ORIGINAL PAPER
Apatite as a monitor of late-stage magmatic processesat Volcan Irazu, Costa Rica
Jeremy W. Boyce Æ Richard L. Hervig
Received: 18 April 2008 / Accepted: 1 July 2008 / Published online: 16 July 2008
� Springer-Verlag 2008
Abstract Apatite phenocrysts from the 1963 and 1723
eruptions of Irazu volcano (Costa Rica) record a volatile
evolution history that confirms previous melt inclusion
studies, and provides additional information concerning the
relative and absolute timing of subvolcanic magmatic
events. Measurements of H, Cl, and F by secondary ion
mass spectrometry reveal multiple populations of apatite in
both 1723 and 1963 magmas. Assuming nominal apatite/
melt partition coefficients allows us to compare the pattern
of melt inclusions and apatites in ternary space, demon-
strating the fidelity of the record preserved in apatite, and
revealing a complex history of magma mixing with at least
two components. The preservation of heterogeneous
populations of apatite and of internally heterogeneous
crystals requires short timescales (days to years) for these
magmatic processes to occur.
Keywords Apatite � Irazu � Costa Rica � Volcano �Volatiles � Magma � Melt Inclusions
Introduction
Volatile elements play a significant role in the evolution of
magmas, so developing new techniques for understanding
volatile budgets and the magmatic processes that affect
them is critical to advance our understanding of how vol-
canoes function. Minerals that incorporate volatiles into
their structure are likely targets for evaluation in this
regard—and chief among them is apatite: Ca5(PO4)3(F, Cl,
OH). In this study we compare the record of H, F, and Cl in
melt inclusions from the literature with measurements of
those elements in apatite phenocrysts from the same hand
samples. This allows us to evaluate the degree to which the
apatite and melt-inclusion records are coherent in a com-
plex natural system, and to guide future research into the
potential of apatite to act as a magma hygrometer, fluo-
rimeter, and chlorimeter.
Apatite
Apatite is a commonly observed mineral in a wide range of
rock types, and is the most abundant mineral reservoir of
phosphate in most terrestrial igneous and metamorphic
rocks. Nominally Ca5(PO4)3(F, Cl, OH), apatite incorpo-
rates much of the periodic table into its structure, making it
useful for a wide range of geochemical and isotopic
applications (Farley and Stockli 2002; Gleadow et al. 2002;
John et al. 2008; Pan and Fleet 2002; Pyle et al. 2001;
Hovis et al. 2007; Schoene and Bowring 2007; Spear and
Pyle 2002). Most relevant to this study is the diverse volatile
chemistry of apatite. Measurements of H, C, F, Cl, and
S have been made in natural apatites (Mathez and Webster
2005; Nadeau et al. 1999; Piccoli and Candela 1994; Streck
and Dilles 1998), but experimental calibrations of apatite
volatile compositions against melts of known composition
Communicated by J. Hoefs.
J. W. Boyce � R. L. Hervig
School of Earth and Space Exploration,
Arizona State University, Tempe, AZ 85287-1404, USA
J. W. Boyce (&)
Department of Earth and Space Sciences,
University of California, Los Angeles, Los Angeles,
CA 90095-1567, USA
e-mail: [email protected]
123
Contrib Mineral Petrol (2009) 157:135–145
DOI 10.1007/s00410-008-0325-x
are less common (Mathez and Webster 2005; Parat and
Holtz 2004; Peng et al. 1997).
Irazu
Volcan Irazu is located near the southern end of the active
expression of the Central American volcanic arc (CAVA),
making it a key location for geochemical observations that
can be related to arc magmatism, in general, and for this
arc in particular. Irazu is also within *25 kilometers of the
capital city of San Jose, Costa Rica (with a metropolitan
population of *2 9 106 persons). The two largest erup-
tions in the past 300 years (1723, 1963) each caused
extensive economic damage and loss of life, making it a
significant volcano from a hazards perspective as well
(Alvarado 1993). In an effort to contribute to both of these
scientific directives, we have undertaken a study of the
late-stage magmatic history of the 1723 and 1963 eruptions
of Irazu. We focus on the volatile records of these magmas
preserved in the chemistry of apatite phenocrysts, demon-
strating both the complexity of the magmatic system at
Irazu and also the utility of apatite in studying magmatic
processes.
A complicated late-stage magmatic history is suggested
for Irazu magmatic products. Alvarado et al. (2006)
describe textural (macroscopic and microscopic), minera-
logical, and geochemical evidence for the existence of at
least two discrete magma bodies over much of the eruption
history of the volcano. They define a rare-earth element
(REE) and high-field-strength element (HFSE) enriched
magma (named ‘‘Haya’’) generally represented by more
primitive, silica-poor, Mg-rich basalts, and a second batch
(‘‘Sapper’’), which is characterized by large-ion-lithophile
element enriched basaltic andesites. The two magmas
define significantly different trends, such that they are not
considered to be simply related by differentiation processes
but rather to have come from different sources. The 1723
and 1963 eruptions are thought to be hybrids of these two
magmas, part of a cycle of eruptions where Haya, mixed,
and Sapper magmas erupt in sequence.
Pre-eruptive volatiles have been measured in Irazu melt
inclusions by Clark et al., (1998), Benjamin et al. (2007),
and Sadofsky et al. (2008). Benjamin et al. (2007) studied
olivine- and clinopyroxene-hosted melt inclusions (MI) by
secondary ion mass spectrometry (SIMS), electron micro-
probe analysis, and laser ablation inductively coupled
plasma mass spectrometry in order to constrain the pre-
eruptive major, trace, and volatile contents of Irazu 1723
and 1963 parent melts. They observed generally higher
CO2, S, Cl, and H2O contents in MI from 1723 olivines
than for 1963 clinopyroxene-hosted MI, an observation
explained by the degassing of the magma prior to clino-
pyroxene appearing as a liquidus phase. Correlations
between S, CO2, and H2O define approximately monotonic
curves, while Cl defines a more complicated, yet generally
decreasing curve against H2O (Fig. 1). They argue that
these curves reflect degassing ‘‘ascent paths’’, with the
most primitive melts represented by the MI with the
highest CO2 (and generally high S, H2O and Cl).
