Chemical Geology 197 (2003) 143 – 159

www.elsevier.com/locate/chemgeo

Do fluid inclusions preserve d18O values of hydrothermal fluids in

epithermal systems over geological time? Evidence from paleo- and
modern geothermal systems, Milos island, Aegean Sea
J. Naden a,b,*, S.P. Kilias a, M.J. Leng c, I. Cheliotis d, T.J. Shepherd b
a
Section of Economic Geology and Geochemistry, Department of Geology, National and Kapodistrian University of Athens,
Panepistimioupolis, Ano Ilisia, 157 84 Athens, Greece
b
British Geological Survey, Keyworth, Notts NG12 5GG, UK
c
NERC Isotope Geosciences Laboratory, British Geological Survey, Keyworth, Notts NG12 5GG, UK
d
Institute for Geology and Mineral Exploration, 70 Messoghion St., 115 27 Athens, Greece
Received 19 December 2001; accepted 27 September 2002

Abstract

Stable isotope compositions of quartz (d18Oquartz) and fluid inclusion waters (d18OFI and dDFI) were analysed from Profitis
Ilias, a low-sulphidation epithermal gold mineralisation deposit on Milos island, Greece, to establish if d18OFI preserve a record
of paleogeothermal processes. Previous studies show that mineralisation at Profitis Ilias resulted from extreme boiling and
vaporisation with a zone located at approximately 430 m above sea level (asl) representing the transition between liquid- and
vapor-dominated systems [Miner. Depos. 36 (2001) 32]. The deposit is also closely associated with an active geothermal
system, whose waters have a well-characterised stable isotope geochemistry [Pflumio, C., Boulegue, J., Liakopoulos, A.,
Briqueu, L., 1991. Source, Transport and Deposition of Metals. Balkema, Rotterdam]. The samples were collected over an
elevation interval of 440 m (210 – 650 m asl) to give information on the liquid and vapor dominated sections of the
paleosystem.
The data show systematic variations with sample elevation. Samples from the highest elevations (ca. 650 m asl) have the
lightest d18OFI (
7.3x) and dDFI (
68.0x) whilst the deepest (ca. 210 m asl) are isotopically heavier (d18OFI,
0.3x;
dDFI,
19.0x). Relative changes in d18OFI closely parallel those in dDFI. d18Oquartz shows an opposite trend, from the
lightest values ( + 13.9x) at the lowest elevations to the heaviest ( + 15.1x) at the highest elevations. d18OFI shows
correlations with other parameters. For example, variable fluid inclusion homogenisation temperatures in the vapor-dominated
part of the system correlate with a rapid shift in dDFI (
33.3xto
50.5x) and d18OFI (
4.1xto
6.2x). Gold
contents also increase in the same zone (up to 50 ppm Au). Comparable correlations in d18Oquartz or d18Ocalculated (estimated
geothermal fluid from fluid inclusion homogenisation data) are absent. d18Ocalculated are always 5 – 10xheavier than d18OFI.
Comparison with the present day geothermal field shows that dDFI and d18OFI are similar. Isotope data for the modern system
and fluid inclusion waters fall on linear trends subparalleling the meteoric water line and project towards seawater values.
Numerical modelling favours kinetically controlled fractionation to explain differences in d18Ocalculated and d18Ofluid rather than
diffusive posttrapping reequilibration. The evidence suggests that in low-temperature epithermal systems, d18OFI may represent

* Corresponding author. British Geological Survey, Keyworth, Notts NG12 5GG, UK. Tel.: +44-115-936-3280; fax: +44-115-936-3200.
E-mail address: [email protected] (J. Naden).

0009-2541/02/$ - see front matter D 2002 NERC. Published by Elsevier Science B.V. All rights reserved.
doi:10.1016/S0009-2541(02)00289-9
144 J. Naden et al. / Chemical Geology 197 (2003) 143–159

a better record of fluid process and the isotopic composition of the paleogeothermal fluid than temperature-corrected quartz
data.
D 2002 NERC. Published by Elsevier Science B.V. All rights reserved.

Keywords: Fluid inclusions; Oxygen and hydrogen isotopes; Active Aegean arc; Epithermal mineralisation

1. Introduction 2. Geological setting and background information

Stable isotope analyses (dD and d18O) of inclusion 2.1. The geology of Milos island
fluids and minerals provide valuable insights into
fluid sources, hydrothermal processes and fluid – rock Milos is located on the active Aegean volcanic arc
interaction in a wide range of geological environments (Fig. 1). Volcanism on Milos is Pliocene to Pleisto-
(e.g. O’Reilly et al., 1997; Wilkinson et al., 1995). cene in age and calc-alkaline in nature, and resulted
One of the main methodologies for determining D/H from the northeastward subduction of the Mediterra-
and 18O/16O focuses on measuring dD on extracted nean seafloor (part of the African plate) below the
inclusion fluids and calculating d18O of the fluid by Aegean microplate (Papazachos and Kiratzi, 1996 and
analysing the host quartz. Recently, procedures that references therein). The geology, geochronology, geo-
analyse both dD and d18O on inclusion fluids have thermal activity and structure of the island are
been developed for quartz and carbonates (Kishima described in detail by Fytikas (1989 and references
and Sakai, 1980; Kazahaya and Matsuo, 1984; Ohba therein), Briqueu et al. (1986), Liakopoulos et al.
and Matsuo, 1988; Lecuyer and O’Neil, 1994). How- (1991) and Tsokas (1996), and its main geological
ever, in quartz, scientific opinions as to which features are shown in Fig. 1. Though the last recorded
approach is the most appropriate are divided (Vityk volcanism ceased 100,000 years ago, Milos is cur-
et al., 1993; Ohba et al., 1995). The indirect approach rently an active geothermal field. Present-day activity
is not ideal as it relies on accurate determination of is concentrated in the central and eastern portions of
quartz precipitation temperatures, usually ascertained the island and is expressed by fumaroles, hot springs,
through fluid inclusion microthermometry. Also, there hot ground and submarine gas escapes (Fig. 1).
is evidence for isotopic disequilibrium between quartz
and the fluid phase at low temperatures ( < 250 jC) 2.2. Epithermal gold mineralisation at Profitis Ilias
(Zhang et al., 1989; Matsuhisa et al., 1978). Direct
measurements also have problems since isotopic The geology, mineralogy and fluid inclusions of
reequilibration can occur after trapping and during Profitis Ilias have been described in detail by Kilias et
cooling of the host mineral (Rye and O’Neil, 1968; al. (2001), and only the salient features are reiterated
Ohba et al., 1995). here.
In this study, we present directly and indirectly The mineralisation is classified as a low-sulphida-
measured dD and d18O data for adularia – sericite tion epithermal gold deposit and the host rocks consist
epithermal-Au mineralisation at Profitis Ilias on Milos of strongly sericitised rhyolitic lapilli-tuffs and ignim-
island (Aegean sea); a deposit closely linked to a brites. The mineralisation, comprising native gold,
modern geothermal reservoir, and whose geology, minor sphalerite, galena, chalcopyrite and pyrite, is
mineralogy and fluid inclusions, are well character- developed in a series of interconnected N – S to NE –
ised (Kilias et al., 2001). In addition, the modern SW trending vein structure that occupy a fault net-
system is well documented mineralogically, geo- work. Vein widths are up to 3 m across and extend to
chemically and isotopically (Liakopoulos, 1987; Lia- depths of at least 300 m below the present-day sur-
kopoulos et al., 1991; Pflumio et al., 1991; Christanis face. Three broad paragenetic stages to the minerali-
and St. Seymour, 1995), and provides an excellent sation are recognised: (i) barren pre-ore comprising
analogue for the epithermal Au mineralisation. microcrystalline quartz and pyrite; (ii) fine-grained
 J. Naden et al. / Chemical Geology 197 (2003) 143–159 145

Fig. 1. Geological map of Milos island showing its position on the Aegean arc plus the locations of mineralisation and the geothermal
manifestations (modified after Fytikas, 1989).

