Earth and Planetary Science Letters 329–330 (2012) 109–121
Contents lists available at SciVerse ScienceDirect
Earth and Planetary Science Letters
journal homepage: www.elsevier.com/locate/epsl
The influence of magmatic differentiation on the oxidation state of Fe in a basaltic
arc magma
Katherine A. Kelley a,⁎, Elizabeth Cottrell b
a
b
Graduate School of Oceanography, University of Rhode Island, Narragansett Bay Campus, Narragansett, RI 20882, USA
Smithsonian Institution, National Museum of Natural History, Washington, DC 20560, USA
a r t i c l e
i n f o
Article history:
Received 27 August 2011
Received in revised form 13 February 2012
Accepted 14 February 2012
Available online xxxx
Editor: R.W. Carlson
Keywords:
oxygen fugacity
XANES
melt Inclusions
subduction
volatiles
degassing
a b s t r a c t
Subduction zone basalts are more oxidized than basalts from other tectonic settings (e.g., higher Fe 3 +/∑Fe),
and this contrast may play a central role in the unique geochemical processes that generate arc and continental
crust. The processes generating oxidized arc magmas, however, are poorly constrained, although they appear
inherently linked to subduction. Near-surface differentiation processes unique to arc settings might drive
oxidation of magmas that originate in equilibrium with a relatively reduced mantle source. Alternatively, arc
magmas could record the oxidation conditions of a relatively oxidized mantle source. Here, we present new
measurements of olivine-hosted melt inclusions from a single eruption of Agrigan volcano, Marianas, in order
to test the influence of differentiation processes vs. source conditions on the Fe3 +/∑Fe ratio, a proxy for system
oxygen fugacity (fO2). We determined Fe3 +/∑Fe ratios in glass inclusions using μ-XANES and couple these data
with major elements, dissolved volatiles, and trace elements. After correcting for post-entrapment crystallization,
Fe3 +/∑Fe ratios in the Agrigan melt inclusions (0.219 to 0.282), and their modeled fO2s (ΔQFM +1.0 to +1.8),
are uniformly more oxidized than MORB, and preserve a portion of the evolution of this magma from 5.7 to
3.2 wt.% MgO. Fractionation of olivine±clinopyroxene±plagioclase should increase Fe3 +/∑Fe as MgO decreases
in the melt, but the data show Fe3 +/∑Fe ratios decreasing as MgO decreases below 5 wt.% MgO. The major
element trajectories, taken in combination with this strong reduction trend, are inconsistent with crystallization
of common ferromagnesian phases found in the bulk Agrigan sample, including magnetite. Rather, decreasing
Fe3 +/∑Fe ratios correlate with decreasing S concentrations, suggesting that electronic exchanges associated
with SO2 degassing may dominate Fe3 +/∑Fe variations in the melt during differentiation. In the case of this
magma, the dominant effect of differentiation on magmatic fO2 is reduction rather than oxidation. Tracing back
Agrigan melts with MgO>5 wt.% (i.e., minimally degassed for S) along a modeled olivine fractionation trend to
primary melts in equilibrium with Fo90 olivine reveals melts in equilibrium with the mantle beneath Agrigan at
fO2s of ΔQFM +1 to +1.6, significantly more oxidized than current constraints for the mantle beneath midocean ridges.
© 2012 Elsevier B.V. All rights reserved.
1. Introduction
The availability of oxygen to participate in chemical reactions
within the Earth's mantle and magmatic systems (i.e., oxygen fugacity;
fO2) is a critical property that controls key igneous processes such as
the speciation and partitioning of multi-valent elements, phase
assemblages and equilibria, and the composition of volcanic gases.
The ability of lavas erupted at the surface to preserve accurate records
of the magmatic and source fO2 they experienced in the Earth's deep
interior, however, is a matter of significant debate (e.g., Ballhaus,
1993; Canil, 2002; Cottrell and Kelley, 2011; Kelley and Cottrell,
2009; Lee et al., 2005, 2010). The observation that basalts erupted at
⁎ Corresponding author. Tel.:+1 401 874 6838; fax: +1 401 874 6811.
E-mail addresses: [email protected] (K.A. Kelley), [email protected] (E. Cottrell).
0012-821X/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsl.2012.02.010
arc volcanoes are notably more oxidized than those erupted at midocean ridges is generally uncontested (e.g., Ballhaus, 1993;
Carmichael, 1991), but the cause of oxidation in the arc environment
remains an open and important question. One hypothesis invokes
oxidation of subduction zone mantle sources by oxidized fluids derived
from subducted slabs (e.g., Brandon and Draper, 1996; Wood et al.,
1990). The oxidation states of Fe in arc basalts have been quantitatively
linked to tracers of slab additions (H2O, Ba/La; Kelley and Cottrell,
2009) in support of this model. An alternate view, based on models
of redox-sensitive element partitioning or isotope fractionation during
mantle melting (e.g., Dauphas et al., 2009; Lee et al., 2005, 2010;
Mallmann and O'Neill, 2009), suggests that mantle source fO2 does
not vary with tectonic setting, and that arc basalts must thus become
oxidized as they differentiate along the path from the mantle source
to the Earth's surface. Few studies have yet explicitly measured the
effects of magmatic crystallization and degassing on the redox
conditions of arc magmas.
110
K.A. Kelley, E. Cottrell / Earth and Planetary Science Letters 329–330 (2012) 109–121
In theory, differentiation is a sensible process for accomplishing
magmatic oxidation. Early-crystallizing mafic minerals (e.g., olivine,
clinopyroxene) preferentially remove Fe 2 + into lattice sites, thereby
enriching the liquid in Fe 3 + if the magma is closed to oxygen
exchange with its surroundings. Mid-ocean ridge basalts (MORB)
show small increases in Fe 3 +/∑Fe ratio (i.e., Fe 3 +/[Fe 2 + + Fe 3 +])
with decreasing MgO, consistent with closed-system removal of
Fe 2 + by the crystallizing phase assemblage (Cottrell and Kelley,
2011). In MORB magmas, Fe 3 +/∑Fe ratios are observed and projected to increase by only 0.03 as MgO decreases from 10 to 5 wt.%
(Cottrell and Kelley, 2011). This is clearly insufficient to explain
the Fe 3 +/∑Fe ratios observed in arc lavas. In MORB systems, H2
degassing has also been proposed as a mechanism for “auto-oxidation” of lavas during eruption (Holloway, 2004), although at fO2 relevant for MORB (e.g., conditions of the quartz–fayalite–magnetite
buffer [QFM]) this process is too inefficient and self-limiting to
drive resolvable changes in MORB fO2 (Cottrell and Kelley, 2011).
The differentiation paths of arc magmas, however, are significantly
different from MORBs. Arc magmas are generally more volatilerich, and when they erupt, they are on average more evolved and
more degassed than MORBs. Yet, geochemical investigation of Mexican andesites and dacites indicates that crystallization and degassing cannot explain the elevated oxidation states of these magmas
either (Crabtree and Lange, 2011). These observations warrant explicit attention to the unique case of arc basalts which, in an
ocean–ocean convergent margin, offer the advantages of avoiding
the potentially complicated passage through thick continental
crust and provide magmas that are less differentiated from their
mantle source. How do these magmas acquire their oxidized
condition?
Here, we present a specific test of the alternate hypotheses of
source vs. differentiation as the cause of oxidation in basaltic arc
magmas by examining the magmatic redox conditions recorded
during differentiation of a single arc magma. Using melt
inclusions, trapped within olivine crystals over a period of time
in the magmatic cooling history, this study captures a significant
range of the liquid line of descent of one eruption of Agrigan volcano,
Marianas, recording both the effects of magmatic degassing and
crystallization processes on the composition and redox state of the
magma.
2. Samples and methods
2.1. Geologic setting and samples
Agrigan volcano is located in the central island providence of the
active Mariana island arc (see the electronic supplement). Trace
element and isotopic signatures of Agrigan lavas indicate significant
incorporation of subducted sediment into the mantle source beneath
Agrigan, making it an important end-member composition among
the Mariana islands (e.g., Elliott et al., 1997; Plank, 2005; Stern and
Ito, 1983; Woodhead, 1988, 1989). Studies of naturally-glassy
olivine-hosted melt inclusions from Agrigan also indicate high
concentrations of dissolved H2O in Agrigan magmas (~5 wt.%;
Kelley et al., 2010; Shaw et al., 2008), which indicate a strong subduction component. These clear slab-derived geochemical signatures,
and the absence of thick continental crust, make Agrigan an ideal
place to conduct this test because the path from the subductioninfluenced mantle wedge to the erupted tephra is more straightforward. The tephra sample selected for this study (AGR19-02) is rich in
euhedral olivine phenocrysts that contain abundant glassy melt inclusions. Inclusions selected for preparation were naturally glassy with no
visible secondary or synchronously trapped crystal phases, petrographically determined to be fully enclosed in the host olivine crystals, and contained either a single vapor bubble or no bubble.
2.2. Analytical methods
We analyzed glass inclusions and host olivines using a variety of
micro-analytical methods to determine major and trace element
composition of glasses and minerals, as well as dissolved volatile
concentrations and Fe 3 +/∑Fe ratios of glasses. At the Smithsonian
Institution, electron microprobe analysis provided major element, S,
and Cl data (Table 1) and Fourier transform infrared (FTIR) spectroscopy
provided dissolved volatile concentrations (Table 1). We determined
Fe 3 +/∑Fe ratios of glass inclusions using micro X-ray absorption
near-edge structure (μ-XANES) spectroscopy (Table 1; Cottrell et al.,
2009) at beamline X26A of the National Synchrotron Light Source,
Brookhaven National Lab, and Fe3 +/∑Fe of the whole rock at the
Smithsonian Institution via micro-colorimetry (Cottrell and Kelley,
2011). Trace element concentrations were determined using laserablation inductively-coupled plasma mass spectrometry (LA-ICP-MS;
Kelley et al., 2003) at the Graduate School of Oceanography, University
of Rhode Island. Details on sample preparation, analytical procedures,
and complete data tables are provided in the electronic supplement.
3. Results
3.1. Assessment of post-entrapment crystallization or modification
As a melt inclusion cools within its host olivine, some quantity of
the host mineral may precipitate from the melt onto the inclusion
walls after entrapment, and/or diffusive processes may drive
exchanges of major elements between an inclusion and the evolving
melt outside the host crystal. The melt inclusion data show remarkable
consistency with whole-rock compositions from Agrigan over a range
of MgO. Using the criteria of Danyushevsky et al. (2000), we find that
post-entrapment diffusive re-equilibration or disequilibrium modification, specifically Fe2 + loss from the inclusions, was minimal (see the
electronic supplement). We assess the extent to which postentrapment crystallization (PEC) may have modified glass inclusion
compositions by comparing the measured host olivine compositions
to the equilibrium olivine compositions predicted by each melt
inclusion, using KDol/liq(Fe 2 +/Mg) = 0.3 (see the electronic supplement).
In this case, the knowledge of the Fe 3 +/∑Fe ratio of each glass
inclusion offers a significant advantage, as this comparison is highly
sensitive to small extents of PEC. The inclusions are hosted by a
broad range of olivine compositions (Fo82-73), and extents of PEC
range from 0 to 3.3%. Many of the analyzed glass inclusions are in
near-perfect Fe–Mg exchange equilibrium with their hosts, and we
consider inclusions indicating b2% PEC as the most faithful records of
the magma composition. These inclusions were corrected for PEC, if
necessary, by assessing the olivine composition in equilibrium with
each inclusion using KDol/liq(Fe 2 +/Mg) = 0.3, then adding 0.1% of the
equilibrium olivine back to the glass composition, and repeating
these steps until equilibrium with the host olivine was reached,
assuming the total moles of Fe3 + in each inclusion remained
unchanged during these very minor extents of PEC (Table 1).
3.2. Fe speciation and oxygen fugacity of the Agrigan Magma
The Fe 3 +/∑Fe ratios of the Agrigan glass inclusions, after correction for PEC, range from 0.219 to 0.282. These ratios are uniformly
higher than in MORBs (0.16 ± 0.01; Cottrell and Kelley, 2011) and
the back-arc basin basalts from the Mariana trough (0.15–0.19;
Kelley and Cottrell, 2009), but overlap with the higher end of Fe 3 +/
∑Fe ratios reported for global arc basaltic melt inclusions (0.18–
0.32) by Kelley and Cottrell (2009). This contrast is broadly consistent
with past observations of whole-rock Fe 3 +/∑Fe ratios of basalts
from ridge and arc settings determined by wet chemistry, and spinel
compositions from peridotites (e.g., Ballhaus, 1993; Carmichael,
1991; Christie et al., 1986; Parkinson and Arculus, 1999; Wood et
Table 1
Major element compositions and Fe3 +/∑Fe ratios of olivine-hosted glass inclusions and host olivines from Agrigan, Marianas.