Supporting evidence comes from the composition of the
olivine hosts, which become more fayalitic with decreasing
volatile content, consistent with crystal fractionation. They
suggest that the MI volatile measurements are consistent
with closed-system degassing (Newman and Lowenstern
2002), and see no evidence in the MI suite for magma
mixing. Sadofsky et al. (2008) performed a study that was
similar in terms of the analytical techniques applied, but
covered a greater number of volcanoes along CAVA. The
stratigraphic context for the Irazu samples in the study of
Sadofsky et al. (2008) suggests that they are from the 1963
eruption, but the obvious difference in F–H2O and Cl–H2O
relationships between the Benjamin et al. (2007) samples
and the Sadofsky et al. (2008) data makes a comparison
between the data of Sadofsky et al. (2008) and this study
suspect. Because our apatite phenocrysts come from the
same hand samples as the MI from the Benjamin et al.
(2007) study, we will focus on those data here as a means
of comparison with the new apatite volatile results.
In the context of considerable evidence of magma mix-
ing at Irazu, the results of the MI study of Benjamin et al.
(2007) are quite surprising. We might have expected com-
plex relationships within and between volatile elements due
to the mixing of discrete magma batches with different
degassing histories. The additional data of Sadofsky et al.
(2008) is not consistent with monotonic degassing, and
suggests that mixing or other complex processes may
influence volatiles. However, this variation from sample to
sample also reaffirms our suggestion that the best data set to
use for the purpose of comparing apatite and melt inclusions
is that of Benjamin et al. (2007). The apparently simple
monotonic degassing path model that is consistent with the
data of Benjamin et al. (2007) will be discussed further in
the context of the additional constraint provided by apatite.
Samples and methods
Samples
Irazu samples IZ-03-17a (*150 g of unconsolidated lapilli
tuff) and IZ-03-19b (*200 g bomb fragment) were
obtained from T. Plank, and are splits from the same
samples used for the melt inclusion (MI) study of Benjamin
et al. (2007). Both the samples were crushed and separated
using standard magnetic and gravimetric techniques by
Apatite to Zircon, Inc. Apatite was present in both the
136 Contrib Mineral Petrol (2009) 157:135–145
123
separates as euhedral to anhedral fragments, but because of
the small sample volume, the number of apatite fragments
liberated was approximately one dozen in each, and less
than half of these were of sufficient size to allow analyses
without risk of contamination. Few fragments showed
unambiguous crystal faces (to allow interior to edge zoning
studies), and as a result, some anhedral crystal fragments
were also analyzed.
Methods
Crystals were mounted in epoxy and polished with alu-
mina-impregnated lapping film to 3 lm grit, then further
polished using 0.2 lm alumina and 0.05 alumina suspen-
sions. Samples were thoroughly cleaned by ultrasonication
prior to application of a gold-coat for SIMS analysis.
Volatile compositions of Irazu apatites were determined
in multiple sessions on the Cameca 6f secondary ion mass
spectrometer in the School of Earth and Space Exploration
at Arizona State University. For H, Cl, and F, we follow the
analytical protocol of Boyce and Hervig (2008), summa-
rized here as follows: we used a *10nA 16O- primary beam
and measured positive secondary ions: 1H+, 12C+, 19F+,31P+, 37Cl+, and 43Ca+. The primary beam was focused to
*20 lm, then rastered over a 20 9 20 lm area to reduce
surface contamination. Further reduction of contamination
was accomplished by including a pre-analytical sputtering
time of 300 s before each analysis. Only ions from the
central *15 lm of the sputtered region were allowed into
the mass spectrometer, and only secondary ions with
75 ± 20 eV excess kinetic energy were detected. Carbon
data are not discussed further, as all are within error of
background values, and are used only to determine which
analyses are compromised by significant contamination.
Volatile element ion intensities were converted to con-
centrations using 31P+ as a reference isotope, while trace
elements were reduced using 43Ca+, both of which are
assumed to be stoichiometric at the percent level. Apatite
standard materials from Durango, Mexico, and Mud Tanks,
Australia, were used to calibrate these analyses. Measured
ratios in apatite standards are generally reproducible within
a single session to better than 3% (2r) for volatiles (except
at near-blank concentrations).
Uncertainties for each analysis are calculated using
standard error propagation equations, and incorporate
uncertainty from counting statistics, as well as from the
linearity of the calibration curve and estimated uncertain-
ties (±10%) in the standard compositions used to derive
the calibration curves.
Fig. 1 Volatile measurements for olivine- and clinopyroxene-hosted
melt inclusions from the 1723 and 1963 eruptions of Irazu. Light graycircles (in all figures) are olivine-hosted 1723 data from Benjamin
et al. (2007). Clinopyroxene-hosted melt inclusions measured by
Benjamin et al. (2007) from the eruptions circa 1963 are represented
by dark gray circles. Olivine-hosted melt inclusions of Sadofsky et al.
(2008) are represented by dark gray crosses. Sample 91-71-16
(white x) was measured by Clark et al. (1998), but was correlated with
the 1723 eruption by Benjamin et al. (2007). a CO2 versus H2O.
b Chlorine versus H2O. c Fluorine versus H2O
c
500
1000
1500
2000
2500
0 1 2 3 4 5500
1000
1500
2000
2500
3000
0
100
200
300
400
500
Benjamin et al. 2007 (1723)
Clark et al. 1998 (1963 or 1723)
Benjamin et al. 2007 (1963)
Sadofsky et al. 2008 (1963)
CO
2 (pp
m)
Cl (
ppm
)F
(pp
m)
H2O (wt. %)
Contrib Mineral Petrol (2009) 157:135–145 137
123
Results
Sixty-eight analyses of H, F, and Cl for 1723 and 1963
Irazu apatites are listed in Table 1, with uncertainties.
Ideally, as OH, F, and Cl all substitute into the same site in
apatite, we would expect that the sum of OH, Cl, and F
(molar) would be constant, and equal to the theoretical
concentration of ions in that site, which is the case within
the error on the individual analyses (average uncertainties
are *13% for OH, *8% for F, and 11% for Cl, two
standard errors of the mean). All grains were homogeneous
within error except one crystal from each eruption descri-
bed in more detail below.
Hydroxyl concentrations in 1963 apatites range from
500 to 2,800 ppm with a mean of 2,200 ppm, while Cl
varies from 0.5 to 2.0 wt %. Apatites from the 1723
eruption display an overlapping range of Cl contents (0.7–
1.5 wt %), but have a significantly higher mean (0.9 wt %
for 1723 versus 0.7 wt % for 1963), and are relatively
enriched in OH (range of 2,700–4,400 ppm, mean
*3,700), in comparison to the 1963 apatites. Fluorine
concentrations in 1963 apatite range from 2.7 wt% to
values within error of pure fluorapatite (3.8 wt %), while F
concentrations in 1723 apatites are considerably lower
(2.3–3.2 wt %), as would be expected considering their
higher OH and Cl contents.
