Blundy Et Al., 2021

The economic potential of
royalsocietypublishing.org/journal/rsos
metalliferous sub-volcanic
brines
Jon Blundy1, Andrey Afanasyev2, Brian Tattitch3,
Perspective
Steve Sparks3, Oleg Melnik2, Ivan Utkin2 and
Cite this article: Blundy J, Afanasyev A, Tattitch
B, Sparks S, Melnik O, Utkin I, Rust A. 2021 The Alison Rust3
Downloaded from https://royalsocietypublishing.org/ on 01 June 2023
economic potential of metalliferous sub-volcanic 1

Department of Earth Sciences, University of Oxford, South Parks Road, Oxford OX1 3AN, UK
brines. R. Soc. Open Sci. 8: 202192. 2
Institute of Mechanics, Moscow State University, 1 Michurinsky Prospekt, Moscow 119192, Russia
https://doi.org/10.1098/rsos.202192 3
School of Earth Sciences, University of Bristol, Wills Memorial Building, Bristol BS8 1RJ, UK
JB, 0000-0001-7263-8925; AA, 0000-0002-2284-7144;
BT, 0000-0002-2725-8031; SS, 0000-0001-7173-2899;
OM, 0000-0002-4655-5269; AR, 0000-0001-5095-7749
Received: 10 December 2020
Accepted: 14 June 2021 The transition to a low-carbon economy will increase demand
for a wide range of metals, notably copper, which is used
extensively in power generation and in electric vehicles.
Increased demand will require new, sustainable approaches
to copper exploration and extraction. Conventional copper
Subject Category:
mining entails energy-intensive extraction of relatively low-
Earth and environmental science grade ore from large open pits or underground mines and
subsequent ore refining. Most copper derives ultimately from
Subject Areas: hot, hydrous magmatic fluids. Ore formation involves phase
volcanology separation of these fluids to form copper-rich hypersaline
liquids (or ‘brines’) and subsequent precipitation of copper
Keywords: sulfides. Geophysical surveys of many volcanoes reveal
copper mining, volcanoes, ore deposits, electrically conductive bodies at around 2 km depth, consistent
with lenses of brine hosted in porous rock. Building upon
geothermal energy
emerging concepts in crustal magmatism, we explore the
potential of sub-volcanic brines as an in situ source of copper
and other metals. Using hydrodynamic simulations, we show
Author for correspondence: that 10 000 years of magma degassing can generate a Cu-rich
Jon Blundy brine lens containing up to 1.4 Mt Cu in a rock volume of a
few km3 at approximately 2 km depth. Direct extraction of
e-mail: jonathan.blundy@earth.ox.ac.uk
metal-rich brines represents a novel development in metal
resource extraction that obviates the need for conventional
mines, and generates geothermal power as a by-product.
1. Introduction
Our modern world is reliant on natural resources of metals, from
steel for construction to rare earth elements for high-tech devices.
Electronic supplementary material is available As we transition from fossil fuel-dominated to more sustainable
online at https://doi.org/10.6084/m9.figshare.c.
5480583. © 2021 The Authors. Published by the Royal Society under the terms of the Creative
Commons Attribution License http://creativecommons.org/licenses/by/4.0/, which permits
unrestricted use, provided the original author and source are credited.
economies, demand for certain metals will rise dramatically. In a world powered by wind, sun and tides, 2
for example, copper demand will increase more than fivefold, surpassing known global reserves well
before 2100 [1,2]. The rise in demand for ‘critical metals’ (e.g. lithium, scandium, cobalt, rare earths)
will be even greater [3]. As existing reserves become depleted and new deposits ever harder to find, it
is unclear how the extra demand can be met; recycling alone will be insufficient [1].
Discovering natural mineral resources is, at heart, a geological problem. The majority of non-ferrous
metals derive ultimately from igneous processes associated with magmatism. Non-ferrous metal ores can
be viewed as extreme end-products of igneous geochemical cycles that begin in Earth’s mantle and
conclude with discharge of hot magmas and gases at the surface. Volcanic volatile phases are
particularly efficient transporters of metals as evidenced by epithermal ore deposits and the chemistry
of fumarole gases. Some active, subduction-related basaltic volcanoes discharge metal-rich gases with
a time-averaged flux to the atmosphere of more than 104 kg day−1 of copper and zinc, along with a
slew of other metals (e.g. silver, tungsten, indium, tin, lead, molybdenum) at fluxes of up to
1000 kg day−1 [4]. This observation re-emphasizes the centrality of igneous processes to formation of
R. Soc. Open Sci. 8: 202192

mineral resources [5], and raises the possibility of recovering metals directly from modern volcanic
fluids as an alternative to mining ancient solid ores. Here, we discuss the potential for in situ mining
of hot, metalliferous volcanic fluids,1 with a particular emphasis on copper, in light of recent
developments in our understanding of the dynamics and architecture of crustal magmatic systems.
2. Porphyry copper deposits

Porphyry copper deposits (PCDs) provide approximately 75% of the world’s copper, plus significant
molybdenum and gold. PCDs are usefully subdivided into hypogene deposits that form primarily by
the action of hydrothermal fluids, and supergene deposits that have been modified and enriched
through reactions of hypogene ore at or close to the water table. Our focus here is exclusively on
hypogene PCDs, on which there is a substantial body of literature; our intention here is not to review
all of this prior work, but instead to summarize some key features of PCDs that place them in the
context of crustal magmatic processes. For comprehensive summaries of hypogene PCDs the reader is
referred to Hedenquist & Lowenstern [5], Sillitoe [6] and Richards [7].
PCDs are typically associated with subduction-related, oxidized, H2O-rich silicic magmas,
themselves derived by differentiation of mantle-derived, hydrous basalts [7]. Magmatic differentiation
within the crust elevates contents of dissolved H2O and other volatiles, including important metal-
complexing ligands, such as Cl– and HS–, while imparting geochemical signatures characteristic of
differentiation at lower- or mid-crustal pressures [8–11]. Unless the parental basalts are unusually
Cu-rich, for which there is little evidence [12], a key process for Cu enrichment and ore formation is
efficient extraction of Cu by hydrous fluids exsolved from evolved magmas as they ascend and
crystallize. Such fluids are associated with active dacitic, andesitic and basaltic andesite volcanoes (e.g.
[5,13–15]) both during and between eruptions of magma [16,17]. The fluids are predominantly H2O
with lesser quantities of CO2, SO2, H2S and halogens and a wide range of trace metals [18,19].
The relationship between magmatic fluids involved in sub-surface PCD formation and those
discharged directly to the atmosphere is complex. Phase changes associated with fluid decompression
and cooling modify the chemistry of the original fluid at the point of exsolution from magma. For
example, phase separation of H2O-NaCl fluids into coexisting low-density (vapour) and high-density
(liquid) fluids can have a profound effect on metal contents. To form an ore deposit, some component
of the original magmatic fluid must become trapped and cool within the crust, depositing its metal
load in the form of ore minerals. Copper mineralization typically takes the form of copper-bearing
sulfides, such as chalcopyrite and bornite. Reduced sulfur (S2−) is therefore a key ingredient in the
formation of PCDs. Similarly, the primacy of the chloride (Cl−) ligand in transporting Cu [20–25]
requires that ore-forming magmatic fluids are saline, that is, they contain chloride, often expressed in
terms of an equivalent amount of sodium chloride (NaCleq), noting that cations in addition to Na+
can be present in significant concentrations (e.g. K+, Ca2+ etc.).
1
We use the term fluid to describe any flowing geological phase, including volcanic gas, low-salinity vapour and hypersaline liquid,
with a density less than that of silicate melt at the same pressure and temperature. Supercritical fluids are those beyond the critical end
point in volatile-dominated systems, i.e. at pressures above phase separation into liquid and vapour. Magmatic volatiles include the
species H2O, CO2, SO2, H2S, NaCl, HCl and other halides. The process of volatile release from silicate melts is described broadly as
degassing regardless of the exact state of the fluid released.
Characteristic and extensive rock alteration haloes around PCDs, containing a variety of secondary 3
minerals, such as sericite mica and clays (see review in [6]), testify to the highly reactive nature of hot
magmatic fluids when they come into contact with cooler igneous rocks. The reactant fluids
responsible for alteration include both primary, hot ore-forming fluids and cooler, acidic exhaust gases
produced once mineralization has occurred. Heated external water (groundwater, seawater) may also
mix with magmatic fluids and participate in hydrothermal reactions (e.g. [26]). Formation of extensive,
hydrothermally altered ‘lithocaps’ above PCDs results from reactions involving fluids of different
provenance [6]. The eventual style of alteration is a complex interplay between the temperature and
composition of these fluids and their ascent paths.
PCDs are spatially associated with hypabyssal intrusions emplaced at depths of 1–4 km [6] that
provide focused pathways for magmatic fluids exsolved deeper in the system [27]. These depths
correspond to the upper reaches of magma reservoirs (or ‘chambers’) that underlie active volcanoes
(e.g. [28]). The hypabyssal intrusions are sourced from these magma reservoirs; some intrusions are
‘blind’ (i.e. do not reach the surface), whereas others may represent magma-filled conduits for

volcanic or phreatomagmatic eruptions, lava dome extrusions or brecciation events. Solidified shallow
magma reservoirs take the form of plutons with a carapace of thermally metamorphosed, and
variously hydrothermally altered, country rock.
3. Transcrustal magmatic systems

PCDs form in the upper crust and hitherto the focus of attention with respect to the magmatic sources of
ore-forming fluids has tended to be a large, degassing shallow magma chamber (e.g. [27,29]). However,
as the paradigm of crustal magmatism evolves it has become clear that processes pertinent to the origin
of PCD-forming magmas and fluids extend over much greater depths. Shallow, melt-rich magma
chambers are increasingly viewed as an ephemeral expression of much larger, vertically extensive, and
longer-lived magma systems in which partially molten rocks, or ‘mushes’, dominate [30–32]. Such
systems may extend throughout the crust as a continuous connected system or as discrete separated
reservoirs that become transiently interconnected during magma ascent. The architecture of
transcrustal magmatic systems has some important implications for ore formation.
In the transcrustal concept (figure 1) magmas and fluids stored in and released from shallow sub-
volcanic reservoirs represent the time-integrated products of magmatic differentiation and fluid
exsolution processes operating over a significant vertical extent. This is particularly important in the
case of magmatic volatile species because of their strongly pressure-sensitive solubility in silicate melts
[33,34]. Consequently, the mass and composition of the magmatic volatile phases exsolved from
magmas are sensitive functions of the pressure (depth) at which degassing occurs.
The transcrustal, hot-zone concepts have arisen from studies in volcanic geology and stratigraphy,
geochemistry, petrology, geophysics and geochronology, with valuable additional insights provided by
numerical models [9,35–40]. In combination, these different approaches show how hot zones can form
over hundreds of thousands to several million years through the sequential emplacement of mantle-
derived magmas into the mid- or lower crust. Over time, temperatures and melt fractions in the hot
zone increase, eventually forming a substantial, vertically extensive body of partially molten rock. The
deep-seated, partially molten nature of hot zones beneath a thick insulating roof confers a longevity
that is not tenable for shallow, liquid-rich magma chambers that are limited by thermal constraints
[41]. Shallow large magma chambers are now considered ephemeral components of the magmatic
system, associated with episodes of relatively high magma transfer from the hot zone into the upper
crust. The generation of silicic melts in transcrustal systems seems to be focused in the middle crust
where intermediate and mafic magmas that differentiated in the lower crust and mantle stall.
Hot zones are not static; they are dynamic mixtures of relatively dense solids and buoyant melts and
fluids. Reaction of ascending melts and fluids with their surrounding rocks, including partially solidified
ancestral magmas, is considered to be an important process in chemical evolution of magmas (e.g.
[37,40]) and fluids (e.g. [42]). Reactive flow enables melts and fluids generated deeper in the system
(i.e. at higher pressure) to encounter and react with shallow-stored (lower pressure) components of the
same magmatic system. The intrinsic gravitational instability of mush systems means that melt and
fluid ascent can be either percolative (steady state) or episodic [31]. Release of melts from hot zones
may lead to eruption at the surface or formation of shallow-level magma chambers and plutons,
themselves constructed incrementally over extended periods of time [43]. Release of highly reactive
fluids from deeper parts of a magmatic system has a number of implications for ore deposits.
gas plume 4
dacite/andesite out
volcano
fumarole
conduit
~5 km
~5 km
brine lens
country rock
brine + vapour
supercrit. fluid

pluton silicic
melt
mafic
basement
mafic crystals
felsic crystals
fluid + melt
melt crustal melting
gas
brine
melt migration
gas migration
crust
mantle
basalt in
Figure 1. Schematic of a typical transcrustal magmatic system showing features described in the text. Note that proportion of melt
and dimensions of crystals and bubbles are exaggerated for clarity. Ascent of melt (green arrows) and fluids (blue arrows) is
decoupled, and may be either steady-state ( percolative) or episodic. Crustal melting may occur in a thermal aureole around the
magmatic system, as shown by black dots. Depths of the crust-mantle transition (Moho), volatile saturation and supercritical
fluid phase-separation are denoted by horizontal dashed lines. Sub-volcanic brine lenses, the focus of this paper, occur at the
shallow extremity of the system ( pale green shading), created and sustained by fluid flow from the underlying system.
Fluids released from hot zones transport heat and fluid-mobile elements, including copper, upwards.
The ascent of melt and fluid may be decoupled in time and space [17]. The endowment of the world’s
most economic PCDs (greater than 10 Mt contained Cu) requires that copper be extracted efficiently
from a much greater volume of magma than the small intrusions associated with the deposits
themselves. The abundance of brecciation and veining features in PCDs [6], and high-resolution
radiometric data suggest that pulses of mineralization tend to occur at the end of relatively long-lived
magmatic episodes (e.g. [44–47]), possibly via catastrophic, large-scale fluid release events. Thermal
modelling of incrementally assembled igneous bodies shows that fluid release can occur in relatively
brief pulses against a background of longer-lived magma accumulation and crystallization [11,48].
Ultimately the upward flux of melt and fluids, and the extent to which fluxes are steady-state or
episodic, are controlled by the evolving permeability of the reservoirs in which they form [49].
2000 5
supercritical
ore fluid
800∞C
1500
700∞C
pressure (bar)
1000
600∞C

500∞C
500
400∞C
halite
0 20 40 60 80 100
wt% NaCl
Figure 2. Phase relations in the system NaCl–H2O calculated using SOWAT [55]. Solid lines show isothermal projections of the solvus
that describe coexisting hypersaline liquid (brine) and low-salinity vapour. Upon decompression a single-phase fluid of intermediate
density will undergo phase separation when it encounters the solvus. The compositions of coexisting phases depend on pressure and
temperature; the compositions of coexisting vapour and hypersaline liquid at 1100 bars and 700°C are shown for reference by the
horizontal blue line. An illustrative supercritical magmatic input fluid with 4 wt% NaCleq (as used in the models) is shown by the
vertical grey bar. The halite saturation field is delineated in red. Addition of CO2 to the NaCl–H2O system expands the solvus and
extends the region of phase separation.
Currently, this is a poorly understood aspect of the multiphase, mushy systems that are thought to
underlie volcanoes, but is likely to be very important in the context of giant ore deposits.
4. Magmatic volatiles
The composition of magmatic fluids depends on the relative solubilities of the key volatile species in
silicate melts that in turn depend on pressure, redox state and composition (and to a much lesser
extent, temperature). Generally speaking, fluids released deeper in the system are richer in CO2 than
those released at shallower levels, due to the greater solubility of H2O than CO2 (e.g. [34]). Thus,
fluids released from magmas at depth in the crust will differ chemically from fluids in equilibrium
with magma stored at shallow depth, with abundant opportunities for chemical reaction, including
flushing of shallow magmatic systems with deep-derived CO2, for which there is compelling
petrological evidence (e.g. [50]). The behaviour of minor fluid species (e.g. NaCl, HCl, SO2, H2S) is
more complex and, in the case of sulfur species, dependent on redox state. For example, in their
experimental study of degassing of oxidized basalt magma Lesne et al. [33] show that magmatic
chlorine is retained in melts to lower pressures than is SO2.
The volcanic fluids emitted from active basaltic volcanoes are dominated by H2O (±CO2) and have
relatively low salinity (less than 1 wt% NaCleq; [4,51]); their composition at the surface closely
resembles that of the fluid at the point of shallow exsolution. The situation is different for magmatic
fluids that exsolve from evolved, more chlorine-rich silicic magmas at greater depths where higher
salinity (greater than 1 wt% NaCleq) fluids are generated [52,53]. Thermodynamics of the NaCl–H2O
system [54,55] define a two-phase region of hypersaline liquid (often referred to as ‘brine’) and low-
salinity vapour. The two-phase region is delimited by a solvus whose location depends on
temperature and pressure (figure 2). At the pressure and temperature of initial volatile exsolution
from H2O-rich silicic magmas, most magmatic fluids probably exist as a single, super-critical fluid of 6
intermediate density (salinity approx. 2 to 12 wt% NaCleq; [24]) due to volatile saturation at pressures
above the NaCl–H2O solvus [29]. However, direct exsolution of high salinity fluids from some silicic
magmas is known to occur [56–59]. The NaCl–H2O solvus is influenced by the presence of other
volatile species, notably CO2 [60,61], which greatly expands the two-phase field. In these cases, not
considered explicitly here, phase separation of hypersaline liquids will occur over a much wider range
of pressure–temperature conditions than for the simple NaCl–H2O system, further expanding their
stability field. Other cations (K+, Ca2+ etc.) also modify phase relations and ore mineral solubilities in
hydrous fluids, although in most mineralizing magmatic systems Na+ is by far the most abundant.
Fluid inclusions from PCDs support involvement of parental, single-phase intermediate-density (ID)
fluids of modest salinity (4–10 wt% NaCleq; [62–65]). Although primary ID fluids typically contain less
than or equal to 2000 ppm Cu ([24]), the proclivity of copper (and many other metals) for the chloride
ligand means that the separated hypersaline liquid phase becomes strongly metal-enriched relative to
its parent fluid. Upon phase separation, Cu partitions into the liquid, in almost direct proportion to

the salinity ratio between liquid and vapour [24,25]. This ratio depends on the pressure and
temperature of phase separation (figure 2), thus higher Cu contents are found in higher salinity
liquids formed at lower pressures and higher temperatures. Globally, Cu contents of hypersaline fluid
inclusions (salinities of 25 to 73 wt% NaCleq) from PCDs range from 3 ppm to 3 wt% [24]. The upper
end of this range is considerably more Cu-rich than the original fluid exsolved from the parent silicic
magma, a testament to the Cu-sequestering ability of brine. The low salinity vapour counterpart that
discharges to the atmosphere is correspondingly depleted in metals.
The importance of brines to ore formation has long been recognized (e.g. [66,67]). Numerical models
of magmatic fluid flow through the shallow crust [68–70] show how the dense, hypersaline liquid formed
upon phase separation can be retained at depth in rock pore space, such as intergranular pores, cavities
and fractures. Thus a significant fraction of the original metal budget of silicic volcanoes can become
trapped within fluids in porous rocks at depth, rather than be discharged to the atmosphere or
retained in the cooled crystallized mass. These considerations raise the possibility of the existence of
metal-rich hypersaline liquid reservoirs beneath active volcanoes with attendant economic potential.
Although few, if any, major PCD-like deposits are likely to underlie modern volcanoes, the ubiquity of
active volcanoes (and associated shallow intrusions) may partly counterbalance their small size,
justifying a careful evaluation of the economic potential of their associated brines.
5. Metals in geothermal systems

