021. An Arsenic driven pump for invisible gold in hydrothermal systems.

Geochem. Perspect. Lett. 17, 39–44.

Pratesi, G., Zoppi, M., 2015. An insight into the inverse transformation of realgar altered

by light. Am. Mineral. 100, 1222–1229.

Prokofiev, V.Y., Tagirov, B.R., Brovchenko, V.D., Reukov, V.L., Filiminova, O.N.,

Zorina, L.D., Abramova, V.D., Kovalchuk, E.V., Mezhueva, A.A., Nikolaeva, I.Y.,

Tarnopolskaya, M.E., Akinfiev, N.N., Kransov, A.N., Komarov, V.B., Bortnikov, N.S.,

2022. Gold in natural and synthetic pyrite: a case of the Darasun Gold District,

Eastern, Transbaikal Region, Russia. Geol. Ore Deposits 64, 329–349.

Putnis, A., 2014. Why mineral interfaces matter. Science 343, 1441–1442.

Qian, G., Brugger, J., Skinner, W.M., Chen, G.R., Pring, A., 2010. An experimental study

of the mechanism of the replacement of magnetite by pyrite up to 300 ◦C. Geochim.

Cosmochim. Acta 74, 5610–5630.

Qian, G., Xia, F., Brugger, J., Skinner, W.M., Bei, J., Chen, G., Pring, A., 2011.

Replacement of pyrrhotite by pyrite and marcasite under hydrothermal conditions

up to 220 ◦C: an experimental study of reaction textures and mechanisms. Am.

Mineral. 96, 1878–1893.

M. Kutzschbach et al.

Geochimica et Cosmochimica Acta 376 (2024) 37–53

Qian, G., Brugger, J., Testemale, D., Skinner, W., Pring, A., 2013. Formation of As (II)-

pyrite during experimental replacement of magnetite under hydrothermal

conditions. Geochm. Cosmochim. Acta 100, 1–10.

Reich, M., Becker, U., 2006. First-principles calculations of the thermodynamic mixing

properties of arsenic incorporation into pyrite and marcasite. Chem. Geol. 225,

278–290.

Reich, M., Kesler, S.E., Utsunomiya, S., Palenik, C.S., Chryssoulis, S.L., Ewing, R.C., 2005.

Solubility of gold in arsenian pyrite. Geochim. Cosmochim. Acta 69, 2781–2796.

Reich, M., Deditius, A.P., Chryssoulis, S., Li, J.W., Ma, C.Q., Parada, M.A., Mittermayr, F.,

2013. Pyrite as a record of hydrothermal fluid evolution in a porphyry copper

system: a SIMS/EMPA trace element study. Geochm. Cosmochim. Acta 104, 42–62.

Rickard, D., Luther, G.W., 2007. Chemistry of iron sulfides. Chem. Rev. 107, 514–562.

Rom´

an, N., Reich, M., Leisen, M., Morata, D., Barra, F., Deditius, A.P., 2019. Geochmical

and micro-textural fingerprints of boiling in pyrite. Geochm. Cosmochim. Acta 246,

60–85.

Ruiz-Agudo, E., King, H.E., Pati˜

no-L´

opez, L.D., Putnis, C.V., Geisler, T., Rodriguez-

Navarro, C., Putnis, A., 2016. Control of silicate weathering by interface-coupled

dissolution-precipitation processes at the mineral-solution interface. Geology 44,

567–570.

Saunder, J.A., Lee, M.-K., Dhakal, P., Ghandehari, S.S., Wilson, T., Billor, M.Z., Uddin,

2018. Bioremediation of arsenic-contaminated groundwater by sequestration of

arsenic in biogenic pyrite. Appl. Geochem. 96, 233–243.

Savage, K.S., Tingle, T.N., O’Day, P.A., Waychunas, G.A., Bird, D.K., 2000. Arsenic

speciation in pyrite and secondary weathering phases, Mother Lode gold district,

Tuolumne County, California. Appl. Geochem. 15 (1219), 1244.