Two of the apatites analyzed (1723–1725 and 1963–1965)
have internal volatile variations that are outside of analytical
uncertainties: Grain 5 from the 1723 eruption has an OH- and
Cl-poor region (OH = 2,750 ppm; Cl = 1 wt %) with a
relatively enriched neighboring region (3,900 ppm OH,
1.3 wt % Cl). However, because this is an anhedral fragment,
we cannot constrain the relative position of the two zones.
The coincidentally named grain 5 from the 1963 eruption,
however, is euhedral and has a core to rim trend of decreasing
Cl (2.0–1.1 wt %) and increasing OH (500–900 ppm)
defined by four analyses. The significance of these grains is
discussed further below.
Discussion
Timescales of apatite homogenization
The preservation of heterogeneous populations and hetero-
geneous crystals (with respect to H, Cl, and F) implies that
these crystals did not coexist for geologically long periods
of time, because the rapid diffusion of F, Cl, and H in
apatite (Brenan 1993) would cause all the apatite to
equilibrate with the host melt (and each other). Even an
apatite crystal armored by inclusion inside an olivine host
would equilibrate internally, and we would not observe
heterogeneous crystals. Consider as an example a very
large (500 lm length in the C-axis) crystal, in a melt at
1,000�C. If the melt volatile contents change (e.g., during
an exsolution event), the original core H, F, and Cl
chemistry in this very large crystal will be 10% compro-
mised in less than 0.5 years, and 90% reset to the new
equilibrium concentrations in \3 years (Brenan 1993). It
should be noted that these diffusion timescales are effec-
tively maxima, as we have chosen the slowest diffusion
coefficients available (1 atm data of Brenan), a large grain
size, and a temperature lower than any calculated by
Benjamin et al. (2007) or Alvarado et al. (2006) for these
magmas. The variations observed in crystal 1963–1965
require a slightly different set of boundary conditions, but
are amenable to a similar analysis. If we hypothesize that
the observed variations are due to diffusive re-equilibration
with a different or modified melt, and not growth, then we
can use the form of the zoning and the observed crystal size
to attempt to constrain the total time between the change in
volatile chemistry of the host magma and its subsequent
eruption. Again assuming a magma temperature of
1,000�C, we can calculate a maximum diffusive timescale
of *170 days by assuming a pure end-member chlorapa-
tite starting composition, and a boundary condition equal to
the mean Cl concentration of the other 1963 apatites. These
timescales decrease to less than 10 days if we choose the
higher pressure diffusion coefficients (Brenan 1993). While
these are only order-of-magnitude estimates, with no con-
trol on initial conditions, and constrained by few analyses,
the existence of heterogeneous crystals clearly indicates
short residence times.
H, F, and Cl in apatite and MI
The pattern of the main trend of Cl versus OH variations
observed in apatite (Fig. 2) is at first glance quite similar to
the Cl versus H2O data of Benjamin et al. (2007; Fig. 1).
As is the case with the MI data, 1723 apatites are consi-
derably more enriched in H and Cl than the 1963 eruptive
products (with the exception of 1963–1965, which will be
discussed in detail below). This qualitative agreement in Cl
versus H between MI and the main trend of the apatite data
is encouraging with regard to the possibility that apatite
volatile concentrations could serve as proxies for pre-
eruptive melt concentrations. Quantitative magma volatile
barometry using apatite requires experimental constraint of
apatite/melt partition coefficients for these particular melts,
which should be a priority for future research efforts. The
relationship between F and OH is not similar to the F and
H2O data from the MI, although there is a steep decline in
F at low OH level (corresponding to high Cl in crystal
1963–1965) that is at least reminiscent of the F–H2O pat-
tern observed in the MI. However, it should be noted that
perfect agreement would not be expected, because
138 Contrib Mineral Petrol (2009) 157:135–145
123
Table 1 SIMS data for H, C, F, and Cl in apatite from the 1723 and 1963 eruptions of Irazu
Analysis Age Grain OH (ppm) dOH (2SE) F (ppm) dF (2SE) Cl (ppm) dCl (2SE) % Occupancy
IZ1963-20 1963 1 2,567 278 27,489 1,593 5,373 567 88
IZ1963_g1-1 1963 1 2,105 235 28,388 1,705 6,108 629 90
IZ1963_g1-2 1963 1 2,169 237 30,397 1,872 6,423 692 96
IZ1963_g1-3 1963 1 2,236 243 31,440 1,911 6,431 685 99
IZ1963_g1-4 1963 1 2,324 247 31,527 1,868 6,430 665 100
IZ1963_g1-5 1963 1 2,390 254 30,762 1,789 6,196 706 98
IZ1963_g1-6 1963 1 2,433 258 30,752 1,731 6,108 611 98
IZ1963_g1-7 1963 1 2,451 256 31,151 1,783 6,126 588 99
IZ1963_g1-8 1963 1 2,464 258 31,350 1,795 6,252 596 99