The heat and fluids associated with shallow magmatic systems are often harnessed through several forms
of geothermal energy production. Geothermal energy is extracted from a range of volcano types and in a
variety of tectonic settings. Conventional geothermal wells are drilled into the upper reaches of magmatic
hydrothermal systems, and provide insights into the composition and distribution of fluid phases.
Typically, these fluids are dominated by a variety of non-magmatic fluids, including sea water, meteoric
water and regional brines, variably modified by heating, water–rock interactions, phase separation and
mixing, with relatively limited chemical input from magma-derived fluids, as evidenced, for example,
by stable isotope chemistry [71].2 This is because, in general, the region of active hydrothermal
convection is separated physically from the conductive magmatic heat source by a seal of relatively low
permeability rock. This physical barrier is often ascribed to the brittle–ductile transition in rocks at
depth at temperatures of around 400°C, although permeability reduction can also be caused by mineral
precipitation (e.g. [67,72]). Below the rock-seal, fluids are predominantly magmatic in origin; above it
they are predominantly non-magmatic. Evidently, fluids of magmatic origin and those of meteoric
origin can be efficiently isolated from each other in geothermal systems; most geothermal wells access
only the shallower, convecting hydrothermal realm. The upper boundary of this realm is defined
typically by a second impermeable barrier, often referred to as the clay cap (e.g. [71]).
A lesser number of high-temperature or ‘supercritical’ geothermal systems [73,74] use wells that
penetrate higher temperature (greater than 400°C) accumulations of magma-derived fluids close to or
below the interface between hydrothermal and magmatic realms. In a few of the deepest wells, drilled
into young, hot magmatic systems, samples of true magmatic hypersaline liquids have been
2
Although it is common in many of these geothermal systems to refer to any fluid with total dissolved solids (TDS) greater than 0.1 to
1 wt% as ‘brine’, these are distinct from the truly hypersaline liquids that are our focus here.
7
(a) salinity:~ 55 wt% (b)
Cu: 6800 ppm [6000(1500)] chalcopyrite
V
halite
L
x
y
salinity:~ 56 wt%
Cu: 4900 ppm [7600(4900)]
10 mm
Ccp+Mb 30 mm
CMA201
(c) salinity:~ 45 wt% (d) salinity:~ 47 wt%

Cu: 2900–3300 ppm Cu: 500–5000 ppm

50 mm 10 mm
Figure 3. Three-dimensional phase maps of experimental and natural metal-bearing, hypersaline fluid inclusions used to estimate
original copper contents. Maps were generated from photomicrographs of hypersaline liquid (L) inclusions from: (a) experiments
[25]; (b) a porphyry copper deposit (Bajo de la Alumbrera; [85]); and the Kakkonda deep geothermal system (c, [84]; d, [83]).
All inclusions show large halite daughter minerals due to high salinity, along with numerous opaque daughter sulfide minerals
(Ccp, chalcopyrite; Mb, molybdenite; x and y are unidentified phases). Inclusions (a) and (b) come from systems with positive
confirmation of chalcopyrite daughter minerals and known Cu contents. Pixel maps of the inclusions yield volume estimates for
halite (red), sylvite (orange), and opaque daughter crystals along with total volume estimates based on assuming elliptical
inclusion geometries and/or thicknesses given by vapour bubble (V) size and shape. Salinity estimates (expressed as wt%
NaCleq) for all inclusions based on volume estimates roughly match those reported from fluid inclusion microthermometry.
Estimates of the possible volume of chalcopyrite (yellow) (34% Cu) or Cu-bearing Fe-sulfide (brown) (approx. 3.5%) have been
compared with total fluid mass in the inclusion to determine the range of Cu concentrations (given in ppm; table 1). For
comparison, Cu concentrations from LA-ICPMS analyses of inclusions in (a) and (b) are given in square brackets (with 1 s.d.
uncertainty in curved brackets), and are a reasonable match to the estimated Cu contents. No explicit measure of uncertainty is
considered but large variations in inclusion thickness from those modelled would result in unrealistic deviations in estimates of
inclusion salinity when compared with the measured values.
recovered. At Larderello (Italy) Cathelinau et al. [75] report quartz-hosted hypersaline liquid inclusions
(30–70 wt% NaCleq) recovered from granite host rocks at depths of 2.7–3.6 km and at temperatures of
425–650°C, overlapping the highest measured bottom-hole temperatures (330–450°C). The studied
samples come from several hundred metres above the seal between convective and conductive
regimes (the ‘K-horizon’; [76]).
At Kakkonda (Japan) active accumulations of hypersaline liquids (39 to 55 wt% NaCleq) have been
identified at depths in excess of 3708 m and temperatures over 500°C in well WD-1a [77,78]. These
depths lie below the conductive–convective interface (3.2 km in WD-1a; [79]) and consequently
represent truly magmatic fluids. Fluids recovered directly from the well-bore contain 3400 ppm Zn
and 1200 ppm Pb, with estimates of greater than 5000 ppm Zn in the original (approx. 55 wt%
NaCleq) hypersaline liquid [77,78,80]. Copper, however, is at concentrations below detection in the
recovered hypersaline fluids; this is thought to be due to rapid Cu-sulfide precipitation during cooling
and decompression of fluids in the well-bore prior to sampling [77]. Measurements of fluid inclusions
from Kakkonda drill-core indicate temperatures above 500°C and salinities above 50 wt% NaCleq [81]
consistent with sampling conditions of the well-bore fluids.
The Larderello and Kakkonda examples indicate that fluid inclusions from drill-core can provide
reliable insights into the temperature and chemistry of deep geothermal fluids. Unfortunately, there
are currently no direct trace metal analyses of fluid inclusions from Kakkonda or Larderello. However,
insights into their chemistry can be gained from published petrographic studies. Detailed examination
of quartz-hosted fluid inclusion images from Kakkonda [82–84] reveals that they contain abundant 8
opaque daughter minerals indicative of substantial chalcophile metal concentrations (figure 3a–d),
similar to those observed in metalliferous hypersaline liquid inclusions from experiments [22,25] and
from ore deposits (e.g. [85]). From the images, we can estimate the bulk composition of the fluid
inclusions using pixel maps and volume calculations for cubic halite crystals, tetrahedral or cubic
sulfides (chalcopyrite or pyrite/pyrrhotite) and spherical bubbles, compared with the remaining
liquid volume of the inclusion. For fluids with known salinities and Cu contents (figure 3a,b), a
combination of volume calculations with densities of observed solid phases recovers salinities (45–
60 wt% NaCleq) and Cu contents (5000–7000 ppm) that lie within 1 sigma of the average for the range
measured by laser-ablation inductively coupled mass spectrometry (LA-ICPMS) on the same
inclusions. This gives us confidence in our methodology when applied to images of fluid inclusions
with unknown Cu concentrations. Assuming, on the basis of crystal shapes, that the opaque phases
from Kakkonda are a mixture of chalcopyrite (34% Cu) and pyrite/pyrrhotite that is saturated in
chalcopyrite (approx. 3.5% Cu), we estimate Cu concentrations in the range of 500–5000 ppm

(figure 3c,d and table 1). No explicit measure of the uncertainty on these Cu estimates is possible.
However, any large variations in inclusion thickness would create a mismatch between the calculated
inclusion salinity and that reported. Our estimated values are not meant to define accurately the Cu
concentration, but simply to show that fluid inclusion petrography is consistent with Cu
concentrations at levels similar to those observed in magmatic brines from PCDs.
Direct analyses of fluid inclusions from high-temperature, magma-driven geothermal systems, such
as those described here, by LA-ICPMS can provide much more accurate constraints of the contents of
Cu and other metals. To date, very few such analyses have been reported in the literature. One set of
fluid inclusion LA-ICPMS data is available for moderate temperature (approx. 400°C) Kakkonda
hypersaline liquids [82]. Measured Zn/Cu ≈ 7 for some of the assemblages. Combined with other
direct measurements and estimates of Zn in the recovered fluids [77], this corresponds to Cu
concentrations from approximately 500 ppm to approximately 1500 ppm in the undisturbed high-
temperature hypersaline liquid reservoir at the well-bottom (table 1). These values are probably
minima given the evidence for substantial precipitation of Cu, and later Zn, as the fluid migrates
through its host rocks and cools (e.g. [82,84]). Copper loss from fluids through precipitation of
sulfides was exacerbated during well-sampling at Kakkonda, consistent with the lack of measurable
Cu in fluid samples extracted to the surface [77,82,84], and the occurrence of abundant Cu-rich well-
bore scales [86]. Scales collected from variable depths in drill holes #13, #19 and #21 contained
chalcocite, Cu-Fe arsenides, sphalerite, galena and pyrite; each scale also contains significant, but
variable, amounts of silica-alumina phases, sulfates and carbonates [84]. All of these mineral phases
are likely to have precipitated directly from hot, ascending fluids, testifying to their metal-rich
character. Cu contents of the analysed Kakkonda scales are in the range 0.27 to 14 wt% Cu, with
highly variable Zn/Cu ratios of 0.002 to 74 [86].
The Kakkonda example of deep geothermal drilling suggests that high-temperature, magmatic brines
can have significant base metal concentrations, including Cu, consistent with the composition of
hypersaline liquids observed in porphyry ore-forming environments. Fluids recovered from Kakkonda
well WD-1b, at depths of greater than or equal to 2625 m, have lower salinity than those from WD-1a,
and show stable isotope evidence for dilution of magmatic fluids with meteoric water [87]. Mixing of
meteoric water or seawater with magmatic fluids is evident in other geothermal reservoirs drilled to
shallower depths and lower temperatures, such as Geysers (USA; [88]) and Mori (Japan; [89]). The
mixing process probably dilutes base metal contents as well as salinity, but no fluid trace metal
analyses are available for these hybrid fluids.
6. Analyses of geothermal brine fluid inclusions

In order to explore further the metal content of shallow magmatic brines, we have performed analyses of
quartz-hosted fluid inclusions from two geothermal wells associated with young magmatic systems:
Larderello, Italy; and Montserrat, Eastern Caribbean. The Larderello-Travale geothermal field is
associated with late Miocene, post-orogenic crustal extension in the Northern Apennines of Tuscany.
Geothermal heat, provided by a suite of 3.8 to 1.3 Ma anatectic granite intrusions [90], has been
exploited at Larderello since the early nineteenth century. It is unclear whether Larderello itself was
ever connected to a volcanic edifice at the surface, although contemporaneous rhyolitic volcanic rocks
outcrop nearby at San Vincenzo and at Roccastrada [90], and the 0.3 to 0.2 Ma Mount Amiata volcano
Table 1. Composition and trapping conditions of quartz-hosted brine inclusions from Montserrat, Larderello and Kakkonda.
location Soufrière Hills, Montserrat Larderello, Italy Kakkonda, Japan
sample no. MSV01-Ha MSV01-HBa MSV01-Ma MSV02-Ma Trav1s-Ba Rad26-Ba Rad26-Ma [85] [84]
core depth 1481 m 1481 m 1481 m 1484 m 4100 m 4600 m 4600 m 2937 m 3729 m
well temp. 210°C 210°C 210°C 210°C >325°C 300–360°C 300–360°C 370°C >500°C
nb (9):(5) (10):(4) (12):(5) (9):(7) (12):(8) (31):(11) (11):(7) (33)c (11)c
TmHalite ð CÞ 280–330e 410–450e 415–470e 415–450e 190–280e 230–275e 300–375e 151–462 194–322
ðLVÞ
Th ð CÞ 200–250 290–350 590–670 550–650 325–385 400–450 440–520 339–562 483–611
salinity (NaCleq)f 36–40 41–54 49–56 45–55 30–36 33–46 38–54 35–55 35–45
Th 280–330g 410–450g 590–670 550–650 325–385 350–450 440–520+h 339–562 483–611
Nad 53 200 (11 200) 79 300 (2600) 57 800 (7600) 73 600 (33 000) 103 500 (17 500) 116 300 (23 200) 82 100 (39 800) 68 000−81 000i
K 57 200 (7600) 73 200 (7500) 103 200 (23 000) 105 100 (48 000) 74 500 (12 000) 61 000 (16 300) 12 1200 (39 200) 42 000−50 000i
Ca BDL BDL BDL BDL 54 700 (13 700) 53 700 (11 700) 36 800 (35 700) 19 000−22 000i
Fe 62 400 (2900) 82 100 (4300) 82 200 (18 200) 52 800 (21 200) 8400 (3200) 6600 (1100) 55 400 (37 900) 49 000−59 000i
Cu 6.2 (5.6) 150 (50) 2700 (700) 1200 (300) 14.1 (9.4) 400 (200) 2500 (1100) 2900–3300j 500–5000j
Zn 1300 (300) 2200 (1100) 1800 (800) 2400 (2100) 1200 (500) 1500 (400) 8700 (7300) 3100–4200i
Pb 400 (100) 600 (300) 400 (200) 500 (400) 500 (200) 500 (100) 2100 (1100) 1000–1400i
a
Fluid inclusions are designated as halite-saturated (-H), hypersaline (-B) and magmatic (-M) brine.
b
Number of analyses, expressed as (microthermometry):(LA-ICPMS).
c
Representative microthermometry data for inclusions from these WD-1a core depths are reported in [82].
d
All compositions reported as ppm with 1 s.d. in parentheses; BDL denotes below detection.
e
Halite was often the only phase that could be assigned a precise dissolution point. Occasionally another phase (presumed sylvite) could be observed dissolving between 80–150°C.
f
Fluid salinity (wt%) was determined by using either halite dissolution or halite and sylvite dissolution. Corrections to actual wt% salt accounting for K and Fe have been made iteratively.
g
Some Montserrat inclusions homogenized by halite disappearance which may indicate trapping under halite-saturated conditions or pressure fluctuations in the brine reservoir.
h
Decrepitation of inclusions prevented homogenization of the highest temperature brine trails. Based on rate of bubble shrinkage Th is probably greater than 550°C for highest temperature brines.
i
Estimated range for major cation chemistry calculated using the recovered fluids from [77] and their estimated dilution factors to back-calculate to brines of 45–55 wt% salt.
j
Concentrations for base metals calculated based on recovered fluids (Zn and Pb) and by image analysis of brine inclusions (Cu, figure 3) due to expectation of Cu precipitation on recovery.
9
R. Soc. Open Sci. 8: 202192 royalsocietypublishing.org/journal/rsos

(and associated geothermal field) lies approximately 50 km to the southeast [76]. By contrast, Montserrat, 10
in the Lesser Antilles volcanic arc, is associated with active westwards subduction of the Atlantic seafloor
beneath the Caribbean Plate. Montserrat volcanic centres have previously erupted a range of magma
types from basalt to dacite. The island’s Soufrière Hills volcano was almost continuously active from
1995 to 2013 [91] erupting andesite lava domes with associated pyroclastic flows. Montserrat is
currently the subject of active geothermal exploration [92].
6.1. Samples
We studied two core samples taken at 4100 and 4600 m depth from, respectively, geothermal wells
Travale 1 sud and Radicondoli 26 in the eastern sector of Larderello geothermal field [58,93,94]. The
sample context for Travale 1 sud is described in detail by Fulignati [58]. Briefly, the well lies on 455 m
of dolomitic limestone and anhydrite of the Tuscan Nappe, underlain to 1630 m by Palaeozoic
phyllites and phyllitic sandstone with intercalations of anhydrites and limestone. From 1630 m to