Sample
AGR19-02
Inclusion #
01
Olivine host
SiO2
FeO
TiO2
MnO
MgO
Cr2O3
NiO
Total
Fo
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
ppm
ppm
ppm
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
Post-entrapment corrected glass
Olivine added
SiO2
wt.%
wt.%
TiO2
Al2O3
wt.%
FeO*
wt.%
FeO
wt.%
Fe2O3
wt.%
MnO
wt.%
MgO
wt.%
CaO
wt.%
Na2O
wt.%
K2O
wt.%
wt.%
P2O5
Total
wt.%
H2O
wt.%
CO2
ppm
S
ppm
Cl
ppm
+3
Fe /∑Fe
Mg/[Mg+Fe*]
02
03
04
46.73
46.68
53.47
47.00
0.77
0.73
0.97
0.66
16.79
17.86
16.23
17.68
10.30
8.58
9.29
9.50
7.40
6.37
7.25
7.15
3.23
2.45
2.26
2.61
0.20
0.21
0.17
0.26
5.19
4.73
3.22
4.59
12.30
13.28
8.36
12.79
1.79
1.91
3.31
1.96
0.43
0.40
1.36
0.38
0.13
0.15
0.32
0.11
94.97
94.77
96.93
95.20
4.12
2.78
2.12
3.25
–
–
–
629
1474
1125
771
1170
727
770
1383
810
0.282
0.257
0.219
0.247
0.806
0.815
0.725
0.792
38.84
18.28
0.03
0.34
42.04
0.01
0.02
99.57
0.804
39.40
17.09
0.04
0.33
43.88
0.01
0.07
100.81
0.821
37.75
24.49
0.04
0.50
36.60
0.00
0.02
99.39
0.727
05
48.41
47.19
0.83
0.85
16.24
18.71
9.87
9.33
7.51
7.27
2.62
2.29
0.27
0.17
4.22
4.34
11.65
12.66
2.21
2.31
0.51
0.54
0.13
0.17
94.61
96.51
4.12
2.72
–
142
813
1404
940
767
0.239
0.220
0.769
0.780
39.30
17.19
38.41
21.08
0.33
42.60
0.41
39.49
0.04
99.45
0.815
0.02
99.41
0.770
0.0%
0.6%
0.1%
2.5%
46.73
46.64
53.46
46.81
0.77
0.72
0.97
0.64
16.79
17.75
16.21
17.25
10.30
8.63
9.30
9.72
7.40
6.43
7.27
7.43
3.23
2.44
2.26
2.55
0.20
0.20
0.17
0.25
5.19
4.96
3.26
5.51
12.30
13.20
8.35
12.47
1.79
1.90
3.31
1.92
0.43
0.39
1.36
0.37
0.13
0.15
0.32
0.11
94.97
94.80
96.93
95.31
4.12
2.76
2.12
3.17
–
–
–
614
1474
1118
771
1141
727
765
1382
790
0.282
0.254
0.219
0.236
0.47
0.51
0.38
0.50
07
38.33
18.07
0.03
0.33
41.23
0.01
0.05
98.05
0.803
0.1%
2.3%
48.40
47.01
0.83
0.84
16.22
18.29
9.89
9.56
7.53
7.55
2.62
2.23
0.27
0.17
4.25
5.18
11.64
12.37
2.21
2.26
0.51
0.53
0.13
0.17
94.61
96.59
4.12
2.66
–
139
812
1372
939
749
0.238
0.210
0.43
0.49
08
09
10
11
12Aa
12B
13
14
15
16
17
18
19
20
47.92
48.41
46.13
46.92
51.54
46.64
47.99
46.49
46.27
46.00
49.68
48.36
47.40
0.77
0.89
0.74
0.77
1.05
0.74
0.81
0.68
0.63
0.67
0.82
0.83
0.84
16.52
16.99
18.09
17.93
17.96
18.31
16.98
17.53
18.17
17.28
16.89
16.48
18.78
9.83
10.12
8.99
9.10
7.98
10.18
10.30
9.83
9.45
10.67
8.70
9.67
8.35
7.54
8.07
6.56
6.90
6.14
7.88
7.91
7.44
6.94
8.10
6.50
7.29
6.28
2.54
2.28
2.69
2.44
2.05
2.56
2.66
2.65
2.79
2.85
2.44
2.64
2.29
0.20
0.23
0.15
0.21
0.18
0.23
0.23
0.18
0.21
0.22
0.18
0.20
0.19
5.27
3.47
4.59
4.62
3.78
4.05
4.26
5.24
4.47
5.36
3.97
4.57
3.96
12.24
11.90
12.64
12.42
10.18
12.61
11.27
11.66
13.50
12.07
10.79
11.35
13.08
1.93
2.34
1.77
2.00
2.07
2.11
2.28
2.01
1.72
2.09
2.38
2.32
2.04
0.46
0.62
0.37
0.43
0.96
0.48
0.57
0.46
0.38
0.42
0.86
0.66
0.40
0.15
0.16
0.14
0.12
0.26
0.09
0.14
0.13
0.12
0.14
0.29
0.19
0.12
95.54
95.37
93.88
94.76
96.16
95.70
95.08
94.48
95.21
95.20
94.80
94.89
95.39
3.61
3.07
4.56
3.49
2.78
3.00
3.30
3.63
4.25
4.36
4.21
3.94
3.25
–
–
485
419
–
431
131
457
903
549
400
–
–
780
1050
1577
1117
1047
1453
1007
1182
1300
1517
1010
953
1089
843
1127
823
813
1167
807
970
793
803
770
1047
1033
690
0.233
0.203
0.270
0.241
0.231
0.226
0.232
0.243
0.266
0.240
0.252
0.246
0.247
0.806
0.719
0.806
0.799
0.786
0.753
0.762
0.807
0.793
0.797
0.784
0.789
0.789
39.21
17.95
0.04
0.36
42.80
0.01
0.05
100.42
0.810
38.69
22.33
39.62
17.11
39.77
43.99
0.01
100.80
0.760
0.05
100.77
0.821
39.25
17.73
0.02
0.35
43.24
0.01
0.04
100.64
0.813
38.77
19.67
39.24
20.58
38.91
21.37
40.92
41.37
40.54
0.03
99.39
0.788
0.03
101.22
0.782
0.03
100.84
0.772
0.4%
3.3%
1.6%
1.5%
0.0%
2.6%
47.89
48.09
46.02
46.81
51.51
46.43
0.77
0.87
0.73
0.76
1.05
0.73
16.45
16.45
17.81
17.66
17.92
17.85
9.86
10.56
9.12
9.24
9.00
10.47
7.58
8.57
6.74
7.07
6.92
8.22
2.53
2.21
2.65
2.40
2.30
2.50
0.20
0.22
0.15
0.20
0.17
0.22
5.42
4.57
5.20
5.18
4.13
4.96
12.19
11.52
12.44
12.24
10.16
12.29
1.92
2.27
1.74
1.98
2.07
2.06
0.46
0.60
0.36
0.42
0.96
0.47
0.15
0.15
0.14
0.12
0.26
0.09
95.55
95.51
93.98
94.84
96.17
95.81
3.59
2.98
4.49
3.44
2.78
2.92
–
–
477
412
–
420
777
1016
1552
1100
1045
1417
840
1091
810
801
1164
786
0.231
0.188
0.261
0.234
0.230
0.214
0.50
0.44
0.50
0.50
0.45
0.46
39.30
18.26
0.03
0.33
42.69
0.00
0.03
100.64
0.807
48.36
0.75
16.83
8.69
6.32
2.62
0.18
4.50
12.31
2.13
0.55
0.16
94.72
3.83
–
793
957
0.272
0.809
39.79
17.15
39.34
18.81
39.35
19.07
39.23
19.75
38.93
16.93
39.64
17.58
44.18
42.59
42.38
42.11
0.29
42.62
43.23
0.06
101.18
0.821
0.03
100.77
0.801
0.03
100.83
0.798
0.03
101.12
0.792
0.06
98.83
0.818
0.05
100.49
0.814
0.9%
0.0%
3.1%
0.4%
1.2%
0.3%
2.8%
47.91
46.49
46.06
45.97
49.55
48.33
47.17
0.80
0.68
0.61
0.66
0.81
0.83
0.82
16.83
17.53
17.63
17.21
16.69
16.43
18.27
10.40
9.83
9.71
10.70
8.82
9.69
8.62
8.03
7.44
7.28
8.14
6.66
7.32
6.61
2.63
2.65
2.71
2.84
2.41
2.64
2.23
0.23
0.18
0.21
0.22
0.17
0.20
0.19
4.57
5.24
5.61
5.51
4.42
4.68
5.01
11.17
11.66
13.09
12.02
10.66
11.32
12.73
2.26
2.01
1.67
2.08
2.35
2.31
1.98
0.57
0.46
0.37
0.42
0.85
0.66
0.39
0.14
0.13
0.12
0.14
0.29
0.19
0.11
95.13
94.48
95.35
95.22
94.86
94.90
95.52
3.27
3.63
4.12
4.34
4.16
3.93
3.16
130
457
876
546
395
–
–
998
1182
1261
1511
998
950
1060
961
793
779
767
1034
1030
671
0.228
0.243
0.251
0.239
0.245
0.245
0.233
0.44
0.49
0.51
0.48
0.47
0.46
0.51
0.5%
48.32
0.75
16.75
8.73
6.38
2.61
0.17
4.69
12.25
2.12
0.55
0.15
94.74
3.81
–
789
952
0.269
0.49
K.A. Kelley, E. Cottrell / Earth and Planetary Science Letters 329–330 (2012) 109–121
Glass inclusion
SiO2
TiO2
Al2O3
FeO*
FeO
Fe2O3
MnO
MgO
CaO
Na2O
K2O
P2O5
Total
H2O
CO2 b
S
Cl
Fe+3/∑Fe
Equil. Fo
(continued on next page)
111
K.A. Kelley, E. Cottrell / Earth and Planetary Science Letters 329–330 (2012) 109–121
0.57
1032
−6.76
1.54
−8.88
1.54
0.57
1057
−7.15
1.15
−8.89
1.16
10
Agrigan, Marianas
Jalopy Cone, Big Pine
Augustine, Aleutians
A
b
)
.2
-0
M
QF
=
0.
05
(~
4
/liq
V in Olivine, ppm
6
V
D ol
2
ol/liq
V
D
1
= 0.0
)
M+2.1
(~QF
0
0
50
100
150
200
250
300
350
400
V in Glass, ppm
0.35
0.30
+2 +2
Canil (2002)
Mallmann & O’Neill (2009)
Mixes V5, 8
Mallmann & O’Neill (2009)
Mixes V1, 7
+2
+1
+1
-0.5
-0.5
0.10
0
QF
5
B
-0.
0.15
M
QFM
0.20
+1
0.25
QF
M
Fe3+/ Fe
Inclusion 12A has been corrected for post-entrapment Fe-loss following the method of Danyushevsky et al. (2000).
A "–" indicates data were below detection limit.
8
a
19
18
0.53
1041
− 6.98
1.32
− 8.73
1.33
0.54
1035
− 6.96
1.34
− 9.03
1.34
17
16
0.55
1065
− 6.99
1.31
− 8.61
1.32
0.58
1048
− 6.89
1.41
− 8.76
1.42
15
0.50
1058
−7.12
1.18
−8.85
1.19
0.52
1077
−7.31
0.99
−8.77
1.00
14
13
12B
0.52
1027
−7.07
1.23
−9.25
1.23
0.57
1063
−7.09
1.21
−8.75
1.22
0.58
1061
− 6.75
1.55
− 8.44
1.56
0.56
1070
− 7.13
1.17
− 8.69
1.18
0.55
1086
−7.42
0.88
−8.75
0.89
Mg/[Mg + Fe ]
Olivine–liquid T
log fO2
ΔQFM
log fO2
ΔQFM
0.56
°C
1050
1 atm., 1200 °C −6.50
1 atm., 1200 °C
1.80
1 kb, ol-liq T
− 8.35
1 kb, ol-liq T
1.80
0.58
1063
−6.94
1.36
−8.60
1.37
0.44
1048
−7.28
1.02
−9.15
1.03
0.57
1083
−7.07
1.23
−8.45
1.24
0.50
1023
−7.05
1.25
−9.18
1.25
0.49
1068
−7.69
0.61
− 9.28
0.62
12Aa
11
10
09
08
07
05
04
03
02
01
Inclusion #
2+
AGR19-02
Sample
Table 1 (continued)
(continued)
al., 1990). Basalt Fe 3 +/∑Fe ratio may be translated into oxygen
fugacity (fO2) using the algorithm of Kress and Carmichael (1991),
which accounts for the effects of melt composition, pressure, and
temperature on the relationship between these two parameters.
Referenced at 1200 °C and 1 bar, the Agrigan magma indicates fO2
conditions from 1.0 to 1.8 log units above the quartz–fayalite–magnetite
buffer (QFM; Table 1).
Other geochemical indices are also sensitive to fO2. In particular,
the olivine–liquid partition coefficient for V has been shown to
depend strongly on fO2, due to the differences in incompatibility of
the V 5 +, V 4 +, and V 3 + species that make V more incompatible as
fO2 increases (e.g., Canil, 1997, 2002; Canil and Fedortchouk, 2001;
Mallmann and O'Neill, 2009). Fig. 1a shows how the concentration
of V in olivine vs. melt varies with tectonic setting, comparing
olivine–melt inclusion pairs from Agrigan, Marianas and Mt. St.
Augustine, Aleutians with olivine-inclusion or glass pairs from the
Basin and Range and MORB pillow glass from the East Pacific Rise
(see the electronic supplement). The values for DVol/liq vary
significantly among these four samples, from ~ 0.01 at Augustine to
0.05 at the mid-ocean ridge. The Fe 3 +/∑Fe ratios of these glasses
also co-vary with DVol/liq (Fig. 1b) along a trajectory most consistent
0.56
1068
−6.93
1.37
−8.51
1.38
20
112
0.02
0.04
0.06
0.08
0.10
DV ol/liq
Fig. 1. Vanadium partitioning between olivine and glass as a function of fO2. (a) Vanadium
concentration in glass vs. V concentration in olivine for four global basalts. Lines show
constant values of DVol/liq that bracket the range observed in the basalts. Each line is labeled
with the approximate corresponding fO2 given by the model of Canil (2002). See the
electronic supplement for detailed information on the samples and data shown. (b) DVol/liq
vs. Fe3 +/∑Fe ratios of natural basalt glasses. Curves shown are modeled using relations
between DVol/liq and fO2 reported by Canil (2002) and Mallmann and O'Neill (2009), in
combination with the major element composition of inclusion AGR19-02-01, P=1 atm,
T=1200 °C to translate Fe3 +/∑Fe ratio into fO2 (in log units relative to the QFM buffer)
using the algorithm of Kress and Carmichael (1991). The model curves thus apply
exclusively to this one composition, although the curves will be similar for the range of
basalt compositions shown. Tick marks for fO2 are therefore considered approximate, and
should be referenced with appropriate caution. Gray shading represents a confidence
envelope for the Canil (2002) model curve that accounts for analytical uncertainty in DVol/liq
(see the electronic supplement). A representative error bar is shown for reference.
K.A. Kelley, E. Cottrell / Earth and Planetary Science Letters 329–330 (2012) 109–121
2.
5
A
2.0
kb
CO2, ppm
800
1.5
k
600
kb
3.5
Co-variations of major elements and dissolved volatiles in the
Agrigan glass inclusions indicate that the magma experienced synchronous crystallization and degassing prior to eruption. Degassing
is assessed by comparing co-variations of dissolved volatiles of differing
vapor/melt solubility. The solubility of CO2, for example, is highly
pressure-dependent and its solubility in basaltic melt decreases significantly with decreasing pressure (Dixon et al., 1995). At crustal
pressures, CO2 solubility is much lower than that of H2O, such that
CO2 will preferentially be removed to the vapor phase until most of
the dissolved CO2 is removed from the degassing melt. Glass inclusions
from Agrigan show a trend in H2O vs. CO2 consistent with closedsystem degassing of a magma with initial H2O content of ~4.5 wt.%
(Fig. 2a). This magmatic H2O content is lower than has been shown
for other Agrigan magmas (up to 5.5 wt.%; Kelley et al., 2010; Shaw et
al., 2008), although the maximum CO2 and H2O of each magma points
to a saturation pressure of ~3 kb, suggesting a common magma storage
depth of about 10 km in the crust beneath Agrigan. The melt–fluid
partitioning of S in basaltic melt is still poorly constrained experimentally, but depends on P, T, fO2, fS2, and melt composition. Empirical
co-variations of H2O and S in arc magmas suggest that melt–fluid partitioning of S may be intermediate between CO2 and H2O in mafic arc
magmas (Kelley et al., 2010; Sisson and Layne, 1993; Wade et al.,
2006) and ranges from 5 to 100, in general agreement with partitioning
experiments (Webster and Botcharnikov, 2011). The AGR19-02 magma
shows S concentrations broadly decreasing with H2O, consistent with
trajectories shown by other Agrigan magmas, in support of the interpretation that S and H2O degassed together (Fig. 2b).
Major element concentrations in Agrigan glass inclusions also
show evidence of multiphase fractional crystallization of olivine ±
clinopyroxene ± plagioclase. The AGR19-02 magma reaches
saturation with plagioclase at ~5 wt.% MgO, at which point Al2O3
kb
3.0
3.3. Degassing and crystallization of the Agrigan Magma
1000
kb
This Study
Kelley et al., 2010
Shaw et al., 2008
Closed System
Open System
b
1.0 k
b
400
0.5 kb
200
0
1800
B
1600
1400
1200
S, ppm
with the modeled relationship between DVol/liq and fO2 from Canil
(2002), which is derived from mineral/melt partitioning experiments
performed using basaltic and komatiitic compositions. More recent
partitioning experiments (Mallmann and O'Neill, 2009) define curves
of similar functional form, but reveal a significant compositional
dependence to DVol/liq as a function of fO2. These experiments,
however, are based on unusual, synthetic melt compositions that
are unlike most terrestrial basalts, and thus may not provide the
optimal reference for natural basalts on Earth. An error analysis
accounting for analytical uncertainty in V concentration in glass and
olivine reveals that uncertainty in the Canil (2002) model may be as
much as ±0.25 log units of fO2 in the range of QFM + 1 to 1.5, and
this uncertainty also increases as fO2 increases (see the electronic
supplement). More importantly, however, the determinations of DVol/
liq
and Fe 3 +/∑Fe ratio are not in perfect agreement with any of the
three models shown on Fig. 1b. When referenced to the model of
Canil (2002), for example, DVol/liq values for individual samples
disagree with Fe 3 +/∑Fe ratios by as much as Fe 3 +/∑Fe = 0.05
(corresponding to 0.7 log units in fO2), which we consider the
practical uncertainty for using DVol/liq as a proxy for fO2. Moreover,
above Fe3 +/∑Fe ~0.25 (DVol/liq b 0.02), the dynamic range in DVol/liq is
small relative to the possible variation in Fe3 +/∑Fe, such that this
method loses resolution for the most oxidized melts. As such, compared
to Fe3 +/∑Fe ratio, DVol/liq provides a relatively coarse, though useful,
index of magmatic redox conditions, and may prove particularly informative at low fO2 where the uncertainty in DVol/liq is smaller and uncertainties in Fe3 +/∑Fe ratio become larger (Cottrell et al., 2009). The
broad agreement between DVol/liq and Fe3 +/∑Fe ratios of these global
basalts further supports the view that the Fe speciation of these melt
inclusions did not change on rapid time scales (faster than V could
diffusively reequilibrate in olivine) and is therefore a faithful record of
magmatic fO2.
113
1000
800
600
400
200
0
0
1
2
3
4
5
6
H2O, wt.%
Fig. 2. Dissolved volatile elements in olivine-hosted glass inclusions from Agrigan,
Marianas, comparing data from this study (shaded circles) with those of prior work
from Shaw et al. (2008; open diamonds) and Kelley et al. (2010; filled crosses).