Schmidt, K., Sch¨

onberg, D.-G., Hannington, M.D., Anderson, M.O., Bühring, B.,

Haase, K., Haruel, C., Lupton, J., Koschinsky, A., 2017. Boiling vapour-type fluids

from the Nifonea vent field (New Hebrides Back-Arc, Vanuatu, SW Pacific):

geochemistry of an early-stage, post-eruptive hydrothermal sytem. Geochm.

Cosmochim. Acta 207, 185–209.

Schoonen, M.A.A., Barnes, H.L., 1991. Mechanisms of pyrite and marcasite formation

from solution: III. Hydrothermal processes. Geochm. Cosmochim. Acta 55,

3491–3501.

Simmons, S.F., White, N.C., John, D.A., 2005. Geological characteristics of epithermal

precious and base metal deposits. Special Publication – Society of Economic

Geologists. Simon, G., Huang, H., Penner-Hahn, J.E., Kesler, S.E., and Kao, L.S.

(1999) Oxidation state of gold and arsenic in gold-bearing arsenian pyrite. Am.

Mineral. 84, 1071–1079.

Sliwinski, M.G., Maicej, G., Kitajima, K., Spicuzza, M.J., Orland, I.J., Ishida, A.,

Fournelle, J.H., Valley, J.W., 2018. SIMS bias on isotope ratios in Ca-Mg-Fe

carbonates (Part III): δ

O and δ

C matrix effects along the magnesite-siderite solid-

solution series. Geostand. Geoanal. Res. 42, 49–76.

Smith, A.H., Smith, M.M.H., 2004. Arsenic drinking water regulations in developing

countries with extensive exposure. Toxicology 198, 39–44.

Stuyfzand, P.J., Bonte, M., 2023. A critical review of arsenic occurrence, fate and

transport in natural and modified groundwater systems in The Netherlands. Appl.

Geochem. 150, 105596.

Su, W., Heinrich, C.A., Pettke, T., Zhang, X., Hu, R., Xia, B., 2009. Sediment-hosted gold

deposits in Guizhou, China: products of wall-rock sulfidation by deep crustal fluids.

Econ. Geol. 104, 73–93.

Su, W.C., Xia, B., Zhan, H.T., Zhang, X.C., Hu, R.Z., 2008. Visible gold in arsenian pyrite

at the Shuiyindong Carlin-type gold deposit, Guizhou, China: implications for the

environment and processes of ore formation. Ore Geol. Rev. 33 (667), 679.

Su, W., Zhang, H., Hu, R., Ge, X., Xia, B., Chen, Y., Zhu, C., 2012. Mineralogy and

geochemistry of gold-bearing arsenian pyrite from the Shuiyindong Carlin-type gold

deposit, Guizhou, China: implications for gold depositional processes. Miner.

Deposita 47, 653–662.

Widodo, S., Oschmann, W., Bechtel, A., Sachsenhofer, R.F., Anggayana, K.,

Puetmann, W., 2010. Distribution of sulfur and pyrite in coal seams from Kutai Basin

(East Kalimantan, Indonesia): implications for paleoenvironmental conditions. Int. J.

Coal Geol. 81, 151–162.

Winderbaum, L., Ciobanu, C.L., Cook, N.J., Paul, M., Metcalfe, A., Gilbert, S., 2012.

Multivariate analysis of an LA-ICP-MS trace element dataset for pyrite. Math. Geosci.

44, 823–842.

Xing, Y., Brugger, J., Tomkins, A., Sharov, Y., 2019. Arsenic evolution as a tool for

understanding formation of pyrite gold ores. Geology 47, 335–338.

Yao, X., Xia, F., Brugger, J., Kartal, M., Adegoke, I.A., 2021. Rapid marcasite to pyrite

transformation in acidic low-temperature hydrothermal fluids and saturation index

control on FeS

precipitation dynamics and phase selection. ACS Earth Space Chem.

5, 2453–2465.

Yu, C., Drake, H., Lopez-Fernandez, M., Whitehouse, M., Dopson, M., Åstr¨

om, M.E.,

2019. Micro-scale isotopic variability of low-temperature pyrite in fractured

crystalline bedrock – a large Fe isotope fractionation between Fe(II)

/pyrite and

absence of Fe-S isotope co-variation. Chem. Geol. 522, 192–207.

M. Kutzschbach et al.