and vuggy quartz with base metal sulphides and ature. This trend only occurs in the upper parts of
native gold and (iii) post-ore fine-grained quartz and the hydrothermal system (>430 m asl).
barite with minor to trace sulphides.
Fluid inclusion homogenisation temperatures In addition to the above, the lowest homogenisa-
show either show narrow (25 – 50 jC) or large tion temperatures at each sample depth closely follow
ranges (>150 jC) that are not evenly distributed a depth-to-boiling curve. This feature is seen in
throughout the depth profile. In the lower levels of modern geothermal systems (e.g. Hedenquist et al.,
the hydrothermal system (below 430 m above sea 1992; Lüders et al., 2001) and provides a reference
level (asl)), only the narrow range is observed, curve for estimating quartz precipitation temperatures
whilst above this level, samples exhibiting widely (Table 1).
varying temperatures predominate (Table 1). In terms Kilias et al. (2001) concluded that the fluid inclu-
of Th –salinity relationships, the data display two sion data coupled with geochemical data strongly
distinct trends: imply that extreme boiling and vaporisation played a
major role in mineralisation. In addition, they noted
1. A high-salinity trend, where rapidly increasing that a zone, located at approximately 430 m asl, marks
salinity (3 –15 wt.% NaCl eq.) is associated with the transition between a liquid- and vapor-dominated
gradual decreases in homogenisation temperature. system. These results show that the studied samples,
This trend is seen throughout hydrothermal system. at the deposit scale, can be treated as being temporally
2. High-Th trend, where a slight tendency of decreas- and compositionally consistent and provide a coherent
ing salinity is associated with increasing temper- framework to interpret the stable isotope data.
146 J. Naden et al. / Chemical Geology 197 (2003) 143–159

Table 1
Sample information and stable isotope results
Sample Elevation Sample FI water d18Oquartz d18OFI dDFI Salinity Th range Boiling
no. (m asl) weight (g) weight (mg) (x) (x ) (x ) (wt.% NaCl eq.) (jC) curve T (jC)
G2185 657 1.4 0.5 15.4
4.5
28.7 6.1 – 6.8 150 – 160 89
G1746 621 0.5 2.5 15.5
7.8
51.0 5.0 – 5.5 213 – 243 149
G1810 613 0.5 0.3 15.1
7.3
47.8 4.5 – 4.5 161 – 284 156
G2374 551 1.5 1.2 14.2
7.8
68.0 3.4 – 5.6 190 – 234 188
G2245 499 0.5 3.7 14.6
6.9
41.5 3.3 – 4.9 205 – 215 204
G2255 457 0.5 1.6 13.9
6.1
39.8 3.1 – 4.7 190 – 219 213
G2424 421 0.7 1.2 14.2
6.2
50.5 0.0 – 7.5 233 – 253 220
G2267 405 1.4 3.7 14.3
4.1
33.3 6.5 – 7.8 223 – 249 223
G2268 403 1.2 2.5 13.7
4.5
34.9 4.4 – 9.2 208 – 231 223
G2283 365 1.2 0.3 13.4
3.7
23.8 3.1 – 6.2 219 – 251 230
PD9380 210 0.5 1.2 14.5
0.3
19.0 5.6 – 5.6 237 – 280 250
Also shown are summary microthermometric data and boiling curve temperatures for each sample elevation (from Kilias et al., 2001).

2.3. The modern geothermal system: chemical and ulation dominated by primary/pseudosecondary inclu-
isotopic characteristics sions and that they covered most of the vertical range
of the deposit. In total, 11 quartz samples were
In the active geothermal system, the reservoir selected for analysis (Table 1). Approximately 10 g
liquid before phase separation is moderately saline of quartz from each sample were crushed and sieved
( f 9 wt.% dissolved salts), and after phase separa- to
1000 + 500 Am. The quartz concentrates were
tion, the condensed vapor has a salinity of 0.12 wt.% then washed in hot (ca. 90 jC) 6 M HCl several times
NaCl. Geochemically, the residual liquid phase is with a final rinse in hot (ca. 90 jC) Milli Q water to
dominated by sodium chloride (up to 14.3 wt.% NaCl) remove ‘‘soluble’’ impurities adhering to the surface
with significant potassium (1.1 wt.%) and calcium of the quartz grains (e.g. iron oxides). This was
(0.5 wt.%). In addition, magnesium (1.36  10
4 mol followed by ultrasonic cleaning in cold Milli Q water
l
1) and sulphate (1.04  10
5 mol l
1) are strongly for 5 min to remove ‘‘insoluble’’ surface impurities
depleted compared to seawater. During its ascent to (e.g. clay minerals). Any contaminants left were then
the surface, the hydrothermal fluid separates into removed by careful handpicking under a binocular
liquid and vapor phases (Liakopoulos, 1987; Liako- microscope to give a pure quartz concentrate weigh-
poulos et al., 1991). In terms of stable (d18O, dD) ing between 0.5 and 1.0 g. Immediately, prior to
isotope systematics, the deep reservoir comprises a analysis, the sample was briefly ultrasonically cleaned
fluid that resulted from a Rayleigh-type distillation of in dichloromethane to remove any remaining organic
seawater accompanied by partial reequilibration with material.
volcanic rocks as the seawater percolated down
through the system (Pflumio et al., 1991). 3.2. d18O and dD from fluid inclusions

d18O and dD in fluid inclusions were determined

3. Materials and methods using a modified methodology based on Kishima and
Sakai (1980), Kazahaya and Matsuo (1984) and
3.1. Sample selection and preparation Lecuyer and O’Neil (1994). Between 0.5 and 1 g of
handpicked quartz grains were degassed at f 20 jC
Sample selection for isotopic analysis was based overnight under vacuum. Samples were then decrepi-
on fluid inclusion microthermometry and elevation tated at 600 jC for 30 min and the fluid inclusion
data. The criteria were that quartz samples contained water collected cryogenically. This water was then
only a single-inclusion type with an inclusion pop- transferred to a microequilibration quartz tube sealed
 J. Naden et al. / Chemical Geology 197 (2003) 143–159 147

at one end, to which 2 cm3 of a standard CO2 gas was 4. Stable isotope results
added, and the tube completely sealed. The tube was
then weighed, and the H2O and standard CO2 gases 4.1. Oxygen data
were left to exchange oxygen isotopes at 25 jC for 7
days. After equilibration, H2O and CO2 were sepa- The oxygen isotope data show correlations
rated cryogenically. The water was converted to between sample elevation and d18O in both fluid
hydrogen by reduction over hot zinc and ratios were inclusion waters (d18OFI) and quartz (d18Oquartz)
measured on a SIRA 10 mass spectrometer. CO2 was (Table 1 and Fig. 2). With the exception of the sample
collected and ratios were determined on an Optima from the highest elevation, d18OFI decreases from
mass spectrometer along with a sample of the stand-
0.3x at 210 m asl to
7.3x at 613 m asl.
ard gas used for the equilibration. The d18O values of d18Oquartz shows an opposite trend, increasing from
the water samples were calculated using the mass + 13.4xto + 15.1x. However, to directly compare
balance equation of Kishima and Sakai (1980) and the the quartz and fluid inclusion water data, d18Oquartz
CO2 – H2O fractionation factor at 25 jC was deter- needs to be corrected for temperature-dependent frac-
mined using O’Neil et al. (1975). The d18O value of tionation effects between quartz and the hydrothermal
standard CO2 used for equilibration was 31.85x fluid (d18Ocorrected). However, in low- to moderate-
(SMOW), a value close to that expected for the temperature epithermal systems ( < 250 jC), correct
CO2 after equilibration. Since the oxygen isotope estimation of quartz precipitation temperatures from
ratio of the CO2 only changes slightly during the fluid inclusion temperatures can be difficult and errors
equilibration, the precision was not compromised. in temperature estimate can result in significant uncer-
The weight of fluid inclusion water released from
the quartz was calculated from the weights of the
collection tubes before and after the H2O – CO2 col-
lection. The water content of each sample was vari-
able, but in most cases, 0.5– 1 g of quartz yielded
between 0.5 and 4 Al of water (Table 1). At each stage
in the extraction procedure, the line pressure was
monitored to check for residual gas. Precisions
obtained for the laboratory-standard water, extracted
and measured at the same time, were 0.1xfor d18O
and 3xfor dD (2r) for samples containing between
1 and 2 Al of water. The data are presented as per mil
(x) deviations from VSMOW.