(a) Plot of H2O vs. CO2 concentrations. Degassing curves and isobars were calculated
at 1100 °C using VolatileCalc (Newman and Lowenstern, 2002). Initial conditions for
the open-system degassing curve are 1000 ppm CO2 and 4.5 wt.% H2O, and the
closed-system curve assumes 2% vapor exsolved. A representative error bar is shown
for reference. (b) Plot of H2O vs. S concentrations. Data show a broad trend towards
coincidentally decreasing H2O and S concentrations during degassing.
begins to decrease and TiO2 begins to increase with decreasing MgO
(Fig. 3a–b). The point of plag-in for this magma may be slightly earlier
than other Agrigan magmas due to its lower H2O content (Fig. 3a;
Kelley et al., 2010; Shaw et al., 2008). At >5 wt.% MgO, the range in
data are limited, and are consistent with saturation of either
olivine-only or olivine + cpx. A clear change in the redox conditions
of the magma is also evident with fractional crystallization (Fig. 3c–
f). As MgO of the inclusions decreases, the Fe 3 +/∑Fe ratios clearly
decrease from a maximum of 0.282 (average 0.25 for all inclusions
with >5 wt.% MgO) to a minimum of 0.219 at ~3.2 wt.% MgO. To
translate these ratios into oxygen fugacity at magmatic conditions,
we assume a mean magmatic pressure of 1 kb, calculate magmatic
temperatures using olivine–liquid thermometry (Médard and Grove,
2007; Putirka et al., 2007), and use the algorithm of Kress and
Carmichael (1991) to calculate fO2 in log units relative to the
quartz–fayalite–magnetite buffer (ΔQFM) at the magmatic conditions
for each inclusion (Table 1). These calculations indicate that magmatic
fO2 conditions decreased from QFM + 1.8 to QFM + 1.0 during the
differentiation period preserved by these glass inclusions.
4. Discussion
The Fe 3 +/ΣFe ratios of the Agrigan glass inclusions decrease
significantly during the recorded period of differentiation, showing
that the magma experienced reduction, not oxidation, by magmatic
processes prior to eruption. In the case of this arc magma, the most
mafic inclusions are the most oxidized. In the sections that follow,
114
K.A. Kelley, E. Cottrell / Earth and Planetary Science Letters 329–330 (2012) 109–121
14
22
This Study
Kelley et al., 2010
Shaw et al., 2008
AGR19-02 WR
Agrigan WR
Olivine only
Multi-phase
Multi-phase + mag.
Al2O3, wt.%
20
19
12
18
17
11
10
9
8
16
A
15
7
6
14
1.2
10
B
1.1
E
9
1.0
8
FeO, wt.%
TiO2, wt.%
D
13
FeO*, wt.%
21
0.9
0.8
7
6
0.7
5
0.6
0.5
4
4.0
C
0.28
F
3.5
Fe2O3, wt.%
Fe3+/ Fe
0.24
0.20
0.16
3.0
2.5
2.0
1.5
Mariana Trough, KC09
MORB, CK11
1.0
0.12
3
4
5
6
7
MgO, wt.%
8
3
4
5
6
7
8
MgO, wt.%
Fig. 3. Major element variations and liquid lines of descent (LLD's) for olivine-hosted glass inclusions from Agrigan, Marianas, comparing data from this study (shaded circles) with
those of prior work from Shaw et al. (2008; open diamonds) and Kelley et al. (2010; filled crosses). Also shown are Mariana arc whole-rock data from Elliott et al. (1997; crossed
squares), and the whole-rock composition of weathered matrix fragments from the AGR19-02 tephra (crossed circle). Note that the glass inclusions are normalized to anhydrous
compositions for comparison with the whole-rock data. Model curves trace trajectories of olivine-only fractional crystallization (thin solid line), multiphase fractionation of ol ± cpx
± plag (dotted line), and multiphase fractionation of ol + cpx + plag + magnetite (thick gray line). Multiphase crystallization was modeled using Petrolog3 (Danyushevsky and Plechov, 2011) at 1 kb, assuming a system closed to oxygen exchange and using the mineral solution models of Ariskin and Barmina (1999) and Danyushevsky (2001). Plots of MgO vs.
(a) Al2O3, (b) TiO2, (c) Fe3 +/∑Fe ratio, (d) FeO* (where * indicates total Fe expressed as FeO), (e) FeO (i.e., the true Fe2 + concentration expressed as FeO), and (f) Fe2O3. Error bars
are smaller than the symbol size, except where shown.
we explore several possible explanations for these observations and
their consequences for the oxygen fugacity of the Agrigan mantle
source.
4.1. Magmatic reduction during differentiation
4.1.1. Magnetite fractionation
Perhaps the most likely potential process for accomplishing
magmatic reduction during arc basalt differentiation involves the
saturation and fractional crystallization of magnetite (Fe3O4) from
the magma. Magnetite is a common phase in arc basalts, and is
present as a phenocryst phase in sample AGR19-02. Because the Fe
in magnetite is mixed valence and dominantly ferric (2Fe3 +:1Fe2 +),
magnetite fractionation in a system closed to oxygen would decrease
both the Fe3 +/∑Fe ratio and the FeO* content of the melt through
the preferential removal of this Fe-rich mineral with Fe3 +/∑Fe=0.67.
The ability of fractional crystallization to accomplish magmatic reduction
in this way, however, requires the magma to be magnetite-saturated
over the recorded segment of the liquid line of descent. Fig. 3 shows
that, when examined together, FeO*, FeO, Fe2O3, and Fe3 +/∑Fe ratios
do not decrease with MgO in a manner consistent with models of magnetite fractionation, although there is some scatter to the data. Magnetite
fractionation is also expected to strongly decrease the TiO2 and V concentrations of the melt (e.g., Jenner et al., 2010), which is not observed
(Fig. 3b) and calls this potential explanation of the overall reduction
trend, starting at 5 wt.% MgO, into question.
In fact, late magnetite crystallization is expected from sample
AGR19-02, which is a hydrous magma that straddles the boundary
between calc-alkaline and tholeiitic magma series. Fig. 4 shows the
ternary AFM discrimination diagram (Irvine and Baragar, 1971),
which segregates tholeiitic and calc-alkaline magma series by
highlighting the point at which FeO* begins to decrease with
K.A. Kelley, E. Cottrell / Earth and Planetary Science Letters 329–330 (2012) 109–121
This Study
FeO*
AGR19-02 WR
80
Agrigan WR
O
Mg
to
N
to
a2 O
+K
2O
80
TH
CA
40
40
30
Fig. 4. A cropped ternary plot of total alkalis vs. FeO* vs. MgO (AFM). The dotted line is
the discriminating line between tholeiitic (TH) and calc-alkaline (CA) fields from Irvine
and Baragar (1971).. Data from Kelley et al. (2010), Kent and Elliott (2002), and Shaw
et al. (2008) are encompassed by the striped field for clarity. Note that the Kent and
Elliott (2002) data are crystallized inclusions that were re-homogenized at
fO2 = QFM, for which FeO* may have been perturbed by the homogenization process.
decreasing MgO and increasing alkalis during crystallization. Early
magnetite crystallization will drive FeO* down in the melt as fractionation progresses, driving basalt compositions into the calc-alkaline
field. Late magnetite crystallization, as in the case of AGR19-02, allows basalts to become relatively enriched in FeO*, keeping them in
the tholeiitic field above the curved discrimination line (Fig. 4). The
evidence for late magnetite saturation in this sample is thus not surprising, given that melt inclusion and whole-rock compositions for
Agrigan are mildly tholeiitic according to the definitions of Irvine
and Baragar (1971) or Miyashiro (1974). To be clear, however, calcalkaline and tholeiitic magma series are two end-members along a
compositional continuum, and assignment of either of these classifications to a given magma depends in part on the criteria used to
classify. For example, Zimmer et al. (2010) assign Agrigan a mildly
calc-alkaline affinity using their “Tholeiitic Index,” but show that its
position near their calc-alkaline/tholeiitic discriminating line is consistent with global arc magmas with dissolved H2O contents between
2 and 4 wt.%. Moreover, models of LLD's for oxidized (QFM + 1) and
variably water-rich (1–6 wt.% H2O) arc magmas, which are appropriate for the Agrigan composition, predict magnetite saturation at
~ 4 wt.% MgO (Zimmer et al., 2010), which is approximately where
we estimate it based on the melt inclusion compositions. We therefore conclude that magnetite fractionation cannot be responsible for
the overall reduction trend recorded by the melt inclusions.
115
The Agrigan data show that decreasing dissolved S concentrations,
indicative of S degassing, correlate with decreasing Fe 3 +/∑Fe ratios
in the glass inclusions (Fig. 5). Moreover, S concentrations in the melt
inclusions decrease with MgO, suggesting that coincident crystal fractionation and S degassing took place in this magma. The magma is not
sulfide-saturated, as the sulfur content at sulfide saturation at these
fO2s (~0.3–1.3 wt.%; Jugo et al., 2010) is significantly higher than
the measured S concentrations of the AGR19-02 melt inclusions
(b1600 ppm), and Cu concentrations are too high (90–157 ppm) for
a melt in equilibrium with sulfide (Jenner et al., 2010). We constructed a simple degassing model to assess whether enough electrons could be supplied by S to accomplish the magnitude of
reduction observed (Fig. 5). The model results suggest that S 2 − may
have lower vapor/melt solubility in this magma than does S 6 +. Although few quantitative constraints currently exist for vapor/melt
partitioning of different S species, this relative sense of solubility is
consistent with the known solubilities of S species in silicate melts.
At high fO2, S solubility in basalt increases significantly because S 6 +
is more soluble (Carroll and Rutherford, 1988; Jugo et al., 2010;
Wallace and Carmichael, 1994). In this highly simplified case, we
find that enough electrons could be supplied by the conversion of dissolved S 2 − to SO2 gas to accommodate the observed change in Fe speciation (Fig. 5). It is important to note, however, that this model
represents an oversimplification of natural systems, where Fe, S, and
other redox-sensitive elements will all respond to the changing electronic balance in the melt, so this example explores an end-member
case. Though the model is speculative by necessity, due to the paucity
of quantitative constraints on the vapor/melt partitioning of S species,
the trend shown by the data evokes a link between S degassing and Fe
redox in this magma.
4.2. Alternative hypotheses for differentiation-related
oxidation mechanisms
Of course, there is a chance that oxidation does happen during
differentiation, but our sampling of the Agrigan magma has failed to
capture the oxidizing segment of the LLD. Here we explore some
outstanding hypotheses for processes by which magmas could
0.30
This Study
0%
10%
0.25
20%
Fe3+/ Fe
4.1.2. Sulfur degassing
Another, perhaps more likely, candidate for changing magmatic
redox conditions without affecting FeO* concentration is magmatic S
degassing. The 8 electron difference between oxidized sulfate (S6 +)
and reduced sulfide (S2 −) species dissolved in basaltic melt gives sulfur great potential to drive changes in redox despite its comparatively
low abundance. Moreover, when S exsolves from magmas as a vapor
phase, its dominant species is SO2 gas (S4 +; Oppenheimer, 2003;
Wallace, 2005). The movement of S from its dissolved state in a basaltic
melt, where it speciates exclusively as either S6 + or S2 − (or as mixtures
of the two), to a vapor phase of dominantly S4 + thus requires electrons
in the melt to be redistributed during S degassing (Métrich et al., 2009).
Evidence of magmatic reduction accompanied by S degassing has been
found at Kilauea volcano (Anderson and Wright, 1972), although SO2
degassing could drive magmatic fO2 to either more reduced or more
oxidized conditions depending on the initial S speciation of the magma
(a function of fO2) and the relative vapor/melt solubilities of S6 + and
S2 − (Métrich et al., 2009).
30%
40%
50%
0.20
0.15
0
500
1000
1500
2000
S, ppm
Fig. 5. Plot of S vs. Fe3 +/∑Fe ratio in olivine-hosted glass inclusions from Agrigan,
Marianas, showing the redox change correlated with S degassing. The thin line with
tick marks is the S degassing model discussed in Section 4.1.2. Tick marks identify
percentages of S degassed from the magma. The model starts with initial conditions
of the magma at 9.64 wt.% FeO*, 1500 ppm S, Fe3 +/∑Fe = 0.28, and S6 +/∑S = 0.83
(S-speciation calculated from fO2 using Wallace and Carmichael (1994)). Sulfur is
assumed to enter the vapor in a S6 +/∑S ratio of 0.35, and all S entering the vapor
was assumed to convert to SO2 (S4 +) which assumes different vapor/melt solubilities
for S2 − and S6 +. Electrons left in the melt at each degassing step were completely
reassigned to Fe to accomplish reduction of the Fe3 +/∑Fe ratio.
116
K.A. Kelley, E. Cottrell / Earth and Planetary Science Letters 329–330 (2012) 109–121
become oxidized during differentiation, and how these relate specifically to the case of the Agrigan magma.
4.2.1. Crystallization of Fe 2 +-rich phases
Cottrell and Kelley (2011) showed that fractional crystallization of
ferromagnesian minerals causes slight oxidation of MORBs during
differentiation. Co-crystallization of Fe-poor plagioclase and Fe 2 +rich olivine and clinopyroxene slightly increase the melt Fe 3 +/∑Fe
ratio, and crystallization of Fe 3 +-rich magnetite (Fe3O4) will drive
liquid Fe 3 +/∑Fe ratios down (Fig. 3c), although crystallization of
any realistic combination of these phases is insufficient to create the
observed arc compositions from a primary magma of Fe 3 +/
∑Fe = 0.14, appropriate for MORB (Cottrell and Kelley, 2011). Low
overall TiO2 concentrations of arc basalts could be interpreted as
evidence that ilmenite (FeTiO3) crystallization may have increased
the arc Fe 3 +/∑Fe ratios. Ilmenite, however, is not observed in our
samples and is an extremely rare phase in arc basalts. Arc basalts
commonly have low TiO2 concentrations due to high extents of
H2O-fluxed mantle melting (Kelley et al., 2006, 2010), withholding
of Ti by residual rutile in the slab (Pearce and Parkinson, 1993),
and/or more depleted mantle sources beneath arcs relative to spreading
ridges (Pearce and Stern, 2006). When present, ilmenite saturates after
titanomagnetite, has a significant hematite (Fe2O3) component (15–
30%; Gill, 1981), and would draw melt TiO2 concentrations down with
progressive crystallization, which is the opposite of our observations
(Fig. 3b).
4.2.2. Degassing of H–C–O–Cl species
Co-variations between volatile species, major and trace elements,
and magmatic oxidation state implicate S degassing as a potential
driving force behind the observed magmatic reduction. It is worth
considering, however, the possible effects of degassing other volatile
species. Water and CO2 were degassing from this Agrigan magma
during inclusion entrapment (Fig. 2); however, we see no evidence
for concomitant oxidation, nor do we expect any. Magmatic
“auto-oxidation” during H2 degassing (Holloway, 2004) can only
drive oxidation if initial magmatic conditions are sufficiently reducing
(fO2 b QFM-1) to allow a significant amount of H2O to dissociate to H2
in the vapor phase and if that vapor phase exits the system out of
equilibrium with the magma it is oxidizing (Cottrell and Kelley, 2011;
Crabtree and Lange, 2011). This process, which becomes increasingly
inefficient and self-limiting as fO2 increases above QFM, cannot drive
magmas to the higher fO2s observed here (Candela, 1986;
Carmichael, 1991; Cottrell and Kelley, 2011; Crabtree and Lange,
2011; Frost and Ballhaus, 1998). Similarly, at fO2 greater than QFM-1,
graphite is not stable and carbon is speciated in the melt as carbonate.
Under these conditions, CO2 vapor loss is fO2 neutral and cannot drive
magmatic fO2s to the values we observe (Ballhaus, 1993; Cottrell and
Kelley, 2011). Chlorine degassing may also drive magmatic oxidation
(Bell and Simon, 2011), but there is no evidence for Cl degassing in
this eruption. We again conclude that sulfur is the only degassing
species for which we see, or expect, a concomitant change in magmatic
redox, and in this case the shift is toward reduction.
4.2.3. Oxidation of melt inclusions via outward H + diffusion
If H2O is lost from a melt inclusion after entrapment, the main
mechanism by which this takes places is proton (H +) diffusion
through the host olivine. This scenario requires a gradient in H2O
between the inclusion and the external melt and, if the melt inclusion
were truly a closed system for fO2, such a process would result in a
simultaneous increase of the inclusion fO2 due to the buildup of
excess oxygen left behind by H2O dissociation and loss of H +. Recent
experimental work, however, has shown that olivine-hosted melt
inclusions are open to fO2 exchange with the external melt, and that
H2O and fO2 re-equilibration through olivine take place independently,
but at roughly similar rates (e.g., Bucholz et al., 2011; Gaetani et al.,
2010). In light of these results, H + diffusion cannot drive oxidation of
melt inclusions because any resultant change in the inclusion fO2 will
be reset by the host magma at the same rate. At equilibrium, the
oxidation state of Fe in inclusions is determined by the oxidation
state imposed by the system, be that the host magma, or the environment of a laboratory experiment. The observation of a range of inclusion fO2s and H2O contents among the AGR19-02 suite suggests that
the pace of re-equilibration of either H2O or fO2 in the inclusions
must have been slower than the rate at which changes in the external
melt occurred, otherwise the inclusions would all have homogeneous
H2O and fO2. We cannot rule out the possibility that diffusive processes
have had minor effects on the inclusion compositions, and emphasize
that the measured H2O concentrations and Fe3 +/∑Fe ratios in the
inclusions are thus robust minima. Most importantly, the melt
inclusions record a trend of Fe reduction coincident with decreasing
H2O in this suite, which is the opposite of the predicted relationship
for H+-loss in a system closed to oxygen exchange.