2680 m, the well passes through intercalated hornfels and skarn in contact with a granite intrusion at
2680 m that extends to the bottom of the hole at 4150 m, where temperatures reached approximately
350°C [93]. The intrusion is a leucogranite comprising quartz, plagioclase, K-feldspar, biotite,
cordierite and tourmaline. The Radicondoli 26 well, roughly 5 km to the northwest of Travale 1 sud
and in a similar setting, is described by Carella et al. [94]. The well passes through limestones until
1350 m depth followed by phyllite and calc-silicate intercalations until intersecting a small number of
metabasite layers at 2350 m. Phyllite gives way to hornfels by 3700 m before reaching granite at
approximately 4000 m at temperatures greater than or equal to 300°C. Quartz crystals from granites of
Travale 1 sud and Radicondoli 26 were prepared for fluid inclusion study.
The Montserrat samples are from core recovered from depths of 1480 m from the MON-1 well, one
of two exploratory geothermal holes drilled into the western flank of Soufrière Hills volcano [95,96].
MON-1 well was drilled through a 530 m thick sequence of andesite lava flows and breccias related to
the young Soufrière Hills volcanic complex overlying a sequence of sandstones, mudstones and clays
(530–1210 m depth). These in turn overlie silicified tuff, breccias and sedimentary rocks, including
limestones, interpreted as older Centre Hills volcanic units and sediments deposited on the volcano’s
flanks, that extend to the bottom of the hole at 2298 m depth. The studied intervals are tuffs from this
lower volcanic segment. Fluid inclusions were identified in small quartz crystals (less than 3 mm)
filling pore spaces in two core samples, on loan from the British Geological Survey, designated MSV01
(MVOGP002-SK77147-1480.8 m) and MSV02 (MVOGP006-SK77151-1484.05 m). The present-day well
temperature at the sampled depth is 210°C [92].
6.2. Methods
Doubly-polished sections of the quartz host minerals (less than 500 µm thick) were thinned until the fluid
inclusion trails of interest were within 50 µm of the surface. Some trails could be matched to growth
bands in the host minerals, but many could not be distinguished as primary or secondary due to the
limited number of inclusions available at this preliminary stage. Microthermometric measurements
were made on brine fluid inclusions using a Linkam TS1400XY stage at University of Bristol. Halite
dissolution measurements were made on all inclusions, whereas sylvite dissolution was only observed
and measured in some inclusions. Homogenization measurements were made for inclusions that
homogenized by halite disappearance as well as by vapour bubble disappearance in different fluid
inclusion assemblages. Halite disappearance temperatures were used to estimate total salinity (wt%
NaCleq) using the SALTY salinity model [97] and were adopted as an internal standard in subsequent
LA-ICPMS analyses. Homogenization temperatures were used to help separate out groups of
inclusion types within the samples where other determinations of fluid generations were not possible.
The major and trace element compositions of fluid inclusions were determined by LA-ICPMS at the
Open University using an Agilent 8900 QQQ ICPMS and a GeoLas 193 nm excimer laser. The ‘triple-
quadrupole’ mass spectrometer allows for near-simultaneous analysis of the relevant major elements
in the brines (Na, K, Ca, Fe) along with several trace metals of interest (Cu, Zn, Pb) [98–100].
Analyses were run using an ARIS (Aerosol Rapid Introduction System) from Teledyne CETAC
designed for the Agilent 8900 that allows for rapid flushing of the sample cell and better signal-to-
background ratio. This helped to ensure good sensitivity for the trace metals despite the small size of
the inclusions. NIST610 was used as a primary element standard and to monitor for instrument drift.
MSV01-HB TRAVIS-B RAD26-B 11
10 mm
MSV01-H
10 mm
MSV01-02
10 mm
MSV02-M

RAD26-M
MSV01-M 10 mm
Figure 4. Photomicrographs of representative, quartz-hosted hypersaline (brine) fluid inclusions recovered from drill-core from
Soufrière Hills volcano, Montserrat (left) and Larderello geothermal field, Italy (right). For sample details, including depths and
well-bore temperatures, see table 1 and text. Brine inclusions in quartz were observed both along apparent crystal growth
boundaries and as secondary fracture fill. The brine inclusions contain halite (and occasionally sylvite) daughter minerals and
approximately 20–40 vol% vapour bubble. Opaque daughter minerals are readily identified in all but the lowest temperature
inclusions trapped under halite-saturated conditions (−H). Some opaque minerals have a triangular shape, and are inferred to
be chalcopyrite. Microthermometry data and LA-ICPMS analyses of these inclusions are given in table 1; histograms of
microthermometric results are given in electronic supplementary material, figure S1.
Only inclusions without significant evidence of necking, and lacking other inclusions within the
analytical volume, were selected for analysis [101]. An energy density of approximately 10 J cm−2 and
a pulse rate of 10 to 50 Hz were used to ablate through the overlying quartz host and into the
individual inclusions while stepwise increasing spot size from 10 µm to just larger than the inclusion
diameter. Element ratios (Na/K/Ca/Fe) for individual inclusions were determined by integrating the
complete analytical signal. The salinity of the inclusions, as determined by microthermometry, was
then used as internal standard to balance (as NaCleq) the Na, K, Ca and Fe present in the inclusions
and so calculate absolute abundances of Cu, Zn and Pb from element ratios determined by LA-
ICPMS. All individual fluid inclusion analyses from each group were averaged to generate the
reported composition of the brines.
6.3. Results
Representative data are presented in table 1; histograms of all fluid inclusion analyses are provided in
electronic supplementary material, figure S1. The samples contained a range of fluid inclusion types
including low-density vapour inclusions (vapour bubble greater than 60 vol%), low-temperature
aqueous inclusions (vapour bubble less than 20 vol%), melt inclusions (glass ± bubble), as well as the
hypersaline brine inclusions of interest. All hypersaline inclusions had vapour bubble fractions of
approximately 20–40 vol% and contained halite daughter minerals along with other daughter
minerals inferred to be sylvite and other chlorides (figure 4). Haematite and opaque daughter
minerals (inferred to include pyrite, magnetite and chalcopyrite) were also observed in many
inclusions. Halite disappearance temperatures (+/− sylvite disappearance temperatures) for the brine
inclusions yield salinities ranging from 30 to 56 wt% NaCleq. Samples MSV01-H, MSV01-B (table 1)
showed homogenization via halite disappearance and may have been trapped under halite-saturated
conditions. The other samples showed indications of near-magmatic trapping conditions with
homogenization to a liquid (brine) at temperatures from 500 to 670°C, indicating brines were initially
present at much higher temperatures than the current reservoir, 210°C (table 1). At Larderello, fluid
inclusion homogenization temperatures range from 325°C to more than 520°C, which overlaps the
upper end of the measured reservoir temperatures (300 to 360°C; [58]).
Major element abundances (Na/K/Ca/Fe) and trace metal analyses (Cu-Zn-Pb) for the analysed 12
brine inclusions are reported in table 1. All of the hypersaline brine inclusions have a broadly similar
Na/K ratio within each sample, ranging from 0.6 to 1.1 for Montserrat and 0.7 to 1.9 for Larderello.
Brines from Montserrat are slightly richer in Fe than those from Larderello, except at the highest
temperature (Rad26-M) where Fe contents are similar. Conversely, Larderello brines contain
appreciable (4 to 5 wt%) Ca, while those from Montserrat have Ca contents below detection; this is
probably due to differences in the composition of the melt from which the fluids exsolved and the
composition of the host rocks in which the fluids were trapped. Overall, ore metal contents are similar
in both sets of brines, and are comparable to those estimated above for Kakkonda brine inclusions
(table 1). At Montserrat, ore metal contents increase significantly with increasing temperature up to
0.24 wt% Cu, 0.27 wt% Zn and 0.05 wt% Pb. At Larderello, ore metal contents show a less obvious
progression with temperature. The most enriched brine inclusions, from sample Rad26-M (with
0.25 wt% Cu, 0.87 wt% Zn and 0.21 wt% Pb), probably record the highest temperatures (greater than
520°C) from Larderello, although decrepitation precluded accurate Th determination (table 1). The

increase in Cu content of the brines from approximately 10 ppm at 310°C to approximately 2750 ppm at
630°C is broadly in line with the temperature dependence of chalcopyrite solubility at elevated salinities
([24]; and below). The Cu, Zn and Pb contents of the analysed brines from these young active geothermal
systems lie within the range of those of brine fluid inclusions associated with PCDs as compiled by
Kouzmanov & Pokrovski [24]: e.g. 0.02 to 1 wt% Cu, 0.1 to 1 wt% Zn and 0.03 to 0.9 wt% Pb at 40 wt% NaCleq.
Our representative fluid inclusion analyses indicate that hot, metalliferous brines existed in both
systems at the time fluid inclusions were trapped, and may still be present today as intergranular
fluids of the type observed in the Kakkonda well-bore. However, in detail, the metal endowment of
brines varies from one field to another, probably in relation to tectonic setting or other magmatic
factors controlling the chemistry (salinity, metal content) of the input fluids. At both Montserrat and
Larderello, the brine inclusion homogenization temperatures are higher than recorded bottom-hole
temperatures suggestive of cooling of the reservoir since the time that the inclusions were trapped.
Cooling is associated with a marked decrease in Cu contents. Montserrat and Larderello wells do not
penetrate the base of their active hydrothermal reservoirs, thus neither of these examples directly
accesses any current reservoir of hot, predominantly magmatic brines. Consequently, in both locations
the brine metal contents of the inclusion fluids probably have been diluted through interactions with
convecting meteoric water. Only at Kakkonda well WD-1a, where the boundary between the
hydrothermal convective zone and thermal conductive zone (approx. 3.2 km depth; [79]) lies above
the sampling depth, do fluid compositions reflect the deeper magmatic reservoir.
7. The importance of sulfur

In addition to metals, reduced sulfur (S2−) is essential to formation of sulfide ore minerals, yet gauging
the concentration and redox state of sulfur in magmatic fluids faces considerable analytical challenges.
Consequently, the sulfur budget of ore-forming fluids remains a critical unknown in our
understanding of PCD formation. A key question is whether ore-forming fluids carry sufficient metals
and sulfur, or whether an external source of sulfur is required (e.g. [102,103]). For relatively oxidized
magmas, such as those commonly associated with PCDs, the dominant sulfur species in the fluid is
SO2. Thus a further question concerns the mechanism and timing of reduction of S4+ to S2−. These
questions have some bearing on the ability of magmatic brines to maintain high dissolved metal
contents as they cool in the sub-volcanic environment.
Limited available data for intermediate density (ID) fluid inclusions from PCDs indicate sulfur
contents in the range 200 ppm to 3 wt% [24] albeit with analytical uncertainties of at least ±20–50%
relative (e.g. [65]). In almost all ID fluid inclusion analyses assembled by Kouzmanov and Pokrovski
[24] sulfur is stoichiometrically in excess of that required to form chalcopyrite (CuFeS2). Similarly, Seo
et al. [65] used analyses of S in fluid inclusions from Bingham Canyon PCD, Utah, to argue that
ample sulfur, of unknown redox state, is available in stoichiometric proportion to dissolved Cu + Fe in
the same inclusions to form chalcopyrite, such that there is no need for any later addition of sulfur. If
primary magmatic, saline fluids contain sufficient reduced sulfur (S2−) to precipitate sulfides, then
phase separation and ore mineralization may be near simultaneous processes, driven by a
combination of fluid depressurization and cooling (e.g. [69]).
Conversely, using data from melt inclusions in volcanic phenocrysts, Blundy et al. [103] argue that the
Cl- and Cu-rich fluids responsible for PCD formation are unlikely to contain sufficient sulfur because of
13
104
Kakkonda
Larderello
102
ppm Cu
SHV
1
4 wt% NaCl

10 wt% NaCl
40 wt% NaCl
10–2
chalcopyrite solubility
(and geothermal brines)
10–4
200 300 400 500 600
temperature (∞C)
Figure 5. Solubility of chalcopyrite (as ppm Cu in solution) in hydrothermal solutions as a function of temperature based on
equation (10.3) for three representative salinities (4, 10 and 40 wt% NaCl). The figure is based on an empirical fit to solubility
curves presented in Fig. 17e of Kouzmanov & Pokrovski [24] for pyrite–magnetite–haematite-saturated fluids with pH = 5 at
0.3 to 1 kbar pressure. Also shown are ranges of homogenization temperatures and Cu contents for brine inclusions from
geothermal wells from Montserrat (SHV), Larderello and Kakkonda (data from table 1).
the different degassing behaviour of Cl and S. In response to this sulfur deficiency, Blundy et al. [103],
following Hattori & Keith [102], proposed that sulfur is delivered at a later, ore-forming stage,
precipitating sulfides through reaction of SO2 gas with metal-rich brines. Based on their observations
of sulfide mineralization at Pinatubo volcano, The Philippines, and Bingham Canyon, Hattori & Keith
[102] proposed that the source of this sulfur was mafic magma. Blundy et al. [103] presented
experimental evidence to support such a proposal.
Although ore minerals are predominantly sulfides, it is not essential for all sulfur in the parent magmatic
fluid to be in its reduced state. Disproportionation of SO2 to form S2− and S6+ is a potent means to create
sulfides by reaction with a metal-bearing solution, especially in the presence of Ca2+ ions when anhydrite
(CaSO4) is an additional product phase [42]. Such reactions are to be expected in transcrustal systems
where SO2-rich fluids from deeper in a system can interact with shallow-stored, Cl-rich fluids in the same
system. Coexistence of anhydrite and metal sulfides is common both in PCDs (e.g. [6,42,104]) and in
geothermal wells, including Kakkonda [105]. Supply of deep sulfur is also consistent with the ‘excess
sulfur’ phenomenon at many volcanoes, whereby the supply of sulfur to the atmosphere exceeds that
which could have been dissolved in the erupted magma [106]. If additional sulfur is required for ore-
formation, then there may be a time lag between phase separation and sulfide precipitation (e.g. [103]).
Regardless of the details of the sulfur source, a main driver for sulfide mineral precipitation from
fluids is cooling, as solubility of both chalcopyrite and bornite is strongly temperature dependent [24].
Sulfide solubility also depends on fluid salinity, pH, redox state and sulfur fugacity. For example,
chalcopyrite solubility in a pyrite–magnetite–haematite-buffered, 40 wt% NaCleq fluid with pH = 5 at
0.3–1.0 kbar falls from 2 wt% Cu at 550°C to 2 ppm at 200°C (figure 5; and [24]). This explains the
very low Cu content of cool, neutralized, low-salinity geothermal fluids. Ultimately the solubility of
chalcopyrite depends on a solubility product of Cu+, Fe3+ and S2− ions in solution. Thus, for different
intensive parameters and/or buffer assemblages, the solubility of chalcopyrite, expressed in terms of
ppm Cu in solution, will change. For example, at fixed salinity (10 wt% NaCleq), pH (5), redox
(magnetite–haematite buffer) and temperature (400°C), but in the absence of pyrite, an increase in the
fluid concentration of H2S from 0.017 to 0.1 mol kg−1 will reduce the Cu content of chalcopyrite-saturated 14
liquids from 160 to 23 ppm ([24], fig. 14d).
In terms of our analyses of fluid inclusions from Montserrat, Larderello and Kakkonda (table 1), the
solubility relationships in figure 5 may account for the overall trends in Cu content with temperature.
However, measured Cu concentrations lie below those of the relevant 40 wt% NaCleq solubility curve.
One explanation is that the brines were not saturated with a pyrite–magnetite–haematite assemblage
at the time of trapping, such that the extrapolated solubility curves are not appropriate for the pH-fO2
conditions of the studied brines. Conversely, the chemical evolution of the brines may reflect
successive episodes of phase separation during protracted volatile fluxing [70], with copper
partitioning into brines at conditions below chalcopyrite saturation, potentially coupled with dilution
by meteoric water. As the analysed brine inclusions do not necessarily represent a single suite of
co-trapped fluids, it is not possible to discriminate between these alternatives without further
chemical information, for example, from stable isotopes. Our observations emphasize the need for
experimentally calibrated thermodynamic models for high-temperature chalcopyrite solubility in high-

ionic-strength fluids. This is an area of active research.
8. Sub-volcanic brine lenses

The feeder zones (conduits) of volcanoes, through which hot, saline magmatic fluids pass, comprise
hypabyssal intrusions (dykes and sills), magma-filled conduits (the likely origin of many stocks),
pyroclastic material and fractured and brecciated rocks, typically with associated hydrothermal
alteration and quartz veins (e.g. [107,108]). High porosities (e.g. greater than or equal to 20 vol%
under Unzen volcano; [109]) are a characteristic of such systems, consistent with reduced P-wave
velocities at depths less than or equal to 8 km beneath some active volcanoes (e.g. [110]). Average
porosity of both dense and fragmental rocks decreases with increasing depth, but cores and well-logs
show that active geothermal systems can maintain porosities (matrix plus fracture) of 10 to 30 vol%
down to depths of 2.5 km (figure 6). Fragmental materials are typically the most porous; other
lithologies, such as vapour-phase-altered pyroclastic materials, are resistant to densification by
physical and chemical processes and may preserve significant primary porosity. Sub-volcanic conduits
with associated damage zones, and fragmental infills related to prior intrusions, eruptions and steam
explosions, have very high porosities (figure 6), that can provide efficient pathways for both fluid flow
and accumulation. The presence of fractures may lead to a significant mismatch between relatively
high porosity at a reservoir scale and relatively low porosity on a drill-core scale, as evidenced by
data and modelling at Kakkonda (figure 6). Precipitation from (and dissolution by) ascending fluids
will modify the original porosity structure in the feeder zone.
To investigate the potential of sub-volcanic conduits as a means of focusing and storing magmatic
fluids, Afanasyev et al. [70] performed hydrodynamic models of discharge through a sub-volcanic
domain of a 738°C supercritical aqueous fluid (4 wt% NaCl) from a magma body located at 7 km
depth. Afanasyev et al. [70] considered explicitly the effect of a volcanic conduit region composed of
high permeability, fractured rock. The porosity–depth distribution used by Afanasyev et al. [70]
throughout the modelled domain is presented in figure 6 for comparison with the available natural
data. For the reference model scenario (total fluid flux of 3 × 109 kg yr−1 through a 0.33 km radius
conduit of permeability k ≤ 10−14 m2), Afanasyev et al. [70] presented calculations for degassing
durations of up to 250 kyr. The permeability-depth distribution adopted by Afanasyev et al. [70] results
in an effective transition from hydrostatic to lithostatic pressures at depths of 5–6 km and temperatures
of 440–520°C, corresponding to the transition between hydrothermal convection and conductive
regimes described above. A region of active hydrothermal convection develops in the upper 3 km of
their modelled domain. Thus the model configuration approximates that of typical geothermal systems.
It is instructive to adapt the simulations of Afanasyev et al. [70] to White Island volcano (New
Zealand), a plausible modern analogue to the scenario being explored. Hedenquist et al. [14] report a
White Island flux of 3.5 × 105 kg day−1 of SO2 via a fumarolic fluid containing, on average, 3.5 wt%
SO2 at temperatures up to 800°C. The corresponding total fluid flux is 3.7 × 109 kg yr−1, a value
similar to that used by Afanasyev et al. [70]. The Cu content of fluids discharged from White Island,
based on the reported Cu/SO2 ratio, is 30 ppm. This is probably a minimum value for the original
magmatic fluid, due to sub-surface deposition of Cu beneath the edifice [14] as well as phase
separation of Cu-rich brines. The total duration of degassing from White Island is not known—for the
purposes of the calculations reported here, we consider only the 10 kyr case. Note that this is a
0 15
500
1000
1500
depth (m)