3.3. d18O from quartz

Silicate ratios were obtained from the quartz after

fluid inclusion extraction using the procedure outlined
by Clayton and Mayeda (1963). Oxygen yields were
monitored and considered acceptable within the range
99 –103% of the theoretical yield. Isotope ratios were
measured on a CJS Sciences mass spectrometer
(phoenix 390, rebuilt VG MM903). d18O values were
normalised through laboratory standards and NBS28,
and corrected according to Craig (1957) and Deines
(1970). The data are presented as per mil (x) Fig. 2. Plot of sample depth versus d18O comparing values obtained
deviations from VSMOW. Overall, analytical repro- from inclusion water, quartz and d18O calculated from fluid
ducibility was of the order of F 0.10x(2r). inclusion temperatures (see text for discussion).
148 J. Naden et al. / Chemical Geology 197 (2003) 143–159

tainties in isotopic compositions. At 200 jC, an tion. Second, there is a difference of about + 8xto
uncertainty of F 20 jC for example in the fluid + 12xbetween d18OFI and d18Ocorrected. Closer
inclusion homogenisation temperature corresponds examination of the corrected data raises a number
to a 2.5xuncertainty in d18Ofluid. Furthermore, at of further points:
Profitis Ilias, where there is a temperature difference
of approximately 100 jC between the lowest and 1. Use of a single temperature value preserves the
highest elevations, uncertainties of this magnitude are d18O – depth trend seen in the quartz data, which is
greater than the total range in the measured quartz still in the opposite sense to the inclusion water
oxygen isotopic composition (1.2x). Several data.
choices of temperature estimate are available to us. 2. Employing mean homogenisation temperature
First, there is a single value, e.g. 225 jC, that results in significant scatter, and makes it difficult
corresponds to estimated boiling temperatures, but to differentiate any data trends.
there is a clear variation in temperature over the 3. Minimum homogenisation temperature again re-
investigated elevation interval (see Table 1). Second, sults in scatter, but trends in the data are now
average sample homogenisation temperatures (Table discernible and the d18O – depth trend is in a similar
2) could also be appropriate. However, in this case, sense to that of the inclusion water.
there is good evidence for heterogeneous trapping 4. Applying temperature estimates calculated from
(Kilias et al., 2001), which will make this choice an the boiling curve, with the exception of the sample
overestimation. Third, the minimum homogenisation from the highest elevation, results in a trend with
temperature of each sample (Table 1), which repre- little scatter that closely parallels the inclusion
sents fluid inclusions that have trapped the least water.
vapor, is also a valid estimate (see Pichavant et al.,
1982; Ramboz et al., 1982; Bodnar et al., 1985 for 4.2. Hydrogen data
discussions that relate homogenisation temperatures,
boiling and heterogeneous trapping). Last, we have dD of inclusion waters (dD FI ) varies between
depth-to-boiling relationships to estimate fluid tem-
68.0xand
19.0x. dDFI shows a reasonable
peratures (Table 1). correlation with sample elevation (Fig. 3), with eight
Fig. 2 compares d18Ocorrected, d18OFI and d18Oquartz of the eleven samples analysed falling on a linear trend
with sample elevation using the above methods for varying from the light values (
47.8x) at high
estimating temperature. First, it is clear that trends in elevations (613 m asl) to heavier values (
19x) at
the data are systematically related to sample eleva- lower altitudes (210 m asl).

Table 2
Comparison of fluid inclusion water d18O with calculated fluid compositions for Profitis Ilias
Sample Elevation d18OFI d18Oquartz d18Ocalculated (x) d18Ocalculated (x) d18Ocalculated (x) Boiling Mean
no. (m asl) (x ) (x ) (T = boiling curve) (T = 225 jC) (T = mean Th) curve T (jC) Th (jC)
G2185 657
4.5 15.4
7.1 4.8 0.0 89 155
G1746 621
7.8 15.5
0.3 4.9 5.2 149 230
G1810 613
7.3 15.1
0.1 4.5 4.4 156 223
G2374 551
7.8 14.2 1.4 3.6 3.0 188 214
G2245 499
6.9 14.6 2.8 4.0 3.2 204 211
G2255 457
6.1 13.9 2.6 3.3 2.3 213 207
G2424 421
6.2 14.2 3.3 3.6 4.4 220 241
G2267 405
4.1 14.3 3.6 3.7 4.2 223 234
G2268 403
4.5 13.7 3.0 3.1 2.5 223 215
G2283 365
3.7 13.4 3.0 2.7 3.4 230 238
PD9380 210
0.3 14.5 5.1 3.9 5.8 250 264
Calculated fluids use Zhang et al. (1989).
 J. Naden et al. / Chemical Geology 197 (2003) 143–159 149

common with present day systematics. It has a similar

range in oxygen isotope composition, and combined
dDFI – d18OFI are close to the vaporisation – mixing
trend described by Pflumio et al. (1991) (Fig. 5). In
addition, the variation in the isotopic composition of
the fluid inclusion waters is modelled by boiling
processes. Fig. 5 shows two steam separation trends
that match the d18OFI data. Here, the heavier samples,
in both dD and d18O, are located at the lowest
elevations and are within the liquid-dominated portion
of the hydrothermal system (below f 430 m asl), and
correspond to the highest temperatures in the system.
Conversely, the lightest samples are located in the
vapor-rich part (above f 430 m asl) and have highly
variable fluid inclusion temperatures indicative of
Fig. 3. Graph of dD versus sample elevation, showing a general heterogeneous trapping of liquid and vapor (Figs. 4
trend of increasing dD with increasing sample depth. and 5). The observed depth – temperature – isotopic
variation in d18OFI can be explained by the trapping
4.3. Correlations between stable isotope, fluid in- of varying physical mixtures (as fluid inclusions) of
clusion and geochemical data residual liquid and steam, with each ‘‘quartz extrac-
tion’’ recording an ‘‘averaged’’ isotopic composition.
In addition to isotope elevation data, it is also In this manner, samples from the higher elevations will
possible to examine the relationships between stable contain more steam and record the lightest dD and
isotope, fluid inclusion and geochemical data (Au and d18O values, whilst those in the deeper portions will be
Ag). It is clear that 430 m asl marks a distinct transition more representative of the reservoir and residual fluid
(Fig. 5). First, dD and d18OFI water show a significant compositions. However, in terms of d18Oquartz compo-
decrease to lighter values and relative changes in dD sitions, irrespective of the temperature used to calcu-
closely parallel those in d18OFI water. Second, fluid late quartz– water fractionation, the d18Ocalculated data
inclusion homogenisation temperatures start to show plot well to the right of the meteoric water line, in a
a high degree of variability above 430 m asl. Here, scattered manner that cannot be related to the modern
individual samples commonly record Th ranges in geothermal system and have no obvious trends.
excess of 100 jC, whilst below the transition temper-
ature, ranges are generally less than 20 jC. This 5.2. The quartz –inclusion water d18O shift
variation correlates with the shift in d18OFI and dDFI to
lighter values. Third, at 430 m asl, metal contents start Table 2 compares d18OFI and d18Ocorrected and
to increase and gold contents are commonly greater clearly shows calculated fluid compositions do not
than 20 ppm. In comparison, the variation in d18Oquartz equate with those measured in the fluid inclusions.
shows a gradual change along the entire depth profile In most cases, irrespective of the choice of temper-
and there is no clearly marked change at 430 m asl. ature, calculated compositions are generally 5 –10x
heavier than the measured fluid inclusion waters. To
address the reasons for this, let us examine the
5. Discussion limited number of cases reported in the literature
where both quartz and inclusion water have been
5.1. Combined dD and d18O, and comparisons with analysed (Table 3).
the modern geothermal system First, Rye and O’Neil (1968) analysed d18O in
inclusion waters hosted in calcite, sphalerite and
Comparison of the paleo- and modern systems on quartz, and concluded that d18OFI undergoes posttrap-
Milos show that d18OFI has a number of features in ping exchange with quartz. However, these conclu-
150 J. Naden et al. / Chemical Geology 197 (2003) 143–159