4.3. Oxygen fugacity of the Agrigan mantle source
4.3.1. Reconstruction of primary melts
To assess the redox conditions of the mantle source beneath
Agrigan, we reconstruct primary mantle melt compositions for the
most mafic inclusions (MgO > 5.0 wt.%; n = 6) of the Agrigan suite
by calculating the equilibrium olivine for each melt inclusion, and
then adding 0.01% of that olivine to each melt composition, and
repeating these steps until equilibrium with mantle olivine for a
range of possible Fo contents is achieved (Fo90, Fo91.5, and Fo93;
Table 2; Kelley et al., 2006, 2010; Stolper and Newman, 1994).
These most mafic melts are sufficiently close to the modeled point
of cpx—in that an assumption of olivine-only on the liquidus results
in relatively small error in the reconstructed melts, which require
~18–25% olivine added back to reach Fo90 equilibrium. For these
calculations, we assumed that the magma evolved in a system closed
to oxygen (as is the case for MORBs; Cottrell and Kelley, 2011), so the
number of moles of Fe 3 + in the system remained unchanged as Fe 2 +
from olivine was added back to each melt. The Fe 3 +/∑Fe ratios of
the Agrigan primary melts are thus lower (0.18–0.22 at Fo90;
Table 2) relative to the fractionated starting compositions (0.23–
0.28). The dissolved H2O of these most mafic inclusions is higher on
average than the more differentiated inclusions, and we thus assume
these are robust minima for the undegassed H2O content of the
magma.
For comparison with the arc, Fig. 6 shows similar reconstructions of
MORB (Cottrell and Kelley, 2011) and back-arc basin pillow basalts
(BABB) from the Mariana trough (Kelley and Cottrell, 2009). The modeled melt compositions clearly show a progressive increase in the Fe3
+
/∑Fe ratios of primary melts that is coincident with increases in geochemical tracers of subduction influence (e.g., dissolved H2O, Ba/La;
Fig. 6a, c), similar to the trends observed for global arc basalts uncorrected for fractionation effects (Fig. 6a inset; Kelley and Cottrell, 2009).
Because composition, pressure, and temperature of magmas all influence
the direct translation of magmatic Fe3 +/∑Fe ratios into oxygen fugacity
(Kress and Carmichael, 1991), fundamental differences in primary melt
composition or P–T conditions of origin that vary with tectonic setting
could potentially offset the relationship between fO2 and Fe3 +/∑Fe ratios of magmas. We assess the extent to which these factors could affect
interpretations of the data by combining constraints on the pressures
and temperatures of last equilibration of primary melts with the mantle
(Table 2; Lee et al., 2009) with the primary melt compositions to determine fO2 for each reconstructed primary melt at relevant P–T conditions
in the mantle. Fig. 6b shows that the modeled fO2s of the mantle sources
indicate the same sense of increase from MORB to BABB to Agrigan as do
Fe3 +/∑Fe ratios. The Agrigan source is modeled at fO2 conditions in the
range of ΔQFM +1 to +1.6 (for a Fo90 source; +0.8 to +1.3 for the most
refractory case), significantly more oxidized than the back-arc basin
Sample
AGR19-02
Inclusion #
01
Target Fo content
Fo90
Olivine added
SiO2
wt.%
wt.%
TiO2
wt.%
Al2O3
FeO
wt.%
wt.%
Fe2O3
MnO
wt.%
MgO
wt.%
CaO
wt.%
wt.%
Na2O
wt.%
K2 O
wt.%
P2O5
wt.%
H2O
Fe+ 3/∑Fe
Temp.
°C
Pressure
GPa
22%
45.56
0.63
13.75
8.42
2.65
0.17
12.70
10.07
1.47
0.35
0.11
3.38
0.220
1299
1.52
08
10
11
14
16
22%
46.51
0.63
13.47
8.55
2.08
0.16
12.90
9.98
1.57
0.38
0.12
2.94
0.179
1308
1.46
18%
45.16
0.62
15.12
7.63
2.25
0.12
11.52
10.56
1.48
0.31
0.12
3.81
0.210
1285
1.37
20%
45.72
0.63
14.72
8.04
2.00
0.17
12.12
10.20
1.65
0.35
0.10
2.87
0.183
1296
1.45
22%
45.36
0.55
14.36
8.45
2.17
0.15
12.75
9.55
1.64
0.38
0.11
2.97
0.188
1310
1.58
25%
44.81
0.53
13.73
9.17
2.26
0.18
13.82
9.59
1.66
0.33
0.11
3.46
0.182
1329
1.86
At 1200 °C and 1 atm
−7.06
log fO2
ΔQFM
1.25
− 7.66
0.65
− 7.24
1.07
− 7.61
0.69
−7.48
0.82
At T and P
log fO2
ΔQFM
− 5.15
0.97
− 5.03
1.38
− 5.22
1.02
−4.86
1.16
−4.58
1.58
01
08
10
11
14
16
08
10
11
14
16
32%
45.21
0.59
12.72
8.51
2.45
0.15
15.46
9.32
1.36
0.33
0.10
3.12
0.205
1368
2.01
32%
46.09
0.58
12.46
8.63
1.92
0.15
15.65
9.23
1.45
0.35
0.11
2.72
0.167
1377
1.95
26%
44.88
0.58
14.13
7.76
2.10
0.12
13.98
9.88
1.38
0.29
0.11
3.56
0.196
1325
1.78
29%
45.39
0.59
13.69
8.15
1.86
0.16
14.72
9.48
1.53
0.33
0.09
2.67
0.171
1362
1.90
32%
45.02
0.51
13.28
8.54
2.01
0.14
15.51
8.84
1.52
0.35
0.10
2.75
0.175
1380
2.08
36%
44.50
0.49
12.66
9.21
2.09
0.16
16.60
8.84
1.53
0.31
0.10
3.19
0.169
1400
2.42
46%
44.83
0.53
11.50
8.43
2.21
0.14
18.87
8.42
1.23
0.30
0.09
2.83
0.191
1458
2.81
46%
45.63
0.53
11.27
8.54
1.74
0.13
19.05
8.35
1.32
0.31
0.10
2.46
0.155
1468
2.75
39%
44.54
0.52
12.81
7.75
1.91
0.10
17.45
8.95
1.25
0.26
0.10
3.23
0.181
1412
2.48
42%
45.01
0.53
12.44
8.11
1.69
0.14
18.05
8.62
1.39
0.30
0.08
2.42
0.158
1450
2.65
46%
44.66
0.46
12.01
8.45
1.82
0.12
18.92
7.99
1.37
0.32
0.09
2.48
0.162
1472
2.91
52%
44.16
0.44
11.33
9.04
1.87
0.15
20.23
7.91
1.37
0.27
0.09
2.86
0.157
1499
3.40
− 6.34
0.76
− 7.21
1.10
− 7.80
0.50
− 7.38
0.92
− 7.75
0.55
− 7.63
0.67
−7.70
0.60
− 7.36
0.94
− 7.95
0.35
− 7.54
0.76
− 7.90
0.40
− 7.79
0.51
− 7.86
0.44
− 4.55
1.11
− 3.74
1.47
− 4.29
0.87
− 4.46
1.28
− 4.42
0.92
− 4.01
1.05
−3.68
1.00
− 2.64
1.34
− 3.19
0.75
− 3.38
1.16
− 3.34
0.80
− 2.90
0.91
− 2.50
0.81
Fo91.5
01
Fo93
K.A. Kelley, E. Cottrell / Earth and Planetary Science Letters 329–330 (2012) 109–121
Table 2
Reconstructed primary melt compositions from Agrigan, Marianas.
117
118
K.A. Kelley, E. Cottrell / Earth and Planetary Science Letters 329–330 (2012) 109–121
H2O (wt.%)
0.0
0.24
4.0
6.0
0.3
Agrigan MI, this Study (Fo90)
Fe3+/ Fe
0.22
Fe3+/ Fe(Fo90)
2.0
Mariana Trough Pillow Glass, KC09
0.2
0.20
MORB Pillow Glass, CK11
Arc MI, KC09
0.1
0.18
0.16
0.14
0.12
A
C
0.10
0
Δ QFM (@ Equil. P, T)
+2.0
10
20
30
Ba/La
+1.5
+1.0
+0.5
0.0
B
-0.5
0
1
2
3
4
H2O (Fo90)
Fig. 6. Modeled primary melt compositions and oxygen fugacities for Agrigan, Marianas. (a) Plot of H2O vs. Fe3 +/∑Fe ratio in reconstructed primary melts in equilibrium with Fo90
olivine. Agrigan data from this study are shaded circles, filtered for the most mafic glass inclusions (MgO > 5 wt.%; see Section 4.3.1). Modeled MORBs (shaded triangles) and
Mariana trough basalts (shaded diamonds) are from Cottrell and Kelley (2011) and Kelley and Cottrell (2009). Inset shows H2O vs. Fe3 +/∑Fe ratios for raw basalt compositions,
uncorrected except for PEC, with the addition of global arc data from Kelley and Cottrell (2009; open circles). (b) Plot of H2O vs. oxygen fugacity (ΔQFM) for reconstructed primary
melts in equilibrium with Fo90 olivine (Table 2). Oxygen fugacity is referenced to the QFM buffer at the P–T conditions of last equilibration of each melt with the mantle, modeled
using melt thermobarometry (Lee et al., 2009). (c) Ba/La ratio vs. Fe3 +/∑Fe ratio for reconstructed primary melts in equilibrium with Fo90 olivine.
(ΔQFM +0.2 to +0.5) or MORB (ΔQFM −0.25 to +0.4) sources, and fO2
increases coincident with increasing primary melt H2O content and
tracers of additions from the subducted slab to the mantle source. Compositional differences among the modeled primary melt compositions
should no longer reflect variations driven by differentiation processes.
Melt inclusions are well-known to preserve compositional diversity
that reflects true heterogeneity of primary melts that aggregated to
form larger magma bodies (e.g., Saal et al., 1998; Sobolev and Shimizu,
1993), and the weak correlation of reconstructed Agrigan primary melt
compositions in Fig. 6 may thus reflect diversity in discrete primary
melts at Agrigan. We emphasize that the H2O content of arc magmas is
a useful tracer of the influence of subduction on the arc mantle source,
but H2O itself is unlikely to be the ultimate cause of oxidation (e.g.,
Frost and Ballhaus, 1998; Kelley and Cottrell, 2009). Although the oxidation state and water content of olivine-hosted melt inclusions may be
reset by the diffusion of H+ and point defects in the presence of chemical
gradients (e.g., Bucholz et al., 2011; Gaetani et al., 2010), this would eliminate, or at least diffuse, the strong correlations we observe between oxidation state and major, trace, and volatile elements. The data arrays seen
in Fig. 6, for example, are not consistent with concomitant loss of H from
and oxidation of the Agrigan inclusions.
4.3.2. Trace element proxies for source fO2
The Agrigan data set offers an opportunity to directly compare
multiple proxies for the oxidation state of the arc mantle source.
The oxidation state of iron in the most mafic inclusions indicates derivation from a mantle source 1–1.5 orders of magnitude more
oxidized than the MORB mantle source (Fig. 6; Cottrell and Kelley,
2011) and we have identified no process during the subsequent
evolution of this magma in the crust to oxidize Fe. Two trace element
proxies for source fO2 provide an independent assessment of our
analysis: the V/Sc ratio (Canil, 1997; Lee et al., 2003, 2005) and the
Zn/Fe ratio (Lee et al., 2010).
The V/Sc ratios of the least-evolved Agrigan inclusions provide a
maximum constraint on V/Sc of primitive Agrigan magmas (and by
proxy on fO2) because clinopyroxene fractionation increases V/Sc
ratios in the melt. These lowest Agrigan ratios overlap the MORB
field and fall between 6 and 8. For these inclusions, representing
15–20% melt fraction (Kelley et al., 2010), the modeled fO2 falls
between ~ QFM and QFM-1 (see the electronic supplement). This
completely overlaps the MORB source, consistent with Lee et al.
(2005), and is in opposition to the fugacities calculated from Fe
speciation. The discrepancy between the Fe and V/Sc proxies is
surprising in this case because our detailed study of this eruptive
suite indicates that neither crystallization nor degassing, the two
processes most often invoked to explain this discrepancy, are
primarily responsible for oxidation of this magma since it last
equilibrated with the mantle source. The reason for this discrepancy
remains unknown, but it is possible that the calibration conditions
for the V/Sc proxy are inappropriate for melting in the hydrous
mantle wedge (Jackson et al., 2010).
The Zn/Fe ratios in the Agrigan glass inclusions broadly increase as
differentiation proceeds (see the electronic supplement) with the
most mafic inclusions bearing ratios of 7–12. Clinopyroxene and
magnetite fractionation can only drive the Zn/Fe ratio up, therefore
the lowest Zn/Fe ratio we observe in this suite (7.6) provides a
minimum fO2 for this system's mantle source of QFM + 2 (Lee et al.,
2010), consistent with the Lee et al. (2010) compilation for Mariana
K.A. Kelley, E. Cottrell / Earth and Planetary Science Letters 329–330 (2012) 109–121
whole rock data and the oxidation state of Fe determined in this
study.
4.4. An oxidized arc mantle wedge
4.4.1. Ferric Fe content of the arc mantle
Reconstructed primary melts for Agrigan may be inverted for
mantle melt fraction, using TiO2 as a proxy for melt fraction and a
simple batch melting model, with constraints from Kelley et al.
(2010). The batch melting equation may then be re-arranged to
solve for the equivalent concentrations of H2O and Fe 3 + of the
Agrigan mantle source, using appropriate mantle/melt partition
coefficients for H2O (0.008; average of Aubaud et al., 2004, and
Hauri et al., 2006), and Fe2O3 (0.1–0.2; Canil et al., 1994; Mallmann
and O'Neill, 2009). These calculations indicate that the arc mantle
relevant for Agrigan basalts contains 0.49–0.87 wt.% Fe2O3 and 0.48–
0.72 wt.% H2O. This Fe2O3 content is bracketed by the Fe2O3
concentration predicted for the background MORB mantle source
(≥0.3 wt.%; Cottrell and Kelley, 2011) and the range predicted for
oxidized, hydrated arc mantle based on similar modeling of Fe2O3 in
whole-rock arc lavas (0.6–1.0 wt.%; Parkinson and Arculus, 1999).
Subtracting the Fe2O3 contribution from the background mantle
(0.3 wt.%) yields excess concentrations of 0.19–0.57 wt.% Fe2O3
added to (or created in) the arc mantle by slab-derived components,
and maximum Fe2O3/H2O ratios for the slab-derived components of
0.4–1.0 on a weight basis (0.04–0.11 Fe2O3/H2O or 0.09–0.23 Fe 3 +/
H2O on a molar basis). Such a high proportion of Fe2O3 to H2O cannot
be added directly by a dilute, oxidized aqueous fluid because Fe 3 +
has low solubility in such fluids (Schneider and Eggler, 1986). As
slab-derived components become more solute-rich, however, Fe 3 +
is likely to become much more mobile, as has been shown for other
fluid-immobile elements (e.g., Johnson and Plank, 1999; Kessel et
al., 2005). Therefore, if all of the observed excess Fe 3 + is transported
directly from the slab into the mantle wedge, these constraints
require that slab-derived, H2O-rich components are hypersaline
brines, supercritical fluids, or silicate melts of subducted sediments
or the basaltic plate.
Alternatively, fluid-mobile elements, such as S, could oxidize the
mantle wedge of subduction zones without requiring direct transport
of Fe 3 +. If the intrinsic oxygen fugacity of fluids released from the descending slab are sufficiently high to carry S as sulfate (SO42 −; S 6 +),
then 1 mol of S has the potential to oxidize 8 mol of Fe 2 + as sulfate
in the fluid is reduced to form sulfide (S 2 −) in the mantle wedge.
Sulfur reduction will take place provided that the oxygen fugacity of
the mantle wedge remains below the sulfur–sulfur oxide (SSO)
buffer, or approximately QFM + 2 (Mungall, 2002). The molar S/H2O
ratio required to oxidize 0.09–0.23 mol of Fe 3 + per mole of H2O is
0.011–0.028. Constraints on the S content of various mantle sources
suggest that arc mantle contains ~ 200 ppm S in excess of MORB
mantle (Wallace, 2005), and given the above constraints on the H2O
content of the Agrigan mantle source, this gives a molar S/H2O ratio
in slab-derived components of ~ 0.016–0.023. From this perspective,
hydrous slab-derived components could deliver enough S to create
the observed Fe 3 + abundance in the arc mantle.