2000
core (dense)
core (fragmental)
well-log (Unzen)
2500 core (Kakkonda)
reservoir (Kakkonda)
Afanasyev et al.
3000
convective
conductive
3500
Kakkonda MT
4000
0 10 20 30 40 50 60
porosity (%)
Figure 6. Porosity–depth relationships in volcanic-hosted geothermal reservoirs, and volcanic conduits. Geothermal well drill-core
data (open and filled black symbols) are redrafted from Fig. 46.14 of Stimac et al. [71], and distinguish between dense (e.g. lavas,
intrusives) and fragmental (breccias, tuffs, volcaniclastics) rock types. Blue line shows well-log data for the Unzen conduit taken from
Fig. 2 of Ikeda et al. [109]. Red squares show porosities of individual drill-core rock samples from Kakkonda [111]; solid red line
shows the depth-porosity relationship adopted in the Kakkonda reservoir model of McGuiness et al. [112]. In this model, porosity is
set at a fixed value in a succession of horizontal layers, decreasing from 18.3% in the top layer to 8.5% in the bottom layer, based
on the sonic and density log results of Kakkonda well WD-1 series and on an analysis of the electric logs of other Kakkonda wells
reported by Sakagawa et al. [113]. The shaded box labelled Kakkonda MT is the range of calculated porosities for the approximately
3.7 km deep, 0.63 to 1.0 S m−1 conductor at Kakkonda [80], as calculated in the text. The boundary between the hydrothermal
convective zone and thermal conductive zone in WD-1a well is located at approximately 3.2 km [79] and shown with the grey line.
Purple dashed line is the depth-porosity relationship adopted in the brine-lens model of Afanasyev et al. [70].
relatively short degassing period in the context of the long-term growth and differentiation of crustal
magmatic systems. Despite the overall similarity of modelled conditions to those at White Island, we
recognize that fluid fluxes and metal contents may vary significantly from one volcanic system to
another. For that reason, our calculations are designed merely to be illustrative.
Afanasyev et al.’s [70] calculations predict accumulation of relatively dense (ca 1350 kg m−3)
hypersaline liquids at approximately 2 km depth in annular lenses surrounding the conduit
(figure 7a), in close agreement with the calculations of Weis et al. [68] and Weis [69], despite some
differences in modelling approaches. Both models concern flows of NaCl–H2O fluid that account for
phase transitions and multi-phase equilibria, and similar governing equations are used. The main
difference concerns the modelling of permeability. Instead of the dynamic permeability model of Weis
[69], we simulate the brittle–ductile transition with the static depth-dependent profile for permeability
described above. Afanasyev et al. [70] show that phase transitions in NaCl–H2O result in fluid
focusing and brine lens formation even in a static permeability field.
(a) r (km) (b) (c) (d) (e) 16
0 1 2
0.4 7000 7000 20 730
Cu concentration in fluid (ppm)
Cu concentration in fluid (ppm)

2
Cu deposition (kg m–3)
temperature (°C)
fluid bulk salinty
z (km)

sulfur sulfur
no sulfur unlimited unlimited
0 0 0 0 20
Figure 7. Modelled distribution of salinity, Cu and temperature in the sub-volcanic region for the reference scenario of Afanaysev
et al. [70] after 10 000 years of degassing in terms of (from left to right): (a) bulk fluid salinity expressed as molar fraction NaCleq;
(b) concentration of dissolved Cu (as ppm Cu) for the no-sulfide case; (c) concentration of dissolved Cu (as ppm Cu) for the
unlimited-sulfide case; (d ) precipitated Cu for the unlimited-sulfide case, expressed as kg Cu per m3 of rock; and (e)
temperature (°C). The conduit radius is 0.33 km, centred on the vertical axis; input flux of fluid (738°C, 4 wt% NaCleq,
700 ppm Cu) is 3 × 109 kg yr−1 originating from 7 km depth. Black dots in all panels indicate halite precipitation. The volume
containing hypersaline liquids with greater than 5000 ppm Cu is outlined in black in (b)–(d). Panels have identical vertical and
horizontal scales, and are axisymmetric about the left-hand axis.
A direct comparison of our models and those of Weis [69] is provided in electronic supplementary
material, figure S2. The depth and shape of modelled brine lenses echo the original proposal of
Fournier [67,72] that brine accumulates below the brittle–ductile transition, which serves as a
permeability barrier. However, the numerical models predict further that the hypersaline liquid lens is
capped by halite precipitation that occludes pore space and restricts further upward movement.
Precipitation of other solutes, such as SiO2, may also help to confine the brine lens, although this has
not yet been explored in detail in our models. Halite precipitation has been described from ore
deposits [56,114,115] and geothermal reservoirs [116,117] and is consistent with the evidence of halite
saturation in some of the brine fluid inclusions in table 1.
In a parametric analysis, Afanasyev et al. [70] showed that the hypersaline liquid lens is not well
developed for lower temperature fluid inputs (less than 700°C) or for very wide (r > 1 km) conduits,
but is enhanced for narrow conduits, and hotter, more saline input fluids. Modelled lenses persist for
up to 1 Myr after degassing ends, eventually sinking, spreading out, and being diluted and dissipated
by any meteoric water convection at the margins [70]. This situation resembles that in many
geothermal reservoirs, including Kakkonda [79,112], where the meteoric and magmatic fluid systems
are efficiently isolated from one another despite active hydrothermal convection.
9. Geophysical observations
Geophysics affords considerable insight into the architecture and dynamics of crustal magma systems [118].
Techniques include potential fields (electrical conductivity, gravity, magnetics), seismic wave speeds and
tomography, and geodesy (ground- and satellite-based). Increasingly these techniques are being applied
to individual volcanic centres to elucidate magma plumbing. Although the images differ from volcano
to volcano, a unifying theme of most geophysical surveys of volcanoes is the presence of relatively large
volumes of reduced seismic wave speed and enhanced electrical conductivity at depths greater than or
equal to 2 km consistent with large distributed volumes of partial melt and/or magmatic fluids, rather
than a single large-liquid-filled magma vat. Real-time geodetic surveys indicate that such systems are
dynamic on timescales of years to decades [119,120], probably due to the relative mobility of fluids (e.g.
[121]). Such images are consistent with the upper reaches of transcrustal magmatic systems (figure 1).
However, the non-uniqueness of geophysical techniques, especially when applied in isolation, makes it
difficult to resolve relatively small system features, and to establish the abundance, composition and 17
distribution of the fluid phase (melt, hypersaline liquid, gas).
From the perspective of sub-volcanic brine lenses, electrical conductivity has the greatest potential as
an imaging and mapping tool. Electromagnetic surveys, including magnetotelluric (MT) and time-
domain (TDEM), have been conducted at numerous active and dormant volcanoes, including Mount
Fuji, Japan [122]; Taal, The Philippines [123]; Kusatsu-Shirane, Japan [124]; Uturuncu, Bolivia [125];
Unzen, Japan [126–128]; Taupo Volcanic Zone, New Zealand [129–131]; and Mount St Helens, USA
[132,133]. These surveys consistently reveal prominent, electrically conductive (greater than or equal to
1 S m−1) regions at depths greater than or equal to 2 km, interpreted by Afanasyev et al. [70] to
represent hypersaline liquid accumulations. Although this is the prevailing interpretation of the
authors of these many studies, other interpretations are possible and have been advanced. For
example, conductive lenses may be smectite clays, although these are stable only at less than
approximately 250°C [71], temperatures well below those expected at such depths in the active
volcanic systems imaged. Moreover, clay layers, where present, e.g. Rotokawa [129], Kusatsu-Shirane

[124] or Montserrat [92], tend to be laterally persistent and lie at depths less than 1 km, with forms
resembling the clay caps of geothermal systems. On the other hand, the conductive lenses might be
magma, although the lower electrical conductivity of silicate melt compared with hypersaline liquid
would require high melt fractions (greater than 50%) at relatively shallow depths to attain the same
bulk conductivity. Such high melt fraction is commonly inconsistent with other geological evidence,
such as shear wave speeds (e.g. [125,128]). Finally, the conductor may be composed, in part, of
electrically conductive minerals such as sulfides or magnetite, which can have conductivities as high
or higher than brines (e.g. [134]). However, conductive minerals would need to be fully connected
through the rock volume in order to generate the observed anomalies; the requisite wetting behaviour
is more consistent with a fluid conductor phase.
Notably, the calculated depth, geometry and electrical conductivity of the modelled and imaged
hypersaline liquid lenses are very similar [70]. In a number of MT studies, the conductor at 2 km
depth extends to the middle or lower crust (e.g. [125,129–133]) suggesting connectivity to a deeper
magma source. In those cases, the upper regions of the conductor would represent fluids exsolved
from ascending hydrous magmas (± precipitated sulfides), creating a vertical, conductive continuum
from regions of hydrous partial melt to accumulated fluids, as shown schematically in figure 1. MT
surveys alone would not be able to resolve this transition.
Despite the potential ambiguities of MT images, the example of Kakkonda provides some important
confirmatory evidence for the existence of brine lenses, as, uniquely, Kakkonda is a location where MT
images are supported by direct fluid sampling. Well WD-1a penetrated the periphery of the main
conductive anomaly as imaged by Uchida et al. [135]. Conductivity at WD-1a well bottom, where
temperature is greater than equal or to 500°C, is in the range 0.10–0.16 S m−1 (MT profiles Line-C and
Line-1 of [135]). The core of the conductor body lies approximately 1.5 km to the southeast of WD-1a
at approximately the same depth (2–4 km) and temperature, but with a conductivity of 0.63 to less
than 1 S m−1. The measured porosity of the brine-hosting core from 3727 m depth is 2.4% [111].
Afanasyev et al. [70] adopt a formulation of Archie’s Law for calculation of the electrical conductivity
of a porous medium containing intergranular saline fluid. For the case where the pore space is filled
exclusively by fluid (i.e. without halite, for which there is no evidence at Kakkonda) the conductivity
of the rock is given by
sl fm
sr ¼ , ð9:1Þ
a
where sr is the electrical conductivity of the fluid-saturated rock, sl is the electrical conductivity of the
fluid, calculated at the pressure–temperature–salinity of interest using the model of Sinmyo & Keppler
[136], f is the porosity, m is the cementation exponent and a is the tortuosity factor. Values of m = 1.9
and a = 0.6 were adopted by Afanasyev et al. [70] and are retained here for consistency.
The calculated conductivity of 39 and 55 wt% NaCleq brine at 1 kbar and 500°C is 53 and 69 S m−1,
respectively. For a porous medium (f ¼ 0:024; [111]) saturated with these two brines the calculated rock
conductivity, from equation (9.1), is 0.08 and 0.10 S m−1, respectively, in excellent agreement with the MT
data for WD-1a well bottom. It is possible to calculate the porosity in the core of the conductor, assuming
that the brine has the same salinity range (39–55 wt% NaCleq), pressure (depth) and temperature. To
obtain the MT value of 0.63 S m−1 requires porosities in the range 6.4 to 7.5 vol%. To attain a
conductivity of 1 S m−1, a value at the upper bound of what is imaged at Kakkonda [135], but in
keeping with conductivity anomalies beneath many volcanoes, the required porosity is 8.2 to
9.5 vol%. The calculated range of porosity values is consistent with those used in the model of Afanasyev 18
et al. [70] at these depths (10 vol% at 3.5 km), and with the McGuinness et al. [112] reservoir model for
Kakkonda (figure 6). We conclude that in the one location where hypersaline fluid accumulation has
been proven by drilling, the electrical conductivity data can be adequately explained by brine at
porosities of several per cent. For reference, the measured permeability of core from WD-1a well
bottom (3727 m depth) is 10−16 m2 [111], a value intermediate between Afanasyev et al.’s [70] model
conduit (10−15 m2) and matrix (10−18 m2) permeabilities at this depth.
10. Copper accumulation in sub-volcanic brine lenses

To assess the potential for metal accumulation in sub-volcanic brine lenses we have incorporated copper
transport into the hydrodynamic simulations of Afanasyev et al. [70] by including partitioning of Cu
between vapour and liquid of differing salinity, and the solubility of chalcopyrite. In the light of the

importance of S2− ions for copper ore mineral precipitation our new model considers the fate of both
sulfide-sufficient and sulfide-free, Cu-bearing magmatic fluids as they discharge through a high-
permeability and porosity conduit, undergo phase separation and cool. We note that Weis et al. [68]
also incorporated Cu solubility and partitioning into their models, concluding that Cu accumulates
below a halite cap, within a region of elevated salinity, similar to the hypersaline liquid lens in figure 7a.
10.1. Methods
We augment the hydrodynamic model equations of Afanasyev et al. [70] with the following conservation
equation for modelling Cu transport
0 1 0 1
@@ X X
f ri ci si þ (f0 f)rCu A þ div@ ri ci wi A ¼ 0, ð10:1Þ
@t i¼g,l i¼g,l
where ϕ is the porosity, ϕ0 is the initial porosity prior to Cu precipitation, ρ is the density, c is the Cu mass
fraction, s is the saturation, w is Darcy’s velocity, and subscripts g, l and Cu refer to parameters of the gas,
liquid and chalcopyrite (solid) phases, respectively. In using equation (10.1), we assume that, providing
reduced sulfur is available, Cu forms a separate solid phase (e.g. chalcopyrite) that precipitates
independently of halite. The copper sulfide mineral density ρCu is set to 4100 kg m−3.
The total concentration of Cu, ct, in gas and liquid phases is given by
X X
ct ¼ ri ci si ri si : ð10:2Þ
i¼g,l i¼g,l
Copper partitioning between vapour and liquid phases is calculated in line with the experimental data in
Fig. 15 of Kouzmanov & Pokrovski [24]
mg rg
log ¼ 3:866 log , ð10:3Þ
ml rl
where m is the number of Cu moles per 1 kg of fluid in the corresponding phase. This expression is
similar to that of Tattitch and Blundy [25] wherein copper partitioning scales with the salinity ratio of
liquid and vapour.
To constrain reasonable upper bounds on brine Cu contents requires an expression for the
equilibrium solubility of Cu, ceq, at brine lens conditions. As noted above, ceq depends on a number of
factors, including intensive parameters and the mineral assemblage that buffers fH2S, and fO2. In the
absence of a robust thermodynamic model at elevated temperatures, we have chosen to simply
interpolate the pyrite–magnetite–haematite-saturated solubility data in Fig. 17e of Kouzmanov &
Pokrovski [24] to yield an empirical relationship for ceq(T,x) as a function of temperature and fluid
salinity x:
log (ceq ) ¼ a00 þ a10 ln (x) þ a01 T þ a20 [ln (x)]2 þ a11 ln (x)T þ a02 T 2 , ð10:4Þ
where T is in °C, x is weight fraction NaCleq in the liquid, a00 = –2.616, a10 = 1.741, a01 = 0.02567,
a20 = 0.07463, a11 = –2.117 × 10−5, a02 = –0.00153. The resulting expression, for three representative
salinities, is shown in figure 5. Copper is completely dissolved in hypersaline liquid if ct < ceq(T,x) and
chalcopyrite precipitates when ct ≥ ceq(T,x). Further experiments are required to refine these solubility
relationships at high temperature (greater than 400°C) and salinity; equation (10.4) is simply a useful 19
starting point for a plausible set of brine lens conditions.
Calculations assume that pH of the fluid is 5, consistent with rock-buffering via hydrolysis reactions
involving K-feldspar and muscovite. For a one molal NaCl solution (5.8 wt% NaCleq) at 225°C these
reactions define pH = 5 ± 0.5 [137]. Hypersaline liquids recovered at depth at 500°C from Kakkonda,
have pH in the range 3–5 [77,78]. If pH deviates from 5 then chalcopyrite solubility will be modified.
For a unit decrease in pH below 5, all other parameters constant, solubility increases by an order of
magnitude; for a unit increase in pH solubility decreases by the same factor [24]. Modelled pH
evolution of the fluids in response to rock reaction would be a future development, as would
modifications to the buffer assemblage. However, comparison of the calculated solubility curves with
our analyses of fluid inclusions from geothermal drill-core (figure 5 and table 1) suggests that
equation (10.4) provides a reasonable approximation to natural systems that contain sufficient S2− to
precipitate chalcopyrite.
Numerical simulation of Cu transport is done using the method of splitting by physical processes.