IZ1963_g1-9 1963 1 2,519 262 31,468 1,958 6,182 580 100
IZ1963_g1-10 1963 1 2,555 283 31,182 1,953 6,177 622 99
IZ1963_g1-11 1963 1 2,548 269 31,140 1,918 6,265 615 99
IZ1963_g1-12 1963 1 2,580 270 31,445 1,819 6,097 660 100
IZ1963_g1-13 1963 1 2,586 268 31,308 1,861 6,350 603 100
IZ1963_g1-14 1963 1 2,584 274 30,854 1,777 6,259 626 98
IZ1963_g1-15 1963 1 2,551 268 31,172 1,932 6,070 638 99
IZ1963_g1-16 1963 1 2,561 275 30,848 1,810 6,023 618 98
IZ1963_g1-17 1963 1 2,544 268 30,785 1,783 6,153 579 98
IZ1963_g1-18 1963 1 2,546 260 30,204 1,781 5,901 618 96
IZ1963_g1-19 1963 1 2,580 273 29,701 1,693 5,904 638 95
IZ1963_g1-20 1963 1 2,605 278 30,377 1,759 5,905 647 97
IZ1963_g1-21 1963 1 2,696 275 31,822 1,866 6,221 666 101
IZ1963_g1-22 1963 1 2,704 280 32,745 1,934 6,374 681 104
IZ1963_g1-23 1963 1 2,729 283 32,390 1,884 6,312 598 103
IZ1963_g1-24 1963 1 2,767 282 31,466 1,791 6,238 721 101
IZ1963_g1-25 1963 1 2,751 283 30,844 1,817 6,007 608 99
IZ1963_g1-26 1963 1 2,754 291 30,899 1,787 5,865 614 99
IZ1963_g1-27 1963 1 2,741 291 30,638 1,747 5,992 630 98
IZ1963_g1-28 1963 1 2,773 279 30,871 1,858 5,899 610 99
IZ1963_g1-29 1963 1 2,788 294 30,917 1,775 5,912 622 99
IZ1963_g1-30 1963 1 2,782 305 30,950 1,769 5,900 664 99
IZ1963-17 1963 2 1,527 246 32,551 2,908 5,396 740 99
IZ1963-18 1963 2 1,508 247 33,700 3,169 5,592 770 102
IZ1963-19 1963 3 1,336 237 33,941 3,097 5,457 617 102
IZ1963-12 1963 5 500 263 32,527 3,430 19,884 2,457 116
IZ1963-09 1963 5 582 264 32,103 3,157 19,121 2,324 114
IZ1963-13 1963 5 600 207 29,749 2,567 15,851 1,707 103
IZ1963-14 1963 5 872 217 30,840 2,611 11,671 1,233 101
IZ1963-10 1963 6 1,087 280 38,540 3,861 6,414 939 115
IZ1963-11 1963 7 1,692 306 35,570 3,828 7,610 1,026 110
IZ1723g01-5 1723 1 3,622 423 28,996 2,739 8,647 1,041 100
IZ1723g01-6 1723 1 3,671 435 28,489 2,515 8,628 1,116 99
IZ1723g01-7 1723 1 3,741 437 29,268 2,615 8,452 982 101
IZ1723g01-8 1723 1 3,805 435 29,002 2,575 8,588 1,074 100
IZ1723g01-9 1723 1 3,747 433 28,831 2,593 8,312 1,053 99
IZ1723g01-10 1723 1 3,790 448 29,115 2,576 8,341 1,036 100
IZ1723g01-11 1723 1 3,776 444 29,024 2,581 8,274 1,107 100
IZ1723g01-12 1723 1 3,821 446 29,338 2,631 8,175 1,040 101
Contrib Mineral Petrol (2009) 157:135–145 139
123
F + Cl + OH are constrained in apatite by the stoichio-
metry of the mineral, whereas no such constraint acts on
F + Cl + H2O in the melt.
Modeling ternary melt chemistry from apatite
Fluorine, chlorine, and hydroxyl ions all occupy the same
site in apatite, nearly to the exclusion of other elements, so
we can plot apatite in the ternary system F–Cl–OH (Fig. 3).
Two trends are clearly defined: a low-OH trend and a high-
OH trend. The former is defined by the low-OH 1963
apatites and the core to rim variations in the strongly zoned
1963 crystal (1963–1965), while the latter is defined by the
1723 and (the majority of the) 1963 apatites. Because of
the relative age difference between 1723 and 1963 erup-
tions, one could argue that the majority of the apatite data
represent a trend of decreasing H2O and increasing F
content with time. The exception to this hypothesis is
found in the apatite crystal 1963–1965, which (based on its
core-to-rim stratigraphy) describes OH-poor volatile evo-
lution towards the most F-rich 1963 apatites. This pattern
may represent the convergence of these two trends (if the
eruption ages correlate in some way to the ages of the
apatites, which is plausible, but far from certain). These
compositions are consistent with the observations of
Alvarado et al. (2006) which support a model with as many
as three magmas or batches of magma with slightly dif-
ferent histories coexisting at shallow levels in the
magmatic system. We suggest that apatite is recording
mixing between at least one mafic melt and an evolved
component with high Cl/F and Cl/H ratios. The latter may
be a mostly crystallized remnant of the felsic magmas
described by Alvarado et al. (2006).
We can plot the MI from Benjamin et al. (2007),
Sadofsky et al. (2008), and Clark et al. (1998) in the ternary
H2O–F–Cl (Fig. 4). Analyses that plot in a scattered and
discontinuous fashion in Cartesian space now converge on
two distinctly linear trends, each comprised of all of the MI
from a given eruption, with no overlap. But making more
detailed interpretations regarding the relationship between
the two magmas or batches of magma is difficult given the
paucity of data. Later, we discuss how combining melt
inclusion and apatite volatile analyses may allow at least
some of the complexity of sub-volcanic systems to be
unraveled.
In order to compare the apatite and MI data sets directly,
we need to have an estimate of the functional relationship
between the volatile composition of the melt and the
volatile content of the apatite in equilibrium with that melt.