4.4.2. Potential sources and causes of oxidation in the mantle wedge
Subducting oceanic lithosphere is highly oxidized (e.g., Alt et al.,
1986; Berndt et al., 1996; Lecuyer and Ricard, 1999). Sediments may
contain Fe 3 +/∑Fe ratios up to 0.82 and altered basalts up to 0.19–
0.24 (Lecuyer and Ricard, 1999). The lithospheric mantle may also
become oxidized through the formation of serpentine and brucite,
which exclude Fe from their olivine protolith that then forms magnetite
(and H+, which escapes the system) at the expense of H2O (e.g., Berndt
et al., 1996). The above discussion (Section 4.4.1) proposes that this
oxidized subducted plate provides the materials, potentially in the form
of Fe3 + or SO42 −, that drive oxidation of the mantle wedge beneath arc
119
volcanoes. Sulfur isotopic studies of Mariana fore-arc serpentinites and
arc lavas suggest that S is delivered by the slab, and specifically by the
subducted sediment, to the mantle wedge (Alt and Shanks, 2006; Alt et
al., 1993), although S abundances and budgets of subducting sediments
in the Marianas and elsewhere remain poorly constrained (Plank and
Langmuir, 1998). Serpentinized lithospheric mantle also has the
potential to produce highly oxidized fluids during de-serpetinization
reactions that consume magnetite during dehydration (Elburg and
Kamenetsky, 2007; Nozaka, 2003).
Oxidized slab-derived fluids, however, have no power to change
the mantle wedge fO2 if the mantle is buffered by mineral equilibria,
which, through changes in mineral mode, may mediate system fO2.
Recent work postulated that the relatively dry and reduced MORB
source may be buffered during melting near QFM by S equilibria
(Cottrell and Kelley, 2011). The buffering capacity of the arc source,
however, may be diminished by the presence of H2O which, through
increased melt fraction, may deplete the mantle wedge of buffering
phases. This could leave arc melts more susceptible to oxidation by
the addition of oxidized fluids (Evans and Tomkins, 2011).
5. Conclusions
Olivine-hosted glass inclusions from Agrigan, Marianas preserve a
segment of the liquid line of descent of a H2O-rich arc magma. Glass
compositions are uniformly more oxidized than MORBs, and indicate
reduction, rather than oxidation, as the major redox change during
differentiation. Correlation between Fe 3 +/∑Fe ratios and dissolved
S concentrations indicate that S degassing may play a major role in
modifying melt redox during degassing. Reconstruction of the most
mafic Agrigan melts to primary mantle melt compositions, which
compensates for the slight oxidation effect of fractionating Fe2 +-rich
crystal phases, shows that primary arc melts are oxidized at
equilibrium with the arc mantle source, and that fO2 correlates
directly with indices of slab additions to the mantle wedge (e.g., H2O,
Ba/La). Simple mass balance calculations suggest that S derived from
the subducted plate is present in the arc wedge in sufficient
abundance to accommodate the magnitude of oxidation
required.
Acknowledgments
We are grateful for thorough reviews from Cin-Ty Lee, Chris
Ballhaus, and Leonid Danyushevsky. We acknowledge constructive
discussions with and inspiration from Cin-Ty Lee, Marc Hirschmann,
Mac Rutherford, and Becky Lange. Terry Plank generously shared
the source sample material and unpublished data to assist with this
study, in addition to valuable thoughts and data on V partitioning.
This work was made possible by the contributions of Benjamin
Parks, who generated a pilot data set for this study through the GSO
SURFO program, and Maryjo Brounce and Christa Jackson, who
assisted in all aspects of data collection. Tony Lanzirotti contributed
invaluable expertise in μ-XANES analysis and beamline operations at
X26A. NSF Award OCE-0644625 provided curatorial support for
marine geological samples at the University of Rhode Island. Use of
the National Synchrotron Light Source, Brookhaven National
Laboratory, was supported by the U.S. Department of Energy, Office
of Science, Office of Basic Energy Sciences, under Contract No. DEAC02-98CH10886. We acknowledge support from Smithsonian's
Scholarly Studies Program (EC), a URI ADVANCE fellowship (KK)
and NSF awards EAR-0838328 (KK), MARGINS-EAR-0841108 (KK)
and MARGINS-EAR-0841006 (EC).
Appendix A. Supplementary data
Supplementary data to this article can be found online at doi:10.
1016/j.epsl.2012.02.010.
120
K.A. Kelley, E. Cottrell / Earth and Planetary Science Letters 329–330 (2012) 109–121
References
Alt, J., Shanks, W., 2006. Stable isotope compositions of serpentinite seamounts in the
Mariana forearc: serpentinization processes, fluid sources and sulfur metasomatism.
Earth Planet. Sci. Lett. 242, 272–285. doi:10.1016/j.epsl.2005.11.063.
Alt, J.C., Honnorez, J., Laverne, C., Emmermann, R., 1986. Hydrothermal alteration of a 1
km section through the upper oceanic crust, Deep Sea Drilling Project Hole 504B:
Mineralogy, chemistry, and evolution of seawater–basalt interactions. J. Geophys.
Res. 91, 10309–10355.
Alt, J.C., Shanks III, W.C., Jackson, M.C., 1993. Cycling of sulfur in subduction zones: the
geochemistry of sulfur in the Mariana Island Arc and back-arc trough. Earth Planet.
Sci. Lett. 119, 477–494.
Anderson, A.T., Wright, T.L., 1972. Phenocrysts and glass inclusions and their bearing
on oxidation and mixing of basaltic magmas, Kilauea volcano, Hawaii. Am. Mineral.
57, 188–216.
Ariskin, A.A., Barmina, G.S., 1999. An empirical model for the calculation of spinel–melt
equilibria in mafic igneous systems at atmospheric pressure: 2. Fe–Ti oxides.
Contrib. Mineral. Petrol. 134, 251–263. doi:10.1007/s004100050482.
Aubaud, C., Hauri, E.H., Hirschmann, M.M., 2004. Hydrogen partition coefficients
between nominally anhydrous minerals and basaltic melts. Geophys. Res. Lett.
31. doi:10.1029/2004GL021341.
Ballhaus, C., 1993. Redox states of lithospheric and asthenospheric upper mantle
Contrib. Mineral. Petrol. 114, 331–348.
Bell, A.S., Simon, A., 2011. Experimental evidence for the alteration of the Fe3 +/Fe of
silicate melt caused by the degassing of chlorine-bearing aqueous volatiles. Geology
39, 499–502. doi:10.1130/g31828.1.
Berndt, M.E., Allen, D.E., Seyfried, W.E., 1996. Reduction of CO2 during serpentinization
of olivine at 300 °C and 500 bar. Geology 24, 351–354.
Brandon, A.D., Draper, D.S., 1996. Constraints on the origin of the oxidation state of
mantle overlying subduction zones: an example from Simcoe, Washington, USA.
Geochim. Cosmochim. Acta 60, 1739–1749.
Bucholz, C.E., Gaetani, G.A., Behn, M.D., 2011. Diffusive re-equilibration of volatiles and
oxygen fugacity in olivine-hosted melt inclusions: experiments and numerical
models. EOS Transactions AGU, V11H-02.
Candela, P.A., 1986. The evolution of aqueous vapor from silicate melts: effect on
oxygen fugacity. Geochim. Cosmochim. Acta 50, 1205–1211.
Canil, D., 1997. Vanadium partitioning and the oxidation state of Archaean komatiite
magmas. Nature 389, 842–845.
Canil, D., 2002. Vanadium in peridotites, mantle redox and tectonic environments:
Archean to present. Earth Planet. Sci. Lett. 195, 75–90.
Canil, D., Fedortchouk, Y., 2001. Olivine–liquid partitioning of vanadium and other
trace elements, with applications to modern and ancient picrites. Can. Mineral.
39, 319–330. doi:10.2113/gscanmin.39.2.319.
Canil, D., O'Neill, H.S.C., Pearson, D.G., Rudnick, R.L., McDonough, W.F., Carswell, D.A.,
1994. Ferric iron in peridotites and mantle oxidation states. Earth Planet. Sci.
Lett. 123, 205–220.
Carmichael, I.S.E., 1991. The redox states of basic and silicic magmas: a reflection of
their source regions? Contrib. Mineral. Petrol. 106, 129–141. doi:10.1007/
BF00306429.
Carroll, M.R., Rutherford, M.J., 1988. Sulfur speciation in hydrous experimental glasses
of varying oxidation state: results from measured wavelength shifts of sulfur Xrays. Am. Mineral. 73, 845–894.
Christie, D.M., Carmichael, I.S.E., Langmuir, C.H., 1986. Oxidation states of mid-ocean
ridge basalt glasses. Earth Planet. Sci. Lett. 79, 397–411.
Cottrell, E., Kelley, K.A., 2011. The oxidation state of Fe in MORB glasses and the oxygen
fugacity of the upper mantle. Earth Planet. Sci. Lett. 305, 270–282. doi:10.1016/
j.epsl.2011.03.014.
Cottrell, E., Kelley, K.A., Lanzirotti, A.T., Fischer, R.A., 2009. High-precision determination
of iron oxidation state in silicate glasses using XANES. Chem. Geol. 268, 167–179.
doi:10.1016/j.chemgeo.2009.08.008.
Crabtree, S.M., Lange, R.A., 2011. An evaluation of the effect of degassing on the oxidation state of hydrous andesite and dacite magmas: a comparison of pre- and posteruptive Fe2+ concentrations. Contrib. Mineral. Petrol.. doi:10.1007/s00410-0110667-7
Danyushevsky, L.V., 2001. The effect of small amounts of H2O on crystallization of mid-ocean
ridge and backarc basin magmas. J. Volcanol. Geotherm. Res. 110, 265–280.
Danyushevsky, L.V., Plechov, P., 2011. Petrolog3: integrated software for modeling
crystallization processes. Geochem. Geophys. Geosyst. 12, Q07021. doi:10.1029/
2011GC003516.
Danyushevsky, L.V., Della-Pasqua, F.N., Sokolov, S., 2000. Re-equilibration of melt inclusions trapped by magnesian olivine phenocrysts from subduction-related magmas:
petrological implications. Contrib. Mineral. Petrol. 138, 68–83. doi:10.1007/
PL00007664.
Dauphas, N., Craddock, P.R., Asimow, P.D., Bennett, V.C., Nutman, A.P., Ohnenstetter, D.,
2009. Iron isotopes may reveal the redox conditions of mantle melting from Archean
to Present. Earth Planet. Sci. Lett. 288, 255–267. doi:10.1016/j.epsl.2009.09.029.
Dixon, J.E., Stolper, E., Holloway, J.R., 1995. An experimental study of water and carbon
dioxide solubilities in mid-ocean ridge basaltic liquids. Part I: calibration and
solubility models. J. Petrol. 36, 1607–1631.
Elburg, M.A., Kamenetsky, V.S., 2007. Dehydration processes determine fO2 of arc and
intraplate magmas. Geochim. Cosmochim. Acta 71, A252.
Elliott, T., Plank, T., Zindler, A., White, W.M., Bourdon, B., 1997. Element transport from
slab to volcanic front at the Mariana arc. J. Geophys. Res. 102, 14991–15019.
Evans, K.A., Tomkins, A.G., 2011. The relationship between subduction zone redox budget
and arc magma fertility. Earth Planet. Sci. Lett. 308, 401–409. doi:10.1016/
j.epsl.2011.06.009.
Frost, B.R., Ballhaus, C., 1998. Comment on “Constraints on the origin of the oxidation
state of mantle overlying subduction zones: an example from Simcoe, Washington,
USA” by A. D. Brandon and D. S. Draper. Geochim. Cosmochim. Acta 62, 329–331.
Gaetani, G.A., O'Leary, J.C., Shimizu, N., Bucholz, C.E., 2010. Decoupling of H2O, oxygen
fugacity and incompatible elements in olivine-hosted melt inclusions by diffusive
re-equilibration. EOS Transactions AGU, Abstract V23E-06.
Gill, J.B., 1981. Orogenic Andesites and Plate Tectonics. Springer-Verlag, New York.
Hauri, E.H., Gaetani, G.A., Green, T.H., 2006. Partitioning of water during melting of the
Earth's upper mantle at H2O-undersaturated conditions. Earth Planet. Sci. Lett.
248, 715–734. doi:10.1016/j.epsl.2006.06.014.
Holloway, J.R., 2004. Redox reactions in seafloor basalts: possible insights into silicic hydrothermal systems. Chem. Geol. 210, 225–230. doi:10.1016/j.chemgeo.2004.06.009.
Irvine, T.N., Baragar, W.R.A., 1971. A Guide to the Chemical Classification of the Common Volcanic Rocks. Can. J. Earth Sci. 8, 523–548.
Jackson, C.M., Cottrell, E., Kelley, K.A., 2010. Mineral-melt partitioning of V and Sc at
arcs: implications for mantle wedge oxygen fugacity. Presented at 2010 Fall Meeting,
AGU, San Francisco, Calif., 13–17 Dec. Abstract V11F-01.
Jenner, F.E., O'Neill, H.S.C., Arculus, R.J., Mavrogenes, J.A., 2010. The magnetite crisis in
the evolution of arc-related magmas and the initial concentration of Au, Ag and Cu.
J. Petrol. 51, 2445–2464. doi:10.1093/petrology/egq063.
Johnson, M.C., Plank, T., 1999. Dehydration and melting experiments constrain the fate
of subducted sediments. Geochem. Geophys. Geosyst. 1.
Jugo, P.J., Wilke, M., Botcharnikov, R.E., 2010. Sulfur K-edge XANES analysis of natural
and synthetic basaltic glasses: implications for S speciation and S content as function
of oxygen fugacity. Geochim. Cosmochim. Acta 74, 5926–5938. doi:10.1016/
j.gca.2010.07.022.
Kelley, K.A., Cottrell, E., 2009. Water and the oxidation state of subduction zone
magmas. Science 325, 605–607. doi:10.1126/science.1174156.
Kelley, K.A., Plank, T., Ludden, J.N., Staudigel, H., 2003. Composition of altered oceanic
crust at ODP Sites 801 and 1149. Geochem. Geophys. Geosyst. 4. doi:10.1029/
2002GC000435.
Kelley, K.A., Plank, T., Grove, T.L., Stolper, E.M., Newman, S., Hauri, E.H., 2006. Mantle
melting as a function of water content beneath back-arc basins. J. Geophys. Res.
111, B09208. doi:10.1029/2005JB003732.
Kelley, K.A., Plank, T., Newman, S., Stolper, E., Grove, T.L., Parman, S., Hauri, E., 2010.
Mantle melting as a function of water content beneath the Mariana arc. J. Petrol.
51, 1711–1738. doi:10.1093/petrology/egq036.
Kent, A.J.R., Elliott, T.R., 2002. Melt inclusions from Mariana arc lavas: implications for
the composition and formation of island arc magmas. Chem. Geol. 183, 263–286.
Kessel, R., Schmidt, M.W., Ulmer, P., Pettke, T., 2005. Trace element signature of
subduction-zone fluids, melts and supercritical liquids at 120–180 km depth.
Nature 437, 724–727. doi:10.1038/nature03971.
Kress, V.C., Carmichael, I.S.E., 1991. The compressibility of silicate liquids containing
Fe2O3 and the effect of composition, temperature, oxygen fugacity and pressure
on their redox states. Contrib. Mineral. Petrol. 108, 82–92.
Lecuyer, C., Ricard, Y., 1999. Long-term fluxes and budget of ferric iron: implication for
the redox states of the Earth's mantle and atmosphere. Earth Planet. Sci. Lett. 165,
197–211.
Lee, C.-T.A., Brandon, A.D., Norman, M.D., 2003. Vanadium in peridotites as a proxy for
paleo-fO2 during partial melting: prospects, limitations, and implications.
Geochim. Cosmochim. Acta 67, 3045–3064. doi:10.1016/S0016-7037(00)
00268-0.
Lee, C.-T.A., Leeman, W.P., Canil, D., Li, Z.-X.A., 2005. Similar V/Sc systematics in MORB
and arc basalts: implications for the oxygen fugacities of their mantle source
regions. J. Petrol. 46, 2313–2336. doi:10.1093/petrology/egi056.
Lee, C.-T.A., Luffi, P., Plank, T., Dalton, H., Leeman, W.P., 2009. Constraints on the depths
and temperatures of basaltic magma generation on Earth and other terrestrial
planets using new thermobarometers. Earth Planet. Sci. Lett. 279, 20–33.
doi:10.1016/j.epsl.2008.12.020.
Lee, C.-T.A., Luffi, P., Le Roux, V., Dasgupta, R., Albaréde, F., Leeman, W.P., 2010. The
redox state of arc mantle using Zn/Fe systematics. Nature 468, 681–685.
doi:10.1038/nature09617.