Fig. 4. Combined stable isotope, fluid inclusion and assay graphs, showing how d18O, dD, fluid inclusion Th and metal (Au, Ag) content vary
with depth, and how they are related to each other. Each parameter shows a distinct change at approximately 430 m asl. Both d18O and dD
decrease rapidly, fluid inclusion homogenisation temperatures change from showing only a narrow range to being highly variable and gold
contents in quartz start to increase and are locally over 30 ppm.

sions are based on limited data and large samples (salinity: 1– 2 wt.% NaCl eq.; Th: 190– 215 jC). In
(Table 3). Thus, it is possible that they did not sample the second, isotope and fluid inclusion temperatures
a single generation of quartz or fluid. This is important disagree and microthermometric data vary consider-
as Hayashi et al. (2001) have shown that in epithermal ably (wt.% NaCl eq.: 0– 16 wt.%; Th: 170 –260 jC).
quartz, d18Oquartz can vary by up to 7xin on a Also, fluid inclusion waters extracted from sulphides
millimetric scale. Thus, without knowing the homo- (sphalerite and galena) and coexisting quartz have
geneity of the material and considering the small very similar d18OFI and d18DFI (Table 3). These data
number of quartz samples analysed, it is difficult to lead to their conclusion that in epithermal systems,
assess the effects of posttrapping reequilibration from inclusion fluids can preserve d18O over geologic time.
this data alone. It is also important to note that Beregovo is a
Second, Vityk et al. (1993) and Ohba et al. (1995) relatively young deposit (ca. 15 Ma).
analysed both d18OFI and d18Oquartz from the Bere- In contrast, Ohba et al. (1995), for relatively high
govo epithermal and Kaneuchi tungsten deposits and temperature (300 – 400 jC) and old (ca. 91 Ma) hydro-
had significantly different results. Vityk et al. (1993) thermal quartz, show that positive D18Ocorrected-FI is
record two distinct types of quartz and fluid inclusion. due to posttrapping diffusional exchange. However,
In the first, isotope and fluid inclusion temperatures it is important to note that in both Rye and O’Neil
agree and microthermometry indicates a composition- (1968) and Ohba et al. (1995), formation temper-
ally and thermally homogeneous hydrothermal fluid atures for hydrothermal quartz were in excess of 300
 J. Naden et al. / Chemical Geology 197 (2003) 143–159 151

itis Ilias and Beregovo, reequilibration by diffusion

may not occur. Here, as the quartz cools rapidly from
hydrothermal to ambient temperatures, slow d18O
diffusion rates may prevent significant posttrapping
exchange. This is indicated qualitatively at Bere-
govo, where one type of quartz has clearly not
undergone any significant posttrapping exchange
(Table 3).

5.3. Modelling of posttrapping exchange by diffusion

We can examine posttrapping exchange by diffu-

sion by looking at Dodson closure temperatures
(Dodson, 1973, 1979) for d18O in quartz. This
approach is particularly successful in systems domi-
nated by volume diffusion (Valley, 2001) and d18O
diffusion between the wall of a fluid inclusion, and
bulk quartz is one example of this.

8 9
< E=R =
Tc ¼
 ð1Þ
: ln
ARTc2 ðD0 =a2 Þ ;
EðyT=ytÞ

The Dodson equation (Eq. (1)) defines the closure

temperature (Tc [K]) in terms of a dimensionless
Fig. 5. Combined D/H – d18O graph comparing directly measured diffusional anisotropy parameter (A), the gas constant
and calculated data with the stable isotope systematics of the (R [J mol
1 K
1]), the preexponential diffusion
modern geothermal system. Included for reference purposes are: (i) coefficient (D0 [m2 s
1]), activation energy (E [J]),
two single-stage steam separation trends for fluid with 5 wt.%
dissolved NaCl (calculated using the methodology of Truesdell,
distance over which diffusion takes place (a [m]) and
1984; the steam tables of Haas, 1977a,b; and the fractionation data cooling rate (yT/yt [K s
1]). As individual fluid
of Horita et al., 1995), (ii) the top left part of the magmatic water inclusions generally have small volumes ( < 0.125
box (Sheppard, 1986), (iii) the Mediterranean meteoric water line Al), then diffusion and exchange of d18O between
and Mediterranean seawater (see text for discussion of the data). host quartz and inclusion water over short distances
will significantly effect the composition of inclusion
water. For example, a 1-Am-thick shell of quartz
jC. It is known from experimental work (Matsuhisa surrounding 10 Am of spherical fluid inclusion repre-
et al., 1978; Zhang et al., 1989) that at temperatures sents over 70% of the inclusion volume (Fig. 6A).
in excess of 300 jC, equilibration between quartz Thus, it is necessary to examine closure temperatures
and water is relatively rapid (ca. 30 – 100 h) whilst at for diffusion distances at the micron and submicron
temperatures lower than 250 jC, isotopic equilibrium scale. Fig. 6B shows quartz closure temperatures at a
is not achieved. Thus, for both of these studies, due variety of diffusion distances and cooling rates. From
to the high precipitation temperature of the hydro- this, it is clear that the closure temperature of d18O in
thermal quartz and the time elapsed since minerali- quartz, even for very short distances (ca. 0.01 Am), is
sation, some posttrapping exchange by diffusion can higher than most epithermal mineralisation (ca. 250
be expected. But in lower temperature ( < 250 jC) jC) for cooling rates in excess of ca. 100 jC Ma
1.
young ( < 15 Ma) epithermal deposits, such as Prof- This means that in low- to moderate temperature
152 J. Naden et al. / Chemical Geology 197 (2003) 143–159

Table 3
Data from previous inclusion water – quartz d18O studies
Data source Sample Sample FI water d18OFI d18Oquartz d18Ocalculated D18Ocalculated FI Temperature dDFI
no. wt. (g) wt. (mg) (x ) (x ) (x ) (x ) (jC) (x )
Rye and O’Neil 63-R-4 60.6 6.0
3.7 n.a. – – – n.a.
(1968) 63-R-23 29.5 6.2
4.5 15.8 8.4 12.9 300 n.a.
63-R-23 29.5 6.2
4.5 15.8 9.0 13.5 315 n.a.
Ohba et al. G-1 n.g. n.g.
0.3 n.a. – – – n.a.
(1995) G-2 n.g. n.g.
1.1 14.5 7.1 8.2 300 n.a.
G-2 n.g. n.g.
1.1 14.5 9.9 11.0 400 n.a.
G-3 n.g. n.g.
5.8 14.5 7.1 12.9 300 n.a.
G-3 n.g. n.g.
5.8 14.5 9.9 15.7 400 n.a.
Y-1 n.g. n.g.
11.2 14.2 6.8 18.0 300 n.a.
Y-1 n.g. n.g.
11.2 14.2 9.6 20.8 400 n.a.
Vityk et al. 486 QI 3–5 n.g.
5.6 9.2
2.9 2.7 200
72
(1993) 1308 QI 3–5 n.g.
5.0 7.5
2.8 2.2 230
80
1522 QI 3–5 n.g.
5.9 9.7 0.3 6.2 250
79
496 QII 3–5 n.g.
7.1 8.7
2.8 4.3 210
56
523 QII 3–5 n.g.
1.9 9.4
2.7
0.8 200
66
515 QII 3–5 n.g.
2.4 8.7
2.8
0.4 210
79
483 QII 3–5 n.g.
2.8 9.1
3.0
0.2 200
84
A n.g. n.g.
3.5 – – – –
82
B n.g. n.g.
2.5 – – – –
79
477 QII 3–5 n.g.
4.0 8.7
3.4 0.6 200
94
100 QIII 3–5 n.g.
2.1 15.1 4.8 6.9 230
55
331 QIII 3–5 n.g.
0.9 17.9 7.6 8.5 230
52
Calculated fluid compositions use Zhang et al. (1989).
n.a. = not analysed; n.g. = not given; A = coexisting sphalerite; B = coexisting galena.