Each time step consists of two separate steps A and B. In Step A transport of the primary NaCl–H2O
fluid is simulated for a given porosity ϕ and permeability, and new distributions of P, T, ρi, si, wi are
evaluated (see [70]). In Step B, these distributions are frozen and Cu transport is calculated according
to equations (10.1)–(10.4). At the end of Step B if ct in a grid block is higher than the equilibrium
value (5) then the fraction ct-ceq of Cu is precipitated and thereby ct is reduced to ceq. Based on the
amount of precipitated Cu the porosity ϕ and the permeability distributions are modified. The
calculation then continues with Step A at the next time step.
Calculations were run with the same initial conditions as the reference model of Afanasyev et al. [70]
in terms of fluid flux, permeability, porosity etc. (figure 7a). The input fluid contains 700 ppm Cu, a
conservative value based on ID fluid inclusions from Bingham Canyon [63,65] and comparable to the
500 ppm Cu used by Weis et al. [68]. As noted above, this value exceeds that in the White Island
fumarole gas (30 ppm), which has been lowered by sub-surface reactions and Cu mineral
precipitation. Sulfide content of the input fluid is deemed either: (i) sufficient to precipitate
chalcopyrite when its solubility is attained according to equation (10.4); or (ii) zero, in which case no
Cu precipitation can occur. Our results, therefore, represent end-member cases where Cu precipitation
is (i) maximized or (ii) minimized at the expense of Cu retained in solution. We do not consider explicitly
sulfide formation via the Ca-mediated disproportionation of SO2 proposed by Mavrogenes & Blundy [42],
nor variations in mineral buffer assemblage.
10.2. Results
Model results for the reference scenario are presented in terms of bulk fluid salinity (figure 7a), Cu
dissolved in hypersaline liquid for the no-sulfide (figure 7b) and unlimited-sulfide (figure 7c) cases,
Cu contained in precipitated solids (unlimited-sulfide case only, figure 7d), and the temperature field
(figure 7e). In panel (b–e), the region containing hypersaline liquids with greater than or equal to
5000 ppm Cu is outlined. Results are shown at 10 000 years after the onset of fluid discharge
(figure 7a). For both precipitated and dissolved Cu, concentrations are highest in an annulus centred
on the conduit at 1.5 to 2 km depth. The shape and depth of the precipitated Cu annulus is similar to
that of Weis et al. [68], despite their use of a different chalcopyrite solubility expression and modelling
approach (electronic supplementary material, figure S2)
The precipitated Cu annulus (figure 7d ) is more restricted in extent than the Cu-rich liquid annulus
(figure 7b,c), reflecting higher temperatures near the conduit. The highest bulk contents of precipitated
Cu, approximately 20 kg m−3 of rock, are contained within an inclined region located at 1.5 km depth
700 m from the centre of the conduit (figure 7d). Very little Cu is precipitated within the conduit
itself. These bulk Cu contents equate to maximum potential ore grades of 0.75 wt% Cu for rock with
2650 kg m−3 density. This grade is comparable to many economic PCD mines. Precipitated Cu
contents decrease rapidly away from this region reflecting temperature decrease. The mass of solid Cu
(as chalcopyrite) increases almost linearly with degassing duration, as the liquid lens becomes
saturated and more and more chalcopyrite precipitates. Thus ore grades of approximately 1.5 wt% Cu
would require degassing for greater than or equal to 20 kyr at the reference flux.
The Cu-in-liquid annulus (figure 7b,c) is slightly broader and deeper than for precipitated Cu,
extending from the axis of symmetry out to 1 km radial distance. The no-sulfide annulus (figure 7b) is
slightly broader than the unlimited-sulfide annulus (figure 7c). The highest hypersaline liquid Cu
contents (greater than or equal to 7000 ppm) occur at the outermost periphery of the annulus where
20 20
solid Cu
dissolved Cu
dissolved Cu (no sulfur)
15
Cu (Mt)
10

0 2 4 6 8 10
t (kyr)
Figure 8. Time evolution (up to 10 000 years) of total mass of dissolved (orange line) and precipitated Cu (dark blue line) for
unlimited-sulfide case, and dissolved Cu ( pale blue line) for the no-sulfide case. Models are for the reference scenario in
figure 7. The total Cu input to the modelled domain is 21 Mt. Copper masses are integrated across the reference volume
defined in the text; there is very little stored Cu in the modelled domain beyond this volume. Note the tendency towards
steady-state mass of dissolved Cu after 10 000 years.
temperature exceeds 400°C (figure 7e) for both the no-sulfide and unlimited-sulfide cases. These
modelled Cu contents are consistent with high-salinity fluid inclusion analyses from PCDs (e.g.
[24,63,65]). Note that, for the unlimited-sulfide case, contours of Cu-in-liquid concentration do not
map simply onto isotherms, because salinity (in addition to temperature) also controls Cu solubility
(figure 5). In the no-sulfide case, the highest Cu contents in the hypersaline liquid lens are controlled
not by chalcopyrite solubility, but by the dynamics of phase separation and partitioning. They are
therefore more sensitive to the model parameters, such as input fluid Cu content, fluid flux and
porosity–permeability characteristics of the conduit region than for the unlimited-sulfide case.
Total Cu input to the system in 10 000 years is 21 Mt for the modelled fluid flux and Cu content. For
the unlimited-sulfide case, the total mass of Cu retained in the modelled domain includes both
precipitated and dissolved forms. For a reference volume with radius 1.5 km extending from the
surface to 3 km depth the mass of precipitated Cu is 20 Mt after 10 000 years (figure 8). The mass of
Cu dissolved in hypersaline liquid at this time (approx. 1 Mt) is less by a factor of 20. Thus, almost no
copper is lost to the atmosphere via the low-salinity exhaust vapour, and very little is stored at low
concentration outside the reference volume. For the no-sulfide case, there is no precipitated Cu, but
the total mass dissolved (approx. 3.5 Mt after 10 000 years; figure 8) is less than the total Cu retained
in the unlimited-sulfide case, due to significant Cu loss to the atmosphere, at a time-averaged flux of
4800 kg day−1. For comparison, copper fluxes to the atmosphere of 200 to 20 000 kg day−1 have been
recorded at several subduction-related basalt volcanoes [4]; at White Island andesite volcano the
copper flux is approximately 300 kg day−1 [14]. The range in Cu flux between volcanoes probably
reflects variation in the initial fluid content of Cu and availability of reduced sulfur, as well as the
efficiency of Cu sequestering by phase separation, sulfide precipitation, and mixing with meteoric
water beneath the volcanic edifice. Our no-sulfide and unlimited-sulfide cases probably provide
bounds on the situation beneath active volcanoes.
The annular form of the modelled high-Cu regions, both precipitated and dissolved (figure 7c–e), is
reminiscent of the three-dimensional form and dimension of ore shells in several PCDs, e.g. Bajo de la
Alumbrera, Argentina [47], Bingham Canyon, USA [138], Batu Hijau, Indonesia [139]. This is
consistent with the notion that hypersaline liquid accumulation is related to ore formation, as
previously suggested by, inter alia, Fournier [72], Weis et al. [68], Weis [69] and Blundy et al. [103]. The
size and copper endowment of the modelled high-Cu regions for the unlimited-sulfide case (approx.
20 Mt) are broadly similar to these world-class PCDs, whereas the endowment for the no-sulfide case
is substantially smaller. This comparison emphasizes the importance of the availability of sulfide (as
well as Cu) in generating PCDs.
As discussed above, a central question in PCD research is the extent to which a single magmatic fluid can
provide both sulfide and copper, and, if so, under what circumstances. Needless to say, a sulfide-poor copper-
rich brine lens will not become a large PCD without the subsequent addition of sulfur, as noted by Blundy et al. 21
[103]. It is well established in volcano studies that the mass of sulfur discharged from active volcanoes
frequently exceeds that dissolved in erupted magma [106]. One explanation for this ‘excess sulfur’ problem
is that significant volumes of unerupted magma contribute to sulfur discharge through the volcanic edifice.
Degassing from a transcrustal magma system, as shown in figure 1, provides a ready means to supply
additional sulfur. The requirement for substantial quantities of sulfur to form sulfide-rich ores may explain
the relative scarcity of world-class PCDs compared with the apparent ubiquity of electrically conductive
regions beneath many volcanoes. If such lenses do represent accumulations of metal-rich hypersaline
liquids, absent an adequate supply of sulfide they may eventually become diluted and dissipated through
groundwater interaction, with low preservation potential.
In our simulations, the conduit region remains at high temperature (figure 7e) because degassing has
been ongoing for 10 000 years. When degassing stops, the system cools; in the unlimited-sulfide case
further Cu-sulfide precipitates. Thus, over time the Cu-sulfide ore body grows at the expense of the
Cu-rich hypersaline liquid lens. Sulfide precipitation can occur in the no-sulfide case only if the

Cu-bearing brine lens interacts with subsequent pulses of sulfur-rich fluid from deeper in the system.
Where this fluid contains predominantly SO2, sulfide precipitation may occur by the Ca-mediated
disproportionation reaction described above. In this case, brine accumulation and lens growth is
decoupled in time from sulfide ore formation, as suggested by Blundy et al. [103].
Working against brine lens growth are dilution and removal of liquid by convecting groundwater, as
suggested by modelling [70,140]. Nonetheless, hypersaline liquid lenses appear to be a persistent feature
of volcanoes even several hundred thousand years after the end of degassing (e.g. Uturuncu, [125]).
Development of a quasi-steady state system after more than 10 000 years (e.g. Figure 8) relates to a
balance between the influx of new magmatic fluid, entrainment of hypersaline liquids into the
associated hydrothermal system, and discharge of Cu to the atmosphere. It is likely that this balance
will vary from volcano to volcano and depend to large extent on the physical separation of the
magmatic liquid reservoir and the convecting hydrothermal system.
11. Implications for in situ mining

Copper mining exploits ore deposits that have long since cooled and exhumed to the (near) surface.
Consequently, the bulk of the Cu in a hypogene PCD is in the form of solid sulfides with vestiges of
Cu-bearing hypersaline liquids confined to tiny fluid inclusions in the host rocks. Because of relatively
low ore grade (less than 1 wt% Cu) mining entails large-volume extraction of Cu-poor rock in open
pits or underground mines. Extracted sulfide ore must then be crushed, concentrated and smelted,
prior to dissolution and electrowinning to generate high-purity copper cathode. Thus, copper
production from hypogene ore is costly, in terms of energy [141], and environmentally impactful, in
terms of excavation, hazardous reagents, disposal of mine tailings, generation of rock dust and
smelter emissions [142].
The possibility that sub-volcanic hypersaline liquid lenses contain substantial dissolved Cu raises an
alternative, in situ approach to mining, namely direct extraction to the surface of hot, Cu-bearing liquids
via boreholes that penetrate the brine lens. Our reference model results indicate that the highest dissolved
Cu concentrations are found in rocks at a radial distance of approximately 1 km from the volcanic
conduit and depths of approximately 2 km (figure 7b,c). Dissolved Cu concentrations are highest
when the transporting fluid is poor in reduced sulfur. Temperatures in this region are very high
(greater than or equal to 400°C, figure 7e), but within the range of the latest generation of geothermal
targets. For example, the Iceland Deep Drilling Project geothermal well IDDP-2 at Reykjanes
penetrated 426°C volcanic rock at depths of 4.5 km [143], the Venelle-2 well at Lardarello reached
504°C at 2.8 km [144], and the Kakkonda WD-1a well reached 500°C at 3.7 km [79]. There is emerging
interest in drilling even deeper, hotter systems to extract power from supercritical fluids associated
with young magmatic systems [73,74,145–147]. For example, the Japan Beyond Brittle Project at
Naruko volcano will target 350–500°C fluids stored in ductile reservoir rocks at 3–5 km depth
[148,149]. A 1 S m−1 conductivity, vertically elongate body (C1) beneath Naruko extends from the
lower crust to approximately 4 km depth [150]. This structure is consistent with a body of melt at
depth that transitions to saline fluids at shallower levels.
The potential of hot hypersaline liquids as a source of metals (e.g. Mn, Zn, Pb) and silica has long
been recognized [151–157], but their attempted exploitation has a chequered history (e.g. [158–162]).
Demand for lithium has led to a recent resurgence of interest in metal recovery from high-salinity 22
geothermal fluids [163].
Provided that deep wells into hot rock can be completed successfully, the difficulty of recovering
Cu-bearing fluids faces two key problems. The first is the relatively low contents of Cu (and other
metals) in conventional geothermal fluids. Indeed, Neupane & Wendt [164], in their assessment of US
geothermal mineral resources, state ‘the presence of low mineral content in the brine could be
prohibitive for extraction’. The available data for geothermal fluids (e.g. [165]) support this statement.
However, almost all geothermal fluids analysed so far are mixtures of magmatic and meteoric fluids
with quite low salinities (average approx. 1 wt% NaCleq; [71]) recovered from actively convecting
hydrothermal systems at temperatures below 360°C, where solubility of copper minerals is low
(figure 5). The lack of suitably hot, solute-rich, hypersaline magmatic liquids, with the exception of
Kakkonda, is largely a consequence of intentionally avoiding such regions when drilling for
geothermal resources. Recovering metal-rich hypersaline liquids requires tapping brine lenses at
significantly higher temperatures than is conventional in geothermal fields, and at depths below the

brittle–ductile transition, remote from dilution by convecting meteoric water.

A second challenge is scaling and corrosion of the well-bore in response to decompression and
cooling of solute-rich, acidic fluids (e.g. [166]). It is for this reason that most geothermal production
wells avoid solute-rich liquid lenses in favour of low-salinity, high-enthalpy vapour [74]. Scales
recovered from a number of geothermal wells are extremely Cu-rich, due to precipitation of sulfides
(and tellurides). For example, at Kakkonda, variable temperature histories in both the hypersaline
liquid reservoir and during the sampling process itself resulted in Cu-Zn-Pb-Fe rich sulfide/oxide
scales with a wide range of chalcophile metal ratios [80] and up to 14 wt% Cu [86]. Despite this
variability, the record of deposition of metals in the scales, combined with a transition from Pb-rich to
Zn-rich to Cu-rich scales with increasing temperature, is consistent with a high-temperature (greater
than 500°C) hypersaline reservoir at Kakkonda that, at depth, may contain dissolved metals at
significant concentrations, as suggested by the fluid inclusion estimates in table 1.
The propensity for scaling will be exacerbated when extracted fluids are even hotter and more solute-
rich than in conventional geothermal wells. There have been a number of attempts to develop fluid
extraction methods to minimize scaling by modifying the chemistry and temperature of the extracted
fluid (e.g. [167–169]). Reducing scaling of silica, an important additional component of geothermal
fluids, has been a particular priority (e.g. [170–172]). Further development of such methods will lead
to more efficient recovery of metals from deep, hypersaline fluid reservoirs.
Finally, drilling into hot rock and extracting metal-bearing fluids represents a major technical
challenge because of uncertainties in the permeability and porosity structure of the reservoir, and the
highly corrosive nature of the fluids themselves. This is largely because there has been very little
drilling into deep portions of geothermal systems with accumulations of high-temperature magmatic
fluids lying below the brittle–ductile transition. Drill-core samples recovered from such wells will tend
to underestimate the in situ porosity and permeability, as cooling and reduction in fluid pressure
cause mineral precipitation and closing of fractures around the well-bore. Also, where hypersaline
magmatic fluids accumulate there may be dynamic permeability variations with spatial and temporal
changes in brittle versus ductile host rock behaviour (e.g. [69]). Data on matrix porosity and
permeability from traditional geothermal reservoirs are often proprietary or not well documented, and
there are few published attempts to reconcile electrical conductivity and reservoir porosity. Our
attempts to do this at Kakkonda, as presented above, do support the existence of hypersaline fluids in
reservoir rocks with less than or equal to 10 vol% porosity, and permeability of the order 10−16 m2.
It is important to understand sub-volcanic porosity–permeability relationships because they control
how much metal-bearing hypersaline fluids can be stored, and the extent to which they will flow into
the borehole with or without additional reservoir stimulation (e.g. [149,173]). Porosity–permeability
relationships also influence the response of the reservoir to fluid extraction, notably variations in
bottom-hole pressures. Loss of control of pressure in the well-bore can lead, for example, to blowouts of
the type that occurred at Wairakei Geothermal Field, New Zealand, in the 1960s [174]; a result of rapid,
uncontrolled decompression of hot geothermal fluids. Although there are fundamental differences
between the 1960s drilling operation at Wairakei and drilling to extract metal-bearing hypersaline fluids,
it is clear that careful monitoring and control of well-bore and reservoir pressures during drilling, use of
dense drilling muds, and careful design of well-bore cements and casings are fundamental to the safe
operation of in situ mining. Corrosion-resistant well-bore casing (or coating) materials are particularly
important, and an area of active research (e.g. [175–177]). These technical challenges require further
investigation by drilling engineers and materials scientists, which is beyond the scope of this article.
12. Prospectives 23
The economic potential of metal-bearing sub-volcanic hypersaline fluids hinges on a number of
unknown parameters that require further evaluation. These include the salinity and composition of
putative magmatic brines (including base metals and deleterious elements, such as arsenic and
mercury), their abundance and distribution beneath volcanoes, and their extractability. At present, we
can only speculate, based on hydrodynamic models and limited geological and geophysical data, on
what such lenses would look like and where they would lie. Uncertainty extends to a variety of
technical challenges associated with extraction and recovery. In these regards detailed reservoir
characterization and careful selection of drilling sites, alongside technical developments, are key to
future development of the concept. The limited data from a few deep magmatic geothermal
exploration sites shows how underrepresented hypersaline magmatic brines are in the general record
of geothermal fluids, largely as a result of the preferential targeting of fluids from the convective
hydrothermal regime for conventional geothermal power production.