Mallmann, G., O'Neill, H.S.C., 2009. The crystal/melt partitioning of V during mantle
melting as a function of oxygen fugacity compared with some other elements
(Al, P, Ca, Sc, Ti, Cr, Fe, Ga, Y, Zr and Nb). J. Petrol. 50, 1765–1794. doi:10.1093/
petrology/egp053.
Médard, E., Grove, T.L., 2007. The effect of H2O on the olivine liquidus of basaltic melts:
experiments and thermodynamic models. Contrib. Mineral. Petrol. 155, 417–432.
doi:10.1007/s00410-007-0250-4.
Métrich, N., Berry, A.J., O'Neill, H.S.C., Susini, J., 2009. The oxidation state of sulfur in
synthetic and natural glasses determined by X-ray absorption spectroscopy.
Geochim. Cosmochim. Acta 73, 2382–2399. doi:10.1016/j.gca.2009.01.025.
Miyashiro, A., 1974. Volcanic rock series in island arcs and active continental margins.
Am. J. Sci. 274, 321–355.
Mungall, J.E., 2002. Roasting the mantle: slab melting and the genesis of major Au and
Au-rich Cu deposits. Geology 30, 915–918.
Newman, S., Lowenstern, J.B., 2002. VolatileCalc: a silicate melt–H2O–CO2 solution
model written in Visual Basic for excel. Comput. Geosci. 28, 597–604.
Nozaka, T., 2003. Compositional heterogeneity of olivine in thermally metamorphosed
serpentinite from Southwest Japan. Am. Mineral. 88, 1377–1384.
Oppenheimer, C., 2003. Volcanic degassing, treatise on geochemistry. , pp. 123–166.
Parkinson, I.J., Arculus, R.J., 1999. The redox state of subduction zones: insights from
arc-peridotites. Chem. Geol. 160, 409–423.
Pearce, J., Parkinson, I.J., 1993. Trace element models for mantle melting: application to
volcanic arc petrogenesis. In: Prichard, H.M., Alabaster, T., Harris, N.B.W., Nears,
C.R. (Eds.), Magmatic Processes and Plate Tectonics, pp. 373–403.
K.A. Kelley, E. Cottrell / Earth and Planetary Science Letters 329–330 (2012) 109–121
Pearce, J., Stern, R.J., 2006. The origin of back-arc basin magmas: trace element and
isotope perspectives. In: Christie, D.M., Fisher, C.R. (Eds.), Back-arc Spreading
Systems: Geological, Biological, Chemical, and Physical Interactions. American
Geophysical Union, Washington, DC, pp. 63–86.
Plank, T., 2005. Constriants from Thorium/Lanthanum on sediment recycling at
subduction zones and the evolution of the continents. J. Petrol.. doi:10.1093/
petrology/egi005
Plank, T., Langmuir, C.H., 1998. The chemical composition of subducting sediment and
its consequences for the crust and mantle. Chem. Geol. 145, 325–394.
Putirka, K.D., Perfit, M., Ryerson, F.J., Jackson, M.G., 2007. Ambient and excess mantle
temperatures, olivine thermometry, and active vs. passive upwelling. Chem. Geol.
241, 177–206. doi:10.1016/j.chemgeo.2007.01.014.
Saal, A.E., Hart, S.R., Shimizu, N., Hauri, E.H., Layne, G.D., 1998. Pb isotopic variability in
melt inclusions from oceanic island basalts, Polynesia. Science 282, 1481–1484.
Schneider, M.E., Eggler, D.H., 1986. Fluids in equilibrium with peridotite minerals:
implications for mantle metasomatism. Geochim. Cosmochim. Acta 50, 711–724.
doi:10.1016/0016-7037(86)90347-9.
Shaw, A.M., Hauri, E.H., Fischer, T.P., Hilton, D.R., Kelley, K.A., 2008. Hydrogen isotopes
in Mariana arc melt inclusions: implications for subduction dehydration and the
deep-Earth water cycle. Earth Planet. Sci. Lett. 275, 138–145. doi:10.1016/
j.epsl.2008.08.015.
Sisson, T.W., Layne, G.D., 1993. H2O in basalt and basaltic andesite glass inclusions from
four subduction-related volcanoes. Earth Planet. Sci. Lett. 117, 619–635.
Sobolev, A.V., Shimizu, N., 1993. Ultra-depleted primary melt included in an olivine
from the Mid-Atlantic Ridge. Nature 363, 151–154.
Stern, R.J., Ito, E., 1983. Trace-element and isotopic constraints on the source of
magmas in the active Volcano and Mariana island arcs, western Pacific. J. Volcanol.
Geotherm. Res. 18, 461–482.
121
Stolper, E., Newman, S., 1994. The role of water in the petrogenesis of Mariana trough
magmas. Earth Planet. Sci. Lett. 121, 293–325.
Wade, J.A., Plank, T., Melson, W.G., Soto, G.J., Hauri, E.H., 2006. The volatile content of
magmas from Arenal volcano, Costa Rica. J. Volcanol. Geotherm. Res. 157,
94–120. doi:10.1016/j.jvolgeores.2006.03.045.
Wallace, P.J., 2005. Volatiles in subduction zone magmas: concentrations and fluxes
based on melt inclusion and volcanic gas data. J. Volcanol. Geotherm. Res. 140,
217–240. doi:10.1016/j.jvolgeores.2004.07.023.
Wallace, P.J., Carmichael, I.S.E., 1994. S speciation in submarine basaltic glasses as
determined by measurements of SKα X-ray wavelengths. Am. Mineral. 79, 161–167.
Webster, J.D., Botcharnikov, R.E., 2011. Distribution of sulfur between melt and fluid in
S–O–H–C–Cl–bearing magmatic systems at shallow crustal pressures and temperatures.
Rev. Mineral. Geochem. 73, 247–283. doi:10.2138/rmg.2011.73.9.
Wood, B.J., Bryndzia, L.T., Johnson, K.E., 1990. Mantle oxidation-state and its relationship
to tectonic environment and fluid speciation. Science 248, 337–345.
Woodhead, J.D., 1988. The origin of geochemical variations in Mariana lavas: a general
model for petrogenesis in intra-oceanic island arcs? J. Petrol. 29, 805–830.
Woodhead, J.D., 1989. Geochemistry of the Mariana arc (western Pacific): source
composition and processes. Chem. Geol. 76, 1–24. doi:10.1016/0009-2541(89)
90124-1.
Zimmer, M.M., Plank, T., Hauri, E.H., Yogodzinski, G.M., Stelling, P., Larsen, J., Singer, B., Jicha,
B., Mandeville, C., Nye, C.J., 2010. The role of water in generating the calc-alkaline trend:
new volatile data for aleutian magmas and a new tholeiitic index. J. Petrol. 51,
2411–2444. doi:10.1093/petrology/egq062.
Supplementary Online Material to Accompany the Manuscript:
“The influence of magmatic differentiation on the oxidation state of Fe in a basaltic arc
magma” by K.A. Kelley and E. Cottrell
Sample Preparation and Petrography
Olivine-hosted melt inclusions were hand-picked from one mafic tephra sample,
AGR19-02, that was collected during a 2004 MARGINS field expedition to the Mariana
islands (http://sio.ucsd.edu/marianas; Figure S1) and was donated to this study by T.
Plank. The tephra is a mixture of loose whole and fragmented phenocrysts of olivine,
plagioclase, clinopyroxene, magnetite, and finely crystalline matrix fragments. Clast sizes
range as large as 20 mm, but the raw tephra was sieved into naturally-occurring size
fractions of >1mm, 0.5-1.0 mm, and 0.5-0.25 mm, rinsed of fine particles with water, and
olivine crystals were hand-picked from these size fractions without any additional
crushing. The olivine separates were immersed in mineral oil for melt inclusion
identification and petrographic analysis. Melt inclusions selected for preparation were
naturally glassy with no visible secondary or synchronously trapped crystal phases,
petrographically determined to be fully enclosed in the host olivine crystals, and
contained either a single vapor bubble or no bubble. Glass inclusions were exposed for
analysis by electron microprobe and then prepared as double-intersected wafers, to fully
expose glass on both sides of each wafer (Fig. S2).
Petrographic data, including photomicrographs and information on vapor bubble
presence and included phases in glass inclusion-hosting phenocrysts, are provided in
Table S1. Most glass inclusions contained a single vapor bubble, although two contained
no bubble. All host olivine phenocrysts were barren of included phases except glass
(Table S1), which is a general feature of olivine in this tephra sample. Olivine
phenocrysts are typically euhedral to subhedral when crystal habit is identifiable, and no
co-entrapped phases, aside from glass, could be clearly identified under magnification.
Host olivines were not strongly zoned approaching melt inclusions; strong zoning would
suggest significant, rapid post-entrapment crystallization or quench modification of the
glass inclusions, albeit without re-equilibration of the olivine or glass inclusion with the
external melt. Complete re-equilibration of olivine with the external melt would
homogenize any previous zoning of olivine phenocrysts, but would also drive olivine Fo
contents towards equilibrium with the outside melt, homogenizing the entire phenocryst
population. Sample AGR19-02 preserves a broad range of olivine compositions, from
Fo73-82, suggesting that olivines in this sample have not completely re-equilibrated with
the external magma.
Detailed Analytical Methods
EMP Analysis
Glass inclusions and host olivines were analyzed for major element, S and Cl
concentrations (Table 1 of main text) by electron microprobe using the using the JEOL8900 5 spectrometer microprobe at the Smithsonian Institution operating at 10 nA, 15 kV
and with a 10 micron beam diameter. Na and K were analyzed first with 20 second peak
count times to minimize alkali loss in hydrous glasses, followed by analysis of Si, Ti, Al,
Fe, Mn, Mg, Ca, and P with 30-40 second peak count times. The glasses were analyzed in
a second round for S and Cl at 80 nA and 15kV also using a 10 micron beam. The sulfur
Kα position was determined for each inclusion using a peak search routine. We
referenced the S concentrations to the VG-2 standard with 1340 ppm sulfur. Adjacent
olivine was analyzed with a point beam, Primary and secondary standards were those
used by Luhr (2001).
FTIR Analysis
Wafered glass inclusions (Figure S2) were analyzed for dissolved CO2 and H2O
concentrations by Fourier Transform Infrared Spectroscopy (FTIR) at the Smithsonian
Institution using a Bio-Rad Excalibur spectrometer. Spectra were collected from 1000 –
6000 cm-1 using a liquid nitrogen-cooled MCT detector, KBr beam splitter and a
tungsten-halogen source. The bench, microscope, and samples were continuously purged
with dry air. Thicknesses were determined directly with a digital piezometric micrometer
(1σ = 1 micron) and, when possible, indirectly using the wavelength of fringes in the
region from 2000-2700 cm-1 (1σ = 0.2 micron; Nichols and Wysoczanski, 2007). The two
techniques agreed to within 5% relative thickness and when there was not perfect
agreement the fringe method was preferred.
Most inclusions (13 out of 20) were analyzed in two rounds of FTIR. In the first
round, thicknesses varied between 30 and 100 microns. Dissolved CO32- was quantified
using the antisymmetric stretching absorptions at 1515 and 1435 cm-1, OH- using the
absorption at 4500 cm-1, and molecular water using the absorption bands at 1635 and
5200 cm-1. Molar absorptivities for C and H species peaks were calculated after Dixon
and Pan (1995) and Dixon et al. (1995) respectively based on the major element
composition of each individual inclusion. Total peak heights above background for H
species were determined by fitting the spectral background with a spline function using
OMNIC software. Total water was calculated by summing the hydroxyl and molecular
components from the 4500 and 5200 cm-1 peaks (or 1635 cm-1 peak). For each inclusion
we checked for internal consistency between measured concentrations of hydroxyl and
molecular water species and total calculated water, and compared these to those expected
from speciation relationships presented in Dixon et al. (1995). In some cases fringe
amplitudes were too large to accurately quantify the 4500 and 5200 cm-1 absorption
bands. Therefore, after completion of all other analyses, inclusions were polished to ~15
microns when possible for a second round of FTIR to constrain total water using the
broad asymmetric peak at ~3535 cm-1. The 3535 cm-1 peak is not very sensitive to
composition and a molar absorptivity of 63/mol-cm was used in all cases (Dixon et al.,
1995).
For most inclusions a 22x22 micron aperture was applied and three spectra, often
partially overlapping, were taken. When total water was quantified using the 3535 cm-1
peak (13 of the 20 inclusions), the reported uncertainties reflect the standard deviation of
the three spectral measurements (≤1-3% relative). When total water was quantified by
summing molecular and hydroxyl components (5 of the 20 inclusions), the error was
calculated in quadrature (≤ 10% relative). For two inclusions (AGR 19-02-04 and 19-0212B) total water concentration was calculated based on quantification of the 1635cm-1
molecular water peak and the speciation relationships presented in Dixon et al. (1995).
No correction was applied to account for loss of volatiles from inclusions to
contraction bubbles. The presence or absence of vapor bubbles in each inclusion is noted
in Table S1.
Fe3+/∑Fe Analysis
Wafered inclusions were also analyzed in situ for Fe3+/∑Fe ratios (i.e., Fe3+/[Fe2+
+ Fe3+]) using micro X-ray Absorption Near Edge Structure (µ-XANES) spectroscopy,
using methods and techniques detailed by Cottrell et al. (2009), at beamline X26A,
National Synchrotron Light Source, Brookhaven National Lab. Wafered glass inclusions
were scanned in two dimensions to ensure that the 9x5 µm XANES beam passed through
glass only. Olivine interference is readily detectable in the spectra, and when evident,
spectra were excluded (e.g., Kelley and Cottrell, 2009). Examples of spectra collected for
the present study are shown in Figure S3. Determinations of Fe3+/∑Fe ratios in basaltic
glasses are highly precise, within ±0.005 (Cottrell et al., 2009).
The Fe3+/∑Fe ratio of the whole rock powder was calculated from a determination
of the FeO content made at the Smithsonian Institution using the micro-colorimetric
procedure of Christie et al. (1986), modified from Wilson (1960) and the total iron
concentration determined by J. Wade (see below). Four USGS standards were analyzed in the
same analytical session (W-2, QLO-1, BCR-1, BIR-1) and were within 0.3 wt.% (absolute)
of the accepted absolute FeO concentration of the standards yielding an uncertainty of ~ 15%
relative in the Fe3+/∑Fe ratio.
LA-ICP-MS Analysis
Glass inclusions and host olivines were also analyzed for trace element
abundances by laser-ablation inductively-coupled plasma mass spectrometry (LA-ICPMS) at the Graduate School of Oceanography, University of Rhode Island. Analyses
were conducted using a Thermo X-Series II quadrupole ICP-MS coupled with a New
Wave UP 213 Nd-YAG laser ablation system, using spot sizes ranging from 30 to 60 µm,
60% energy output, and 5 Hz repeat rate to maximize ablation time in thin, wafered
samples. Typical ablation duration in glass inclusions and host olivines lasted from 30-60
seconds, depending on sample thickness, and care was taken to preserve areas of
inclusion glass in each sample, where possible, for future work. An example of a typical
LA-ICP-MS ablation spectrum is provided in Figure S4. Glasses and olivines were
analyzed for 36 trace elements (Table S2), although for olivines, we report only 5 minor
and 11 trace elements because these were consistently above the detection limit.
Procedures for reducing LA-ICP-MS data follow those outlined by Kelley et al. (2003),
using 43Ca as the internal standard for glasses and 26Mg as the internal standard for
olivine. Calibration curves were generated using eight natural-composition glasses from
the USGS (BIR-1G, BCR-2G, BHVO-2G) and the Max Planck Institute (KL2-G, ML3BG, StHs6/80-G, GOR132-G, T1-G; (Jochum et al., 2006), and were linear to R2>0.990.
Reproducibility between replicate spots, where possible, was typically within 5% rsd for
all elements reported in glass, and 10% rsd for all elements reported in olivine. A single
crystal of San Carlos olivine (Fo88) was also analyzed periodically as a check on the
determination of olivine forsterite content.
Detailed Assessment of Post-Entrapment Modification of Inclusion Compositions
Glass inclusion compositions were assessed in detail for post-entrapment Fe-loss
using the method of Danyushevsky et al. (2000), which relies on comparison of inclusion
and whole-rock compositions. We compare the equilibrium olivine compositions of
inclusions and whole rocks to their total FeO* contents (Figure S5), assuming an average
Fe3+/∑Fe ratio for the whole-rock lavas of 0.23 (average of the MI data). We find that all
but one inclusion fall remarkably close to the trend defined by the whole-rock data for
Agrigan. Given the range of scatter among the whole-rock data and the uncertainty
contributed by assuming an average Fe3+/∑Fe ratio for the whole-rocks, we expect that
these inclusions are within the uncertainty of the whole-rock trend for Agrigan. One
inclusion falls below the whole-rock trend, indicating possible Fe loss. This inclusion was
corrected for Fe loss following the method of Danyushevsky et al. (2000), using the
correction module in Petrolog 3.0 (Danyushevsky and Plechov, 2011), to FeO*=9.0 wt.%
(Table 1). Overall, however, we find that the glass inclusions from sample AGR19-02
show remarkable fidelity with whole-rock lavas from Agrigan, strengthening the case that
these inclusions are faithful records of the magmatic evolution.