epithermal deposits, the d18O of inclusion water will and to this, Vityk et al. (1993) show that quartz – fluid
not be significantly modified by posttrapping inclusion isotope temperatures associated with a spe-
exchange with quartz. However, in higher temper- cific quartz type are in good agreement with fluid
ature (300 –400 jC) hydrothermal systems, Fig. 6B inclusion microthermometry. However, for the Prof-
shows that quartz is not closed to d18O diffusion. itis Ilias samples and other quartz types at Beregovo,
Hence, posttrapping diffusion will take place and the isotope temperatures do not agree (see Tables 2 and
d18O composition of the inclusion water will be 3). Thus, in the epithermal examples, the fact that
modified. Also, in higher temperature systems, the some isotope temperatures agree and others do not
degree of exchange will be dependent on inclusion suggests that another process must be responsible for
size. As for a given distance and cooling rate, smaller the differences.
inclusions will have proportionally more quartz to
exchange with than larger inclusions (see Fig. 6A). 5.4. Kinetically controlled fluid –mineral exchange
This relationship between inclusion size and degree of during quartz formation
posttrapping exchange has been recorded by Ohba et
al. (1995). Thus, for high temperature (>300 jC) Dubinina and Lakshtanov (1997) provide a differ-
hydrothermal systems, d18Oquartz data coupled with ent perspective into reasons for the quartz– inclusion
accurate knowledge of quartz precipitation temper- water d18O shift. They show that providing no other
atures are better estimators of d18Ofluid than d18OFI. isotopic equilibration mechanism operates (e.g. diffu-
However, in lower temperature systems ( < 250 jC), sion), complete isotopic equilibrium between a min-
quartz closure temperatures indicate that fluid inclu- eral and a fluid phase cannot be achieved through
sions do not reequilibrate with quartz during cooling, kinetically controlled solution – precipitation pro-
 J. Naden et al. / Chemical Geology 197 (2003) 143–159 153

cesses. Dubinina and Lakshtanov (1997) expressed (De). Additionally, they described the degree of
the degree of isotopic exchange ( F) between quartz exchange ( F) in terms of total mass of water (w),
and fluid, where the quartz has a silica gel precursor, total mass of mineral (M) and mass of dissolved and
as reprecipitated mineral (m) as

D0
Di h
m i
w
F¼ ð2Þ F ¼ 1
exp
1þ ð3Þ
D0
De w M

where D0, Di and De are differences in d values

between fluid and (i) silica gel (D0), (ii) freshly Eq. (3) means that during solution – precipitation,
precipitating quartz (Di) and (iii) quartz at equilibrium isotopic fractionation between mineral and fluid can
exceed the predicted equilibrium value of Dmineral –
fluid. This has been recorded experimentally by Zhang
et al. (1989) who documented Di
De values of
+ 8.3xduring the early stages of a 250 jC silica
gel –water isotope equilibration experiment.
The modelling predicts changes in isotopic com-
position as an unstable mineral is transformed iso-
chemically into a stable one and is particularly
relevant to epithermal systems where crystalline
quartz can have amorphous precursors (e.g. Dong et
al., 1995). The following equation describes varia-
tions in the bulk isotopic composition of quartz (d̄Bs)
formed during solution –precipitation processes (for a

Fig. 6. Inset diagram shows a cross-section through a hypothetical

spherical fluid inclusion (diameter r) with a halo of quartz
(thickness a) that inclusion water can exchange d18O through
diffusion. At distances >a, the inclusion does not exchange with the
quartz. (A) Graph of the volume ratio of a spherical quartz shell and
fluid inclusion at different inclusion sizes and quartz shell
thicknesses—inset graph shows volume ratio variation at low
( < 0.1) thicknesses. At low volume ratios ( < 0.05), any posttrapping
diffusive exchange will not significantly affect the isotopic
composition of the fluid inclusion as the volume of the inclusion
is significantly larger than the amount of quartz available for
diffusive exchange. Thus, for inclusions not to be affected by
posttrapping exchange, effective diffusion distance must not be
greater than f 0.01 Am. (B) Graph of Dodson closure temperatures
of quartz as a function of effective diffusion distance (a in inset).
Line I is a cooling rate reflecting the maximum age and temperature
(225 jC) of the Profitis Ilias mineralisation and present-day ambient
temperature (25 jC)—this represents the lowest possible cooling
rate. Line II reflects a typical life span of a hydrothermal system
(100 ka) and a temperature drop of 200 jC (calculations use data of
Farver and Yund, 1991 and an infinite cylinder diffusion model).
From A and B, it is clear that for diffusion distances in the region of
0.01 Am, the closure temperature for cooling rates applicable to
Profitis Ilias (between Line I and II) is greater than the temperature
of mineralisation (225 jC).
154 J. Naden et al. / Chemical Geology 197 (2003) 143–159

detailed discussion of its derivation see Dubinina and compares it with fractionation factors for various
Lakshtanov, 1997): equilibrium silica –water pairs. From these studies, it
w h
m i is clear that silica – water oxygen isotope fractionation,
d̄sB ¼ d0A
ðD0
Di Þ 1
exp
ð4Þ under evaporative conditions, is non-equilibrium.
m w
Moreover, D18Osilica – water appears to increase with
where dA0 is the initial isotopic composition of the temperature. Though it would be difficult to extrap-
amorphous (unstable) phase, D0 is derived from the olate these data to higher temperatures, it would be
initial isotopic compositions of the fluid (dw0) and the reasonable to use the higher temperature value
unstable silica phase (D0 = dA0
dw0), Di is calculated ( + 14x) as a minimum estimate for D0. Using the
from the initial fluid composition and quartz –water above information and Eq. (4), we can model the bulk
fractionation at the temperature at which the solution – isotopic composition of quartz formed from a ‘‘colloi-
precipitation process take place (Di = dw0 + 1000Lna) dal’’ precursor at 225 jC. Fig. 8A illustrates this
and m/w represents the reprecipitated mineral –fluid approach and three features are evident: (i) the final
ratio as the amount of quartz formed (m) changes from composition of transformed quartz is dependent on
0 to M. To model how kinetic processes relate to our fluid– mineral ratios, (ii) compositions are approxi-
measured quartz compositions, we need to estimate the mately 3x heavier at low mineral –fluid ratios (w/
initial isotopic compositions of the fluid, the unstable M = 0.05) than at isotopic equilibrium and (iii) with w/
silica phase and temperature. A first approximation of M>10, kinetic processes start to approximate equili-
solution – precipitation temperatures can be estimated brium quartz – water fractionation. In terms of our
from fluid inclusion data (ca. 225 jC). For the initial d18Oquartz data, if we take the deepest and hottest
fluid, the composition of the modern reservoir sample (PD9380 in Table 1) to be most representative
( + 3x, Pflumio et al., 1991) provides a good ana- of the reservoir composition, d18Oquartz and d18OFI can
logue for the paleosystem at Profitis Ilias. Kita and be modelled by solution – precipitation processes.
Taguchi (1986) measured isotopic fractionation factors Using Eq. (4), the measured d18Oquartz ( + 14.5x)
between colloidal silica and a geothermal fluid at 76 predicts an initial fluid composition of 0.7x
and 88 jC, where silica was allowed to precipitate by (T = 225 jC, w/M = 0.05, dA0= + 14x), which is very
evaporation of the geothermal fluid. This is broadly close to the measured d18OFI (
0.3x), whereas
analogous to Profitis Ilias, where mineralisation is equilibrium fractionation predicts a fluid composition
associated with extreme boiling and vaporisation, of + 3.9x.
(Kilias et al., 2001). Fig. 7 reproduces their data and At Profitis Ilias, the fluids were boiling and tem-
peratures of quartz formation vary systematically with
depth (Table 1). Fig. 8B models this variation and
depicts a number of scenarios where d18Oquartz varies
with temperature, w/M, and initial fluid composition
(dw0), and from this, a number of controls on d18Oquartz
can be seen:

1. Where quartz precipitates in equilibrium with a

hydrothermal fluid (dw0 = 1x), which varies in
temperature from 250 to 100 jC. d18Oquartz shows a
relatively large systematic variation from 9.9xto
22.0x.
2. Where quartz results from a dissolution –precip-
itation process at low water-to-solid ratios (w/
M = 0.01, dw0 = 1x), d18Oquartz is relatively con-
stant and approximates dw0 + D0 (15x).
Fig. 7. Silica – water fractionation factors for quartz, amorphous 3. Going from equilibrium to low w/M and at constant
silica and colloidal silica. dw0 + D0, d18Oquartz varies as family of curves of
 J. Naden et al. / Chemical Geology 197 (2003) 143–159 155

increasing radius of arc that pivot in an anticlock-

wise fashion about the intersection between the
quartz equilibrium curve and d18Oquartz = dw0 + D0.
4. Changes in dw0 + D0 at constant w/M simply shift
the curve in the direction of changing dw0 + D0
without changing the shape of the curve.

In terms of actual d18Oquartz compositions, we can

compare the predictive curves with the measured
values through the boiling curve data (Table 1). These
provide a reasonable estimate of temperature –depth
variation within the Profitis Ilias hydrothermal system
and locate the quartz samples in temperature – d18O
space. The data show an sympathetic relationship
between d18Oquartz and temperature that is modelled
by solution –precipitation processes with a water-to-
solid ratio of 0.25 and, with the exception of the
lowest and highest temperature samples, an initial
fluid varying in composition by about 1x(Fig.
8B). Moreover, the data do not conform to an equi-
librium model of quartz precipitation. Though kinetic
processes can explain variation in d18Oquartz with
respect to estimated paleotemperatures, relating
d18Oquartz to d18Ofluid through solution – precipitation
is more challenging, as this requires knowledge of
D18Osilica – water for the quartz precursor. In our case, a
value of + 14x, taken from Kita and Taguchi
(1986), predicts a fluid composition in the region of
0– 1x (Fig. 8B). This is in approximate agreement
with the sample that best reflects the reservoir com-
position of the hydrothermal fluid (PD9380, Table 1)
and the estimated composition of the modern geo-
thermal reservoir, but samples from higher levels
record lighter values (
7.8x) that are not predicted
by solution – precipitation processes. However, these
can be explained by heterogeneous trapping of liquid
(d18O f 0 – 1x) and varying proportions of isotopi-
cally light vapor during boiling events (Fig. 5).
In addition, when we consider Vityk et al. (1993),
kinetically controlled isotopic exchange is a much
Fig. 8. Calculated variations in the bulk isotopic composition of
quartz formed from the progressive dissolution – precipitation of a
better process than posttrapping diffusion for explain-
‘‘colloidal’’ precursor. (A) Variation in d18Oquartz as the water to ing variations in D18Ocorrected-FI. Here, variation in
solid ratio changes from 0.05 to 10 at constant temperature (225 D18Ocorrected-FI is systematically related to different
jC), fluid composition ( + 3x ) and D0 ( + 14x ). (B) Variation in quartz types (Table 3). For samples where isotope and
d18Oquartz as a function of temperature, water to solid ratio and fluid inclusion temperatures agree, quartz precipitated
initial fluid composition at constant D0 ( + 14x ). Numbers on the
curves refer to initial composition of the fluid (x). Filled circles are in equilibrium with the hydrothermal fluid, or if it
quartz compositions and their corresponding boiling curve temper- formed from a precursor silica phase, then fluid-to-
atures given in Table 1 (see text for discussion). solid ratios were high (i.e. the conditions of open
156 J. Naden et al. / Chemical Geology 197 (2003) 143–159

flow). For samples where isotope and fluid inclusions 4. The modelling of solution – precipitation processes
record different temperatures (D18Ocalculated-FI: 2.2 – show that kinetic factors are a valid alternative to
8.5x), D18Ocalculated-FI can be explained by quartz posttrapping reequilibration for explaining differ-
forming through solution –precipitation processes at ences between d18Ocalculated and d18OFI.
low fluid-to-solid ratios (Fig. 8A). This corresponds to
conditions where fluid flow was restricted and/or it Thus, the above lend support to the assertion that,
underwent extensive boiling. This feature has been in epithermal systems, d18O determined on fluid
observed at lower temperatures by Kita and Taguchi inclusion waters may be a better guide to the original
(1986). When we consider the alternative of posttrap- isotopic composition of the hydrothermal fluid than
ping diffusional exchange as a mechanism for the d18O temperature-corrected quartz data (Fig. 5). This is
quartz – inclusion water shift in Vityk et al. (1993), it is because d18Oquartz in solution – precipitation processes
difficult to explain the variation. First, it would be can be significantly different from d18Oquartz at equi-
expected to affect all samples equally. That is, for a librium for a given temperature and d18Ofluid. How-
given size of fluid inclusion, quartz age and mineral- ever, in higher temperature (>300 jC) hydrothermal
isation temperature, D18Ocorrected-FI should be constant systems, such as porphyry– Cu, Sn –W and orogenic–
in all samples. This is not the case. Some show no shift Au deposits, Dodson closure temperatures show that
whilst others show D18Ocorrected-FI to be + 8x(Table quartz and fluid inclusion water undergo posttrapping
3). In terms of a diffusion mechanism, this can only be exchange. Thus, in these cases, the indirect method
explained by significantly different quartz ages or that will always provide the best estimate of fluid d18O.
fluid inclusions are significantly larger in one type of Also, at Profitis Ilias, d18OFI contains a record of
quartz. Vityk et al. (1993) do not indicate that either of fluid processes not seen in d18Ocalculated (Fig. 5 and
these are applicable to the Beregovo quartzes. Section 5.1). First, this is probably because temper-
ature estimates within paleohydrothermal systems are
5.5. What does fluid inclusion d18O represent? not precise enough to allow accurate reconstruction
of variations in d18O of the hydrothermal fluid. For
The modelling of Dodson closure temperatures for example, in a boiling epithermal system fluid, tem-
quartz show, theoretically, that for inclusions larger peratures can vary by up to 50 –70 jC over a vertical
than about 5 Am and at temperatures less than 250 jC, interval of only 100– 200 m (Table 1). This can
quartz does not significantly exchange d18O with fluid represent differences of nearly 10xin fluid d18O
inclusion water. A number of independent lines of (i.e. the difference in boiling-curve-estimated d18O
evidence support this: between 457 and 657 m elevations, Table 1). Sec-
ond, formation of quartz via an amorphous precursor
1. The epithermal mineralisation at Profitis Ilias (e.g. Dong et al., 1995) is a non-equilibrium process
shows a consistency between d18OFI, dDFI and and can result in estimates of d18Ofluid that are too
fluid inclusion and geochemical data that is not heavy.
seen in the d18Ocalculated data (see Fig. 5).
2. On Milos, there is close isotopic and chemical 5.6. Implications for d18O analysis of quartz
similarity between the composition of the fluid
inclusion waters and that of the modern geo- At Profitis Ilias, combined dD –d18Oquartz and dD –
18
thermal system (see Fig. 5), and there are sound d OFI and comparison with the modern geothermal
geological reasons that a continuum exists between system coupled with kinetic modelling of the data
the ancient and modern systems (Kilias et al., (Figs. 5 and 8) show that the differences in d18Oquartz
2001). and d18OFI can be related though a combination of
3. The data of Vityk et al. (1993) record quartz –fluid extensive boiling and quartz formation via an amor-
inclusion isotope temperatures that agree with fluid phous precursor. Whereas, calculation of equilibrium
inclusion data and cannot be explained by diffusion fluid compositions appears to overestimate the fluid
alone. Also, quartz and coexisting sulphides have composition by + 8xto + 10x. If the kinetic
similar dDFI and d18OFI (see Table 3). model is valid, this places limitations on the use of
 J. Naden et al. / Chemical Geology 197 (2003) 143–159 157