The relationship could be defined thermodynamically as a
Table 1 continued
Analysis Age Grain OH (ppm) dOH (2SE) F (ppm) dF (2SE) Cl (ppm) dCl (2SE) % Occupancy
IZ1723g01-13 1723 1 3,811 444 28,803 2,622 8,274 1,056 100
IZ1723g01-14 1723 1 3,782 438 29,299 2,685 8,267 1,052 101
IZ1723g01-15 1723 1 3,748 436 29,297 2,552 8,097 1,048 100
IZ1723g01-16 1723 1 3,812 436 29,190 2,652 8,278 1,038 101
IZ1723g01-17 1723 1 3,869 451 29,115 2,710 8,151 1,086 100
IZ1723g01-18 1723 1 3,795 444 29,056 2,679 7,978 956 100
IZ1723g01-19 1723 1 3,742 433 29,195 2,552 8,127 996 100
IZ1723g01-20 1723 1 3,755 443 29,352 2,581 8,223 1,026 101
IZ1723g01-21 1723 1 3,810 442 29,005 2,531 8,291 993 100
IZ1723g01-22 1723 1 3,870 446 29,563 2,581 8,258 1,055 102
IZ1723g01-23 1723 1 3,886 447 29,041 2,561 8,292 1,015 100
IZ1723g01-24 1723 1 3,917 449 29,396 2,654 8,364 1,010 102
IZ1723-14 1723 1 3,659 361 24,310 1,506 6,514 631 85
IZ1723-11 1723 2 3,190 355 28,468 2,550 7,359 845 95
IZ1723-12 1723 2 3,174 354 28,351 2,522 7,384 894 95
IZ1723-15 1723 3 3,189 323 24,952 1,509 7,128 682 86
IZ1723-08 1723 4 4,301 476 26,827 2,661 14,591 1,759 105
IZ1723-09 1723 4 4,427 481 25,040 2,171 13,522 1,516 99
IZ1723-10 1723 5 2,750 368 32,014 3,323 9,763 1,212 107
IZ1723-13 1723 5 3,855 406 23,511 2,256 13,051 1,521 92
Mean molar occupancy of X site 99.7%
Two standard deviations from the mean 10.3%
Mean propagated uncertainty into total from individual analyses 18.1%
140 Contrib Mineral Petrol (2009) 157:135–145
123
function of P and T (Piccoli and Candela 1994) or by
experimental determination for the melt composition of
interest (Mathez and Webster 2005). Whatever the theo-
retical basis for the relationship, if the ratio of apatite
composition to melt composition (for a given element) is a
constant value, it would generally be termed a ‘‘partition
coefficient’’, and we will use this term in the broadest sense
for the purposes of discussion in the paper (labeled as DF,
DCl, and DOH). Unfortunately, no partition coefficients
exist for the melt compositions observed for these Irazu
melts. Mathez and Webster (2005) studied a range of mafic
melt compositions, but none approximating the materials
represented in Table 1. However, they were able to deter-
mine that DF is approximately a factor of four higher than
DCl for fluorine-rich apatites, as would be expected. Con-
straint is even more tenuous for DOH, as there are zero
Cl (
0%)
Cl (
10%
)
Cl (
20%
)
Cl (
30%
)
Cl (
40%
)
F (60%)
F (70%)
F (80%)
F (90%)
F (100%)
OH (0%)
OH (10%)
OH (20%)
OH (30%)
OH (40%)
OH
ClF
1963
1723
Fig. 3 Apatite volatile measurements plotted in F–OH–Cl ternary
space, representing mole fraction occupancy of the X site in apatite.
Subplot to right of main ternary depicts region of full ternary space
(scaled to 40%) represented by the main diagram. Color scheme same
as in Fig. 2. Note that the data fall on two discrete linear trends,
converging at high-F compositions (lower left of diagram)
Cl (
0%)
Cl (
10%
)C
l (20
%)
Cl (
30%
)C
l (40
%)
Cl (
50%
)
F (0%)
F (10%)
F (20%)
F (30%)
F (40%)
F (50%)
H2O (50%)
H2O (60%)
H2O (70%)
H2O (80%)
H2O (90%)
H2O (100%)
H2O
ClF
1723 MI
91-71-16 MI
P2-72 MI
1963 MI
Fig. 4 Melt inclusion volatile measurements of Benjamin et al.
(2007) plotted in the ternary space F–H2O–Cl. Color scheme is that of
Fig. 1. Subplot to right of main ternary depicts region of full ternary
space (50% scaled) represented by the main diagram. Note the two
distinct trends which are not observable in Cartesian space
4000
6000
8000
10000
12000
14000
16000
18000
20000
22000C
l (pp
m)
1963A
1723
0 1000 2000 3000 4000 500020000
25000
30000
35000
40000
45000
OH (ppm)
F (
ppm
)
B
Fig. 2 Chlorine (A) and fluorine (B) plotted against OH, as measured
in this study in apatite phenocrysts from the 1723 (red squares) and
1963 (yellow diamonds) deposits. These apatite crystals were derived
from an aliquot of the same samples used in the study of Benjamin
et al. (2007). Error bars are two standard errors of the mean and
include calibration errors, which are considered conservative given
the reproducibility of measured ratios in standard materials is *3%.
All crystals were homogeneous within error, except for 1963–1965
and 1723–1725, which are joined by colored tie lines in all figures.
Crystal 1963–1965 is euhedral and has decreasing Cl towards the rim,
while 1723–1725 lacks the morphology to allow us to determine the
relative core and rim positions
Contrib Mineral Petrol (2009) 157:135–145 141
123
published measurements of H in coexisting apatite and
melt. The data of Mathez and Webster (2005) allows us to
estimate this value, if we are willing to make the
assumptions that the molar site occupancy in the X site in
apatite is 100% filled by F, Cl, and OH, and that the H2O in
the melt (quenched glass) is equivalent to the difference
between the electron probe totals and the theoretical sum of
100%. Given these assertions, we can calculate the H
partition coefficient by difference (with correspondingly
large uncertainties). This analysis tells us that the apatite/
melt partition coefficient for hydrogen is, in general,
approximately an order of magnitude lower than that of Cl.
However, this exercise falls well short of the goal of being
able to perform quantitative volatile barometry on Irazu
magmas.
An alternative to an independent determination of the
partition coefficients for F, Cl, and OH between apatite and
melt is to use the melt inclusion and apatite data here to
determine if the MI and apatite data sets are mutually
compatible for any constant DF, DCl, and DOH values.
Figure 5 shows the result of this exercise, with estimated
partition coefficients of DF = 15, DCl = 5, and
DOH = 0.15. We observe that the pattern of MI data is
compatible with the melts predicted to be in equilibrium
with the apatite. These estimated partition coefficients are
consistent with those calculated by simply dividing the
mean F, Cl, and OH in the 1723 apatite by the mean F, Cl,
and H2O in the 1723 melt inclusions (DF *15, DCl *5,
and DOH *0.15). Both data sets define two discrete, quasi-
linear trends, with the only significant differences being the
extension of the melt defined by apatite analyses into much
more Cl-rich model melt compositions, and the partial
overlap of 1963 and 1723 apatites. The first observation is
likely due to the assumption of a single partition coefficient
for all the apatites measured here, which is unlikely to be
true. As melt Cl/F ratios increase, there must be a point
where chlorapatite becomes the stable phase (or dominant
component in the case of complete solid solution) at the
expense of fluorapatite, and the partitioning relationship
between the melt and the apatite would be in the opposite
sense as it is for fluorine-rich apatites: Chlorine would be
more favorably partitioned into the chlorine-rich apatite,
and fluorine would be less favored. So we argue that the
high-chlorine compositions predicted by our modeling (up
to 40 mole% of the volatile components of the melt) are
maximum values necessitated by the paucity of experi-
mental data and reinforce the need for measurements on
apatite-bearing systems exploring a wide range of halogen
contents. The second observation, that 1963 and 1723
apatite populations lap over each other in ternary space,
while the MI data do not, is probably a function of sparse
data sets. We speculate that large numbers of apatite and
MI measurements would generate the same trends observed
here, but with more overlap highlighting the relationship
between the two magmas.