Whole-Rock Composition of AGR19-02 Tephra Matrix
Hand-picked matrix shards were crushed and analyzed for whole-rock major and
trace elements at Boston University by J. Wade, following methods outlined in Wade et
al. (2005). These unpublished data, along with the tephra sample, were contributed to this
study by T. Plank (Table S2).
Vanadium Partitioning
Natural samples examined for vanadium partitioning include AGR19-02
(Agrigan, Marianas; this study), AUNY17 (Augustine, Aleutians; Zimmer et al., 2010),
DF-BP-08-19b (Jalopy Cone, Big Pine, Basin & Range; this study), and EN113-13D-1g
(East Pacific Rise MORB; Kelley and Cottrell, 2009). The Agrigan and Jalopy samples
were analyzed for V concentrations in glass inclusions and host olivines by LA-ICP-MS
at GSO/URI using the methods outlined above (see Tables S1 and S3), and the Jalopy
inclusions and olivines were also analyzed for major element composition by electron
microprobe at the Smithsonian (Table S4). The MORB and Augustine samples were
analyzed using comparable LA-ICP-MS methods (Cooper et al., 2010) at LamontDoherty Earth Observatory by T. Plank (Table S3).
Analytical uncertainty in DVol / liq , which was used to derive the error envelope on
Figure 2b (see the main text), was determined assuming 5% error on concentrations of V
! in olivine V content from 10 to 35% as V concentration
in glass, and scaled errors
decreases from 10 ppm to 2 ppm. This results in an increase in the size of model error
bars from ±0.0020 at DVol / liq =0.1 to ±0.0027 at DVol / liq =0.008.
!
List of Tables
!
Table S1. Petrograhic data for inclusion-hosting olivines from sample AGR19-02
Table S2. Trace element compositions of olivine-hosted glass inclusions and host
olivines from Agrigan, Marianas determined by LA-ICP-MS
Table S3. Whole-rock major and trace element composition of AGR19-02 tephra,
Agrigan, Marianas
Table S4. Trace element compositions of glasses, matrix, olivine-hosted glass inclusions
and olivines from Jalopy, Augustine, and the East Pacific Rise determined by LA-ICPMS
Table S5. Major element compositions and Fe3+/∑Fe ratios of olivine-hosted glass
inclusions and host olivines from Jalopy Cone, Big Pine
Figure Captions
Figure S1. Regional map of the Mariana subduction zone. The inset shows the location
of Agrigan with respect to other neighboring volcanoes in the Mariana central island
province.
Figure S2. Photomicrographs of a wafered olivine-hosted glass inclusion from this study,
imaged in (a) plane polarized, transmitted light and (b) crossed polars. The glass
inclusion has been intersected and polished on both sides of the wafer, leaving only
isotropic glass that appears black in crossed polars. The inclusion shown is sample
AGR19-02-11.
Figure S3. Raw and processed µ-XANES fluorescence spectra. (a) Raw energy vs. edgestep normalized intensity for two reference glasses, equilibrated at QFM+1 (LW_10) and
QFM+2 (LW_20; Cottrell et al., 2009), and an olivine-hosted glass inclusion from
Agrigan volcano (AGR19-02-05, this study). (b) Energy vs. background-subtracted
intensity for the pre-edge region of the Fe K-edge for the three glasses shown in (a).
Figure S4. Raw LA-ICP-MS spectra for select trace elements. Shaded regions show the
portions of the spectra that were averaged to generate mean signal and background
intensities. Both spectra were collected using a 60 µm spot, 5 Hz repeat rate, and 60%
beam energy output. (a) Intensity vs. time for select trace elements in host olivine for
inclusion AGR19-02-14. (b) Intensity vs. time for select trace elements in glass inclusion
AGR19-02-14.
Figure S5. (a) Plot of forsterite content measured in olivines hosting glass inclusions vs.
calculated equilibrium olivine from Agrigan glass inclusion compositions, using
K Dol / liq (Fe 2+ / Mg) = 0.3 . The 1:1 line indicates perfect Fe2+-Mg exchange equilibrium
between glass inclusions and their host crystals. Inclusions falling below and to the right
!
of the line have experienced post-entrapment crystallization (PEC) of olivine. Inclusions
falling above and to the left may have experienced post-entrapment Fe2+ loss. Inclusions
indicating >2% PEC were excluded from consideration. (b) Plot of equilibrium olivine
composition vs.FeO* for Agrigan whole-rocks and olivine-hosted glass inclusions, after
Danyushevsky et al. (2000). Whole-rock olivine compositions were calculated assuming
an average Fe3+/∑Fe ratio of 0.23. Inclusion shown in bold may have experienced Fe
loss, and a correction is applied to its composition in Table 1 (following methods of
Danyushevsky et al., 2000).
Figure S6. Comparison of trace element ratio proxies for source oxygen fugacity
with Fe3+/ Fe ratios. (a) Plot of V/Sc ratios vs. Fe3+/ Fe ratios in MORBS
(triangles; Cottrell and Kelley, 2011) and Agrigan glass inclusions with MgO >5
wt.% (shaded circles). Curves show oxygen fugacity modeled from V/Sc ratios
from Lee et al. (2005) relative to Fe3+/∑Fe ratios modeled using the melt
composition of inclusion AGR19-02-01 at variable fO2, P=1 atm, T=1200°C;
solid curve is for spinel lherzolite melting at F=10%, dashed curve is for spinel
lherzolite melting at F=17% (more appropriate for Agrigan; Kelley et al., 2010).
(b) Plot of Zn/ Fe ratios vs. Fe3+/ Fe ratios in MORBs and Agrigan glass
inclusions, modified from Lee et al. (2010). The range of Zn/ Fe ratios in MORB
is from Lee et al. (2010), and the range of MORB Fe3+/ Fe ratios is from Cottrell
and Kelley (2011). Agrigan glass inclusions (circles) are color-coded for MgO
content; higher MgO inclusions on average have lower Zn/ Fe ratios.
References
Christie, D.M., Carmichael, I.S.E., Langmuir, C.H., 1986. Oxidation states of mid-ocean
ridge basalt glasses. Earth Planet. Sci. Lett. 79, 397-411.
Cooper, L.B., Plank, T., Arculus, R.J., Hauri, E.H., Hall, P.S., Parman, S.W., 2010. HighCa boninites from the active Tonga Arc. J. Geophys. Res. 115,
doi:10.1029/2009jb006367.
Cottrell, E., Kelley, K.A., 2011. The oxidation state of Fe in MORB glasses and the
oxygen fugacity of the upper mantle. Earth Planet. Sci. Lett. 305, 270-282,
doi:10.1016/j.epsl.2011.03.014.
Cottrell, E., Kelley, K.A., Lanzirotti, A.T., Fischer, R.A., 2009. High-precision
determination of iron oxidation state in silicate glasses using XANES. Chem.
Geol. 268, 167-179, doi:10.1016/j.chemgeo.2009.08.008.
Danyushevsky, L.V., Della-Pasqua, F.N., Sokolov, S., 2000. Re-equilibration of melt
inclusions trapped by magnesian olivine phenocrysts from subduction-related
magmas: petrological implications. Contrib. Mineral. Petrol. 138, 68-83,
doi:10.1007/PL00007664.
Danyushevsky, L.V., Plechov, P., 2011. Petrolog3: Integrated software for modeling
crystallization processes. Geochem. Geophys. Geosys. 12, Q07021,
doi:10.1029/2011GC003516.
Dixon, J.E., Pan, V., 1995. Determination of the molar absorptivity of dissolved
carbonate in basanitic glass. Am. Mineral. 80, 1339-1342.
Dixon, J.E., Stolper, E., Holloway, J.R., 1995. An experimental study of water and
carbon dioxide solubilities in mid-ocean ridge basaltic liquids. Part I: Calibration
and solubility models. J. Pet. 36, 1607-1631.
Jochum, K.P., Stoll, B., Herwig, K., Willbold, M., Hofmann, A.W., Amini, M., Aarburg,
S., Abouchami, W., Hellebrand, E., Mocek, B., Raczek, I., Stracke, A., Alard, O.,
Bouman, C., Becker, S., Dücking, M., Brätz, H., Klemd, R., de Bruin, D., Canil,
D., Cornell, D., de Hoog, C.-J., Dalpé, C., Danyushevsky, L.V., Eisenhauer, A.,
Gao, Y., Snow, J.E., Groschopf, N., Günther, D., Latkoczky, C., Guillong, M.,
Hauri, E.H., Höfer, H.E., Lahaye, Y., Horz, K., Jacob, D.E., Kasemann, S.A.,
Kent, A.J.R., Ludwig, T., Zack, T., Mason, P.R.D., Meixner, A., Rosner, M.,
Misawa, K., Nash, B.P., Pfänder, J., Premo, W.R., Sun, W.D., Tiepolo, M.,
Vannucci, R., Vennemann, T., Wayne, D., Woodhead, J.D., 2006. MPI-DING
reference glasses for in situ microanalysis: New reference values for element
concentrations and isotope ratios. Geochem. Geophys. Geosys. 7, Q02008,
doi:10.1029/2005GC001060.
Kelley, K.A., Cottrell, E., 2009. Water and the oxidation state of subduction zone
magmas. Science 325, 605-607, doi:10.1126/science.1174156.
Kelley, K.A., Plank, T., Ludden, J.N., Staudigel, H., 2003. Composition of altered
oceanic crust at ODP Sites 801 and 1149. Geochem. Geophys. Geosys. 4,
doi:10.1029/2002GC000435.
Kelley, K.A., Plank, T., Newman, S., Stolper, E., Grove, T.L., Parman, S., Hauri, E.,
2010. Mantle melting as a function of water content beneath the Mariana arc. J.
Pet. 51, 1711-1738, doi:10.1093/petrology/egq036.
Lee, C.-T.A., Leeman, W.P., Canil, D., Li, Z.-X.A., 2005. Similar V/Sc systematics in
MORB and arc basalts: Implications for the oxygen fugacities of their mantle
source regions. J. Pet. 46, 2313-2336, doi:10.1093/petrology/egi056.
Lee, C.-T.A., Luffi, P., Le Roux, V., Dasgupta, R., Albaréde, F., Leeman, W.P., 2010.
The redox state of arc mantle using Zn/Fe systematics. Nature 468, 681-685,
doi:10.1038/nature09617.
Luhr, J.F., 2001. Glass inclusions and melt volatile contents at Parícutin Volcano,
Mexico. Contrib. Mineral. Petrol. 142, 261–283, doi:10.1007/s004100100293.
Nichols, A.R.L., Wysoczanski, R.J., 2007. Using micro-FTIR spectroscopy to measure
volatile contents in small and unexposed inclusions hosted in olivine crystals.
Chem. Geol. 242, 371-384, doi:10.1016/j.chemgeo.2007.04.007.
Wade, J.A., Plank, T., Stern, R.J., Tollstrup, D.L., Gill, J.B., O'Leary, J.C., Eiler, J.M.,
Moore, R.B., Woodhead, J.D., Trusdell, F., Fischer, T.P., Hilton, D.R., 2005. The
May 2003 eruption of Anatahan volcano, Mariana Islands: Geochemical evolution
of a silicic island-arc volcano. J. Volc. Geother. Res. 146, 139-170,
doi:10.1016/j.jvolgeores.2004.11.035.
Wilson, A.D., 1960. The micro-determination of ferrous iron in silicate minerals by a
volumetric and a colorimetric method. Analyst 85, 823-827.
Zimmer, M.M., Plank, T., Hauri, E.H., Yogodzinski, G.M., Stelling, P., Larsen, J.,
Singer, B., Jicha, B., Mandeville, C., Nye, C.J., 2010. The Role of Water in
Generating the Calc-alkaline Trend: New Volatile Data for Aleutian Magmas and
a New Tholeiitic Index. J. Pet. 51, 2411-2444, doi:10.1093/petrology/egq062.