d18Oquartz in low-temperature epithermal systems to between d18OFI and d18Oquartz can be explained by
estimate d18Ofluid where solution – precipitation pro- kinetic processes. However, insufficient data and
cesses operate. Failure to recognise this limitation can knowledge concerning the d18O composition of pos-
result in significant errors in the estimation of sible silica precursors at temperatures above 150 jC
d18Ofluid. Observations that indicate d18Oquartz is and fluid-to-solid ratios preclude a generalised recon-
kinetically controlled and mitigate against using it struction of fluid compositions using this method.
to calculate fluid compositions are: The results of diffusion and kinetic modelling
coupled with independent lines of evidence, such as
1. Textural information indicating formation from an the consistency of dDFI and d18OFI with other geo-
amorphous precursor (e.g. Dong et al., 1995). This chemical data and their similarity to the isotope
can be documented through careful petrography systematics of the modern systems, lead us to believe
and cathodoluminescence studies. that fluid inclusion d18O has the potential to be a
2. Fluid inclusion data that indicate mineralisation is better estimator of oxygen isotope systematics in low-
associated with extensive boiling and vaporisation temperature hydrothermal systems than d18Oquartz and
(e.g. Simmons and Browne, 1997; Scott and fluid inclusion temperatures. However, as there are
Watanabe, 1998; Kilias et al., 2001). only two published fluid inclusion d18O studies in the
3. Where mineralisation temperatures are less than epithermal environment, further data from other sys-
250 jC and data are available. Systematic differ- tems are required to extend the applicability our
ences between d18Oquartz and d18OFI can also findings. In the first instance, these should concentrate
indicate a kinetic control. on boiling systems as these provide the best temper-
ature constraints. In addition, good petrographic and
geologic control is a prerequisite for reliable interpre-
6. Summary and conclusions tations. In our case, this was achieved through detailed
petrography and fluid inclusion microthermometry
In summary, dD and d18O in extracted fluid inclu- within a well-constrained spatial and geological
sion waters from the Profitis Ilias epithermal miner- framework (Kilias et al., 2001) that enabled compar-
alisation were found to show a linear correlation that isons between a mineralised system and its modern
closely parallels the stable isotope systematics in the analogue. At a smaller scale, or in the absence of a
active Milos geothermal system, with dD varying modern analogue for comparison, careful cathodolu-
from
23.8xto
68.0xand d18O from
0.3 minescence petrography coupled with localised iso-
to
7.8.x. d18O analysis of the quartz, using the topic analysis of the quartz will also fulfil the
same samples, shows a restricted variation of 13.4– requirement. Finally, recent analytical developments
15.9xand correction of this data, employing a for the rapid determination of dD and d18O in small
variety fluid inclusion temperature estimates, does (0.1 Al) amounts of water (Sharp et al., 2001) indicate
not match d18O in the corresponding fluid inclusion it should be possible to routinely analyse inclusion
waters. D1 8 O c o rr ec te d -F I ranges from + 8xto waters without the need for the lengthy sample
+ 12xand calculated Dodson closure temperatures preparation and analytical procedures. Factors that
for quartz, at scales appropriate to fluid inclusions and have so far hindered systematic investigations into
at cooling rates compatible with the lifetime of a the applicability of oxygen isotope determination of
geothermal systems, indicate that at temperatures fluid inclusion water.
below 250 jC, these differences cannot be accounted
for by posttrapping diffusional exchange of 18O
between inclusion water and host quartz. Thus, for Acknowledgements
the Profitis Ilias mineralisation, posttrapping reequili-
bration is not thought to be an appropriate process for This work was funded by the British Council, The
explaining the observed D18Ocorrected-FI. In our case, Greek General Secretariat of Research and a Marie
solution – precipitation processes are preferred as Curie Experienced Researcher Fellowship to JN
modelling shows that differences, in relative terms, (HPMF-CT-2000-00762). Special thanks go to J.A.
158 J. Naden et al. / Chemical Geology 197 (2003) 143–159