Reservoir characterization requires integration of a variety of geophysical techniques including MT

(to identify conductive bodies), gravity (to constrain the distribution of low-density fluids versus
dense sulfides), geodesy (to monitor ground inflation and deflation in real time) and seismology (to
characterize porosity–depth relationships). Integration of different types of geophysical data is
necessary to resolve some of the potential ambiguities that come with a single method. Few active
volcanoes are sufficiently characterized using these different techniques, although there is growing
interest in applying joint geophysical inversions to address problems of sub-volcanic magma storage
at a number of active volcanoes: Laguna del Maule, Chile [178]; Mount St Helens, USA [179,180];
Uturuncu, Bolivia [181]; Santorini, Greece [182]. There is also an argument for scientific drilling
directly into sub-volcanic regions to better understand their magmatic architecture [183,184]. There
is considerable commonality in the objectives of these drilling proposals, e.g. melt distribution
and connectivity, fluid/melt chemistry, temperature and dynamics, and the key unknown parameters
around the proposed exploitation of sub-volcanic hypersaline fluid lenses. Direct drilling of
potential target volcanoes is undoubtedly the best method to reduce some of the uncertainties
inherent in the in situ extraction of metal-rich brines, not least to establish their presence in the
volumes predicted by our models. Drilling would also allow for recovery and chemical
characterization of any such fluids.
13. Conclusion
Sub-volcanic regions can trap hypersaline liquids that form on phase separation of ascending
supercritical magmatic fluids. There is permissive evidence for the existence of brine lenses from
geophysical studies of the roots of active or dormant volcanoes, for example from electrical
conductivity surveys. However, these geophysical signals have ambiguities, and further work is
required to resolve the contributions to elevated electrical conductivity due to brines, sulfide minerals
and silicate melts, and combinations thereof.
Hydrodynamic calculations show that Cu concentrations in dynamic, hypersaline liquid lenses
beneath active or dormant volcanoes may be as high as 7000 ppm at temperatures over 400°C. For our
reference model, after 10 000 years of degassing the hypersaline liquid lens contains 1.0 to 1.4 Mt of
dissolved Cu, depending on the availability of reduced sulfur. These expectations are consistent with
limited measurements of high Cu content in fluid inclusions from geothermal drill-core (table 1),
together with analyses of recovered hypersaline fluids at Kakkonda. In order to become potential
economic metal resources, hot, metal-rich brines would need to be extracted efficiently without
significant reservoir clogging or well-bore scaling. The extent to which such a resource could be
exploited depends on the technological challenges and costs of drilling into hot reservoir rocks and
recovering hypersaline liquid. If Cu can be extracted in solution form, Cu processing costs may be
greatly reduced. Similarly, the co-production of geothermal energy to power fluid extraction and
processing would confer additional economic and environmental benefit (e.g. [162]). We have focused
only on Cu; the presence of other metals in magmatic fluids, including Li, Zn, Pb, Au and Ag [4]
allows for significant co-recovery. We propose, on the basis of recent developments in our
understanding and geophysical imaging of crustal magmatic systems, that mining of sub-volcanic
metal-bearing brines may be a fruitful avenue for addressing the impending growth in demand for
certain metal resources that will arise during the transition to a low-carbon economy. The economic
potential of such brines hinges critically on the geophysical challenge of locating them and the 24
technological challenge of extracting them to the surface. Regardless of the economic outcomes, direct
drilling into hot, sub-volcanic systems is increasingly feasible technically and will provide invaluable
scientific insights into both magmatic and ore-forming processes.
Data accessibility. Model results are available from the following link: (https://doi.org/10.5061/dryad.0cfxpnw0w).
Further data are provided in the electronic supplementary material [185].
Authors’ contributions. J.B. devised the original brine lens mining concept in 2016. It was subsequently refined through
discussions with all other authors. B.T. performed the fluid inclusion metal content calculations, microthermometry
and analyses. A.A. performed the hydrodynamic simulations; O.M. and I.U. post-processed the results to include
copper. All authors helped to draft the final manuscript, gave final approval for publication and agree to be held
accountable for the work performed therein.
Competing interests. We declare we have no competing interests.
Funding. J.B. acknowledges funding through a Royal Society Research Professorship (RP\R1\201048); A.A. and O.M.
acknowledge funding from Russian Science Foundation under grant no. 16-17-10199. Samples from Larderello were

kindly provided by P. Fulignati (University of Pisa).
Acknowledgements. We thank A. Stinton (University of West Indies) for help in obtaining drill-core samples from
Montserrat MON-1 well, financial and technical support for which were provided by: Government of Montserrat;
UK Foreign, Commonwealth and Development Office, Natural Environment Research council; British Geological
Survey; Iceland Drilling and ISOR Iceland Geosurvey; and Baker Hughes. We thank B. Kunz and F. Jenner for
assistance with LA-ICPMS analyses, G. Pokrovski for help with the chalcopyrite solubility calculations, J. Stimac
for a primer on geothermal systems, and S. Gatehouse, S. Simmons, J. Hedenquist, J. Lowenstern, J. Mavrogenes
and N. White for reviews of the manuscript. Analytical work was supported through a research grant from BHP.
References
1. Schipper BW, Lin HC, Meloni MA, Wansleeben K, 10. Loucks RR. 2014 Distinctive composition of copper- Montserrat. Geochem. Geophys. Geosyst. 16,
Heijungs R, van der Voet E. 2018 Estimating ore-forming arc magmas. Austral J. Earth Sci. 61, 2797–2811. (doi:10.1002/2015GC005791)
global copper demand until 2100 with 5–16. (doi:10.1080/08120099.2013.865676) 18. Williams-Jones AE, Heinrich CA. 2005 100th
regression and stock dynamics. Resour. Conserv. 11. Chiaradia M, Caricchi L. 2017 Stochastic Anniversary special paper: vapour transport of
Recycl. 132, 28–36. (doi:10.1016/j.resconrec. modelling of deep magmatic controls on metals and the formation of magmatic-
2018.01.004) porphyry copper deposit endowment. Sci. Rep. hydrothermal ore deposits. Econ. Geol. 100,
2. Valenta RK, Kemp D, Owen JR, Corder GD, Lèbre 7, 1–11. (doi:10.1038/s41598-016-0028-x) 1287–1312. (doi:10.2113/gsecongeo.100.7.1287)
É. 2019 Re-thinking complex orebodies: 12. Chiaradia M. 2014 Copper enrichment in arc 19. Simmons SF, Brown KL. 2007 The flux of gold
consequences for the future world supply of magmas controlled by overriding plate and related metals through a volcanic arc,
copper. J. Cleaner Prod. 220, 816–826. (doi:10. thickness. Nat. Geosci. 7, 43–46. (doi:10.1038/ Taupo Volcanic Zone, New Zealand. Geology 35,
1016/j.jclepro.2019.02.146) ngeo2028) 1099–1102. (doi:10.1130/G24022A.1)
3. Grandell L, Lehtilä A, Kivinen M, Koljonen T, 13. Giggenbach WF. 1992 Magma degassing and 20. Candela PA, Holland HD. 1984 The partitioning
Kihlman S, Lauri LS. 2016 Role of critical metals mineral deposition in hydrothermal systems of copper and molybdenum between silicate
in the future markets of clean energy along convergent plate boundaries. Econ. Geol. melts and aqueous fluids. Geochim. Cosmochim.
technologies. Renew. Energy 95, 53–62. (doi:10. 87, 1927–1944. Acta 48, 373–380. (doi:10.1016/0016-
1016/j.renene.2016.03.102) 14. Hedenquist JW, Simmons SF, Giggenbach WF, 7037(84)90257-6)
4. Edmonds M, Mather TA, Liu EJ. 2018 A distinct Eldridge CS. 1993 White Island, New Zealand, 21. Simon AC, Pettke T, Candela PA, Piccoli PM,
metal fingerprint in arc volcanic emissions. Nat. volcanic-hydrothermal system represents the Heinrich CA. 2006 Copper partitioning in a
Geosci. 11, 790. (doi:10.1038/s41561-018-0214-5) geochemical environment of high-sulfidation Cu melt–vapour–brine–magnetite–pyrrhotite
5. Hedenquist JW, Lowenstern JB. 1994 The role of and Au ore deposition. Geology 21, 731–734. assemblage. Geochim. Cosmochim. Acta 70,
magmas in the formation of hydrothermal ore (doi:10.1130/0091-7613(1993)021<0731: 5583–5600. (doi:10.1016/j.gca.2006.
deposits. Nature 370, 519–527. (doi:10.1038/ WINZVH>2.3.CO;2) 08.045)
370519a0) 15. Taran YA, Hedenquist JW, Korzhinsky MA, 22. Frank MR, Simon AC, Pettke T, Candela PA,
6. Sillitoe RH. 2010 Porphyry copper systems. Econ. Tkachenko SI, Shmulovich KI. 1995 Piccoli PM. 2011 Gold and copper partitioning
Geol. 105, 3–41. (doi:10.2113/gsecongeo.105.1.3) Geochemistry of magmatic gases from Kudryavy in magmatic-hydrothermal systems at 800°C
7. Richards JP. 2015 The oxidation state, and sulfur volcano, Iturup, Kuril Islands. Geochim. and 100 MPa. Geochim. Cosmochim. Acta 75,
and Cu contents of arc magmas: implications for Cosmochim. Acta 59, 1749–1761. (doi:10.1016/ 2470–2482. (doi:10.1016/j.gca.2011.02.012)
metallogeny. Lithos 233, 27–45. (doi:10.1016/j. 0016-7037(95)00079-F) 23. Zajacz Z, Seo JH, Candela PA, Piccoli PM, Tossell
lithos.2014.12.011) 16. Shinohara H. 2008 Excess degassing from JA. 2011 The solubility of copper in high-
8. Hildreth W, Moorbath S. 1988 Crustal volcanoes and its role on eruptive and intrusive temperature magmatic vapours: a quest for the
contributions to arc magmatism in the Andes of activity. Rev. Geophys. 46, 4. (doi:10.1029/ significance of various chloride and sulfide
central Chile. Contrib. Mineral Petrol. 98, 2007RG000244) complexes. Geochim. Cosmochim. Acta 75,
455–489. (doi:10.1007/BF00372365) 17. Christopher TE, Blundy J, Cashman K, Cole P, 2811–2827. (doi:10.1016/j.gca.2011.02.029)
9. Annen C, Blundy JD, Sparks RSJ. 2006 The Edmonds M, Smith PJ, Sparks RSJ, Stinton A. 24. Kouzmanov K, Pokrovski GS. 2012 Hydrothermal
genesis of intermediate and silicic magmas in 2015 Crustal-scale degassing due to magma controls on metal distribution in porphyry Cu
deep crustal hot zones. J. Petrol. 47, 505–539. system destabilisation and magma-gas (-Mo-Au) systems. Econ. Geol. Spec. Publ. 16,
(doi:10.1093/petrology/egi084) decoupling at Soufrière Hills Volcano, 573–618. (doi:10.5382/SP.16.22)
25. Tattitch BC, Blundy JD. 2017 Cu-Mo partitioning 39. Blundy JD, Annen CJ. 2016 Crustal magmatic silicate melt and hydrothermal solutions: 25
between felsic melts and saline-aqueous fluids as systems from the perspective of heat transfer. I. Partition of NaCl-KCl. Geochim. Cosmochim. Ac.
a function of XNaCleq, fO2, and fS2. Am. Miner. Elements 12, 115–120. (doi:10.2113/gsele 53, 2617–2630. (doi:10.1016/0016-7037(89)
102, 1987–2006. (doi:10.2138/am-2017-5998) ments.12.2.115) 90133-6)
26. Dilles JH, Einaudi MT, Proffett JM, Barton MD. 40. Jackson MD, Blundy J, Sparks RS. 2018 Chemical 53. Tattitch B, Chelle-Michou C, Blundy J, Loucks
2000 Overview of the Yerington porphyry copper differentiation, cold storage and remobilization RR. 2021 Chemical feedbacks during magma
district: magmatic to nonmagmatic sources of of magma in the Earth’s crust. Nature 564, degassing control chlorine partitioning and
hydrothermal fluids: their flow paths and 405–409. (doi:10.1038/s41586-018-0746-2) metal extraction in volcanic arcs. Nat. Commun.
alteration effects on rocks and Cu-Mo-Fe-Au ores. 41. Annen C, Blundy JD, Leuthold J, Sparks RSJ. 12, 1774. (doi:10.1038/s41467-021-21887-w)
Soc. Econ. Geol. Guidebook Series 32, 55–66. 2015 Construction and evolution of igneous 54. Henley RW, McNabb A. 1978 Magmatic vapour
27. Cloos M. 2001 Bubbling magma chambers, bodies: towards an integrated perspective of plumes and ground-water interaction in
cupolas, and porphyry copper deposits. Int. crustal magmatism. Lithos 230, 206–221. porphyry copper emplacement. Econ. Geol. 73,
Geol. Rev. 43, 285–311. (doi:10.1080/ (doi:10.1016/j.lithos.2015.05.008) 1–20. (doi:10.2113/gsecongeo.73.1.1)
00206810109465015) 42. Mavrogenes J, Blundy J. 2017 Crustal 55. Driesner T, Heinrich CA. 2007 The system H2O–
28. Huber C, Townsend M, Degruyter W, Bachmann sequestration of magmatic sulfur dioxide. NaCl. Part I: correlation formulae for phase
O. 2019 Optimal depth of subvolcanic magma Geology 45, 211–214. (doi:10.1130/G38555.1) relations in temperature–pressure–composition
chamber growth controlled by volatiles and 43. Annen C. 2009 From plutons to magma space from 0 to 1000°C, 0 to 5000 bar, and 0 to