Figure S1
26°N
Mariana Arc
Philippine
Sea Plate
Mariana Trough
Pacific
Plate 22°N
18°N
14°N
Asuncion
10°N
19°N
Agrigan
Pagan
18°N
Alamagan
145°E
146°E
146°E
150°E
Figure S2
200µm
A
B
Figure S3
Normalized Intensity
1.4
A
1.2
1.0
0.8
0.6
0.4
QFM+1
QFM+2
Agr19-02-05
0.2
0.0
7100
7120
7140
7160
7180
7200
Energy (eV)
Normalized Intensity
0.06
B
0.05
0.04
0.03
0.02
0.01
0
7109
7110
7111
7112
7113
Energy (eV)
7114
7115
7116
Figure S4
107
A
26
106
Intensity (CPS)
Background
Host Olivine
AGR19-02-14
Mg
57
Fe
60
Ni
105
45
104
Sc
51
V
103
Ca
43
102
Signal
101
106
10
88
51
5
Intensity (CPS)
Glass Inclusion
AGR19-02-14
Sr
B
V
Ca
43
Background
45
Sc
10
4
La
139
Pb
208
103
U
238
102
Signal
10
1
0
10
20
30
40
50
Time Slice
60
70
80
90
100
Figure S5
Equilibrium Olivine (Fo)
0.84
0.82
A
0.80
Post-Entrapment
Fe2+ loss
0.78
0.76
1:1
0.74
.3)
=0
(K D
Post-Entrapment
Crystallization
0.72
0.70
0.70
<2% PEC
>2% PEC
0.72
0.74
0.76
0.78
Host Olivine (Fo)
0.80
0.82
0.84
13
12
11
FeO*, wt.%
10
9
8
7
6
5
4
0.70
Agrigan MI, This Study
Agrigan WR
0.72
0.74
0.76
0.78
0.80
0.82
Equilibrium Olivine (Fo)
0.84
0.86
0.88
Figure S6
0.24
A
Fe3+/∑Fe
0.22
QFM+1
QFM+1
0.20
0.18
0.16
0.14
0.12
4
0.5
6
7
8
9
5.4
5.0
V/Sc
10 11 12 13
4.6 4.2 3.8
MgO, wt.%
3.4
+4
+3
0.3
+2
0.2
+1
MORB
4
6
8
10
–1
.9
2
K D=0
0
.0
0.0
0
K D=1
0.1
Zn/∑Fe (x10 )
4
12
14
16
fO2 (∆QFM;
approximate)
Fe /∑Fe
5
B
0.4
3+
This Study
MORB, CK11
QFM
QFM
Table S1. Petrographic data for olivine-hosted glass inclusions from sample
AGR19-02
Sample
Olivine #
Photo (PPL)
Photo (XPOL)
Photo (RL)
Vapor Bubble
Included
Phases in
Olivine
AGR19-02
01
Glass
AGR19-02
02
Glass
AGR19-02
03
Glass
AGR19-02
04
Glass
AGR19-02
05
Glass
Table S1. Petrographic data for olivine-hosted glass inclusions from sample
AGR19-02
Sample
Olivine #
Photo (PPL)
Photo (XPOL)
Photo (RL)
Vapor Bubble
Included
Phases in
Olivine
AGR19-02
07
Glass
AGR19-02
08
Glass
AGR19-02
09
Glass
AGR19-02
10
Glass
AGR19-02
11
Glass
Table S1. Petrographic data for olivine-hosted glass inclusions from sample
AGR19-02
Sample
Olivine #
Photo (PPL)
Photo (XPOL)
Photo (RL)
Vapor Bubble
Included
Phases in
Olivine
AGR19-02
12
Glass
AGR19-02
13
Glass
AGR19-02
14
Glass
AGR19-02
15
Glass
AGR19-02
16
Glass
Table S1. Petrographic data for olivine-hosted glass inclusions from sample
AGR19-02
Photo (PPL)
Photo (XPOL)
Photo (RL)
Vapor Bubble
Included
Phases in
Olivine
Sample
Olivine #
AGR19-02
17
Glass
AGR19-02
18
Glass
AGR19-02
19
Glass
AGR19-02
20
Glass
Table S2. Trace element compositions of olivine-hosted glass inclusions and host olivines from Agrigan, Marianas determined by LA-ICP-MS
Sample
Inclusion #
Glass
Inclusion
Li
Be
Sc
V
Cr
Co
Ni
Cu
Zn
As
Rb
Sr
Y
Zr
Nb
Cs
Ba
La
Ce
Pr
Nd
Sm
Eu
Gd
Tb
Dy
Ho
Er
Tm
Yb
Lu
Hf
Ta
Pb
Th
U
Olivine Host
Li
Na2O
Al2O3
P2O5
CaO
Sc
TiO2
V
Cr
Co
Ni
Cu
Zn
Sr
Y
Zr
AGR19-02
San Carlos
01
03
04
08
09
10
12A
12B
13
14
15
16
17
18
19
20
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
8.45
7.76
41.1
327
17.7
41.5
7.63
143
61.0
0.583
11.2
341
14.6
36.3
0.77
99.9
4.87
9.21
1.29
6.05
3.22
0.839
2.15
0.332
2.00
0.504
1.47
0.279
1.21
0.289
0.445
0.0468
1.48
0.368
0.218
8.15
1.06
36.4
252
7.23
24.0
4.39
136
86.3
1.40
32.9
298
29.3
104
1.83
0.709
229
11.6
22.0
3.42
15.5
4.24
1.42
5.14
0.723
4.97
1.06
3.07
0.442
2.58
0.404
2.54
0.119
4.30
1.35
0.448
5.30
0.653
43.9
287
32.2
28.1
12.8
124
128
0.288
7.94
274
14.2
28.7
0.612
0.250
100
3.75
7.41
1.10
6.10
1.98
0.590
2.20
0.404
2.25
0.501
1.26
0.176
1.31
0.145
0.598
0.0548
5.71
0.236
0.107
4.94
0.744
46.6
274
22.0
35.0
13.2
110
76.5
0.842
10.6
285
16.7
37.1
0.619
0.370
104
4.76
9.29
1.45
7.74
2.06
0.850
2.81
0.443
2.71
0.623
1.59
0.226
1.64
0.231
0.984
0.0375
1.30
0.458
0.149
6.61
0.223
45.4
392
14.2
29.9
3.93
147
98.4
1.21
14.3
322
18.6
43.1
0.885
0.355
132
5.59
12.6
1.88
8.88
2.61
0.968
3.09
0.490
2.93
0.640
1.91
0.268
1.71
0.255
1.04
0.0407
1.65
0.482
0.222
4.30
0.343
41.7
287
8.42
28.9
9.09
107
79.4
0.668
6.50
305
13.1
31.3
0.566
0.213
85.2
3.92
8.62
1.34
6.16
1.76
0.745
2.25
0.351
2.24
0.468
1.29
0.201
1.31
0.179
0.769
0.0327
1.32
0.451
0.156
5.18
1.56
37.6
264
9.29
25.1
4.33
111
69.6
1.41
22.7
319
22.0
64.2
1.29
0.638
167
7.91
16.8
2.51
11.9
3.09
0.977
4.00
0.622
3.69
0.767
2.20
0.312
1.86
0.298
1.48
0.0837
2.34
0.937
0.343
5.30
0.366
42.2
348
11.5
33.7
5.90
132
87.3
0.973
12.2
333
14.0
30.2
0.584
0.284
100
4.32
9.55
1.46
6.90
1.92
0.795
2.23
0.375
2.34
0.502
1.42
0.205
1.23
0.214
0.700
0.0280
1.34
0.334
0.167
5.68
0.640
43.0
365
13.6
32.8
7.22
136
99.0
1.06
13.2
292
15.7
37.0
0.725
0.377
118
4.79
11.0
1.63
7.70
2.34
0.828
2.62
0.411
2.67
0.560
1.55
0.247
1.44
0.240
0.862
0.0391
1.55
0.421
0.212
5.46
0.491
40.6
280
17.3
35.6
15.3
107
82.8
0.801
10.8
305
15.2
35.0
0.653
0.273
101
4.52
9.47
1.48
7.10
2.17
0.830
2.63
0.402
2.44
0.525
1.53
0.228
1.51
0.213
0.931
0.0355
1.30
0.604
0.159
4.74
44.3
296
32.6
27.4
6.35
114
79.9
0.972
8.02
292
13.9
28.5
0.566
0.322
91.6
3.70
7.43
1.18
6.07
1.76
0.728
2.40
0.390
2.37
0.472
1.05
0.210
1.44
0.183
0.576
0.0277
1.04
0.307
0.117
5.91
2.40
44.0
303
9.94
35.5
7.21
157
78.5
0.733
8.98
323
13.7
31.5
0.543
0.203
81.0
4.22
8.64
1.40
6.83
1.68
0.733
2.21
0.386
2.27
0.509
1.60
0.263
1.30
0.227
0.842
0.0545
1.06
0.411
0.142
4.96
0.126
37.3
230.5
8.85
28.6
4.72
105.0
75.4
1.06
20.5
274
15.4
56.3
1.12
0.448
145
7.32
15.2
2.28
10.3
2.71
0.974
2.96
0.454
2.61
0.516
1.36
0.198
1.24
0.194
1.32
0.0446
1.92
0.730
0.314
5.83
0.428
41.8
313
16.1
31.5
9.66
121
84.0
0.927
14.0
292
16.9
41.2
0.739
0.388
124
5.32
11.4
1.74
8.50
2.29
0.896
2.72
0.437
2.63
0.598
1.65
0.232
1.62
0.267
1.05
0.0429
1.59
0.506
0.225
4.22
0.886
38.0
282
22.7
29.0
9.66
101
73.7
0.718
7.65
350
12.4
33.3
0.590
0.245
96.0
4.01
8.66
1.29
5.97
1.76
0.690
2.30
0.373
2.21
0.460
1.19
0.187
1.29
0.186
0.851
0.0383
1.84
0.425
0.159
4.11
0.332
48.0
260
26.5
27.3
6.71
90.9
57.8
1.00
11.2
291
18.4
41.7
0.688
0.342
107
5.11
9.87
1.67
8.09
2.60
0.881
2.93
0.478
2.89
0.641
1.93
0.297
1.78
0.298
0.801
0.0357
1.57
0.507
0.137
ppm
wt.%
wt.%
wt.%
wt.%
ppm
wt.%
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
0.866
0.027
0.074
0.022
0.26
9.98
0.010
5.94
13.7
240
328
7.10
73.1
1.05
0.118
0.580
2.84
0.018
0.042
0.034
0.22
12.5
0.011
6.40
7.65
260
251
7.19
161
0.416
0.231
0.164
1.30
0.058
0.488
0.024
0.47
11.3
0.020
11.7
34.2
241
458
7.64
96.3
6.081
0.430
0.880
1.29
0.010
0.030
0.016
0.23
11.1
0.005
4.29
23.9
247
435
4.13
101
0.149
0.073
0.030
1.91
0.016
0.036
0.032
0.24
11.1
0.008
6.38
16.0
278
262
6.71
137
0.423
0.137
0.064
1.34
0.031
0.156
0.022
0.27
10.3
0.011
8.10
33.6
243
612
5.56
90.7
2.24
0.146
0.202
1.85
0.010
0.021
0.010
0.14
8.95
0.007
5.48
12.2
290
401
4.74
107
0.102
0.103
0.033
1.74
0.016
0.063
0.016
0.19
9.62
0.008
6.51
13.3
289
396
5.26
114
0.602
0.087
0.060
1.85
0.019
0.054
0.014
0.21
10.6
0.007
6.21
17.0
291
334
5.96
131
0.606
0.139
0.105
1.47
0.015
0.057
0.027
0.24
10.5
0.009
5.60
21.4
245
491
4.22
97.0
0.648
0.089
0.089
1.19
0.013
0.025
0.017
0.17
8.90
0.004
4.14
30.9
271
622
4.81
97.9
0.110
0.060
0.059
1.26
0.013
0.025
0.016
0.22
10.5
0.005
4.38
11.4
267
347
5.96
95.2
0.045
0.080
0.020
1.39
0.019
0.061
0.018
0.22
9.72
0.009
6.16
11.2
289
363
4.66
108
0.695
0.101
0.147
1.62
0.028
0.107
0.029
0.23
10.7
0.011
7.80
20.6
256
367
6.89
111
1.61
0.180
0.300
1.29
0.013
0.038
0.016
0.19
8.59
0.006
4.78
23.2
251
657
4.34
101
0.248
0.056
0.056
1.10
0.011
0.039
0.017
0.26
11.2
0.006
4.89
28.3
243
432
3.92
92.7
0.289
0.101
0.062
2.15
0.014
0.015
0.018
0.07
5.25
0.005
4.42
160
169
4089
2.26
73.8
0.015
0.038
0.024
Table S3. Whole-rock major
and trace element composition
of AGR19-02 tephra, Agrigan,
Marianas
Sample
ICP-AES
SiO2
TiO2
Al2O3
Fe2O3*
MnO
MgO
CaO
Na2O
K2O
P2O5
Total
H2OLOI
Sr
Ba
Ni
Cu
Zr
Y
AGR19-02
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
ppm
ppm
ppm
ppm
ppm
ppm
Wet Chemistry
Fe+3/∑Fe
ICP-MS
Li
Be
Sc
TiO2
V
Cr
Co
Νι
Cu
Ga
Zn
Rb
Sr
Y
Zr
Nb
Cs
Ba
La
Ce
Pr
Nd
Sm
Eu
Gd
Tb
Dy
Ho
Er
Yb
Lu
Hf
Ta
Pb
Th
U
56.04
0.89
15.57
11.60
0.24
3.57
7.94
3.33
1.34
0.23
100.75
1.21
0.11
352
240
0.649
73.0
107
27.8
0.29
ppm
ppm
ppm
wt.%
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
9.83
0.884
31.8
0.877
243
1.57
28.0
3.39
75.6
18.2
105
31.5
347
28.5
94.2
1.93
0.735
242
11.1
24.0
3.53
15.9
4.19
1.38
4.86
0.796
4.87
1.02
2.90
2.86
0.452
2.53
0.128
2.87
1.38
0.569
Table S4. Trace element compositions of glasses, matrix, olivine-hosted glass inclusions and olivines
from Jalopy, Augustine, and the East Pacific Rise determined by LA-ICP-MS
Sample
DF-BP-08-19b (Jalopy, Big Pine)
Inclusion or
Chip #
Li
Be
Sc
TiO2
V
Cr
MnO
Co
Ni
Cu
Zn
Rb
Sr
Y
Zr
Nb
Cs
Ba
La
Ce
Pr
Nd
Sm
Eu
Gd
Tb
Dy
Ho
Er
Tm
Yb
Lu
Hf
Ta
Pb
Th
U
ppm
ppm
ppm
wt.%
ppm
ppm
wt.%
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
Li
Na2O
Al2O3
P2O5
CaO
Sc
TiO2
V
Cr
Co
Ni
Cu
Zn
Sr
Y
Zr
ppm
wt.%
wt.%
wt.%
wt.%
ppm
wt.%
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
ppm
01
Glass
Inclusion
9.50
1.73
30.3
1.72
217
111
0.094
24.7
106
87.9
68.4
21.0
1683
25.8
218
11.5
0.562
1460
54.7
123
15.9
62.1
10.3
2.81
7.25
1.01
5.00
0.893
2.38
0.323
2.20
0.336
4.46
0.568
13.6
4.41
1.26
02
Glass
Inclusion
9.04
1.98
23.3
1.78
199
34.75
0.094
23.5
57.8
103
62.6
28.4
2086
29.3
274
15.9
0.571
1869
72.5
142
18.6
75.2
11.6
3.04
8.13
1.16
5.64
1.07
2.62
0.341
2.27
0.308
5.06
0.801
15.4
5.86
1.31
03
Glass
Inclusion
17.3
3.07
23.1
1.89
234
60.9
0.139
31.6
74.5
111
125
43.4
1700
25.7
238
15.7
0.768
1713
62.1
151
17.4
63.9
10.5
2.78
6.80
0.95
4.86
0.949
2.49
0.314
1.98
0.305
4.43
0.755
22.6
5.47
1.85
EN113-13D-1g (East Pacific Rise)
04
Glass
Inclusion
11.1
2.15
23.7
1.83
233
63.0
0.113
31.4
67.9
100
88.1
29.7
1928
21.8
202
15.1
0.360
1857
60.4
155
19.0
68.3
10.3
2.74
6.75
0.881
4.37
0.763
2.05
0.267
1.70
0.217
3.62
0.614
16.9
4.06
1.35
Olivine Host Olivine Host Olivine Host Olivine Host
3.68
3.04
3.72
3.57
0.034
0.019
0.065
0.015
0.062
0.048
0.119
0.039
0.041
0.062
0.052
0.049
0.28
0.27
0.23
0.18
6.79
7.73
6.33
6.44
0.010
0.010
0.018
0.010
3.70
3.45
4.98
3.83
267
74.8
138
97.8
163
178
193
198
2813
1773
2373
1881
5.40
4.85
5.64
4.54
97.6
95.5
110
101
3.03
1.65
7.43
1.20
0.133
0.152
0.210
0.111
0.401
0.364
1.023
0.275
1
2
AUNY17
(Augustine)
3
Pillow Glass Pillow Glass Pillow Glass
6.07
6.17
6.51
0.645
0.668
0.586
39.4
40.9
35.4
1.50
1.50
1.50
222
220
229
276
271
280
0.177
0.173
0.178
43.6
42.3
43.0
138
134
139
75.8
72.6
74.9
72.1
70.2
77.7
0.0702
0.0784
0.079
132
135
133
34.6
36.8
29.6
106
115
96.4
0.431
0.437
0.439
0.0017
0.0013
0.763
0.771
0.795
1.87
1.96
1.80
7.93
7.96
8.32
1.68
1.71
1.68
9.78
10.2
9.59
3.59
3.79
3.43
1.35
1.39
1.32
5.12
5.55
4.62
0.881
0.948
0.807
5.83
6.28
5.21
3.49
3.79
3.08
3.38
0.526
2.66
0.0433
0.411
0.0286
0.0122
3.65
0.576
2.90
0.0448
0.419
0.0311
0.0193
3.07
0.459
2.33
0.0389
0.443
0.0275
0.0126
Olivine
1.32
Olivine
1.31
Olivine
1.26
0.29
8.28
0.006
7.59
296
129
1677
2.30
54
0.0061
0.221
0.0372
0.28
7.95
0.006
7.74
290
129
1678
2.26
53
0.0022
0.204
0.0258
0.28
8.10
0.006
8.51
314
134
1751
2.25
55
0.0037
0.202
0.0380
Ave. Glass
Inclusion
208
Ave. Olivine
2.61
Table S5. Major element compositions and Fe3+/∑Fe ratios of
olivine-hosted glass inclusions and host olivines from Jalopy
Cone, Big Pine
Sample
Inclusion #
Glass Inclusion
SiO2
TiO2
Al2O3
FeO*
FeO
Fe2O3
MnO
MgO
CaO
Na2O
K2O
P2O5
Total
S
Cl
Fe+3/∑Fe
Equil. Fo
Olivine Host
SiO2
FeO
MgO
NiO
Total
Fo
01
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
wt.%
ppm
ppm
48.40
1.54
16.51
5.83
4.37
1.63
0.22
5.73
13.61
3.31
1.35
0.83
97.50
3313
297
0.252
0.886
47.37
1.87
17.98
6.54
4.87
1.85
0.18
5.34
9.92
3.69
2.15
1.41
96.63
3930
437
0.254
0.867
46.74
1.89
17.80
8.01
6.07
2.15
0.17
4.90
10.09
3.55
2.12
1.34
96.82
3453
383
0.242
0.827
46.69
1.84
17.71
7.14
5.34
2.01
0.23
5.57
10.82
3.45
2.08
1.35
97.08
2610
400
0.253
0.861
wt.%
wt.%
wt.%
wt.%
wt.%
40.27
10.61
48.26
0.28
99.42
0.890
39.93
11.88
46.95
0.17
98.93
0.876
39.63
13.78
45.87
0.14
99.41
0.856
39.66
11.53
47.20
0.18
98.57
0.879
0.6%
48.35
1.54
16.51
5.87
4.40
1.63
0.22
5.99
13.61
3.31
1.35
0.83
97.74
3313
297
0.250
1.4%
47.27
1.84
17.73
6.62
4.98
1.82
0.18
5.92
9.78
3.64
2.12
1.39
96.67
3876
431
0.248
3.9%
46.48
1.82
17.13
8.28
6.41
2.07
0.16
6.40
9.71
3.42
2.04
1.29
96.94
3324
369
0.225
2.9%
46.51
1.79
17.21
7.30
5.54
1.95
0.23
6.74
10.51
3.35
2.02
1.31
97.16
2536
389
0.241
0.88
1.09
0.91
1.01
Post-Entrapment Corrected Glass
Olivine Added
wt.%
SiO2
wt.%
TiO2
wt.%
Al2O3
FeO*
wt.%
FeO
wt.%
wt.%
Fe2O3
MnO
wt.%
MgO
wt.%
CaO
wt.%
wt.%
Na2O
wt.%
K2O
wt.%
P2O5
Total
wt.%
S
ppm
Cl
ppm
Fe+3/∑Fe
Δ QFM
DF-BP-08-19b (Jalopy, Big Pine)
02
03
04
1 atm., 1200°C