Crossing and Y. Simos, geologists at Midas S.A., who phases and thermochemical properties of the NaCl component in
allowed access to drill-core and proprietary company boiling NaCl solutions. U.S. Geol. Surv. Prof. Pap., B 1421-B,
B1 – B71.
data. Drs. T.H.E. Heaton and B. Spiro are thanked for Hayashi, K.-I., Maruyama, T., Satoh, H., 2001. Precipitation of gold
providing impetus and encouragement during the in a low-sulfidation epithermal gold deposit: insights from a
research. J.N., M.J.L. and T.J.S. publish with the submillimeter-scale oxygen isotope analysis of vein quartz.
permission of the Director, British Geological Survey Econ. Geol. 96, 211 – 216.
Hedenquist, J.W., Reyes, A.G., Simmons, S.F., Taguchi, S., 1992.
(NERC). [CA]
The thermal and geochemical structure of geothermal and epi-
thermal systems: a framework for interpreting fluid inclusion
data. Eur. J. Mineral. 4, 989 – 1015.
References Horita, J., Cole, D.R., Wesolowski, D.J., 1995. The activity – com-
position relationship of oxygen and hydrogen isotopes in aque-
Bodnar, R.J., Reynolds, T.J., Kuehn, C.A., 1985. Fluid inclusion ous salt solutions: III. Vapor – liquid water equilibration of
systematics in epithermal systems. In: Berger, B.R., Bethke, NaCl solutions to 350jC. Geochim. Cosmochim. Acta 59,
P.M. (Eds.), Geology and Geochemistry of Epithermal Systems. 1139 – 1151.
Rev. Econ. Geol., vol. 2, pp. 73 – 97. Kazahaya, K., Matsuo, S., 1984. A new ball-milling method for
Briqueu, L., Javoy, M., Lancelot, J.R., Tatsumoto, M., 1986. Iso- extraction of fluid inclusions from minerals. Geochem. J. 19,
tope geochemistry of recent magmatism in the Aegean arc: Sr, 45 – 54.
Nd, Hf, and O isotopic ratios in the lavas of Milos and San- Kilias, S., Naden, J., Cheliotis, I., Shepherd, T.J., Constandinidou, H.,
torini—geodynamic implications. Earth Planet. Sci. Lett. 80, Crossing, J., Simos, I., 2001. Epithermal gold mineralisation in
41 – 54. the active Aegean Volcanic arc: the Profitis Ilias deposit, Milos
Christanis, K., St. Seymor, K.S., 1995. A study of scale deposi- island, Greece. Miner. Depos. 36, 32 – 44.
tion an analog of mesothermal to epithermal one formation in Kishima, N., Sakai, H., 1980. Oxygen-18 and deuterium determi-
the volcano of Milos, Agean-arc, Greece. Geothermics 24, nations on a single water sample of a few milligrams. Anal.
541 – 552. Chem. 52, 356 – 358.
Clayton, R.N., Mayeda, T.K., 1963. The use of bromine pentafluor- Kita, I., Taguchi, S., 1986. Oxygen isotopic behavior of precipitat-
ide in the extraction of oxygen from oxide and silicates for ing silica from geothermal water. Geochem. J. 20, 153 – 157.
isotopic analysis. Geochim. Cosmochim. Acta 27, 43 – 52. Lecuyer, C., O’Neil, J.R., 1994. Stable isotope compositions of fluid
Craig, H., 1957. Isotopic standards for carbon and oxygen and inclusions in biogenic carbonates. Geochim. Cosmochim. Acta
correction factors for mass spectromatic analysis of carbon di- 58, 353 – 363.
oxide. Geochim. Cosmochim. Acta 12, 133 – 149. Liakopoulos, A., 1987. Hydrothermalisme et mineralisations metal-
Deines, P., 1970. Mass spectrometer correction factors for the de- liferes de l’ile de Milos (Cyclades – Grece). Mem. Sci. Terre.
tection of small isotopic variation of carbon and oxygen. Int. J. Universite Paris, VI, No. 36 – 87.
Mass Spectrom. Ion Phys. 4, 283 – 295. Liakopoulos, A., Katerinopoulos, A., Markopoulos, T., Boulegue,
Dodson, M.H., 1973. Closure temperature in cooling geochrono- J., 1991. A mineralogical, petrographic and geochemical study
logical and petrologic systems. Contrib. Mineral. Petrol. 40, of samples from wells in the geothermal field of Milos Island
259 – 274. (Greece). Geothermics 20, 237 – 256.
Dodson, M.H., 1979. Theory of cooling ages. In: Jaeger, E., Hun- Lüders, L., Pracejus, B., Halbach, P., 2001. Fluid inclusion and
ziker, J.C. (Eds.), Lectures in Isotope Geology. Springer-Verlag, sulfur isotope studies in probable modern analogue Kuroko-type
Heidelberg, pp. 194 – 202. ores from the JADE hydrothermal field (Central Okinawa
Dong, G., Morrison, G., Jaireth, S., 1995. Quartz textures in epi- Trough, Japan). Chem. Geol. 173, 45 – 58.
thermal veins, Queensland-classification, origin and implication. Matsuhisa, Y., Goldsmith, J.R., Clayton, R.N., 1978. Mechanisms
Econ. Geol. 90, 1841 – 1856. of hydrothermal crystallization of quartz at 250 jC and 15 kbar.
Dubinina, E.O., Lakshtanov, L.Z., 1997. A kinetic model of iso- Geochim. Cosmochim. Acta 42, 173 – 182.
topic exchange in dissolution – precipitation processes. Geo- Ohba, T., Matsuo, S., 1988. Precise determination of hydrogen and
chim. Cosmochim. Acta 61, 2265 – 2271. oxygen isotope ratios of water in fluid inclusions of quartz and
Farver, J.R., Yund, R.A., 1991. Oxygen diffusion in quartz: de- halite. Geochem. J. 22, 55 – 68.
pendence on temperature and water fugacity. Chem. Geol. 90, Ohba, T., Kazahaya, K., Matsuo, S., 1995. Diffusional 18O loss
55 – 70. from inclusion water in a natural hydrothermal quartz from
Fytikas, M., 1989. Updating of the geological and geothermal re- the Kaneuchi tungsten deposit, Japan. Geochim. Cosmochim.
search on Milos island. Geothermics 18, 485 – 496. Acta 59, 3039 – 3047.
Haas Jr., J.L., 1977a. Physical properties of the coexisting phases O’Neil, J.R., Adami, L.H., Epstein, S., 1975. Revised value for the
18
and thermochemical properties of the H2O component in boil- O fractionation between CO2 and water at 25 jC. J. Res. U.S.
ing NaCl solutions. U.S. Geol. Surv. Prof. Pap., B 1421-A, Geol. Surv. 3, 623 – 624.
A1 – A73. O’Reilly, C., Jenkin, G.R.T., Feely, M., Alderton, D.H.M., Fallick,
Haas Jr., J.L., 1977b. Thermodynamic properties of the coexisting A.E., 1997. A fluid inclusion and stable isotope study of fluid
 J. Naden et al. / Chemical Geology 197 (2003) 143–159 159

evolution in the Galway Granite, Connemara, Ireland. Contrib. Sheppard, S.M.F., 1986. Characterization and isotopic variations in
Mineral. Petrol. 129, 120 – 142. natural waters. In: Valley, J.W., Taylor, Jr., H.P., O’Neil, J.R.
Papazachos, C.B., Kiratzi, A.A., 1996. A detailed study of the (Eds.), Stable Isotopes in High Temperature Geological Pro-
active crustal deformation in the Aegean and surrounding area. cesses. Reviews in Mineralogy, vol. 16. Mineralogical Society
Tectonophysics 253, 129 – 153. of America, Washington, DC, pp. 165 – 183.
Pflumio, C., Boulegue, J., Liakopoulos, A., Briqueu, L., 1991. Oxy- Simmons, S.F., Browne, P.R.L., 1997. Saline fluid inclusions in
gen, hydrogen, strontium isotopes and metals in the present-date sphalerite from the Broadlands-Ohaaki geothermal system: a
and past geothermal systems of Milos island (Aegean arc). In: coincidental trapping of fluids being boiled toward dryness.
Pagel, M., Leroy, J.L. (Eds.), Source, Transport and Deposition Econ. Geol. 92, 485 – 489.
of Metals. Balkema, Rotterdam, pp. 107 – 112. Truesdell, A.H., 1984. Stable isotopes in hydrothermal systems. In:
Pichavant, M., Ramboz, C., Weisbrod, A., 1982. Fluid immiscibil- Henley, R.W., Truesdell, A.H., Barton, P.B. (Eds.), Fluid – Min-
ity in natural processes; use and misuse of fluid inclusion data: I. eral Equilibria in Hydrothermal Systems. Rev. Econ. Geol., vol.
Phase equilibria analysis; a theoretical and geometrical ap- 1, pp. 129 – 142.
proach. Chem. Geol. 37, 1 – 27. Tsokas, G.N., 1996. Interpretation of the Bouguer anomaly of Milos
Ramboz, C., Pichavant, M., Weisbrod, A., 1982. Fluid immiscibil- island (Greece). J. Volcanol. Geotherm. Res. 72, 163 – 181.
ity in natural processes; use and misuse of fluid inclusion data: Valley, J.W., 2001. Stable isotope thermometry at high temperatures.
II. Interpretation of fluid inclusion data in terms of immiscibility. In: Valley, J.W., Cole, D.R. (Eds.), Stable Isotope Geochemistry.
Chem. Geol. 37, 29 – 48. Rev. Mineral. Geochem., vol. 43, pp. 365 – 413.
Rye, R.O., O’Neil, J.R., 1968. The O18 content of water in primary Vityk, M.O., Krouse, H.R., Demihov, Y.N., 1993. Preservation of
fluid inclusions from Providencia, north-central Mexico. Econ. d18O values of fluid inclusion water over geological time in an
Geol. 63, 232 – 238. epithermal environment: Beregovo deposit, Transcarpathia, Uk-
Scott, A.M., Watanabe, Y., 1998. ‘‘Extreme boiling’’ model for raine. Earth Planet. Sci. Lett. 119, 561 – 568.
variable salinity of the Hokko low-sulfidation epithermal Au Wilkinson, J.J., Jenkin, G.R.T., Fallick, A.E., Foster, R.P., 1995.
prospect, southwestern Hokkaido, Japan. Miner. Depos. 33, Oxygen and hydrogen isotopic evolution of Variscan crustal
568 – 578. fluids, south Cornwall, UK. Chem. Geol. 123, 239 – 254.
Sharp, Z.D., Atudorei, V., Durakiewicz, T., 2001. A rapid method Zhang, L., Liu, J.X., Zhou, H., Chen, Z.S., 1989. Oxygen isotope
for determination of hydrogen and oxygen isotope ratios from fractionation in the quartz – water – salt system. Econ. Geol. 84,
water and hydrous minerals. Chem. Geol. 178, 197 – 210. 1643 – 1650.