Combining apatite and MI data predicts considerably
higher partition coefficients for F and Cl in these magmas
than the experiments of Mathez and Webster (2005) would
support, but because of the large compositional differences
between Irazu major element compositions and the experi-
mental compositions of Mathez and Webster (2005) we
consider the change in partition coefficients unsurprising.
For our purposes, the absolute values are not a serious
concern, because the goodness of agreement of the two
data sets in ternary space is only a function of the relative
partition coefficients. However, the DF, DCl, and DOH
values used here are consistent (by definition) with the
mean ratios of apatite and melt compositions, and predict
mean melt compositions that agree within a few percent.
Apatite and MI data sets yield a compatible pattern of
volatile data, suggesting that apatite can be used as a
supplement for MI when relative concentrations are suffi-
cient to solve the geologic problem of interest. Apatite will
also be useful in cases where melt inclusions are rare or not
present in the geologic record. Further experimental
determination of the relationship between melt chemistry
and apatite chemistry is essential for quantitative volatile
barometry.
Cl (
0%)
Cl (
15%
)C
l (30
%)
Cl (
45%
)C
l (60
%)
Cl (
75%
)
F (0%)
F (15%)
F (30%)
F (45%)
F (60%)
F (75%)
H2O (25%)
H2O (40%)
H2O (55%)
H2O (70%)
H2O (85%)
H2O (100%)
ClF
1963 apatite1723 apatite
DF = 15
DCl = 5
DOH = 0.15
1723 MI
1
2
91-71-16 MI
P2-72 MI
1963 MI
H2O
Fig. 5 Melt inclusion data (as in Fig. 4) plotted with melt volatile
compositions predicted from apatite compositions (Fig. 3) using
estimated partition coefficients: DF = 15, DCl = 5, and DOH = 0.15.
Subplot to right of main ternary depicts region of full ternary space
(75% scaled) represented by the main diagram. Note that the pattern
of apatite and MI data are quite similar, suggesting that the two data
sets record compatible volatile histories. Ellipses approximate Trends
1 and 2 as defined in the text, with arrows showing relative time
progression. Analyses of crystal 1963–1965 define a core to rim
stratigraphy towards decreasing chlorine
142 Contrib Mineral Petrol (2009) 157:135–145
123
Magma evolution at Irazu
We consider even the qualitative level of agreement
observed here between apatite and MI somewhat surpri-
sing. Slow diffusion and low solubility of many elements in
olivine causes olivine-hosted MI to re-equilibrate slowly
with their host magma, making them useful for studying
the most primitive melts in a magmatic system (Anderson
1974, and others), although the fidelity of melt inclusions is
not guaranteed in all cases (Danyushevsky et al. 2002).
Volatile diffusion in apatite, as previously discussed, is
relatively fast, and apatites should equilibrate with their
host melt given a few years worth of time. So we should
expect that apatite is recording the ‘‘latest’’ stage magma
volatile content, as opposed to a more primitive melt vol-
atile content that might be preserved in MI. In the case of
zoned apatites, the most recent volatile information is
present at the edges of the crystals. The fact that the
measured and modeled melt volatile patterns agree means
that volatile ‘‘character’’ of the magma could not have
changed significantly in the time between MI entrapment
and apatite equilibration, and that both must have occurred
within a few years of eruption (or olivine- and cliopyro-
xene-hosted MI are also exchanging with the melt over
similar timescales).
It is safe to say that whatever volatile variations are
recorded by the different apatite crystals from Irazu must
have been present immediately prior to eruption. If these
variations are observed among the MI from the same
eruption, then the variations must have persisted from the
time of MI entrapment to the time of apatite equilibration
prior to eruption. However, MI volatile trends have been
interpreted as a path, which implies a temporal connection
between MI. If this is true, and the apatite is recording this
same path, then the timescale of the ‘‘travel’’ along this
path must be on the order of days to years, in order to
prevent the more Cl-rich apatites from equilibrating with
the host magma. This could be interpreted as placing loose
constraints on the significant major-element magma evo-
lution that occurred during this time (SiO2 increases from
47 to 56 wt % within the 1723 MI suite), as all timescales
that allow preservation of heterogeneous populations of
apatite are measured in days to years.
However, the heterogeneous volatile chemistry observed
in both the apatite and MI can be preserved over a longer
timescale if the data represent a mixture of separately
evolving magma bodies. Melt inclusions and apatite enri-
ched in Cl and H may source from a deeper or simply less
degassed batch of magma, while the Cl- and H-poor apa-
tites and MI come from a shallower or more degassed
magma body. Late stage mixing of these components
yields distinctive populations of apatite and MI that
resemble a degassing path, even though they equilibrated
with their host magmas at different places and over dif-
ferent timescales. This is consistent with the magma
mixing model of Alvarado et al. (2006), which would in
fact predict at least two populations of apatites in the 1963
and 1723 magmas, provided the timescales of mixing,
storage, and eruption were shorter than the timescales of
diffusive re-equilibration.
When we plot the MI data in H2O–F–Cl ternary space
(Fig. 4), it is clear that the same single process cannot be
responsible for the generation of the two discrete trends for
the 1723 and 1963 MI. Combining these data with the melt
compositions derived from apatite (Fig. 5) suggests that the
1723 and 1963 volatile histories overlap significantly.
Again, we suggest that it is more appropriate to divide the
data into populations based on the trend of the data, not just
the eruption age. Doing this results in two overlapping
subsets, each perhaps representing a discrete set of pro-
cesses in magma evolution: Trend 1 consists of increasing
F at the expense of H2O, with only a slight increase in Cl,
and contains all the 1723 MI, all the 1723 apatite, and the
more OH-rich 1963 apatite. This may represent mixing
between two components, or volatile evolution driven by
ascent, exsolution, or other processes. Trend 2 contains the
1963 MI, and the low OH 1963 apatite including 1963–
1965, and is defined by increasing Cl at the expense of
H2O, with only a slight increase in F. An important con-
clusion to draw from this is that Trend 1 apatites appear in
both 1723 and 1963 eruptions, confirming that the two
eruptions are likely the products of related magma cham-
bers, and placing constraints on the time–temperature
history of the apatite erupted in 1963, as such a heteroge-
neous population could not have survived a prolonged
magma residence.
In evaluating the possibility that multiple magma batches
or bodies were present at different depths at Irazu, we must
re-evaluate the saturation pressures calculated by Benjamin
et al. (2007) from the analyses of C and H in the 1723 MI
suite. Instead of using VolatileCalc (Newman and Lowen-
stern 2002), we use the solubility model of Papale et al.