crust rheology. Nat. Geosci. 12, 762–768. chambers: thermal constraints on the 1 XNaCl. Geochim. Cosmochim. Acta 71,
(doi:10.1038/s41561-019-0415-6) accumulation of eruptible silicic magma in the 4880–4901. (doi:10.1016/j.gca.2006.01.033)
29. Cline JS, Bodnar RJ. 1991 Can economic upper crust. Earth Planet. Sci. Lett. 284, 56. Koděra P, Heinrich CA, Wälle M, Lexa J. 2014
porphyry copper mineralization be generated by 409–416. (doi:10.1016/j.epsl.2009.05.006) Magmatic salt melt and vapour: extreme fluids
a typical calc-alkaline melt? J. Geophys. Res. 44. Barra F, Alcota H, Rivera S, Valencia V, Munizaga forming porphyry gold deposits in shallow
Solid Earth 96, 8113–8126. (doi:10.1029/ F, Maksaev V. 2013 Timing and formation of subvolcanic settings. Geology 42, 495–498.
91JB00053) porphyry Cu–Mo mineralization in the (doi:10.1130/G35270.1)
30. Cashman KV, Sparks RSJ, Blundy J. 2017 Chuquicamata district, northern Chile: new 57. Webster JD, Vetere F, Botcharnikov RE, Goldoff
Vertically extensive and unstable magmatic constraints from the Toki cluster. Mineral B, McBirney A, Doherty AL. 2015 Experimental
systems: a unifying view of igneous processes. Deposita 48, 629–651. (doi:10.1007/s00126- and modeled chlorine solubilities in
Science 355, 1280. (doi:10.1126/science.aag3055) 012-0452-1) aluminosilicate melts at 1 to 7000 bars and 700
31. Sparks RSJ, Annen C, Blundy JD, Cashman KV, 45. Chelle-Michou C, Chiaradia M, Ovtcharova M, to 1250°C: applications to magmas of Augustine
Rust AC, Jackson MD. 2019 Formation and Ulianov A, Wotzlaw JF. 2014 Zircon Volcano, Alaska. Am. Miner. 100, 522–535.
dynamics of magma reservoirs. Phil. petrochronology reveals the temporal link (doi:10.2138/am-2015-5014)
Trans. R. Soc. A 377, 20180019. (doi:10.1098/ between porphyry systems and the magmatic 58. Fulignati P. 2017 Fluid inclusion evidence on the
rsta.2018.0019) evolution of their hidden plutonic roots (the direct exsolution of magmatic brines from a
32. Hildreth W. In press. Comparative rhyolite Eocene Coroccohuayco deposit. Peru). Lithos 198, granite intrusion beneath the eastern sector of
systems: inferences from vent patterns and 129–140. (doi:10.1016/j.lithos.2014.03.017) Larderello geothermal field (Italy). Periodico
eruptive episodicities: Eastern California and 46. Tapster S, Condon DJ, Naden J, Noble SR, Minerral 86, 201–211.
Laguna del Maule. J. Geophys. Res. Solid Earth Petterson MG, Roberts NMW, Saunders AD, 59. Fulignati P. 2018 Hydrothermal fluid evolution
(doi:10.1029/2020JB020879) Smith DJ. 2016 Rapid thermal rejuvenation of in the ‘Botro ai Marmi’ quartz-monzonitic
33. Lesne P, Kohn S, Blundy J, Witham F, high-crystallinity magma linked to porphyry intrusion (Campiglia Marittima, Tuscany, Italy:
Botcharnikov R, Behrens H. 2011 Experimental copper deposit formation; evidence from the evidence from a fluid inclusion investigation.
simulation of closed-system degassing in the Koloula Porphyry Prospect, Solomon Islands. Mineral Mag. 82, 1169–1185. (doi:10.1180/
system basalt-H2O-CO2-S-Cl. J. Petrol. 52, Earth Planet. Sci. Lett. 442, 206–217. (doi:10. mgm.2018.116)
1737–1762. (doi:10.1093/petrology/egr027) 1016/j.epsl.2016.02.046) 60. Duan Z, Møller N, Weare JH. 1995 Equation of
34. Burgisser A, Alletti M, Scaillet B. 2015 47. Buret Y, von Quadt A, Heinrich C, Selby D, Wälle state for the NaCl-H2O-CO2 system: prediction of
Simulating the behavior of volatiles belonging M, Peytcheva I. 2016 From a long-lived upper- phase equilibria and volumetric properties.
to the C–O–H–S system in silicate melts under crustal magma chamber to rapid porphyry Geochim. Cosmochim. Acta 59, 2869–2882.
magmatic conditions with the software D- copper emplacement: reading the geochemistry (doi:10.1016/0016-7037(95)00182-4)
Compress. Comput. Geosci. 79, 1–14. (doi:10. of zircon crystals at Bajo de la Alumbrera (NW 61. Gottschalk M. 2007 Equations of state for
1016/j.cageo.2015.03.002) Argentina). Earth Planet. Sci. Lett. 450, complex fluids. Rev. Mineral. Geochem. 65,
35. Bergantz GW. 1989 Underplating and partial 120–131. (doi:10.1016/j.epsl.2016.06.017) 49–97. (doi:10.2138/rmg.2007.65.3)
melting: implications for melt generation and 48. Schöpa A, Annen C, Dilles JH, Sparks RSJ, 62. Audétat A, Pettke T, Heinrich CA, Bodnar RJ.
extraction. Science 245, 1093–1095. (doi:10. Blundy JD. 2017 Magma emplacement rates 2008 The composition of magmatic-
1126/science.245.4922.1093) and porphyry copper deposits: thermal hydrothermal fluids in barren and mineralized
36. Annen C, Sparks RS. 2002 Effects of repetitive modeling of the Yerington Batholith, Nevada. intrusions. Econ. Geol. 103, 877–908. (doi:10.
emplacement of basaltic intrusions on thermal Econ. Geol. 112, 1653–1672. (doi:10.5382/ 2113/gsecongeo.103.5.877)
evolution and melt generation in the crust. econgeo.2017.4525) 63. Landtwing MR, Pettke T, Halter WE, Heinrich
Earth Planet. Sci. Lett. 203, 937–955. (doi:10. 49. Degruyter W, Parmigiani A, Huber C, Bachmann CA, Redmond PB, Einaudi MT, Kunze K. 2005
1016/S0012-821X(02)00929-9) O. 2019 How do volatiles escape their shallow Copper deposition during quartz dissolution by
37. Solano JM, Jackson MD, Sparks RS, Blundy JD, magmatic hearth? Phil. Trans. R. Soc. A 377, cooling magmatic–hydrothermal fluids: the
Annen C. 2012 Melt segregation in deep crustal 20180017. (doi:10.1098/rsta.2018.0017) Bingham porphyry. Earth Planet. Sci. Lett. 235,
hot zones: a mechanism for chemical 50. Caricchi L, Sheldrake TE, Blundy JD. 2018 229–243. (doi:10.1016/j.epsl.2005.02.046)
differentiation, crustal assimilation and the Modulation of magmatic processes by CO2 64. Klemm LM, Pettke T, Heinrich CA. 2008 Fluid
formation of evolved magmas. J. Petrol. 53, flushing. Earth Planet. Sci. Lett. 491, 160–171. and source magma evolution of the Questa
1999–2026. (doi:10.1093/petrology/egs041) (doi:10.1016/j.epsl.2018.03.042) porphyry Mo deposit, New Mexico, USA. Mineral
38. Karakas O, Dufek J. 2015 Melt evolution and 51. Aiuppa A, Baker DR, Webster JD. 2009 Halogens Deposita 43, 533. (doi:10.1007/s00126-008-
residence in extending crust: thermal modeling in volcanic systems. Chem. Geol. 263, 1–18. 0181-7)
of the crust and crustal magmas. Earth Planet. (doi:10.1016/j.chemgeo.2008.10.005) 65. Seo JH, Guillong M, Heinrich CA. 2012
Sci. Lett. 425, 131–144. (doi:10.1016/j.epsl. 52. Shinohara H, Iiyama JT, Matsuo S. 1989 Separation of molybdenum and copper in
2015.06.001) Partition of chlorine compounds between porphyry deposits: the roles of sulfur, redox,
and pH in ore mineral deposition at Bingham plutonic-hydrothermal system around Montserrat from 2000 to 2010. Geol. Soc. Lond. 26
Canyon. Econ. Geol. 107, 333–356. (doi:10. Quaternary granite in the Kakkonda geothermal Memoirs 39, 1–40. (doi:10.1144/m39.1)
2113/econgeo.107.2.333) system, Japan. Geothermics 27, 663–690. 92. Ryan GA, Peacock JR, Shalev E, Rugis J. 2013
66. Roedder E. 1984 Reviews in mineralogy, volume (doi:10.1016/S0375-6505(98)00039-X) Montserrat geothermal system: a 3D conceptual
12: fluid inclusions. Mineralogical Society of 80. Uchida T, Akaku K, Yanagisawa N, Kamenesono model. Geophys. Res. Lett. 40, 2038–2043.
America. H, Sasaki M, Miyazaki S, Doi U. 1998 Deep (doi:10.1002/grl.50489)
67. Fournier RO. 1987 Conceptual models of brine geothermal resources survey project in the 93. Barelli A, Cei M, Lovari F, Romagnoli P. 2010
evolution in magmatic-hydrothermal systems. Kakkonda geothermal field. Energy Sources 20, Numerical modeling for the Larderello-Travale
US Geol. Surv. Prof. Pap. 1350, 1487–1506. 763–777. (doi:10.1080/00908319808970096) geothermal system, Italy. In Proc. World
68. Weis P, Driesner T, Heinrich CA. 2012 Porphyry- 81. Sasaki M, Sasada M, Fujimoto K, Muramatsu Y, Geothermal Congress.
copper ore shells form at stable pressure- Komatsu R, Sawaki T. 1995 History of post- 94. Carella M, Fulignati P, Musumeci G, Sbrana A.
temperature fronts within dynamic fluid intrusive hydrothermal systems indicated by 2000 Metamorphic consequences of Neogene
plumes. Science 338, 1613–1616. (doi:10.1126/ fluid inclusions occuring in the young granitic thermal anomaly in the northern Apennines
science.1225009) rocks at the Kakkonda and Nyuto geothermal (Radicondoli-Travale area, Larderello geothermal
69. Weis P. 2015 The dynamic interplay between systems. Shigen-Chishitsu. 45, 303–312. field–Italy). Geodinamica Acta 13, 345–366.
saline fluid flow and rock permeability in 82. Sasaki M et al. 1998 Characterization of a (doi:10.1080/09853111.2000.11105379)
magmatic-hydrothermal systems. Geofluids 15, magmatic/meteoric transition zone at the 95. Brophy P, Poux B, Suemnicht G, Hirtz P, Ryan G.

350–371. (doi:10.1111/gfl.12100) Kakkonda geothermal system, northeast Japan. 2014 Preliminary results of deep geothermal
70. Afanasyev A, Blundy J, Melnik O, Sparks S. 2018 In Proc. of the 9th Int. Symp. on Water-Rock drilling and testing on the island of Montserrat.
Formation of magmatic brine lenses via Interaction, Taupo, New Zealand (eds Arehart, In Proc. of the 39th Workshop on Geothermal
focussed fluid-flow beneath volcanoes. Earth Hulston), pp. 483–486. Rotterdam, The Reservoir Engineering, Stanford, CA (Vol. 2426,
Planet. Sci. Lett. 486, 119–128. (doi:10.1016/j. Netherlands: Balkema. p. 11).
epsl.2018.01.013) 83. Sasaki M, Fujimoto K, Sawaki T, Tsukamoto H, 96. Ryan GA, Shalev E. 2014 Seismic velocity/
71. Stimac J, Goff F, Goff CJ. 2015 Intrusion-related Kato O, Komatsu R, Doi N, Sasada M. 2003 temperature correlations and a possible new
geothermal systems. In The encyclopedia of Petrographic features of a high-temperature geothermometer: insights from exploration of a
volcanoes (eds H Sigurdsson), pp. 799–822. granite just newly solidified magma at the high-temperature geothermal system on
Cambridge, MA: Academic Press. Kakkonda geothermal field, Japan. J. Volcanol. Montserrat, West Indies. Energies 7, 6689–6720.
72. Fournier RO. 1999 Hydrothermal processes Geotherm. Res. 121, 247–269. (doi:10.1016/ (doi:10.3390/en7106689)
related to movement of fluid from plastic into S0377-0273(02)00428-6) 97. Bodnar RJ, Sterner SM, Hall DL. 1989 SALTY: a
brittle rock in the magmatic-epithermal 84. Fujimoto K, Sasaki M, Sawaki T, Yanagisawa N. FORTRAN program to calculate compositions of
environment. Econ. Geol. 94, 1193–1211. 2000 Structure and evolution process of the fluid inclusions in the system NaCl–KCl–H2O.
(doi:10.2113/gsecongeo.94.8.1193) Kakkonda geothermal system from the Comput. Geosci. 15, 19–14. (doi:10.1016/0098-
73. Dobson P, Asanuma H, Huenges E, Poletto F, viewpoint of water-rock interaction. Geologic 3004(89)90053-8)
Reinsch T, Sanjuan B. 2017 Supercritical Survey of Japan, Report No. 284, pp. 105–116. 98. Gunther D, Audétat A, Frischknecht R, Heinrich
geothermal systems—a review of past studies 85. Ulrich T, Gunther D, Heinrich CA. 1999 Gold CA. 1998 Quantitative analysis of major, minor
and ongoing research activities. In 42nd concentrations of magmatic brines and the and trace elements in fluid inclusions using
Workshop on Geothermal Reservoir Engineering, metal budget of porphyry copper deposits. laser ablation inductively coupled plasma mass
Stanford University, Stanford, California, Nature 399, 676–679. (doi:10.1038/21406) spectrometry. J. Anal. At. Spectrom. 13,
February 13–15, 2017, SGP-TR-212 86. Yanagisawa N, Fujimoto K, Hishi Y. 2000 Sulfide 263–270. (doi:10.1039/A707372K)
74. Agostinetti NP, Licciardi A, Piccinini D, Mazzarini scaling of deep-geothermal well at Kakkonda 99. Heinrich CA et al. 2003 Quantitative multi-
F, Musumeci G, Saccorotti G, Chiarabba C. 2017 geothermal field in Japan. Proc. World element analysis of minerals, fluid and melt
Discovering geothermal supercritical fluids: a Geothermal Congress, pp. 1969–1974. inclusions by laser-ablation inductively-coupled-
new frontier for seismic exploration. Sci. Rep. 7, 87. Kasai K, Hishi Y, Fukuda D, Kato O, Doi N, Akaku plasma mass-spectrometry. Geochim.
14592. (doi:10.1038/s41598-017-15118-w) K, Ominato T, Tosha T. 2000 The fluid Cosmochim. Acta 67, 3473–3497. (doi:10.1016/
75. Cathelineau M, Marignac C, Boiron MC, Gianelli geochemistry and reservoir model for the S0016-7037(03)00084-X)
G, Puxeddu M. 1994 Evidence for Li-rich brines Kakkonda geothermal system, obtained by 100. Mutchler S, Fedele L, Bodnar R. 2008 Analysis
and early magmatic fluid-rock interaction in the NEDO’s deep-seated geothermal reservoir management system (AMS) for reduction of
Larderello geothermal system. Geochim. survey, Japan. In Proc. World Geothermal laser ablation ICPMS data. Laser-ablation-ICPMS
Cosmochim. Acta 58, 1083–1099. (doi:10.1016/ Congress, Kyushu, Tohoku, Japan. in the earth sciences: current practices and
0016-7037(94)90574-6) 88. Moore JN, Gunderson RP. 1995 Fluid inclusion outstanding issues. Mineral. Assoc. Canada Short
76. Batini F, Brogi A, Lazzarotto A, Liotta D, Pandeli and isotopic systematics of an evolving Course Series 40, 318–327.
E. 2003 Geological features of Larderello-Travale magmatic-hydrothermal system. Geochim. 101. Bodnar RJ. 2003 Introduction to fluid inclusions.
and Mt. Amiata geothermal areas (southern Cosmochim. Acta 59, 3887–3907. (doi:10.1016/ In Fluid inclusions: analysis and interpretation
Tuscany, Italy). Episodes 26, 239–244. (doi:10. 0016-7037(95)00289-C) (eds I Samson, A Anderson, D Marshall),
18814/epiiugs/2003/v26i3/015) 89. Muramatsu Y, Komatsu R. 1999 pp. 1–8. Mineralogical Association of Canada,
77. Kasai K, Sakagawa Y, Komatsu R, Sasaki M, Microthermometric evidence for the formation Short Course 32.
Akaku K, Uchida T. 1998 The origin of of Ca-rich hypersaline brine and CO2–rich fluid 102. Hattori KH, Keith JD. 2001 Contribution of mafic
hypersaline liquid in the Quaternary Kakkonda in the Mori geothermal reservoir. Japan. Res. melt to porphyry copper mineralization:
granite, sampled from well WD-1a, Kakkonda Geol. 49, 27–37. (doi:10.1111/j.1751-3928. evidence from Mount Pinatubo, Philippines, and
geothermal system, Japan. Geothermics 27, 1999.tb00029.x) Bingham Canyon, Utah, USA. Mineral Deposita
631–645. (doi:10.1016/S0375-6505(98)00037-6) 90. Dini A, Gianelli G, Puxeddu M, Ruggieri G. 2005 36, 799–806. (doi:10.1007/s001260100209)
78. Kasai K, Sakagawa Y, Miyazaki S, Akaku K, Origin and evolution of Pliocene–Pleistocene 103. Blundy J, Mavrogenes J, Tattitch B, Sparks S,
Uchida T. 1998 Supersaline and metal-rich brine granites from the Larderello geothermal Gilmer A. 2015 Generation of porphyry copper
obtained from the Quaternary Kakkonda granite field (Tuscan Magmatic Province, Italy). deposits by gas–brine reaction in volcanic arcs.
by NEDO WD-1a in the Kakkonda geothermal Lithos 81, 1–31. (doi:10.1016/j.lithos.2004. Nat. Geosci. 8, 235. (doi:10.1038/ngeo2351)
field, Japan. Mineral Deposita 33, 298–301. 09.002) 104. Gustafson LB, Hunt JP. 1975 The porphyry
(doi:10.1007/s001260050148) 91. Wadge G, Voight B, Sparks RSJ, Cole PD, copper deposit at El Salvador, Chile.
79. Doi N, Kato O, Ikeuchi K, Komatsu R, Miyazaki Loughlin SC, Robertson REA. 2014 An overview Econ. Geol. 70, 857–912. (doi:10.2113/
SI, Akaku K, Uchida T. 1998 Genesis of the of the eruption of Soufriere Hills Volcano, gsecongeo.70.5.857)
105. Sasaki M et al. 2003 Geochemical features of Andes. Geochem. Geophys. Geosyst. 5, Q02002. 2015 Structure of the Tongariro volcanic system: 27
vein anhydrite from the Kakkonda geothermal (doi:10.1029/2003GC000610) insights from magnetotelluric imaging. Earth
system, Northeast Japan. Res. Geol. 53, 120. Biggs J, Ebmeier SK, Aspinall WP, Lu Z, Planet. Sci. Lett. 432, 115–125. (doi:10.1016/j.
127–142. (doi:10.1111/j.1751-3928.2003. Pritchard ME, Sparks RSJ, Mather TA. 2014 epsl.2015.10.003)
tb00164.x) Global link between deformation and volcanic 132. Hill GJ, Caldwell TG, Heise W, Chertkoff DG,
106. Wallace PJ, Edmonds M. 2011 The sulfur budget eruption quantified by satellite imagery. Nat. Bibby HM, Burgess MK, Cull JP, Cas RA. 2009
in magmas: evidence from melt inclusions, Commun. 5, 3471. (doi:10.1038/ncomms4471) Distribution of melt beneath Mount St Helens
submarine glasses, and volcanic gas emissions. 121. Gottsmann J, Blundy J, Henderson S, Pritchard and Mount Adams inferred from
Rev. Mineral Geochem. 73, 215–246. (doi:10. M, Sparks S. 2017 Thermomechanical modeling magnetotelluric data. Nat. Geosci. 2, 785.
2138/rmg.2011.73.8) of the Altiplano-Puna deformation anomaly: (doi:10.1038/ngeo661)
107. Stasiuk MV, Barclay J, Carroll MR, Jaupart C, multiparameter insights into magma mush 133. Bedrosian PA, Peacock JR, Bowles-Martinez E,
Ratté JC, Sparks RSJ, Tait SR. 1996 Degassing reorganization. Geosphere 13, 1042–1065. Schultz A, Hill GJ. 2018 Crustal inheritance and
during magma ascent in the Mule Creek vent (doi:10.1130/ges01420.1) a top-down control on arc magmatism at
(USA). Bull. Volcanol. 58, 117–130. (doi:10. 122. Aizawa K et al. 2005 Hydrothermal system Mount St. Helens. Nat. Geosci. 11, 865–870.
1007/s004450050130) beneath Mt. Fuji volcano inferred from (doi:10.1038/s41561-018-0217-2)
108. Goto Y, Nakada S, Kurokawa M. 2008 magnetotellurics and electric self-potential. 134. Shuey RT. 2012 Developments in economic
Character and origin of lithofacies in the conduit Earth Planet. Sci. Lett. 235, 343–355. (doi:10. geology, volume 4: semiconducting ore minerals.