(2006), as implemented by OFM Research (http://www.
ofm-research.org), which is more appropriate for melts of
these compositions. Recalculated saturation pressures
(Fig. 6) differ from those reported in Benjamin et al. (2007)
by -37% to +76% (-572 to +130 bars), resulting in a
considerably narrower range of model pressures for 1723
MI: (300–990 bars, as compared to the previous range of
170–1,560 bars). In addition, we can estimate model satu-
ration pressures for the 1963 MI for which there are CO2
and H2O measurements. Using the temperature range
(1,004–1,095�C) reported by Alvarado et al. (2006), we
calculate a model pressure range for three of the four 1963
MI of 73 ± 1, 294 ± 6, and 318 ± 83 bars (uncertainties
incorporate only variations in model pressures due to the
Contrib Mineral Petrol (2009) 157:135–145 143
123
range of temperatures, and are gross minimum estimates of
the total uncertainty).
These data do not provide a unique conclusion, but
certain aspects seem quite clear. For example, it is likely
that the eruptions sampled at least two, and possibly three
magma components: a deep mafic magma batch corre-
sponding to P [800 bars and SiO2 \52 wt%, a shallow
mafic magma batch corresponding to 300 B P B 450 bars
and 63 C SiO2 C52 wt% (which may correspond to the
low-P equivalent of the deep mafic melt), and a third felsic
component represented by just one MI at 73 bars
and [63% SiO2. The 1723 MI at P = 550 bars,
SiO2 = 55 wt% is somewhat enigmatic, and may represent
either a low-pressure version of the deep magma batch, or
the deepest sample recorded by the MI suite from the
shallow mafic batch, or it may simply be noise in the data.
However, it should be noted that the MI in question (IZ03-
17a-5 of Benjamin et al.) is the only MI that was large
enough for the authors to reproduce during their SIMS
session, and demonstrated excellent reproducibility. We
therefore conclude that analytical uncertainties are an
unlikely source of error in this case, although inaccuracies
in the saturation pressure model could explain the devia-
tion from the trend.
This sparse pressure data set is consistent with the
hypothesis that there were two batches of magma com-
prising the 1723 erupted products, but it is possible to draw
a single monotonic curve through the pressure-SiO2 data,
provided one is willing to invoke very large random errors,
on the order of 250 bars, or [3.5 wt% SiO2. If a single
polybaric magma model were preferred, or if the required
errors are reasonable (which is difficult to evaluate for
saturation pressures, which are derived from a model), then
one might suggest that the 1723 apatites and melt inclu-
sions represent a single (noisy) degassing path with a
period of isobaric evolution at a depth corresponding to
*300 bars (Fig. 6). The 1963 apatite and melt inclusions
can then be explained as a mixture of a Cl-rich felsic
component and a mafic component, likely a ‘‘leftover’’
1723 magma residing at a depth equivalent to *300 bars
of pressure, assuming saturation. This mafic magma pro-
vided the OH-rich apatites found in the 1963 eruption that
we suggest were in equilibrium with the 1723 MI (Fig. 5).
Conclusions
Apatite phenocrysts from the 1723 and 1963 eruptions of
Irazu volcano record the same volatile evolution history as
MI from the same units, confirming that apatite can be a
useful companion to MI studies. Not only are apatites
easier to locate and prepare than MI, they can provide
additional chronometric information by way of core-to-rim
relationships, diffusion profiles, and isotope geochrono-
logy. Experimental constraints on the partitioning behavior
of F, Cl, and H between apatite and melts will likely allow
apatite to function as a stand-alone magma volatile
barometer.
45 50 55
Benjamin et al. 2007 (1723)
SiO2 (wt. %)60 65
0
200
400
600
800
1000
P (
bar)
Benjamin et al. 2007 (1963)
Fig. 6 Pressure—SiO2 diagram derived from melt inclusion data of
Benjamin et al. (2007), but with CO2–H2O saturation pressures
recalculated using the model of Papale et al. (2006). Data may be
consistent with a single degassing path with a period of isobaric
evolution at a depth corresponding to *300 bars (broad gray path),
or may represent discrete low and high pressure batches of magma
that mix at a late stage in the magmatic history with a third, felsic
magma at very low pressure (thin dark gray paths)
Table 2 Whole rock compositional data, reproduced from Benjamin
et al. (2007), Sadofsky et al. (2008), and Clark et al. (1998)
Sample IZ03-17a IZ03-19b1 P2-72 91-71-16
Rock type Lapilli Bomb Ash
Age 1723 1963–1965 1963 1963 or 1723
UTM (E) 552.7 552.6 – –
UTM (N) 218.2 218.1 – –
Latitude – – 9.98072N –
Longitude – – 83.8537W –
SiO2 54.23 57.09 54.14 54.09
TiO2 1.2 0.99 0.94 1.15
Al2O3 17.34 16.83 16.06 17.72
Fe2O3 8.15 7.48 7.84 8.09
MnO 0.138 0.124 0.13 0.135
MgO 4.66 5.55 6.29 4.75
CaO 8.94 8.06 7.77 8.83
Na2O 3.5 3.41 3.13 3.4
K2O 2.02 1.97 1.98 2.2
P2O5 0.49 0.39 0.34 0.45
Total 100.67 101.89 99.01 100.82
144 Contrib Mineral Petrol (2009) 157:135–145
123
Integrating the updated saturation-model pressure esti-
mates with the volatile patterns observed in MI and apatite-
derived melt compositions elucidates the complex mag-
matic history of the 1723 and 1963 eruptions at Irazu. Melt
inclusion data indicate a wide range of entrapment pres-
sures for otherwise indistinguishable 1723 MI, implying
that they must have been trapped at markedly different
depths, and further suggest that these MI may correlate to
different batches of magma (or a multiply stagnated single
batch). These two different populations of 1723 MI may
also be reflected in the two different populations of apatite
observed for the 1723 eruption (Fig. 2). Apatite provides
additional information by demonstrating the existence of a
Cl-rich felsic component which explains the trend towards
higher Cl in the 1963 MI (Fig. 5). Additional MI and
apatite analyses, and refined CO2–H2O pressure models
should permit discrimination between single-magma,
multiple-stagnation models and models invoking multiple-
magma batches. Table 2.
Acknowledgments This project was funded by a National Science
Foundation MARGINS program Postdoctoral Fellowship (NSF-EAR-
0549082). The SIMS laboratory at Arizona State University is funded
by NSF EAR 0622775. Samples were generously provided by
T. Plank, and collected by E. Benjamin. Apatites used as standards
were provided by J. Hanchar, and S. Bergman. Many people con-
tributed to this project via hearty discussion, including attendees of
the SEIZE-SubFac Workshop 2007, and petrologists at ASU and
UCLA. The article benefited from constructive criticism from
J. Wade, G. Moore, M. Portnyagin, and an anonymous reviewer, and
we thank them for their efforts.
References
Alvarado G (1993) Volcanology and petrology of Irazu Volcano.
Costa Rica. Ph.D. thesis, Christian-Albrechts Universitat zu Kiel,
Kiel, p 261
Alvarado G, Carr MJ, Turrin BD, Swisher CC III, Schmincke H-U,
Hudnut KW (2006) Recent volcanic history of Irazu volcano,
Costa Rica: alternation and mixing of two magma batches, and
pervasive mixing. In: Rose WI, Bluth GJS, Carr MJ, Ewert J,
Patino LC, Vallance J (eds) Volcanic hazards in Central
America, vol 412, pp 259–276
Anderson AT (1974) Evidence for a picritic, volatile-rich magma
beneath Mt. Shasta, California. J Petrol 15:243–267
Benjamin E, Plank T, Wade J, Kelley K, Hauri E, Alvarado G (2007)
High water contents in basaltic magmas from Irazu Volcano,
Costa Rica. J Volcanol Geotherm Res 168:25
Boyce J, Hervig R (2008) Magmatic degassing histories from apatite
volatile stratigraphy. Geology 36(1):63. doi:10.1130/G24184A.1
Brenan J (1993) Kinetics of fluorine, chlorine, and hydroxyl exchange
in fluorapatite. Chem Geol 110:195–210. doi:10.1016/0009-2541
(93)90254-G
Clark SK, Reagan MK, Plank T (1998) Trace element and U-series
systematics for 1963–1965 tephras from Irazu Volcano, Costa
Rica: Implications for magma generation processes and transit
times. Geochim Cosmochim Acta 62:2689–2699. doi:10.1016/
S0016-7037(98)00179-3
Danyushevsky LV, McNeill AW, Sobolev AV (2002) Experimental
and petrological studies of melt inclusions in phenocrysts from
mantle-derived magmas; an overview of techniques, advantages,
and complications. Chem Geol 183(1–4):5–24. doi:10.1016/
S0009-2541(01)00369-2
Farley KA, Stockli D (2002) (U-Th)/He dating of phosphates: apatite,
monazite, and xenotime. Rev Mineral Geochem 48:559–577
Gleadow AGW, Belton DX, Kohn BP, Brown RW (2002) Fission
track dating of phosphate minerals and the thermochronology of
apatite. Rev Mineral Geochem 48:579–630
Hovis G, Harlov D, Hahn A, Seigert H (2007) Enthalpies and
volumes of F-Cl mixing in fluorapatite–chlorapatite crystalline
solutions: Geophys Res Abstr 9:01748
John T, Klemd R, Gao J, Garbe-Schonberg C-D (2008) Trace-element
mobilization in slabs due to non steady-state fluid–rock interac-
tion: constraints from an eclogite-facies transport vein in
blueschist (Tianshan, China). Lithos 103:1–24
Mathez E, Webster J (2005) Partitioning behavior of chlorine and
fluorine in the system apatite-silicate melt-fluid. Geochim Cos-
mochim Acta 69(5):1275–1286. doi:10.1016/j.gca.2004.08.035
Nadeau SL, Epstein S, Stolper E (1999) Hydrogen and carbon
abundances and isotopic ratios in apatite from alkaline intru-
sives, with a focus on carbonatites. Geochim Cosmochim Acta
63(11–12):1837–1851. doi:10.1016/S0016-7037(99)00057-5
Newman S, Lowenstern JB (2002) VolatileCalc: a silicate melt-H2O-
CO2 solution model written in Visual Basic for Excel. Comput
Geosci 28:597–604. doi:10.1016/S0098-3004(01)00081-4
Pan YM, Fleet ME (2002) Compositions of the apatite-group
minerals: substitution mechanisms and controlling factors. Rev
Mineral Geochem 48:13–49
Papale P, Moretti R, Barbato D (2006) The compositional dependence
of the saturation surface of H2O + CO2 fluids in silicate melts.
Chem Geol 229(1–3):78–95. doi:10.1016/j.chemgeo.2006.01.013
Parat F, Holtz F (2004) Sulfur partitioning between apatite and
melt and effect of sulfur on apatite solubility at oxidizing
conditions. Contrib Mineral Petrol 147:201–212. doi:10.1007/
s00410-004-0553-7
Peng G, Luhr JF, McGee JJ (1997) Factors controlling sulfur
concentrations in volcanic apatite. Am Mineral 82:1210–1224
Piccoli P, Candela P (1994) Apatite in felsic rocks: a model for the
estimation of initial halogen concentrations in the Bishop Tuff
(Long Valley) and Tuolumne intrusive suite (Sierra Nevada
batholith) magmas. AJS 294:92–135
Pyle JM, Spear FS, Rudnick RL, McDonough WF (2001) Monazite-
xenotime-garnet equilibrium in metapelites and a new monazite-
garnet thermometer. J Petrol 42:2083–2107. doi:10.1093/
petrology/42.11.2083
Sadofsky S, Portnyagin M, Hoernle K, Bogaard P (2008) Subduction
cycling of volatiles and trace elements through the Central
American volcanic arc: evidence from melt inclusions. Contrib
Mineral Petrol 155(4):433–456. doi:10.1007/s00410-007-0251-3
Schoene B, Bowring S (2007) Determining accurate temperature—
time paths from U–Pb thermochronology: an example from the
Kaapvaal craton, southern Africa. Geochim Cosmochim Acta
71(1):165–185. doi:10.1016/j.gca.2006.08.029
Spear FS, Pyle JM (2002) Apatite, monazite, and xenotime in
metamorphic rocks. Rev Mineral Geochem 48:293–335
Streck MJ, Dilles JH (1998) Sulfur evolution of oxidized arc
magmas as recorded in apatite from a porphyry copper batholith.
Geology 26:523–526. doi:10.1130/0091-7613(1998)026\0523:
SEOOAM[2.3.CO;2
Contrib Mineral Petrol (2009) 157:135–145 145
123