of Unzen volcano, Japan. J. Volcanol. Geotherm. 1016/j.epsl.2005.03.023) Amsterdam, The Netherlands: Elsevier.
Res. 175, 45–59. (doi:10.1016/j.jvolgeores.2008. 123. Yamaya Y, Alanis PKB, Takeuchi A, Cordon JM, 135. Uchida T, Ogawa Y, Takakura S, Mitsuhata Y.
03.041) Mogi T, Hashimoto T, Sasai Y, Nagao T. 2013 A 2000 Geoelectrical investigation of the Kakkonda
109. Ikeda R, Kajiwara T, Omura K, Hickman S. 2008 large hydrothermal reservoir beneath Taal geothermal field, northern Japan. In Proc. World
Physical rock properties in and around a conduit Volcano (Philippines) revealed by magnetotelluric Geothemal Congress, Tohoku-Kyushu, Japan,
zone by well-logging in the Unzen Scientific resistivity survey: 2D resistivity modeling. Bull. pp. 1893–1898.
Drilling Project, Japan. J. Volcanol. Geotherm. Volc. 75, 729. (doi:10.1007/s00445-013-0729-y) 136. Sinmyo R, Keppler H. 2017 Electrical
Res. 175, 13–19. (doi:10.1016/j.jvolgeores.2008. 124. Nurhasan OY, Ujihara N, Tank SB, Honkura Y, conductivity of NaCl-bearing aqueous fluids to
03.036) Onizawa SY, Mori T, Makino M. 2006 Two 600°C and 1 GPa. Contrib. Mineral. Petrol. 172,
110. Kiddle EJ, Edwards BR, Loughlin SC, Petterson electrical conductors beneath Kusatsu-Shirane 4. (doi:10.1007/s00410-016-1323-z)
M, Sparks RSJ, Voight B. 2010 Crustal structure volcano, Japan, imaged by 137. Yardley BW. 2005 Metal concentrations in
beneath Montserrat, Lesser Antilles, constrained audiomagnetotellurics, and their implications for crustal fluids and their relationship to ore
by xenoliths, seismic velocity structure and the hydrothermal system. Earth, Planets Space formation. Econ. Geol. 100, 613–632. (doi:10.
petrology. Geophys. Res. Lett. 37, L00E11. 58, 1053–1059. (doi:10.1186/BF03352610) 2113/gsecongeo.100.4.613)
(doi:10.1029/2009GL042145) 125. Comeau MJ, Unsworth MJ, Cordell D. 2016 New 138. Redmond PB, Einaudi MT, Inan EE, Landtwing
111. Fujimoto K, Takahashi M, Doi N, Kato O. 2000 constraints on the magma distribution and MR, Heinrich CA. 2004 Copper deposition by
High permeability of Quaternary granites in the composition beneath Volcán Uturuncu and the fluid cooling in intrusion-centered systems: new
Kakkonda geothermal area, northeast Japan. In southern Bolivian Altiplano from insights from the Bingham porphyry ore
World Geothermal Congress, Kyushu–Tohoku, magnetotelluric data. Geosphere 12, deposit, Utah. Geology 32, 217–220. (doi:10.
Japan, pp. 1139–1144. 1391–1421. (doi:10.1130/GES01277.1) 1130/G19986.1)
112. McGuinness M, White S, Young R, Ishizaki H, 126. Srigutomo W, Kagiyama T, Kanda W, Munekane 139. Arif J, Baker T. 2004 Gold paragenesis and
Ikeuchi K, Yoshida Y. 1995 A model of the H, Hashimoto T, Tanaka Y, Utada H, Utsugi M. chemistry at Batu Hijau, Indonesia: implications
Kakkonda geothermal reservoir. Geothermics 24, 2008 Resistivity structure of Unzen Volcano for gold-rich porphyry copper deposits. Mineral
1–48. (doi:10.1016/0375-6505(94)00020-D) derived from time domain electromagnetic Deposita 39, 523–535. (doi:10.1007/s00126-
113. Sakagawa Y, Aoyama K, Ikeuchi K, Takahashi M, (TDEM) survey. J. Volcanol. Geotherm. Res. 175, 004-0433-0)
Kato O, Doi N, Tosha T, Ominato T, Koide K. 231–240. (doi:10.1016/j.jvolgeores.2008.03.033) 140. Gruen G, Weis P, Driesner T, Heinrich CA, de
2000 Natural state simulation of the Kakkonda 127. Komori S, Kagiyama T, Utsugi M, Inoue H, Ronde CE. 2014 Hydrodynamic modeling of
geothermal field, Japan. In Proc. of World Azuhata I. 2013 Two-dimensional resistivity magmatic–hydrothermal activity at submarine
Geothermal Congress, Kyushu–Tohuku, Japan, structure of Unzen Volcano revealed by AMT arc volcanoes, with implications for ore
pp. 2839–2844. and MT surveys. Earth Planets Space 65, formation. Earth Planet. Sci. Lett. 404, 307–318.
114. Cloke PL, Kesler SE. 1979 The halite trend in 759–766. (doi:10.5047/eps.2012.10.005) (doi:10.1016/j.epsl.2014.07.041)
hydrothermal solutions. Econ. Geol. 74, 128. Triahadini A, Aizawa K, Teguri Y, Koyama T, 141. Palacios J-L, Abadias A, Valero A, Valero A,
1823–1831. (doi:10.2113/gsecongeo.74.8.1823) Tsukamoto K, Muramatsu D, Chiba K, Uyeshima Reuter M. 2019 The energy needed to
115. Lecumberri-Sanchez P, Steele-MacInnis M, Weis M. 2019 Magnetotelluric transect of Unzen concentrate minerals from common rocks: the
P, Driesner T, Bodnar RJ. 2015 Salt precipitation graben, Japan: conductors associated with case of copper ore. Energy 181, 494–503.
in magmatic-hydrothermal systems associated normal faults. Earth Planets Space 71, 28. (doi:10.1016/j.energy.2019.05.145)
with upper crustal plutons. Geology 43, (doi:10.1186/s40623-019-1004-z) 142. Northey S, Haque N, Mudd G. 2013 Using
1063–1066. (doi:10.1130/g37163.1) 129. Bertrand EA, Caldwell TG, Bannister S, sustainability reporting to assess the
116. Hesshaus A, Houben G, Kringel R. 2013 Halite Soengkono S, Bennie SL, Hill GJ, Heise W. 2015 environmental footprint of copper mining.
clogging in a deep geothermal well– Using array MT data to image the crustal J. Cleaner Prod. 40, 118–128. (doi:10.1016/j.
geochemical and isotopic characterisation of salt resistivity structure of the southeastern Taupo jclepro.2012.09.027)
origin. Phys. Chem. Earth 64, 127–139. (doi:10. Volcanic Zone, New Zealand. J. Volcanol. 143. Friðleifsson GÓ, Elders WA, Zierenberg RA,
1016/j.pce.2013.06.002) Geotherm. Res. 305, 63–75. (doi:10.1016/j. Stefánsson A, Fowler AP, Weisenberger TB,
117. Nitschke F, Held S, Himmelsbach T, Kohl T. 2017 THC jvolgeores.2015.09.020) Harðarson BS, Mesfin KG. 2017 The Iceland
simulation of halite scaling in deep geothermal 130. Heise W, Caldwell TG, Bertrand EA, Hill GJ, Deep Drilling Project 4.5 km deep well, IDDP-2,
single well production. Geothermics 65, 234–243. Bennie SL, Palmer NG. 2016 Imaging the deep in the seawater-recharged Reykjanes
(doi:10.1016/j.geothermics.2016.09.009) source of the Rotorua and Waimangu geothermal field in SW Iceland has successfully
118. Magee C et al. 2018 Magma plumbing systems: geothermal fields, Taupo Volcanic Zone, New reached its supercritical target. Sci. Drilling 23,
a geophysical perspective. J. Petrol. 59, Zealand. J. Volcanol. Geotherm. Res. 314, 1–12. (doi:10.5194/sd-23-1-2017)
1217–1251. (doi:10.1093/petrology/egy064) 39–48. (doi:10.1016/j.jvolgeores.2015.10.017) 144. Bertani R et al. 2018 The first results of the
119. Pritchard ME, Simons M. 2004 An InSAR-based 131. Hill GJ, Bibby HM, Ogawa Y, Wallin EL, Bennie DESCRAMBLE project. In Proc, 43rd Workshop on
survey of volcanic deformation in the central SL, Caldwell TG, Keys H, Bertrand EA, Heise W. Geothermal Reservoir Engineering, Stanford
University, Stanford, California, February 12–14, proposed cementation process. Geothermics 11, geothermal reservoirs by thermal and 28
2018, SGP-TR-213. 239–258. (doi:10.1016/0375-6505(82)90031-1) mechanical stimulation: the case of Krafla
145. Elders WA, Friðleifsson GÓ, Albertsson A. 2014 159. Duyvesteyn WP. 1992 Recovery of base metals volcano, Iceland. J. Volcanol. Geotherm.
Drilling into magma and the implications of the from geothermal brines. Geothermics 21, Res. 391, 106351. (doi:10.1016/j.jvolgeores.
Iceland deep drilling project (IDDP) for high- 773–799. (doi:10.1016/0375-6505(92)90030-D) 2018.04.008)
temperature geothermal systems worldwide. 160. Gallup DL. 1998 Geochemistry of geothermal 174. Bolton RS, Hunt TM, King TR, Thompson GE.
Geothermics 49, 111–118. (doi:10.1016/j. fluids and well scales, and potential for mineral 2009 Dramatic incidents during drilling at
geothermics.2013.05.001) recovery. Ore Geol. Rev. 12, 225–236. (doi:10. Wairakei Geothermal field, New Zealand.
146. Scott S, Driesner T, Weis P. 2015 Geologic 1016/S0169-1368(98)00004-3) Geothermics 38, 40–47. (doi:10.1016/j.
controls on supercritical geothermal resources 161. Clutter TJ. 2000 Mining economic benefits from geothermics.2008.12.002)
above magmatic intrusions. Nat. Commun. 6, geothermal brine. Geo-Heat Center Quart. Bull. 175. Naganawa S, Tsuchiya N, Okabe T, Kajiwara T,
7837. (doi:10.1038/ncomms8837) 21, 1–3. Shimada K, Yanagisawa N. 2017 Innovative
147. Tsuchiya N, Yamada R, Uno M. 2016 Supercritical 162. Bourcier WL, Lin M, Nix G. 2005 Recovery of drilling technology for supercritical geothermal
geothermal reservoir revealed by a granite– minerals and metals from geothermal fluids (No. resources development. In Proc. 42nd Workshop
porphyry system. Geothermics 63, 182–194. UCRL-CONF-215135). Livermore, CA: Lawrence on Geothermal Reservoir Engineering Stanford
(doi:10.1016/j.geothermics.2015.12.011) Livermore National Laboratory. University, Stanford, CA, SGP-TR-212.
148. Muraoka H, Asanuma H, Tsuchiya N, Ito T, Mogi 163. Harrison S. 2014 Technologies for extracting 176. Nogara J, Zarrouk SJ. 2018 Corrosion in

T, Ito H. 2014 The Japan beyond-brittle project. valuable metals and compounds from geothermal environment Part 2: metals and
Sci. Drilling 17, 51–59. (doi:10.5194/sd-17-51- geothermal fluids (No. DOE-SIM-001). Simbol alloys. Renew. Sustain. Energy Rev. 82,
2014) Materials. 1347–1363. (doi:10.1016/j.rser.2017.06.091)
149. Watanabe N, Numakura T, Sakaguchi K, Saishu 164. Neupane G, Wendt DS. 2017 Assessment of 177. Karlsdottir SN, Geambazu LE, Csaki I,
H, Okamoto A, Ingebritsen SE, Tsuchiya N. 2017 mineral resources in geothermal brines in the Thorhallsson AI, Stefanoiu R, Magnus F, Cotrut
Potentially exploitable supercritical geothermal US. In Proc 42nd Workshop on Geothermal C. 2019 Phase evolution and microstructure
resources in the ductile crust. Nat. Geosci. 10, Reservoir Engineering, Stanford University, analysis of CoCrFeNiMo high-entropy alloy for
140. (doi:10.1038/ngeo2879) Stanford, CA, pp. 13–15. electro-spark-deposited coatings for geothermal
150. Ogawa Y, Ichiki M, Kanda W, Mishina M, 165. Yardley BW, Bodnar RJ. 2014 Fluids in the environment. Coatings 9, 406. (doi:10.3390/
Asamori K. 2014 Three-dimensional continental crust. Geochem. Perspect. 3, 1–2. coatings9060406)
magnetotelluric imaging of crustal fluids and (doi:10.7185/geochempersp.3.1) 178. Singer BS et al. 2014 Dynamics of a large,
seismicity around Naruko volcano, NE Japan. 166. Reyes AG, Trompetter WJ, Britten K, Searle J. restless, rhyolitic magma system at Laguna del
Earth, Planets Space. 66, 1–3. (doi:10.1186/ 2002 Mineral deposits in the Rotokawa Maule, southern Andes, Chile. GSA Today 24,
1880-5981-66-1) geothermal pipelines, New Zealand. J. Volcanol. 4–10. (doi:10.1130/GSATG216A.1)
151. Kimura K. 1953 On the utilization of hot springs Geotherm. Res. 119, 215–239. (doi:10.1016/ 179. Kiser E, Palomeras I, Levander A, Zelt C, Harder
in Japan. Proc. 7th Pacific Sci. Congress, New S0377-0273(02)00355-4) S, Schmandt B, Hansen S, Creager K, Ulberg C.
Zealand 2, 500. 167. Harrar JE, Otto CH, Deutscher SB, Ryon RW, 2016 Magma reservoirs from the upper crust to
152. Kennedy AM. 1961 The recovery of lithium and Tardiff GE. 1979 Studies of brine chemistry, the Moho inferred from high-resolution Vp and
other minerals from geothermal water at precipitation of solids, and scale formation at Vs models beneath Mount St. Helens,
Wairakei. In Proc. UN Conf. on New Sources of the Salton Sea geothermal field (No. UCRL- Washington State, USA. Geology 44, 411–414.
Energy, Rome, Italy, pp. 21–31. 52640). Lawrence Livermore Nat Lab. (doi:10.1130/G37591.1)
153. White DE, Anderson ET, Grubbs DK. 1963 168. Gallup. 1992 Recovery of precious metals from 180. Ulberg CW et al. 2020 Local source Vp and Vs
Geothermal brine well: mile-deep drill hole may aqueous media. US Patent 5,082,492. tomography in the Mount St. Helens region
tap ore-bearing magmatic water and rocks 169. Baba A, Demir MM, Koç GA, Tuğcu C. 2015 with the iMUSH broadband array. Geochem.
undergoing metamorphism. Science 139, Hydrogeological properties of hyper-saline Geophys. Geosyst. 21, 2019GC008888. (doi:10.
919–922. (doi:10.1126/science.139. geothermal brine and application of inhibiting 1029/2019GC008888)
3558.919) siliceous scale via pH modification. Geothermics 181. Pritchard ME et al. 2018 Synthesis:
154. Skinner BJ, White DE, Rose HJ, Mays RE. 1967 53, 406–412. (doi:10.1016/j.geothermics.2014. PLUTONS: investigating the relationship
Sulfides associated with the Salton Sea 08.007) between pluton growth and volcanism in the
geothermal brine. Econ. Geol. 62, 316–330. 170. Patterson MC. 2006 Geothermal brines: high Central Andes. Geosphere 14, 3. (doi:10.1130/
(doi:10.2113/gsecongeo.62.3.316) value mineral extraction. In Sohn Int. Symp.; GES01578.1)
155. Ellis AJ. 1968 Natural hydrothermal systems and Advanced Processing of Metals and Materials 182. Hooft EE et al. 2017 Backarc tectonism,
experimental hot-water/rock interaction: Volume 6: New, Improved and Existing volcanism, and mass wasting shape seafloor
reactions with NaCl solutions and trace metal Technologies: Aqueous and Electrochemical morphology in the Santorini-Christiana-Amorgos
extraction. Geochim. Cosmochim. Acta 32, Processing, pp. 579–588. region of the Hellenic Volcanic Arc.
1356–1363. (doi:10.1016/0016-7037(68) 171. Gallup DL, Sugiaman F, Capuno V, Manceau A. Tectonophysics 712, 396–414. (doi:10.1016/j.
90035-5) 2003 Laboratory investigation of silica removal tecto.2017.06.005)
156. Werner HH. 1970 Contribution to the mineral from geothermal brines to control silica scaling 183. Lowenstern JB, Sisson TW, Hurwitz S. 2017
extraction from supersaturated geothermal and produce usable silicates. Appl. Geochem. 18, Probing magma reservoirs to improve volcano
brines Salton Sea area, California. Geothermics 1597–1612. (doi:10.1016/S0883-2927(03) forecasts. Eos 98. (doi:10.1029/2017eo085189)
2, 1651–1655. (doi:10.1016/0375-6505(70) 00077-5) 184. Eichelberger J. 2019 Planning an international
90492-X) 172. Premuzic ET, Lin MS, Jin JZ, Hamilton K. 1997 magma observatory. Eos 100. (doi:10.1029/
157. Barnea J. 1979 Geothermal minerals–the Geothermal waste treatment biotechnology. 2019EO125255)
neglected minerals. Geotherm. Energy Mag. Energy Sources 19, 9–17. (doi:10.1080/ 185. Blundy J, Afanasyev A, Tattitch B, Sparks S,
7, 12. 00908319708908828) Melnik O, Utkin I, Rust A. 2021 The economic
158. Maimoni A. 1982 Minerals recovery from Salton 173. Eggertsson GH, Lavallée Y, Kendrick JE, potential of metalliferous sub-volcanic brines.
Sea geothermal brines: a literature review and Markússon SH. 2020 Improving fluid flow in FigShare.

Blundy Et Al., 2021

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Blundy Et Al., 2021

Uploaded by

Copyright:

Available Formats

The economic potential of

economic potential of metalliferous sub-volcanic 1

R. Soc. Open Sci. 8: 202192

2. Porphyry copper deposits

R. Soc. Open Sci. 8: 202192

3. Transcrustal magmatic systems

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

5. Metals in geothermal systems

(c) salinity:~ 45 wt% (d) salinity:~ 47 wt%

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

6. Analyses of geothermal brine fluid inclusions

location Soufrière Hills, Montserrat Larderello, Italy Kakkonda, Japan

R. Soc. Open Sci. 8: 202192 royalsocietypublishing.org/journal/rsos

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

7. The importance of sulfur

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

ionic-strength fluids. This is an area of active research.

8. Sub-volcanic brine lenses

R. Soc. Open Sci. 8: 202192

Cu concentration in fluid (ppm)

Cu concentration in fluid (ppm)

Cu deposition (kg m–3)

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

10. Copper accumulation in sub-volcanic brine lenses

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

11. Implications for in situ mining

R. Soc. Open Sci. 8: 202192

brittle–ductile transition, remote from dilution by convecting meteoric water.

R. Soc. Open Sci. 8: 202192

Reservoir characterization requires integration of a variety of geophysical techniques including MT

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

R. Soc. Open Sci. 8: 202192

You might also like